The Iron Gates in Prehistory:

New perspectives

Edited by

Clive Bonsall
Vasile Boroneanţ
Ivana Radovanović

BAR International Series 1893

2008
This title published by

Archaeopress
Publishers of British Archaeological Reports
Gordon House
276 Banbury Road
Oxford OX2 7ED
England
bar@archaeopress.com
www.archaeopress.com

BAR S1893

The Iron Gates in Prehistory: New perspectives

© the individual authors 2008

Cover and Title Page illustration: ‘The Nymph’ drawn by N. Mitri!, based on a gural sculpture from Lepenski Vir now in the
National Museum collection in Belgrade ((with permission)

ISBN 978 1 4073 0373 4

Printed in England by Blenheim Colour Ltd

All BAR titles are available from:

Hadrian Books Ltd

122 Banbury Road
Oxford
OX2 7BP
England
bar@hadrianbooks.co.uk

The current BAR catalogue with details of all titles in print, prices and means of payment is available
free from Hadrian Books or may be downloaded from www.archaeopress.com
 Contents

Acknowledgements ii

A Note on Dating ii

List of Contributors iii

Introduction 1

Regional studies

Dušan Mihailović Lithic technology and settlement systems of the Final Palaeolithic and 11
Early Mesolithic in the Iron Gates.

Dragana Antonović The development of the ground stone industry in the Serbian part of 19
the Iron Gates.

László Bartosiewicz, Clive Bonsall Sturgeon fishing along the Middle and Lower Danube. 39
& Vasile Şişu

Mirjana Roksandic The Mesolithic–Neolithic in the Ðerdap as evidenced by non-metric 55

anatomical variants.

Mary Jackes, Mirjana Roksandic Demography of the Ðerdap Mesolithic–Neolithic transition. 77

& Christopher Meiklejohn

Joni L. Manson Approaches to Starčevo culture chronology. 89

Haskel Greenfield Faunal assemblages from the Early Neolithic of the central Balkans: 103
methodological issues in the reconstruction of subsistence and land
use.

Site studies

Vesna Dimitrijević Lepenski Vir animal bones: what was left in the houses? 117

Sofija Stefanović & Dušan Borić New-born infant burials underneath house floors at Lepenski Vir: in 131
pursuit of contextual meanings.

Biljana Čuljković, Sofija Stefanović DNA-based sex identification of the infant remains from Lepenski Vir. 171
& Stanka Romac

Clive Bonsall, Ivana Radovanović, Dating burials and architecture at Lepenski Vir. 175
Mirjana Roksandic, Gordon Cook,
Thomas Higham & Catriona Pickard

Haskel Greenfield Reanalysis of the vertebrate fauna from Hajdučka Vodenica in the 205
Danubian Iron Gates: subsistence and taphonomy from the Early
Neolithic and Mesolithic.

Rastko Vasić Velesnica and the Lepenski Vir culture. 227

Mirjana Roksandic The human osteological material from Velesnica. 243

Paolo Biagi, Elisabetta Starnini The Mesolithic–Neolithic transition in the Trieste Karst (north-eastern 251
& Barbara Voytek Italy) as seen from the excavations at the Edera Cave.

i
 Acknowledgements
Many individuals and organizations have helped directly or indirectly in the preparation of this volume. We thank all the
contributors for their cooperation and forbearance throughout the long production process. Each of the papers has been re-
viewed by at least two independent referees, and we have many colleagues to thank for their expert opinions on the
manuscripts. The camera-ready copy was prepared using David Pilling’s excellent desktop publishing software, Ovation Pro
for Windows. Financial support for the Edinburgh symposium was provided by the British Academy and The University of
Edinburgh Development Fund, while practical help at the meeting was provided by Sarah Davison, Brian Dean, Elizabeth
Hargreaves, Ian Morrison, Mary Jane Nicholls, Ishbel Ogilvie, Pat Storey, Diana Şişu, and Kirsten Thompson.

A Note on Dates
Throughout this volume dates followed by ‘BP’ are in radiocarbon years before present. Dates followed by ‘cal BC’ are cal-
ibrated radiocarbon ages (calendar ages). Dates followed simply by ‘BC’ are calendar ages derived by some other method, e.g.
luminescence dating. The following table shows the approximate correspondence between uncalibrated and calibrated 14C ages
over the time-range from 2000 to 13,900 BP. The calibrations were performed with CALIB 5.0.2 (Stuiver & Reimer 1993;
Stuiver et al. 2005) using the IntCal04 curve (Reimer et al. 2004) and rounded to the nearest five years.

Radiocarbon date calibration table

BP cal BC BP cal BC BP cal BC BP cal BC

2000 0 5000 3775 8000 6930 11,000 10,965

2100 125 5100 3850 8100 7065 11,100 11,060
2200 295 5200 3995 8200 7230 11,200 11,160
2300 390 5300 4130 8300 7400 11,300 11,240
2400 460 5400 4295 8400 7500 11,400 11,310
2500 625 5500 4345 8500 7560 11,500 11,390
2600 795 5600 4410 8600 7595 11,600 11,475
2700 835 5700 4525 8700 7685 11,700 11,600
2800 955 5800 4660 8800 7875 11,800 11,735
2900 1085 5900 4760 8900 8075 11,900 11,815
3000 1255 6000 4890 9000 8250 12,000 11,900
3100 1395 6100 5020 9100 8290 12,100 12,005
3200 1470 6200 5130 9200 8395 12,200 12,110
3300 1570 6300 5265 9300 8565 12,300 12,210
3400 1715 6400 5375 9400 8680 12,400 12,430
3500 1820 6500 5475 9500 8790 12,500 12,705
3600 1950 6600 5540 9600 8965 12,600 12,915
3700 2085 6700 5625 9700 9220 12,700 13,055
3800 2250 6800 5690 9800 9270 12,800 13,160
3900 2405 6900 5765 9900 9320 12,900 13,280
4000 2535 7000 5895 10,000 9515 13,000 13,395
4100 2645 7100 5995 10,100 9765 13,100 13,535
4200 2795 7200 6050 10,200 9950 13,200 13,680
4300 2905 7300 6155 10,300 10,130 13,300 13,830
4400 3015 7400 6295 10,400 10,330 13,400 13,970
4500 3215 7500 6400 10,500 10,570 13,500 14,105
4600 3365 7600 6450 10,600 10,740 13,600 14,240
4700 3420 7700 6530 10,700 10,835 13,700 14,370
4800 3550 7800 6630 10,800 10,880 13,800 14,490
4900 3675 7900 6735 10,900 10,920 13,900 14,735

References
Reimer P.J., Baillie, M.G.L., Bard, E., Bayliss, A., Beck, J.W., et al., 2004: IntCal04 terrestrial radiocarbon age calibration, 0–26 cal kyr BP.
Radiocarbon 46: 1029–1058.
Stuiver, M. & Reimer, P.J. 1993: Extended 14C data base and revised CALIB 3.0 14C age calibration program. Radiocarbon 35: 215–230.
Stuiver, M., Reimer, P.J. & Reimer, R. 2005: CALIB Radiocarbon Calibration (rev. 5.0.2): Online Manual.
http://radiocarbon.pa.qub.ac.uk/calib/manual/

ii
 List of Contributors
Dragana Antonović Institute of Archaeology, Knez Mihailova 35/IV, 11000 Belgrade, Serbia.
Email: D.Antonovic@ai.sanu.ac.yu
László Bartosiewicz Institute of Archaeological Sciences, Loránd Eötvös University, Múseum körút 4/B, H-1088
Budapest, Hungary. Email: bartwicz@yahoo.com
Paolo Biagi Dipartimento di Scienze dell’Antichità e del Vicino Oriente, Università Ca’ Foscari Venezia,
Palazzo Malcanton Marcorà, Dorsoduro 3484/D, I-30123 Venezia, Italy.
Email: pavelius@unive.it
Clive Bonsall School of History, Classics and Archaeology, University of Edinburgh, Old High School,
Infirmary Street, Edinburgh, EH1 1LT, U.K. Email: Clive.Bonsall@ed.ac.uk
Dušan Borić Department of Archaeology, University of Cambridge, Downing Street, Cambridge, CB2
3DZ, U.K. Email: db231@cam.ac.uk
Vasile Boroneanţ Bucharest Museum of History and Art, Bd I.C. Bratianu no. 2, sector 3, 70058 Bucharest,
Romania. Email: vasileboroneant@yahoo.com
Biljana Čuljković Institute for Research in Immunology and Cancer, P.O. Box 6128, Station Centre-Ville,
Montreal, Quebec, H3C 3J7, Canada. Email: biljana.culjkovic@umontreal.ca
Gordon Cook Radiocarbon Laboratory, S.U.E.R.C., Scottish Enterprise Technology Park, Rankine Avenue,
East Kilbride, G75 0QF, U.K. Email: g.cook@suerc.gla.ac.uk
Vesna Dimitrijević Department of Archaeology, Faculty of Philosophy, University of Belgrade, Čika Ljubina
18–20, 11000 Belgrade, Serbia. Email: vesnadim@beotel.yu
Haskel Greenfield Department of Anthropology, University of Manitoba, Fletcher Augue 435, Winnipeg,
Manitoba, R3T 5V5, Canada. Email: Greenf@cc.umanitoba.ca
Thomas Higham Oxford Radiocarbon Accelerator Unit, Research Laboratory for Archaeology and the History
of Art, Dyson Perrins Building, University of Oxford, Oxford, OX1 3QY, U.K.
Email: thomas.higham@rlaha.ox.ac.uk
Mary Jackes Department of Anthropology, University of Waterloo, Waterloo, Ontario, N2L 3G1, Canada.
Email: mkjackes@watarts.uwaterloo.ca
Joni Manson Heritage Education and Research Services, 13269 Bevelheimer Road, Westerville, OH
43081, U.S.A. Email: joni.manson@sbcglobal.net
Christopher Meiklejohn Department of Anthropology, University of Winnipeg, Winnipeg, Manitoba, R3B 2E9,
Canada. Email: c.meiklejohn@uwinnipeg.ca
Dušan Mihailović Department of Archaeology, Faculty of Philosophy, University of Belgrade, Čika Ljubina
18-20, 11000 Belgrade, Serbia. Email: dusan.mihailovic@sbb.co.yu
Catriona Pickard School of History, Classics and Archaeology, University of Edinburgh, Old High School,
Infirmary Street, Edinburgh, EH1 1LT, U.K. Email: Catriona.Pickard@ed.ac.uk
Ivana Radovanović Department of Archaeology, University of Kansas, 616 Fraser Hall, 1415 Jayhawk
Boulevard, Lawrence, KS 66045, U.S.A. Email: ivana@ku.edu
Mirjana Roksandic Department of Anthropology, University of Winnipeg, Winnipeg, Manitoba, R3B 2E9,
Canada. Email: m.roksandic@uwinnipeg.ca
Stanka Romac Faculty of Biology, University of Belgrade, Studentski trg 16, P.O. Box 52, Belgrade 11000,
Serbia. Email: stanka@bf.bio.bg.ac.yu
Vasile Şişu Muzeul Regiunii Porţilor de fier, Strada Independentei Nr 2, Drobeta Turnu-Severin 1500,
Mehedinţi, Romania. Email: vasilesisu@yahoo.com
Elisabetta Starnini Dipartimento di Scienze dell’Antichità e del Vicino Oriente, Università Ca’ Foscari Venezia,
Palazzo Malcanton Marcorà, Dorsoduro 3484/D, I-30123 Venezia, Italy.
Email: elisabetta.starnini@unive.it
Sofija Stefanović Department of Archaeology, Faculty of Philosophy, University of Belgrade, Čika Ljubina
18-20, 11000 Belgrade, Serbia. Email: smstefan@f.bg.ac.yu
Rastko Vasić Institute of Archaeology, Knez Mihailova 35/IV, 11000 Belgrade, Serbia.
Email: R.Vasic@ai.sanu.ac.yu
Barbara Voytek Archaeological Research Facility, 2251 College Building, University of California, Berkeley,
CA 94720-1076, U.S.A. Email: bvoytek@berkeley.edu

iii
 Introduction
Clive Bonsall, Vasile Boroneanţ & Ivana Radovanović
This book had its origins in a symposium held at the
University of Edinburgh from 30 March to 2 April 2000,
which was attended by 54 archaeologists with a shared in-
terest in the prehistory of the small but distinctive region of
Southeast Europe known as the Iron Gates (Fig. 1). It is by no
means a complete compilation of the papers delivered at the
symposium. Some contributions dealt with aspects of the
Romanian–British excavations at Schela Cladovei, and will
form the basis of a separate publication. Moreover, of the 15
articles included in the present volume, only ten are based on
papers presented at the Edinburgh meeting, while five —
those by Dragana Antonović, Vesna Dimitrijević, Joni
Manson, László Bartosiewicz et al., and Clive Bonsall et al.
— are later additions.
As a geographical label, the term ‘Iron Gates’ has acquired
several different meanings.1 In the broad sense it refers to the
section of the Danube valley where the river forms the mod-
ern political border between Serbia and Romania, and it is
this definition we have adopted for the present volume. The
230 km long Iron Gates section marks the beginning of the
Lower Danube. It divides naturally into two linear zones —
Figure 1. Principal Mesolithic and Early Neolithic sites of the Iron Gates.
1
the Iron Gates ‘gorge’ (Fig. 2) and the ‘downstream area’
(Fig. 3). The gorge, really a series of narrow, steep-sided
canyons separated by sections where the valley is wider and
the sides less steep, cuts a winding course for 134 km
between the southern Carpathian Mountains and the north-
western foothills of the Balkan Mountains. In Serbia the
gorge is known as Đerdap (from the Turkish ‘girdap’ mean-
ing whirlpool — literally, ‘gorge of whirlpools’) reflecting
the former presence of several sets of rapids and the generally
turbulent flow of the river along this stretch of the Danube. In
contrast, the downstream area is a zone of much more mod-
erate relief, marking the beginning of the Danube’s journey
across the Wallachian Plain toward the Black Sea. Here the
river is flanked by a broad alluvial plain comprising a series
of terraces. Notwithstanding the differences in topography,
microclimates and natural vegetation between the two zones,
the ‘gorge’ and the ‘downstream area’ have many features in
common archaeologically.
The Danube is both an important route way and a natural
barrier, and historically the Iron Gates has had great strategic
importance. The region is exceptionally rich in sites dating to
 The Iron Gates in Prehistory
the Roman and later periods, especially military installations.
Roman achievements in the Iron Gates include forts, a canal
built to bypass the rapids at the lower end of the gorge, and
Trajan’s Bridge constructed in AD 103–105 to supply Roman
legions during the Second Dacian War.
While there is a long tradition of Roman archaeology in
the Iron Gates extending back more than a century, the pre-
history of the region was virtually unknown until the second
half of the twentieth century when the first systematic ar-
chaeological surveys were undertaken in advance of the im-
pounding of the Danube by the Iron Gates I and II dams. The
survey work and rescue excavations were concentrated in the
parts of the valley that would eventually be submerged be-
neath the reservoirs created by the two dams — essentially a
narrow zone along both banks of the Danube as well as some
low-lying islands in the river. The fieldwork progressed in
two phases. The first between 1960 and 1971 focused on the
areas affected by the Iron Gates I dam; these were very
largely within the gorge, although some work was under-
taken in the immediately downstream area at Ostrovul
Banului and Schela Cladovei. In the second phase between
1977 and 1984 fieldwork was concentrated in the down-
stream area between the Iron Gates I and II dams.
Altogether, over two hundred archaeological sites of vari-
ous periods were discovered. Among them were a number of
Mesolithic sites, the first to be identified in the entire central
Balkan region (Fig. 1). Excavations were undertaken at many
of these sites — seven on the Serbian bank (Padina, Stubica,
Lepenski Vir, Vlasac, Hajdučka Vodenica, Velesnica, and
Kula) and 16 on the Romanian side (Privod, Alibeg, Vodneac,
Ilişova, Cuina Turcului, Climente I and II, Veterani Cave,
Veterani Terrace, Răzvrata, Icoana, Ostrovul Banului, Schela
Cladovei, Ostrovul Corbului, and Ostrovul Mare km 875 and
km 873). Some were major projects extending over several
field seasons, but many were small, exploratory excavations
lasting just a few days or weeks, and some of these were
never completed, being abandoned when the level of the
Danube became too high. The published record of these ex-
cavations is variable. Comprehensive reports are available
for just two sites — Vlasac (Srejović & Letica 1978) and
Cuina Turcului (Păunescu 1970, 1978). But there are many
other publications that offer either interim excavation reports
or studies of particular bodies of material, especially from
Lepenski Vir, Padina, Hajdučka Vodenica, and Schela
Cladovei. There have also been several attempts to provide
regional syntheses, most notably by Radovanović (1996),
Boroneanţ (2000), and Bonsall (2008).
Although a number of important Bronze and Iron Age
sites were also discovered, it is the Mesolithic that has tended
to dominate the literature on the Iron Gates in Prehistory —
and it is a recurrent theme throughout the present volume.
The reasons are not hard to identify; the range and quality of
the evidence from the Iron Gates relating to Mesolithic ar-
chitecture, art, burial practices, bone and stone technology,
and subsistence patterns, far exceeds that from any other re-
gion of Southeast or Central Europe west of the Black Sea.
Moreover, some Mesolithic sites continued to be used into
the Early Neolithic, making the Iron Gates an area of critical
importance for understanding the processes involved in the
transition from hunting and gathering to farming in Europe.
2
Since the completion of the dam projects (and the con-
sequent ‘loss’ of many of the valley floor sites recorded in the
1960s to 1980s) the emphasis in research has shifted, al-
though the focus has remained firmly on the Mesolithic and
the Neolithic transition. While some fieldwork has continued
along the banks of the Danube, notably at Schela Cladovei in
Romania (Boroneanţ et al. 1999; Bonsall 2008) and Vlasac in
Serbia (Borić 2007), many publications have been based on
re-analyses of existing archaeological collections, taking ad-
vantage of advances in archeological science such as AMS
14C dating (e.g. Cook et al. 2002; Bonsall et al. 2004; Borić
& Miracle 2004; Borić & Dimitrijević 2007), stable isotope
analysis (e.g. Bonsall et al. 1997, 2000; Grupe et al. 2003),
and techniques for the recovery of DNA from animal and
human bones (e.g. Larson et al. 2007). The data generated by
these new analyses in turn have permitted archaeologists to
reassess the models of prehistoric settlement that were pro-
posed on the basis of initial field observations. The papers
that make up this volume exemplify both strands of this new
phase of research.
Objectives and organization of the volume
The objectives of this volume are broadly those of the ori-
ginal Edinburgh symposium. First and foremost the volume
is intended to illustrate the immense research potential of the
Iron Gates region, even though those areas along the
Danube’s banks that proved attractive for prehistoric settle-
ment are now largely flooded and there has been comparat-
ively little additional fieldwork undertaken since the mid-
1980s. A second objective is to provide case studies that
illustrate the nature of current research and the rich possibil-
ities offered by the growing range of scientific techniques
available to archaeologists and their application to existing
archaeological collections.
When editing the volume it soon became evident that there
was considerable overlap between the papers in terms of
chronological and/or geographical coverage, or theoretical
and/or methodological emphasis. Therefore, we decided on a
basic division between those papers that include an element
of regional synthesis, and those that deal largely or exclus-
ively with the evidence from an individual site:
Regional studies
This section begins with the paper by Dušan Mihailović,
which focuses on the Late Glacial (Final Palaeolithic) and the
Early Holocene (Early Mesolithic) in the Iron Gates.
Mihailović identifies clear trends in settlement pattern, sub-
sistence and technology during this time range, including a
reduction in the size of annual territories, increased use of
local resources, and a decline in the range and quality of
chipped stone artefacts. He argues that these perceived trends
reflect a reduction in residential mobility in response to
changing environmental conditions and resource availability.
While the technological emphasis in Dusan Mihailović’s
paper is on chipped stone artefacts, Dragana Antonović re-
views the evidence for a ground stone industry in the Iron
Gates and its development during the Mesolithic and Early
Neolithic. Assigned to the category of ‘ground stone arte-
 Clive Bonsall, Vasile Boroneanţ & Ivana Radovanović: Introduction

Figure 2. Two views of the Iron Gates gorge. Top: The gorge near the sites of Hajdučka Vodenica (Serbia) and Icoana
(Romania); the photograph was taken from the Serbian bank looking upriver (© Mirjana Roksandic, 2000). Bottom: The
gorge photographed from Lepenski Vir, looking upriver (© Ivana Radovanović, 2008).

facts’ are “all stone implements worked by
grinding/polishing, as well as unfinished examples with
traces of flaking or pecking … [and] … implements that are
naturally polished or polished through use” (Antonović, this
volume: 19). Her analysis is restricted to sites on the Serbian
bank of the Danube — proceeding downstream, Padina,
Lepenski Vir, Vlasac, Hajdučka Vodenica, Ajmana,
Velesnica, Knjepište, and Ušće Kameničkog Potoka. Within
the ground stone industry she recognizes two major com-
ponents, local and ‘imported’. These are distinguishable not
only in terms of typology and technique, but also raw mater-
ial choices and ascribed function. The local component is
seen to have developed independently within the Iron Gates
region and comprises a wide range of tool forms including

3
 The Iron Gates in Prehistory

Figure 3. Two views of the ‘downstream area’ of the Iron Gates. Top: The Danube floodplain at Schela Cladovei (Romania),
8 km downstream from the Iron Gates I dam. The Schela Cladovei archaeological site extends for about a kilometre along
the riverbank. The town of Kladovo in Serbia can be seen on the opposite bank of the Danube (© Clive Bonsall, 2008).
Bottom: The confluence of the Zamna river with the Danube on the Serbian side, 8 km upstream from the Iron Gates II dam.
Opposite the confluence lies Ostrovul Mare (The Big Island). Several Mesolithic sites were found in the vicinity — cf.
Figure 3 (© Ivana Radovanović, 2008).

4
 Clive Bonsall, Vasile Boroneanţ & Ivana Radovanović: Introduction
various kinds of grinding tools, percussive instruments,
weights, and ornamental artefacts, invariably made from
locally available raw materials. The ‘imported’ component
comprises mainly ground edge implements (axes, adzes, and
chisels), which have clear parallels in the Starčevo and Vinča
cultures of the central Balkans, and the materials are both
local and imported. This, and the fact that both the local and
‘imported’ elements are found together in some sites, leads
Antonović to the conclusion that the ‘imported’ element was
an adopted technology. Her paper is both an important con-
tribution to knowledge of the Iron Gates in Prehistory, and a
valuable demonstration that ground stone tools (in the broad
sense of non-chipped stone tools) were an important, and of-
ten abundant, component of Mesolithic and Neolithic sites
throughout the Iron Gates. This is equally true of the
Romanian bank of the Danube, although there they appear
not to have been collected so systematically.
Somewhat broader in its chronological and geographical
scope is the paper by László Bartosiewicz, Clive Bonsall &
Vasile Şişu, which provides an overview of the zooarchae-
ological and ethnohistorical evidence for sturgeon fishing
along the middle and lower Danube as represented by two
contrasting sections of the river, in the Carpathian Basin and
the Iron Gates. The exploitation of migratory sturgeon is
documented from the Late Glacial onwards, although inter-
pretation of the osteoarchaeological evidence is rarely
straightforward. The authors attempt to provide a framework
for future investigations, highlighting the many variables that
archaeologists need to consider when seeking to interpret the
role of sturgeon fishing in the lives of prehistoric peoples
living along the Danube. These include sturgeon biology and
behaviour, taphonomy, river conditions, fishing techniques,
as well as local beliefs and customs.
Two papers focus on the effects of the
Mesolithic–Neolithic transition in the Iron Gates region, and
specifically the issue of population change. Both use data
gathered by Mirjana Roksandic for her PhD study of the
skeletal populations from four sites in the Serbian part of the
Iron Gates — Hajdučka Vodenica, Lepenski Vir, Padina, and
Vlasac (Roksandic 1999). Mirjana Roksandic’s first paper
in this volume examines the incidence of selected skeletal
non-metric traits among the populations from different time
periods and, based on the assumption that the traits are linked
to ancestry, uses the data as a measure of the degree of inter-
action between local foragers and non-local farmers.2 The
259 adult skeletons examined are divided into three chrono-
logical groups (Mesolithic, Mesolithic–Neolithic ‘Contact’
period, and Neolithic), and the groups compared. The results
are interpreted as reflecting a substantial degree of population
continuity between Mesolithic and Neolithic. It is suggested
there may have been some ‘seeping in’ of immigrants during
the contact period before the establishment of a farming eco-
nomy in the Iron Gates, but evidence of population replace-
ment at the transition is lacking. From this the author con-
cludes that the Iron Gates Neolithic was primarily the result
of local adoption. The paper by Mary Jackes, Mirjana
Roksandic & Christopher Meiklejohn applies standard-
ized palaeodemographic methods of analysis to Roksandic’s
age/sex data for skeletal populations from the Iron Gates, in
order to test for changes in fertility across the
5
Mesolithic–Neolithic transition. Whilst stressing that their
interpretations are severely constrained by inadequate
sample sizes and uncertainty over the dating of many skelet-
ons, the authors offer the tentative conclusion that population
levels in the Iron Gates were stable during the Mesolithic
with an increase in fertility in the Neolithic, which is con-
sistent with findings elsewhere in Europe.
The last two papers in this section are concerned with
specific aspects of the Early Neolithic in the wider region of
the central–northern Balkans. Joni Manson discusses the
chronology of the Starčevo culture. Several chronological
frameworks are in current use, all of which rely on pottery
typology. But none of these has been adequately tested by
independent dating procedures. Most attempts to establish
the typological sequence of stylistic change in Starčevo
ceramics have relied on stratigraphy and/or 14C dating of as-
sociated organic materials. A complementary approach
would be to use the pottery itself to obtain direct age meas-
urements, since there are more assumptions involved in 14C
dating of material found near pottery than in trying to date the
pottery directly (Bonsall et al. 2002). In a pioneering study
based on her PhD research (Manson 1990), Dr Manson de-
scribes the results of archaeomagnetic intensity dating of
potsherds from 12 Starčevo sites. By combining these data
with Aranđelović-Garašanin’s (1954) typological sequence
and the available radiocarbon dates for Starčevo sites, she is
able to propose a revised chronology of the Starčevo culture.
In the following paper, Haskel Greenfield reviews the
state of archaeozoological research in the central Balkans
pertaining to the Early Neolithic. He points out that there are
relatively few Early Neolithic sites with analyzed faunal as-
semblages, and the quality of the information from those sites
is highly variable. Greenfield’s discussion ranges over issues
such as the effects of site location, sample size and taxonomic
diversity, taphonomy, recovery methodology, quantification
methods, curation, and state of publication, leading to the
conclusion that the majority of faunal assemblages from the
region have only limited potential for reconstructing animal
exploitation strategies and land use patterns in the Early
Neolithic. Greenfield’s paper serves as both a warning of the
fragility of conclusions based on existing archaeozoological
reports, and a plea for a more rigorous approach to the ana-
lysis of Early Neolithic animal bone assemblages from
Southeast Europe.
Site studies
Of the six papers that deal with individual sites, four focus on
Lepenski Vir, arguably the most famous archaeological site
in the Iron Gates region. All four papers present the results of
new analyses of the finds from Dragoslav Srejović’s excava-
tion between 1965 and 1971. Not all of the finds survive from
those excavations and this applies particularly to the animal
remains, which in any case were mainly hand-collected (rel-
atively little sieving was undertaken at Lepenski Vir). Vesna
Dimitrijević focuses on the small surviving collection of an-
imal bones stored in the National Museum in Belgrade, and
specifically those recovered from within the famous
trapezoidal buildings. In a thought-provoking analysis she
distinguishes between remains deposited while a building
was in use, and those deposited when it was abandoned. In so
 The Iron Gates in Prehistory
doing she identifies episodes of food preparation, bone man-
ufacturing, and ritual deposition. Consideration is also given
to the question of seasonality of building abandonment, al-
though the evidence is limited.
The other papers focus on Lepenski Vir are concerned
primarily with the human remains from the site. An unusual
feature of Lepenski Vir was the occurrence of the remains of
neonates underneath the floors of the trapezoidal buildings or
immediately to the rear of the buildings. The papers by Sofija
Stefanović & Dusan Borić and Biljana Čjulković, Sofija
Stefanović & Stanka Romac are both directly concerned
with these remains. Stefanović & Borić provide a detailed
osteoarchaeological evaluation of the burials, including a
discussion of taphonomy, supported by plans and photo-
graphs many of which have not been published previously.
This leads on to a wide-ranging and thought-provoking dis-
cussion of the possible chronological and social significance
of the neonate burials, which draws inspiration from archae-
ological, ethnographic, and ethnohistorical sources.
Knowledge of the sex of the infants is fundamental to the in-
terpretation of the burial evidence. However, determining the
sex of neonate skeletons using standard osteometric proced-
ures is notoriously unreliable. An alternative approach is to
use ancient (aDNA) techniques. Čuljković et al. have done
just this. Of the 41 infant skeletons excavated from under
buildings, 30 individuals were tested and successful sex
identification is reported in nearly all cases. These results
have important implications for the related study by
Stefanović & Borić, and thus the two papers should be con-
sidered together. Although by no means the first successful
amplification of aDNA from ancient human remains,
Čuljković et al.’s paper stands as a pioneering study for the
Iron Gates and demonstrates the potential of the Iron Gates
sites for archaeogenetic research.
The burial record from Lepenski Vir is not, of course,
confined to neonates; there were also many burials of older
children, adolescents, and adults, and these are the subject of
the paper by Clive Bonsall, Ivana Radovanović, Mirjana
Roksandic, Gordon Cook, Thomas Higham & Catriona
Pickard. AMS 14 C dates and stable isotope values are
presented for 24 burials. The data are then used to establish
the chronological contexts of different forms of burial rep-
resented at Lepenski Vir, and to refine the dating of the
trapezoidal buildings where these occur in a clear strati-
graphic relationship with directly dated burials. The authors
suggest that their data also have implications for dating the
appearance of farming in the Lepenski Vir catchment, al-
though they acknowledge that this is a controversial subject,
which is further complicated by curatorial problems and a
lack precision in the 14C dates.
Three other papers in this section are concerned with sites
downriver of Lepenski Vir. Haskel Greenfield provides an
account of the (hitherto unpublished) animal bone as-
semblages from Late Mesolithic and Early Neolithic contexts
at the site of Hajdučka Vodenica in the gorge, excavated by
Borislav Jovanović between 1966 and 1969. In interpreting
the results, Greenfield takes into account excavation meth-
odology, uncertainties over stratigraphy, and the extent of ta-
phonomic loss, in keeping with the approach advocated in his
other contribution to this volume (pp. 103–114).
6
Comparisons are made with other sites in the Iron Gates
gorge. A key finding is that bones of apparently domestic
cattle and pig occur in both the Mesolithic and Neolithic
samples which begs the important question, is this evidence
of Mesolithic animal domestication, exchange with farmers,
or ‘stratigraphic mixing’?
The important site of Velesnica in the downstream area is
the focus of separate but related papers by Rastko Vasić and
Mirjana Roksandic. Excavated by Rastko Vasić between
1980 and 1984, Velesnica is a multi-period open-air site on
the Serbian bank of the Danube (Fig. 1). Vasić reviews the
evidence from the main settlement phases. The most abund-
ant remains belong to the Early Neolithic (Starčevo culture)
and include three graves, one of which contained seven skel-
etons; the burials are discussed in more detail in the com-
panion paper by Mirjana Roksandic. Vasić draws attention to
several features of the Starčevo settlement at Velesnica that
he believes demonstrate a cultural connection with Lepenski
Vir and other Late Mesolithic/Early Neolithic sites in the
gorge, including the presence of carved stone ‘altars’. An
earlier, Mesolithic occupation is suspected at Velesnica but
could not be distinguished stratigraphically and has yet to be
confirmed by radiocarbon dating.
The final paper in the volume, by Paolo Biagi, Elisabetta
Starnini & Barbara Voytek, stands apart in that it does not
deal with the Iron Gates per se. Their article reviews the
evidence from Edera Cave on the karst plateau at the head of
the Adriatic Sea, which has a long occupation sequence in-
cluding Mesolithic and Neolithic deposits. It is included here
because it touches on the question of Mesolithic ‘survival’
and contact between hunter-gatherers and farmers — still a
major point of controversy among prehistorians working in
the Iron Gates, with differing views aired by, e.g., Voytek &
Tringham (1989), Radovanović & Voytek (1997), Tringham
(2000), Borić (2002), Radovanović (2006), and Bonsall
(2007, 2008). The co-occurrence of pottery, wild and do-
mestic fauna, and chipped stone artefacts of Mesolithic char-
acter in ‘layer 3a’ at Edera is reminiscent of the situation de-
scribed for some sites in the Iron Gates and evokes similar
questions regarding the nature and timing of the transition
from Mesolithic to Neolithic.
This volume presents new information and new perspect-
ives on the prehistoric settlement of the Iron Gates. We also
hope that it points out directions for future research. Many
areas of uncertainty and controversy remain, and we look
forward to further investments in AMS radiocarbon dating,
isotopic analysis, and aDNA research in the future to help
resolve some of these issues. In the years ahead we would
also hope to see an extension of fieldwork into the hinterland
on both sides of the Danube, beyond the areas affected by
dam construction, in order to achieve a more representative
picture of the human use of the Iron Gates in Prehistory.
Notes
1. The original ‘Iron Gate’ of the Danube was the cliffs on either side
of the rapids at the lower end of the gorge, where subsequently the
Iron Gates I dam was built. In the Danube Convention of 1948 the
term ‘Iron Gates’ was formally applied to “the section between
Vince and Kostol on the right bank and between Moldova-Veche
and Turnu-Severin on the left bank of the Danube”, which in-
 Clive Bonsall, Vasile Boroneanţ & Ivana Radovanović: Introduction
cludes most of the gorge and the first 8 km of the downstream area.
A second dam was constructed 80 km downstream from Iron
Gates I and given the official name ‘Iron Gates II’, thus effectively
extending the term ‘Iron Gates’ to more or less the whole of the
border region between Serbia and Romania.
2. The term ‘non-metric trait’, or ‘discrete trait’, refers to any minor
(non-pathological) anomaly observed in bones and teeth that is
not normally recorded by measurement. Non-metric traits may
be recorded as being either present or absent or, less often, scored
according to their degree of development (Mays 1998: 102).
References
Aranđelović-Garašanin, D. 1954: Starčevačka kultura. Ljubljana.
Bonsall, C. 2007: When was the Neolithic transition in the Iron
Gates? In Spataro, M. & Biagi, P. (eds), A Short Walk through the
Balkans: the First Farmers of the Carpathian Basin and
Adjacent Regions. Trieste: Società per la Preistoria e Protostoria
della Regione Friuli-Venezia Giulia, 53–65.
— 2008: The Mesolithic of the Iron Gates. In Bailey, G. & Spikins,
P. (eds), Mesolithic Europe. Cambridge: Cambridge University
Press, 238–279.
Bonsall, C., Lennon, R., McSweeney, K., Stewart, C., Harkness, D.,
Boroneanţ, V., Bartosiewicz, L., Payton, R. & Chapman, J. 1997:
Mesolithic and Early Neolithic in the Iron Gates: a palaeodietary
perspective. Journal of European Archaeology 5: 50–92.
Bonsall, C., Cook, G., Lennon, R., Harkness, D., Scott, M.,
Bartosiewicz, L. & McSweeney, K. 2000: Stable isotopes, ra-
diocarbon and the Mesolithic–Neolithic transition in the Iron
Gates. Documenta Praehistorica 27: 119–132.
Bonsall, C., Cook, G., Manson, J. & Sanderson, D. 2002: Direct
dating of Neolithic pottery: progress and prospects. Documenta
Praehistorica 29: 47–59.
Bonsall, C., Cook, G., Hedges, R., Higham, T., Pickard, C. &
Radovanović, I. 2004: Radiocarbon and stable isotope evidence
of dietary change from the Mesolithic to the Middle Ages in the
Iron Gates: new results from Lepenski Vir. Radiocarbon 46(1):
293–300.
Borić, D. 2002: The Lepenski Vir conundrum: reinterpretation of
the Mesolithic and Neolithic sequences in the Danube Gorges.
Antiquity 76: 1026–1039.
— 2007: New discoveries at the Mesolithic–Early Neolithic site of
Vlasac: preliminary notes. Mesolithic Miscellany 18(1): 7–14.
Borić, D. & Miracle, P. 2004: Mesolithic and Neolithic (dis)con-
tinuities in the Danube Gorges: new AMS dates from Padina and
Hajdučka Vodenica (Serbia). Oxford Journal of Archaeology 23:
341–371.
Borić, D. & Dimitrijević, V. 2007: When did the ‘Neolithic package’
reach Lepenski Vir? Radiometric and faunal evidence.
Documenta Praehistorica 34: 53–72.
Boroneanţ, V. 2000: Paléolithique Supérieur et Épipaléolithique
dans la Zone des Portes de Fer. Bucureşti: Silex.
Boroneanţ, V., Bonsall, C., McSweeney, K., Payton, R. & Macklin,
M. 1999: A Mesolithic burial area at Schela Cladovei, Romania.
1
7
In Thévenin, A. (ed.) L’Europe des Derniers Chasseurs:
Épipaléolithique et Mésolithique. Paris: Éditions du Comité des
Travaux Historiques et Scientifiques, 385–390.
Cook, G.T., Bonsall, C., Hedges, R.E.M., McSweeney, K.,
Boroneanţ, V., Bartosiewicz, L. & Pettitt, P.B. 2002: Problems of
dating human bones from the Iron Gates. Antiquity 76: 77–85.
Grupe, G., Mikić, Ž., Peters, J. & Manhart, H. 2003: Vertebrate food
webs and subsistence strategies of Meso- and Neolithic popula-
tions of central Europe. In Grupe, G. & Peters, J. (eds)
Decyphering Ancient Bones: The Research Potential of
Bioarchaeological Collections. Rahden (Westf): Marie Leidorf,
193–214.
Larson, G., Albarella, U., Dobney, K., Rowley-Conwy, P., Schibler,
J., Tresset, A., Vigne, J-D., Edwards, C.J., Schlumbaum, A.,
Dinu, A., Balacsescu, A., Dolman, G., Tagliacozzo, A.,
Manaseryan, N., Miracle, P., Van Wijngaarden-Bakker, L.,
Masseti, M., Bradley, D.G. & Cooper, A. 2007: Ancient DNA,
pig domestication, and the spread of the Neolithic into Europe.
Proceedings of the National Academy of Sciences of the United
States of America 104(39): 15276–15281.
Manson, J.L. 1990: A Reanalysis of Starčevo Culture Ceramics:
Implications for Neolithic Development in the Balkans.
Unpublished PhD dissertation, Southern Illinois University,
Carbondale (University Microfilms, Ann Arbor).
Mays, S. 1998: The Archaeology of Human Bones. London:
Routledge.
Păunescu, A., 1970. Epipaleoliticul de la Cuina Turcului-Dubova.
Studii şi Cercetări de Istorie Veche 21(1): 3–47.
— 1978: Cercetările archeologice de la Cuina Turcului-Dubova
(Jud. Mehedinţi). Tibiscus Istorie, Volum Închinat Celei de-a 60
Aniversari a Unirii: 11–56.
Radovanović, I. 1996: The Iron Gates Mesolithic. Ann Arbor:
International Monographs in Prehistory.
— 2006: Further notes on Mesolithic–Neolithic contacts in the Iron
Gates region and the central Balkans. Documenta Praehistorica
33: 107–124.
Radovanović, I. & Voytek, B. 1997: Hunters, fishers and farmers:
sedentism, subsistence and social complexity in the Iron Gates
Mesolithic. Analecta Praehistorica Leidensia 29: 19–31.
Roksandic, M. 1999: Transition from Mesolithic to Neolithic in the
Iron Gates Gorge: Physical Anthropology Perspective.
Unpublished PhD thesis, Department of Anthropology, Simon
Fraser University, Burnaby, B.C.
Srejović, D. & Letica, Z. 1978: Vlasac: A Mesolithic Settlement in
the Iron Gates. Vol. 1, Archaeology. Belgrade: Serbian Academy
of Arts and Sciences.
Tringham, R. 2000: Southeastern Europe in the transition to agri-
culture in Europe: bridge, buffer, or mosaic. In Price, T.D. (ed.)
Europe’s First Farmers. Cambridge: Cambridge University
Press, 19–56.
Voytek, B. & Tringham, R. 1989: Rethinking the Mesolithic: the
case of South-east Europe. In Bonsall, C. (ed.) The Mesolithic in
Europe. Edinburgh: John Donald, 492–500.
 Section 1

REGIONAL STUDIES
 Lithic technology and settlement systems of the
Final Palaeolithic and Early Mesolithic in the Iron Gates

Dušan Mihailović

Abstract: This paper deals with the changes that took place in the settlement system, economy and lithic technology during the Final
Palaeolithic and Early Mesolithic of the Iron Gates. An attempt is made to explain the correlations, frequency and variability of various types
of material remains in terms of the level of specialization in their acquisition, exploitation and usage. It is argued that the Final Palaeolithic
communities of the Iron Gates were characterized by a high level of specialization in the production of artefacts, an organized system of
settlements, and an organized system of supply of all resources at the regional level. Changes occurred at the beginning of the Holocene,
reflected in: a) significantly narrower site catchments, b) intensification in the exploitation of local resources, and c) technological decline
in stone artefact production. The technological decline was a consequence of cultural adaptation to the environmental changes during that
period. Indirectly, it is manifested in reduced mobility of the communities and different activities conducted within the settlements.
Key words: lithic technology, settlement systems, Palaeolithic, Mesolithic, Iron Gates

Introduction Gates Mesolithic, especially in its earlier phase, were caused

Changes in the settlement system are fundamental to the in-
terpretation of the cultural, social and ideological transform-
ation of Mesolithic communities in the Iron Gates. However,
in spite of the many features that point to a semi-sedentary or
even sedentary way of life (Radovanović & Voytek 1997),
the problem of establishing the degree of mobility of the Iron
Gates Mesolithic communities is far from solved. The reas-
ons are only partially theoretical in nature, related to the im-
possibility of the precise definition of the process of sedent-
arization (Voytek & Tringham 1989; Tchernov 1991;
Bar-Yosef & Meadow 1995; Radovanović & Voytek 1997).
Much greater difficulties result from fieldwork methods that
were constrained by the ‘rescue’ character of the excavations
and the uneven scope of investigation of certain sites; the
perplexities of the stratigraphy and chronology of some of the
key sites; and the lack of specialist analyses that are neces-
sary for reliable conclusions regarding seasonality, mobility
and functional differences among the settlements. However,
since investigations of the Iron Gates Mesolithic have in-
tensified in all respects over the past decade, it is reasonable
to hope that in the near future most of the data necessary for
reconstructing the settlement system will be at our disposal.
This paper is an attempt to reconstruct the activities that
resulted in the specific repertoire and patterning of the ma-
terial remains in the Iron Gates sites from the perspective of
lithic technology. The specific character of the Iron Gates
Mesolithic necessitates a different approach to the relation-
ship between lithic technology and settlement than that usu-
ally applied in ethnoarchaeological studies (Binford 1979;
Binford & O’Connell 1984; Gamble 1993) or studies of other
Mesolithic assemblages from Southeast Europe (Perlès 1987;
Kozłowski 1999; Sinclair 1999). Specifically, exotic raw
materials, curated technology and other features indicative of
greater mobility of human groups and direct acquisition or
exchange of goods over a wider area, in fact are not decisive
in understanding the Iron Gates Mesolithic. Most of the
available data lead to the conclusion that changes in the Iron
primarily by internal cultural, social and ecological factors.
For this reason, I shall try to explain the correlations, fre-
quency and variability of various types of material remains in
terms of the level of specialization in their acquisition, ex-
ploitation and usage. The degree of specialization in techno-
logy and the mode of resource acquisition reflect to a con-
siderable extent the social, economic, and even ideological
demands of the community in the exploitation of the envir-
onment. Therefore, analysis of the various forms of special-
ization may lead to a better understanding of the changes in
culture and way of life of Mesolithic hunter-gatherers in the
Iron Gates.
Final Palaeolithic
The chipped stone industries from level I of the Cuina
Turcului rockshelter and Climente II cave, attributed to the
very end of the Pleistocene, are comparable with the Late
Glacial industries of the Balkan and Apennine peninsulas, as
discussed by several previous authors (Păunescu 1970, 1989;
Boroneanţ 1999). However, from the point of view presented
here, it is important to stress that over the wider region of the
eastern and southern Mediterranean in this period, there is an
increase in intensity of habitations in caves and rockshelters
in mountainous areas. This phenomenon has been explained
by stressing the importance of palaeoecological, palaeogeo-
graphical, demographical or cultural factors (Bailey et al.
1983; Bar-Yosef & Meadow 1995; Miracle 1996). It may be
suggested, however, that the quantity and diversity of the
finds in the Final Palaeolithic horizons of sites in Southeast
Europe testify above all to an organized settlement system
based on intensive exploitation of resources at the regional
level (Mihailović 1999b). The technology and economy of
the period are characterized primarily by specialization (eco-
nomizing on resources and maximizing their exploitation) on
the one hand, and diversification on the other. This is mani-
fested in technology by the presence of a wider repertoire of

11
 The Iron Gates in Prehistory
Figure 1. Cuina Turcului, level I: frequency of ungulate
species identified.
microliths, and in economy by the addition of new food re-
sources (Miracle 1996).
The most comprehensive data on the palaeoecology of the
Iron Gates and the mode of exploitation by human groups
have been gathered by pollen and faunal analyses of levels I
and II at Cuina Turcului (Bolomey 1970, 1973; Pop et al.
1970; Păunescu 1978). In the lower levels of this rockshelter
species from various biomes have been recorded, reflecting
well the contrasts in the natural environment within the Iron
Gates gorge and its immediate hinterland during this period.
For example, in Cuina Turcului level I, along with the bones
of ibex, chamois, bison and horse (assumed to be species ad-
apted to an environment lacking forested habitats), a very
significant percentage of woodland fauna also appears (Fig.
1). On the other hand, a lower incidence of woodland fauna
and an increase in steppe and montane elements in level II
(Fig. 2) indicate that this level formed under colder climatic
conditions, most probably during the Younger Dryas.
However, it has to be acknowledged that the avifauna and
pollen analyses point to the opposite trend toward climatic
amelioration (Pop et al. 1970; Păunescu 1978).
Figure 2. Cuina Turcului, level II: frequency of ungulate
species identified.
12
The frequency of the faunal remains from different biomes
(forest, steppe and montane) in the lower levels of Cuina
Turcului indirectly suggest that the rockshelter during this
period was a base camp. The organization of hunting activit-
ies, the mode of hunting and, most probably, the weapons
used in hunting the species from these environments differ
significantly (for example, hunting ibex and chamois requires
a high level of specific skills — Gamble 1999: 230–240).
Therefore, in spite of the relatively short distances between
different ecological zones in the area of Veliki Kazan (‘the
Great Cauldron’) and their mutual connections (Bolomey
1973: 43–45), it may be assumed that the faunal remains in
the lower levels of the site are the result of a number of forays
with more-or-less specialized intentions. The remains from
level I indicate intensive hunting of wild boar, ibex and
chamois, while those from level II show an emphasis on ibex
and chamois (and the important role played by steppe anim-
als) with a very low ratio (c. 10%) of forest animals.
The quantity and character of the lithic industries from the
Final Palaeolithic sites of the Iron Gates offer additional in-
formation on the degree of mobility, and nature and duration
of human occupations of natural shelters. Mobility and con-
tacts at the regional level are attested by a significant pres-
ence of high quality raw materials at Cuina Turcului, espe-
cially in level II (Dinan 1996a, 1996b). Artefacts of local
grey flint are prevalent in both levels, while quartz is used in
only small percentages in both occupation phases.
Concerning technological indicators of the character of
the settlement, Dinan (1996b) suggested that cores had been
brought to the site in a partially prepared state and that most
of the artefacts (including tools and finished blanks) had been
produced outside the habitation. She based this proposition
on the fact that her analysis of a small sample of the artefacts
from Cuina Turcului (representing only 5–10% of the total
assemblage) revealed a small percentage of cortical pieces
(23% in level I and only 13% in level II). However, the huge
amount of chipping debris reported by Păunescu (1978) un-
doubtedly indicates that manufacturing of the majority of the
artefacts was executed inside the rockshelter.
Compared to the other Iron Gates sites, Cuina Turcului is
also characterized by a lower percentage of cores and re-
touched artefacts (Fig. 3), as well as a high percentage of
unretouched and retouched blades and bladelets. This pattern
is surely influenced by the intensive exploitation of high
quality cores, as well as by the fact that the bipolar flaking
technique was not widely implemented during this phase.
The typological composition of the retouched artefacts
from levels I and II of Cuina Turcului is presented in detail in
the works of Păunescu (1970, 1978). On the basis of the data
presented, the tool assemblages from these levels are charac-
terized by very high percentages of end-scrapers (the highest
of all the Iron Gates sites) — up to 60%. Retouched blades
and bladelets are well represented in the assemblage (over
20%) while backed tools make up 13% in level I, falling to
6% in level II. These data are well known, and it is the in-
tention here only to emphasize the very small percentage of
typologically undifferentiated (informal) tools. According to
Păunescu’s classification (Păunescu 1978: 19–20), only the
notched and denticulated tools may be listed as such, while
side-scrapers do not appear at all. However, the tendency to-
 Dušan Mihailović: Lithic technology and settlement systems
Figure 3. Tripolar graph representing a seriation of the cores, blades/flakes and tools at Iron Gates sites: A–Alibeg,
P–Padina, VL–Vlasac, VT–Veterani Terrace, IC–Icoana, R–Răzvrata, OB–Ostrovul Banului, SC–Schela Cladovei.
wards a highly selective use of blanks in the production of
tools is very marked, especially in level II. End-scrapers
dominate among the tools made from flakes, while blades
and bladelets were used in the production of backed and
geometric tools, end-scrapers (11.4% in level I and 16.5% in
level II), notched (>50% of notched pieces in both levels) and
denticulated tools (25.3% in level I and 13.0% in level II).
Early Mesolithic
The settlement of Padina, by Gospođin Vir (‘the Lady’s
Whirlpool’), has been dated by a series of radiometric 14C
ages on human bone (not corrected for the freshwater reser-
voir effect, cf. Cook et al. 2002) to the second half of the
ninth millennium cal BC (Burleigh & Živanović 1980). It
belongs to the earliest phase of habitation on the Danube
bank, following level II of Cuina Turcului (Jovanović 1971,
1974, 1987; Radovanović 1981, 1996). One should however
bear in mind that most of these dates have been established
for skeletal remains from the so-called ‘stone construction of
the necropolis’ in sector III, assigned to phase A2, so the finds
from phase A1 could be somewhat older (Radovanović 1996:
173, 359). Layers I and II of the site of Ostrovul Banului may
also belong to the initial phase of the establishment of open-
air settlements in the Iron Gates area (Boroneanţ 1999). The
13
finds from these layers are not dated, but they surely belong
to the period before the beginning of the seventh millennium
cal BC — the date of layer III. This site is not very well
known, as is the case with Veterani Terrace, having elements
of material culture of both this and the subsequent phase of
the Iron Gates Mesolithic (Boroneanţ 1973, 1989, 1999;
Radovanović 1996: 8–12).
The remains of the earliest settlement (A1) at Padina are
represented only by the finds from the central part of sector II
(Radovanović 1981). The majority of the finds assigned to
phase A2 also come from sector II and most of them were
found above the construction of flat stones and pebbles in
block 2a. In sector I a small number of artefacts were recor-
ded below the hearth attributed to the hypothetical phase AB.
There are indications, however, that the original settlement in
this sector was situated on the very edge of the river and was
probably destroyed by river erosion (Radovanović 1996: 62).
Therefore, it is possible that excavation of this part of the site
uncovered only a peripheral part of the settlement, and this
may explain the small number of artefacts belonging to phase
A. Artefacts from phase AB have been recorded in the hearth
in block 1b of sector II, and have been compared to the in-
dustry from Alibeg owing to the presence of typical micro-
liths and by-products of blade débitage (Radovanović 1981:
63). In any case, there is no certain archaeological evidence
of settlement continuity at Padina between phases A and B.
 The Iron Gates in Prehistory
Figure 4. Padina A: frequency of ungulate species
identified.
The fauna of Padina A is represented by the remains of a
large number of animal species and more than a thousand in-
dividual bone specimens (Clason 1980). Red deer remains
are the most frequent, while wild boar, roe deer and aurochs
(Bos primigenius) are represented by less than 15% (Fig. 4).
The quantity of fish bones is very large, up to about 25% of
the total faunal assemblage from this phase (in phase B fish
bones constitute 87%). This fact, along with the kind of fish
(a substantial number of large specimens), the context of the
majority of bones, and the very position of the settlement
(near Gospođin Vir), points to the importance of fishing in the
economy of the early inhabitants of Padina. However, the
quantity of red deer remains is a clear indication that Padina,
especially in its early phase, was not a specialized fishing
settlement. Unfortunately, the seasonality and the true nature
of the habitation at Padina cannot be reconstructed at present,
since the data indicating that the site was inhabited all
through the year apply to Padina A as a whole (Clason 1980:
171), not its individual phases. Phase A1 may therefore be
judged only on the basis of the chipped stone artefacts.
Figure 5. Percentage of bipolar cores in relation to the
total number of cores: A–Alibeg, P–Padina, VL–Vlasac,
LV–Lepenski Vir, VT–Veterani Terrace, IC–Icoana,
R–Răzvrata, OB–Ostrovul Banului, SC–Schela Cladovei.
14
Padina A is different from the other sites of the Iron Gates
owing to the very poorly preserved architectural features and
the large quantity of faunal remains and chipped stone arte-
facts concentrated in a small area (Radovanović 1981: 26).
For example, in block 2a of sector II, as many as 75 lithic
artefacts per square metre were recovered — more than in
level I of Cuina Turcului, where the average density of finds
was 72.9 per square metre (in level II as high as 143.9 per
square metre). However, the composition of the finds from
phases A1 and A2 in sector II does not meet the criteria for
interpretation as a lithic workshop (Kozłowski 1980) — i.e.
the percentage of retouched tools in these levels (23.7% in
phase A1 and 15.1% in phase A2) is among the highest in the
Iron Gates and is only matched by levels I and II of Ostrovul
Banului (Radovanović 1996: 233). The high percentage of
retouched tools on Palaeolithic and Mesolithic sites is usually
an indication of short-term habitations, but in the case of
Padina A it seems more probable that it reflects the nature and
intensity of the activities performed in that part of the site.
This question, just as with the issue of seasonality, will
hopefully be solved after analysis of the faunal remains has
been completed.
The industry of Padina A differs from that in the lower
levels of Cuina Turcului in having a lower frequency of high
quality raw materials, a much higher percentage of flakes and
tools on flakes, and a much smaller proportion of backed
pieces in phase A2. A somewhat smaller percentage of unre-
touched artefacts in the overall inventory of Padina A1 and
A2 may be explained by the higher percentage of bipolar
cores (Fig. 5) and, on the other hand, by the less intensive use
of this type of core (i.e. the smaller number of flakes removed
from them). A similar pattern appears in layer II of Ostrovul
Banului, but the frequency of cores (100% bipolar) is some-
what lower, retouched tools are more frequent, and the per-
centage of unretouched artefacts is lower, compared to
Padina.
A preliminary analysis of the raw material composition
indicates that at Padina the same raw materials were used as
on the neighbouring sites of Vlasac and Lepenski Vir (Fig. 6).
Artefacts made from grey and grey-brown flint predominate.
Low quality siliceous brown and greenish stones are much
less frequent (including siltstone and mudstone). A very
small percentage of the artefacts are of chalcedony, and red
and green jasper, while finds of obsidian are very few. Quartz
is poorly represented in the form of unretouched flakes, but
not cores or retouched tools. All this points to a decrease in
mobility and longer periods of occupation at the settlements.
It may be argued that the differences between Cuina Turcului
and Padina, noted in all three categories (raw materials, flak-
ing technology, tools inventory), do not result simply from
the fact that Padina is an open-air settlement while Cuina
Turcului is a rockshelter. Rather, the differences probably re-
flect primarily the change in the mode of habitation and ex-
ploitation of resources taking place in this period. These
changes are clearly recorded in the evolution of the industries
at numerous Palaeolithic and Mesolithic sites in Montenegro
and Greece (Mihailović 1999a, 2001) — especially at Crvena
Stijena (Mihailović 1998) and Franchthi Cave on the
Peloponnese (Perlès 1987, 1990, 1999).
Further analyses of the material from Padina have shown
 Dušan Mihailović: Lithic technology and settlement systems
Figure 6. Flint artefacts as a percentage of the total num-
ber of artefacts of flint, quartz/quartzite, and other siliceous
rocks in Iron Gates sites: A–Alibeg, P–Padina, VL–Vlasac,
LV–Lepenski Vir, VT–Veterani Terrace, IC–Icoana,
R–Răzvrata, OB–Ostrovul Banului, SC–Schela Cladovei.
that selective use of raw materials in the production of certain
types of blanks and retouched tools is very poorly expressed,
but is present nonetheless. Grey flint is the basic material in
bipolar débitage — attested by the relatively high number of
bipolar cores of this material and a large number of flakes
with a linear, almost unrecognizable striking platform. It is
worth noting that the implementation of this technology
seems to be more evident from the characteristics of the
striking platform than the scars on the dorsal surface. A kind
of specialization and economization of raw material re-
sources is attested by the characteristic products of flaking:
flakes with two ventral sides and blades resembling burin
spalls, detached from so-called plate cores. All this clearly
indicates that all the available supports were used for knap-
ping (as cores), including very small thick flakes. An
identical situation has been recorded at Vlasac (Kozłowski &
Kozłowski 1982: 33–36, 43–45). Also, as at Cuina Turcului
(Dinan 1996b: 19), many cores are secondarily processed —
in fact, turned into retouched implements.
It needs to be stressed that bipolar technology and micro-
lithization represent the major traits of the chipped stone in-
dustry, not only of Padina, but also the other Iron Gates sites
(Boroneanţ 1989, 1999). This is important because these two
elements are not so well expressed on the sites in Montenegro
and Greece, where the proportion of bipolar cores is lower
and the tendency toward microlithization of tools made on
flakes is not so pronounced (Perlès 1987; Mihailović 1996,
1998, 1999a). In the next phase of the Iron Gates Mesolithic,
on the sites of the lower Gorge, the industries acquire a
‘quartzose’ character; bipolar technology is used on quartz
too, and quartz is also used in the production of tools. As the
technology of quartz flaking is mastered and tools of this
material appear, the need for systematic acquisition of flint
decreases. The composition of the raw material of the indus-
tries from the Veterani Terrace and Ostrovul Banului clearly
shows a gradual decrease in acquisition of flint from the hill
of Sviniţa in the upper Gorges (Boroneanţ 1989: 477;
Radovanović 1996: 228–234), while in the case of Alibeg
quartz represents the dominant material, although flint from
15
the Privod hill, in the close vicinity of the site, is used
(Radovanović 1996: 316). All this indicates that mobility
and, most probably, the need to acquire high quality raw ma-
terials decreases markedly in this period. Underlying this ap-
proach to the exploitation of lithic resources must have been
the achievement of a certain level of technological specializ-
ation, necessary for successful knapping of quartz cores.
Conclusion
Changes in settlement and subsistence practices may be ob-
served in the Iron Gates from the Final Palaeolithic to the
beginning of the Neolithic (Tables 1 & 2). In the Final
Palaeolithic of Cuina Turcului the ratio of faunal remains and
the lithic raw materials used point above all to the greater
exploitation of resources in the immediate vicinity of settle-
ments, although there are elements that point to regional mo-
bility and, even, inter-regional contacts (obsidian).
According to the available data, it may be assumed that dur-
ing this period the rockshelter was a base camp, and that the
activities of the human groups were directed toward the Iron
Gates hinterland. The differences in the span and nature of
the habitations point to functional differences among Cuina
Turcului and the caves of Climente II and Hoţilor Băile
Herculane (Bolomey 1973; Dinan 1996a), although their
contemporaneity and role in the settlement system cannot be
established with certainty. As far as subsistence is concerned,
there is no evidence of specialization on particular animal
species. Economy in the use of resources and the level of
elaboration of technological process are attested by micro-
lithization of artefacts, standardization of tools, and selective
use of particular raw materials and especially blanks in the
production of certain categories of artefacts.
The possibility cannot be ruled out that as early as the
Younger Dryas, which over the last decade has been estab-
lished as an abrupt, cold oscillation of global character
(Sherratt 1997; Adams & Otte 1999), the Iron Gates region
was a natural refugium that was intensively inhabited. There
is, however, no evidence to back up this assumption. The
settlements on the river bank, even if they existed, were
washed away by river erosion, either by the beginning of
Holocene, after the increase in the Danube level, or in the
subsequent periods. However, in level II of Cuina Turcului,
no matter whether it belongs to the Younger Dryas (as indic-
ated by several lines of evidence) or to the very beginning of
Holocene, there are features that point to the exploitation of
a poor environment (orientation toward hunting ibex and
chamois, the additional supply of resources from the steppe
biome). Also the total quantity and composition of the arte-
facts and faunal remains in this level indicate that Cuina
Turcului in this period had a clearer role as a base camp, but
at the same time its economic territory was somewhat larger,
embracing some elements of a logistical system of settlement.
With regard to the Holocene and the establishment of
open-air settlements, changes in the settlement system, ac-
quisition of food and mineral resources, and flaking techno-
logy are gradual but radical — as seen clearly in the finds
from Padina A, Ostovul Banului I and II, and Veterani
Terrace. The continuity with the Final Palaeolithic is marked
 The Iron Gates in Prehistory

Table 1. Comparison of Padina A and Cuina Turcului I–II in terms of flaking technology.

Lithics Variability Degree of operationalization Intensification

- various raw materials, from different sources, - high degree of raw material - intensive exploitation of
collected and transported in different ways selectivity in blade/flake blade/bladelet cores
- different types of blanks (including production (decreasing) - rejuvenation of cores during
FINAL PALAEOLITHIC microbladelets) - high degree of blank selectivity exploitation
(Cuina Turcului) - pronounced variability and standardisation of in tool production - microlithisation
tools (especially in microlithic group) - secondary modification of
cores and tools (repairing
and retooling)

- local, bad quality raw materials prevail
- flakes predominate in artefact assemblages
EARLY MESOLITHIC
- low variability of tools; low level of tool
(Padina)
standardisation
- high – but only in bipolar core
exploitation
- low level of raw material se-
lectivity (with exception of grey
flint and mudstone)
- low level of blank selectivity in
tool production
- intensive exploitation of
bipolar and other (irregular
flake) cores
- characteristic by-products
- microlithisation

Table 2. Comparison of Padina A and Cuina Turcului I–II in terms of subsistence strategy.

Fauna Diversity in environment and Degree of operationalization of Intensity in exploitation of

archaeological record hunting technique, game selection environment, some faunal species
and prey processing and prey

- relatively poor environment ? - different hunting strategies - unknown – confirmed in other

- species from different biomes (Cuina
FINAL PALAEOLITHIC Turcului I)
(Cuina Turcului) - orientation towards exploitation of
species from mountainous
environment (Cuina Turcului II)
- site location near or at moderate
distance from area of exploitation
- activities directed to the hinterland
- hunting equipment: projectiles sites in Balkans (Badanj, Klithi)
with inserted backed points and/or
geometric microliths
* selective hunting (size, age, sex
groups) confirmed for some other
Late Upper Palaeolithic sites in the
Balkans (Badanj, Klithi); no data
available for Iron Gates sites.

- relatively rich environment - new hunting/fishing techniques – - no available data

- orientation towards fishing and deer probably highly developed
EARLY MESOLITHIC hunting - hunting by bone points ?
(Padina) - location of settlement near zone of * Intensification is visible only in
exploitation * no data available for selective hunt- the number of hunted specimens
ing in this period; in fishing – orienta- (i.e. in the quantity of finds)
tion to bulky specimens

solely in the repertoire of tools — by the presence of backed
pieces, geometric artefacts, and circular and thumbnail
scrapers. On the other hand, in terms of the composition of
raw materials there is a sudden decrease in the presence of
high-quality stones, while in technological terms flakes and
tools on flakes show a growing trend. The changes in sub-
sistence are even more marked. Compared to the preceding
period, when survival of the communities, living in a relat-
ively poor environment, depended on maximal exploitation
of diverse resources, in the Early Mesolithic there was an in-
tensification of the exploitation of selected resources (deer
1
hunting and fishing), in a (probably) much richer environ-
ment. This appears to have been accompanied by a similar
trend in technology — the narrowing of the repertoire of the
artefacts and intensification in the production of implements
of low-quality flint. This was a consequence of several
factors: (i) the disintegration of cultural identity caused by
the abandonment of the settlement system practised in the
preceding period, (ii) changes in hunting techniques, and (iii)
a reduction in mobility of human communities, with longer
stays at habitation sites and, probably, different activities be-
ing performed there.

16
 Dušan Mihailović: Lithic technology and settlement systems
References
Adams, J. & Otte, M. 1999: Did Indo-European languages spread
before farming? Current Anthropology 40: 73–76.
Bailey, G., Carter, P., Gamble, C. & Higgs, H. 1983: Epirus revis-
ited: seasonality and inter-site variation in the Upper Palaeolithic
of north-west Greece. In Bailey, G. (ed.) Hunter-Gatherer
Economy in Prehistory. Cambridge: Cambridge University
Press, 64–78.
Bar-Yosef, O. & Meadow, R.H. 1995: The origins of agriculture in
the Near East. In Price, T.D. & Gebauer, A.B. (eds) Last
Hunters–First Farmers: New Perspectives on the Prehistoric
Transition to Agriculture. Santa Fe: School of American
Research Press, 39–94.
Binford, L.R. 1979: Organization and formation processes: looking
at curated technologies. Journal of Anthropological Research 35:
255–273.
Binford, L.R. & O’Connell, J. 1984: An Alawyara day: the stone
quarry. Journal of Anthropological Research 40: 406–432.
Bolomey, A. 1970: Anexa nr. III — Cîteva observaţii asupra faunei
de mamifere din straturile romanello-aziliene de la Cuina
Turcului. Studii şi Cercetări de Istorie Veche 21(1): 37–39.
— 1973: An outline of the Late Epipalaeolithic economy at the ‘Iron
Gates’: the evidence on bones. Dacia n.s. 27: 41–52.
Boroneanţ, V. 1973: Recherches archéologiques sur la culture
Schela Cladovei de la zone des ‘Portes de Fer’, Dacia n.s. 27:
5–39.
— 1989: Thoughts on the chronological relations between the Epi-
Palaeolithic and the Neolithic of the Low Danube. In Bonsall, C.
(ed.) The Mesolithic in Europe. Edinburgh: John Donald,
475–480.
— 1999: The Mesolithic habitation complexes in the Balkans and
Danube Basin. Living Past 1. http://www.cimec.ro/livingpast/
mesolithic.htm
Burleigh, R. & Živanović, S. 1980: Radiocarbon dating of a Cro-
Magnon population from Padina, Yugoslavia, with some general
recommendations for dating human skeletons. Zeitschrift für
Morphologie und Anthropologie 70(3): 269–274.
Clason, A.T. 1980: Padina and Starčevo: game, fish and cattle.
Palaeohistoria 22: 141–173.
Cook, G.T., Bonsall, C., Hedges, R.E.M. , McSweeney, K.,
Boroneanţ, V., Bartosiewicz, L. & Pettitt, P. 2002: Problems of
dating human bones from the Iron Gates. Antiquity 76: 77–85.
Dinan, E.H. 1996a: A preliminary report on the lithic assemblage
from the early Holocene level at the Iron Gates site of Băile
Herculane. Mesolithic Miscellany 17(2): 15–24.
— 1996b: Preliminary lithic analysis of the Epigravettian levels
from the Iron Gates site of Cuina Turcului. Mesolithic
Miscellany 17(2): 25–40.
Gamble, C. 1993: Exchange, foraging and local hominid networks.
In Scarre, C. & Healy, F. (eds) Trade and Exchange in
Prehistoric Europe. Oxford: Oxbow Books, 35–44.
— 1999: The animal bones from Klithi. In Bailey, G.N., Adam, E.,
Perlès, C., Panagopoulou, E. & Zachos, K. (eds) The
Palaeolithic Archaeology of Greece and Adjacent Areas.
London: British School at Athens, 207–244.
Jovanović, B. 1971: Elements of the Early Neolithic architecture in
the Iron Gate gorge and their functions. Archaeologica
Iugoslavica 9: 19.
— 1974: Praistorija gornjeg Ðerdapa, Starinar n.s. 22: 122.
— 1987: Die Architektur und Keramik der Siedlung Padina B am
Eisernen Tor, Jugoslawien. Germania 65: 116.
Kozłowski, J.K. 1980: Technological and typological differentiation
of lithic assemblages in the Upper Palaeolithic: an interpretation
attempt. In Schild, R. (ed.) Unconventional Archaeology: New
1
17
Approaches and Goals in Polish Archaeology. Wrocław: Polska
Akademia Nauk Instytut Historii Kultury Materialnej, 33–55.
— 1999: Gravettian/Epigravettian sequences in the Balkans: envir-
onments, technologies, hunting strategies and raw material pro-
curement. In Bailey, G.N., Adam, E., Perlès, C., Panagopoulou,
E. & Zachos, K. (eds) The Palaeolithic Archaeology of Greece
and Adjacent Areas. London: British School at Athens, 319–329.
Kozłowski, J.K. & Kozłowski, S.K. 1982: Lithic industries from the
multi-layer site Vlasac in Yugoslavia. In Kozłowski, J.K. (ed.)
Origin of the Chipped Stone Industries of the Early Farming
Cultures in Balkans. Warszawa–Kraków: Państwowe Wydaw-
nictwo Naukowe, 11–109.
Mihailović, D. 1996: Upper Palaeolithic and Mesolithic chipped
stone industries from the rockshelter of Medena Stijena. In
Srejović, D. (ed.) Prehistoric Settlements in Caves and Rock-
shelters of Serbia and Montenegro — Fascicule I. Belgrade:
Centre for Archaeological Research, 9–60.
— 1998: Gornji paleolit i mezolit Crne Gore. Unpublished PhD
dissertation, Faculty of Philosophy, University of Belgrade.
— 1999a: The Upper Palaeolithic and Mesolithic stone industries of
Montenegro. In Bailey, G.N., Adam, E., Perlès, C.,
Panagopoulou, E. & Zachos, K. (eds) The Palaeolithic
Archaeology of Greece and Adjacent Areas. London: British
School at Athens, 343–356.
— 1999b: Intensification of settlement in the Late Glacial of south-
western Balkans. In Kobusiewicz, M. & Kozłowski, J.K. (eds)
Post-pleniglacial Re-colonisation of the Great European
Lowland. Folia Quaternaria 70. Kraków: Polska Akademia –
Komisja Paleogeografii Czwartorzędu, 385–392.
— 2001: Technological decline of the early Holocene chipped stone
industries in South-east Europe. In Kertész, R. & Makkay, J.
(eds) From the Mesolithic to the Neolithic. Proceedings of the
International Archaeological Conference held in the Damjanich
Museum of Szolnok, September 22–27, 1996. Budapest:
Archaeolingua, 339–347.
Miracle, P.T. 1996: Diversification in Epipalaeolithic subsistence
strategy along the eastern Adriatic coast: a simulation approach
applied to zooarchaeological assemblages. Atti della Società per
la Preistoria e Protostoria della Regione Friuli-Venezia Giulia 9
(1994–1995): 33–62.
Păunescu, A. 1970: Epipaleoliticul de la Cuina Turcului-Dubova.
Studii şi Cercetari de Istorie Veche 21(1): 3–47.
— 1978: Cercetarile arheologice de la Cuina Turcului-Dubova (Jud.
Mehedinţi). Tibiscus Istorie, Volum Închinat Celei de-a 60
Aniversari a Unirii: 11–56.
— 1989, Le Paléolithique et le Mésolithique de Roumanie (un bref
aperçu). L’Anthropologie 93(1): 123–158.
Perlès, C. 1987: Les Industries Lithiques Taillées de Franchthi
(Argolide, Grèce). Tome I: Présentation Générale et Industries
Paléolithiques. Bloomington, Indianopolis: Indiana University
Press.
— 1990: Les Industries Lithiques Taillées de Franchthi (Argolide,
Grèce). Tome II: Les Industries du Mésolithique et du
Néolithique Initial. Bloomington, Indianopolis: Indiana
University Press.
— 1999: Long term perspectives on the occupation of the Franchthi
Cave. In Bailey, G.N., Adam, E., Perlès, C., Panagopoulou, E. &
Zachos, K. (eds) The Palaeolithic Archaeology of Greece and
Adjacent Areas. London: British School at Athens, 311–318.
Pop, E., Boscaiu, N. & Lupsa, V. 1970: Anexa nr. 1 — Analiza
sporo-polenica a sedimentelor de la Cuina Turcului-Dubova.
Studii şi Cercetari de Istorie Veche 21(1): 31–34.
Radovanović, I. 1981: Ranoholocenska kremena industrija sa
lokaliteta Padina u Ðerdapu. Belgrade: Arheološki Institut.
 The Iron Gates in Prehistory
— 1996: The Iron Gates Mesolithic. Ann Arbor, International
Monographs in Prehistory.
Radovanović, I. & Voytek, B. 1997: Hunters, fishers and farmers:
sedentism, subsistence and social complexity in the Iron Gates
Mesolithic. Analecta Praehistorica Leidensia 29: 19–31.
Sherratt, A. 1997: Climatic cycles and behavioural revolutions: the
emergence of modern humans and the beginning of farming.
Antiquity 71: 271–287.
Sinclair, A. 1999: Technological decision making: the case of
Megalakkos. In Bailey, G.N., Adam, E., Perlès, C., Panago-
poulou, E. & Zachos, K. (eds) The Palaeolithic Archaeology of
18
Greece and Adjacent Areas. London: British School at Athens,
188–196.
Tchernov, E. 1991: Biological evidence for human sedentism in
Southwest Asia during the Natufian. In Bar-Yosef, O. & Valla, F.
(eds) The Natufian Culture in the Levant. Ann Arbor, MI:
International Monographs in Prehistory, Archaeological Series 1,
315–340.
Voytek, B. & Tringham, R. 1989: Rethinking the Mesolithic: the
case of South-east Europe. In Bonsall, C. (ed.) The Mesolithic in
Europe. Edinburgh: John Donald, 492–499.
 The development of the ground stone industry in the
Serbian part of the Iron Gates

Dragana Antonović

Abstract: The early Holocene ground stone industry of the Iron Gates comprises two components – local and ‘imported’. The local com-
ponent was developed independently against a background of Mesolithic sedentary settlement, according to the basic economic needs
(hunting and fishing) of the inhabitants of this area. The tools were manufactured by limited modification of natural pebbles. Alongside the
artefacts typical of the Iron Gates ground stone industry, occur ground edge tools (axes, adzes and chisels) characteristic of the Neolithic
of central Serbia that were used exclusively for woodworking – these are referred to as the ‘imported’ component. The tools of the ‘imported’
component have traits that are very different from the tools of the local component. The parallel appearance of two ground stone industries
of differing character can be regarded as another line of evidence for the parallel development of the Iron Gates Mesolithic and the Neolithic
of the neighbouring areas which, after all, is indicated by absolute dating.
Key words: ground stone industry, Mesolithic, Neolithic, Iron Gates, Serbia

Introduction
The early Holocene ground stone industry of the Iron Gates
is an exceptional phenomenon in Serbia. Nothing resembling
it has been discovered elsewhere in the country. On present
evidence, there would appear to have been an autochthonous
and continuous development of the industry from the Early
Mesolithic, independent of the origin and development of the
Neolithic in the other parts of Serbia and, it seems, from the
Neolithic in Romania and Bulgaria.
In the course of archaeological research in the Iron Gates,
the ground stone tools have been analysed from Mesolithic
and Early Neolithic (VIII–VI millennium cal BC) horizons at
the following sites: Padina, Lepenski Vir, Vlasac, Hajdučka
Vodenica, Ajmana, Velesnica, Knjepište, and Ušće
Kameničkog Potoka (Fig. 1). Included in the category
‘ground stone artefacts’ are all stone implements worked by
grinding/polishing, as well as unfinished examples with
traces of flaking or pecking. Also included under this heading
are implements that are naturally polished or polished
through use. This group includes various polishers, hammer-
stones and working plates,1 made from unmodified pebbles
that were suitable for use.
All implements were studied from two main aspects: pet-
rographic and functional–typological.
Petrographic analysis. The raw materials from which the
ground stone implements are made were examined only
macroscopically by non-destructive methods. They were
tested for hardness and compactness — hardness is given in
terms of the Mohs’ scale, tested by scratching with a steel
needle; compactness is estimated by simple visual observa-
tion of the specimen. Then the presence of calcium carbonate
was determined by testing the specimen with dilute hydro-
chloric acid. The mineral composition, texture and structure
of the rock were determined by inspection with a hand lens.
Some of the stones recovered in the excavations at Lepenski
Vir and Vlasac were identified by Miomir Babović of the
Mining-Geology Faculty in Belgrade (Srejović & Babović
1981, 1983). Other stones from Lepenski Vir, as well as from
Padina and Velesnica, were analysed by the present author.
The field documentation from some Iron Gates sites contains
descriptive rock identifications, but these are not sufficiently
detailed or accurate for present purposes.
Functional–typological analysis. The method of manu-
facture of the artefacts, their shape and use (determined by
microwear analysis) were considered. Striations on some of
the tools from Lepenski Vir and Velesnica were examined
under a binocular microscope at 50X magnification, and in-
terpreted on the basis of previous experience and examina-
tion (cf. Semenov 1976: 126–142). Typological analysis of
the ground edge tools (axes, adzes and chisels) was done ac-
cording to the classification of the material from Vinča
(Antonović 1992: 7–16). Analysis of other artefact types
made use of the observations of previous researchers in the
Iron Gates.
The raw materials
The raw materials are extremely varied, which is not surpris-
ing since most of the ground stone implements are made from
rounded, river pebbles. Magmatic, sedimentary and meta-
morphic rocks were used equally for their manufacture. The
physical-technical characteristics of the stone were decisive
in the selection of materials. Thus, predominantly fine-
grained, blow-resistant rocks without cracks and natural
flaws were chosen for the manufacture of ground stone
implements.
Magmatic rocks
Granite
Granites are the most widespread plutonic rocks, most com-
monly exhibiting various shades of grey and reddish colour.
Only hard rocks, resistant to abrasion, were selected. In the
Lepenski Vir locality fine-grained, very compact, grey-
coloured varieties, are represented. They were used for the

19
 The Iron Gates in Prehistory
Figure 1. Early Holocene sites in the Serbian part of the Iron Gates where ground stone implements have been recorded.
manufacture of hammers, which were found in all horizons
of the site. At Velesnica fine- and medium-grained massive
pebble varieties are found, which were used either as ham-
merstones or shaped as mallet-weights. Larger granite blocks
served as querns or stationary grindstones/polishers for
working hard materials. At Vlasac granite was used for the
manufacture of polishers for working bone needles (Srejović
& Letica 1978: 99). At Padina granite pebbles were utilized
as anvils.
Gabbro
Like granites, gabbros are tough rocks — i.e. resistant to
breakage. Artefacts of gabbro are reported from Vlasac and
Velesnica. At Vlasac fine-grained gabbro was used for the
manufacture of grooved polishers (Srejović & Babović 1981:
76). At Velesnica fine-grained, compact and hard varieties
(gabbro-diabases?) are represented; these were used for the
manufacture of axes for woodworking, pounders, and small
fishing weights. At Padina uralitized and saussirized gabbro
was used for making axes.
20
Diabase
Diabases are rocks characterized by ophitic texture. They are
hard, compact and abrasion resistant. At Lepenski Vir a fine-
grained, dark-grey variety is represented, which was used for
the manufacture of axes and chisels for woodworking. At
Velesnica dark-grey fine-grained diabase was used as
pebble-hammerstones.
Andesite
Andesites are compact and tough rocks exhibiting clearly
visible porphyritic texture. At Lepenski Vir there are fine-
grained, compact, dark-grey varieties that were used for the
manufacture of woodworking axes. At Vlasac small pebbles
of this material were utilized as sling-balls (Srejović &
Babović 1981: 76).
Aplite
Aplites are fine-grained rocks (related to granites) composed
mainly of quartz and feldspar, and therefore white in colour.
At Velesnica small fishing weights were made from aplites.
 Dragana Antonović: The ground stone industry in the Serbian Iron Gates
Tuff
These are compact rocks that are easily working. At Padina
slightly silicified tuff (hardness 6) exhibiting an irregular
conchoidal fracture and various shades of grey and ochreous
colour, was used for the manufacture of tranchets and ham-
merstones. It is likely that the tranchets from Vlasac were
also made of tuff, and not of some undefined siliceous rock
(cf. Srejović & Letica 1978: 98).
Other magmatic rocks
It is possible that in the Iron Gates other magmatic rocks were
also utilized, but precise identifications have not been made.
At Lepenski Vir various kinds of hard, very fine-grained, ba-
sic rocks were used, which could not be characterized mac-
roscopically. These were used for the manufacture of axes,
both blunt forms (Srejović 1972: 133) and ground edge
forms. At Velesnica poorly defined magmatic rocks were also
noted. These comprise fine-grained, dense, compact and hard
rocks, of uniform ochreous colour, which occurred as
pebbles, as well as fine-grained, atypical, compact and hard
rocks, ochre-grey in colour, whose constituents are not re-
cognizable owing to the presence of a deposit on the surface
of the pebble. Such pebbles were used, slightly reworked, as
small fishing weights and as anvils for stone tool working. At
Vlasac various magmatic rocks (not identified more precisely
— Srejović & Letica 1978: 99–100) were used for the man-
ufacture of sceptres, hammerstones, working plates and
sling-balls, which occurred in all horizons of the Vlasac
Mesolithic. From the site of Ušće Kameničkog Potoka a
‘green stone of volcanic origin’ is mentioned, which was used
for the manufacture of axes and chisels (Stanković 1986a:
468). Most probably, this is a gabbroic rock.
Sedimentary rocks
Sandstones
These rocks may be used, especially when quartzose, as an
abrasive material. Medium- and fine-grained varieties can be
distinguished on the basis of grain size, and on the basis of
the cement, they can be classified as calcareous, marly, fer-
ruginous or siliceous sandstones. These rocks can readily be
accepted as the leading raw material used during the
Mesolithic and Early Neolithic in the Iron Gates area. All art
objects (sculptures, altars) are made from sandstone pebbles
(Srejović & Babović 1983: 107–181). Sandstones were
largely used in the production of ground stone artefacts. In all
levels at Lepenski Vir there are sandstones with siliceous and
calcareous cement, as well as very fine-grained, weakly
metamorphosed varieties. Pebbles of quartzose sandstones
are very abundant. They served for the manufacture of altars
and sculptures. From the fine-grained varieties were made
axes, mallet-weights, hammerstones and trapezoidal
grindstones/polishers embedded in the floors of houses, as
well as pounders, which were found lying on the house floors
beside the grindstones (Srejović 1972: 133; Srejović &
Babović 1981: 76–77; Srejović & Babović 1983: 186–189).
At Velesnica sandstones were the most common raw materi-
al. Varieties with calcareous, marly and siliceous cement
were used — most commonly these were fine-grained, occa-
sionally they were nearly fine-grained, compact and hard
varieties. Medium- to coarse-grained sandstones and com-
21
pletely silicified fine-grained types were only occasionally
recorded. At this site sandstones were used for the manufac-
ture of pounders, whetstones, polishers, millstones, working
plates, mallets, pebble-axes, weights, line winders, and axes;
all of these artefacts, except the last two types, were manu-
factured from partially modified pebbles. At Vlasac various
coarse- and fine-grained varieties were used for the manu-
facture of sceptres, pebble-axes, hammerstones, and polish-
ers for awls (Srejović & Letica 1978: 99–100; Srejović &
Babović 1981: 76).
Limestones
These are compact, fine-grained rocks, of hardness 4 (except
silicified limestone with hardness grading up to 6.5).
However, they are brittle and, therefore, not suitable for the
manufacture of hammering tools. At Lepenski Vir several
varieties of these calcareous rocks were in use. Weakly sili-
cified, compact limestone with hardness 5, exhibiting various
shades of grey and light greyish-green colour, was utilized
for the manufacture of axes, adzes and hammers. Dense, hard
limestones, as well as marly varieties, were used for the
manufacture of sceptres (Srejović & Babović 1983:
186–189); whereas ornamental artefacts — beads and amu-
lets — were made from whitish and reddish crystalline lime-
stones (Srejović & Babović 1981: 72, 92; Srejović &
Babović 1983: 191–197). Limestones are present in all hori-
zons of Lepenski Vir. At Velesnica fine-grained, silicified,
compact and hard, grey-coloured varieties were utilized for
the manufacture of small fishing weights that retained the
form of the pebbles. At Vlasac artefacts of limestone are also
represented (Srejović & Letica 1978: 98).
Other sedimentary rocks
In addition to the sandstones and limestones mentioned
above, which were abundantly used for the manufacture of
ground stone artefacts, other sedimentary rocks are present in
the Iron Gates sites in much smaller quantities. Marlstone
and mudstone without calcite were used at Lepenski Vir for
the manufacture of sceptres (Srejović & Babović 1983: 188).
Silicified sedimentary rocks were also used to a lesser extent;
for example; at Velesnica, where silicified pelite is recorded.
This rock closely resembles chert; it is off-white in colour
with alternating thin darker and lighter bands, compact, fine-
grained, has a hardness around 6, a conchoidal fracture, and
contains calcium carbonate. Axes were made from this rock,
on account of its hardness and toughness. At Vlasac there
were numerous sceptres made from sedimentary rocks the
character of which has not been clearly defined (Srejović &
Letica 1978: 99).
Metamorphic rocks
Gneiss and granite gneiss
The gneisses are distinguished from granites by parallel
(schistose) structure. They are mostly medium- to coarse-
grained, compact, hard rocks. When fine grained and with the
poorly developed schistosity, especially granite gneisses,
they differ slightly from granite macroscopically. Granite
gneiss is a rock from which at Lepenski Vir sceptres were
made (Srejović & Babović 1983: 185). Gneiss is recorded at
Velesnica, where pebbles of it were used as hammerstones.
 The Iron Gates in Prehistory
Phyllite and phyllite–mica schist
These are fine-grained (slightly crystalline) schistose rocks.
Phyllite–mica schist is a phyllite grading to mica schist. At
Lepenski Vir axes and chisels were made from these rocks,
these being found in all levels of the site.
Amphibolite
These rocks may exhibit both schistose and massive struc-
ture. Most are fine-grained and compact; when massive or
with reduced schistosity, they could represent a high quality
raw material for implement production. Amphibolite arte-
facts are present at Vlasac (pebble-axes; Srejović & Babović
1981: 75) and Lepenski Vir (sceptres; Srejović & Babović
1983: 185).
Other types of schist
These are medium- to coarse-grained rocks of medium to
high crystallinity, with perfect schistosity, which makes them
easily recognizable. The various schists were preferred for
the manufacture of ground stone implements, especially
sceptres. Compact rocks were always used, but it should be
stressed that some of these are not exceptionally hard rocks,
as is the case with chlorite schists. Amphibole schist was used
at Padina for sceptre manufacture. Some sceptres from
Vlasac are made of albite-chlorite-muscovite schists
(Srejović & Babović 1981: 74), but also of other varieties of
schists. The sceptres from Lepenski Vir are most commonly
made from muscovite-chlorite schist (Srejović & Babović
1983, 185–187). At Velesnica amphibole-muscovite schist is
recorded, which was utilized for the production of axes.
Eclogite
This is a massive, metamorphic rock of high crystallinity
produced by metamorphosis of basic rocks. It is hard and
tough, suitable for the manufacture of implements, but was
not common in the early Holocene Iron Gates settlements. It
is recorded at Lepenski Vir, where it was used for the pro-
duction of chisels.
Marbles
These are white, granular rocks, exhibiting a wide range in
grain size. They are of relatively low hardness (3–4 on Mohs
scale), and therefore easy to work. When polished they ex-
hibit a high lustre. Because of these characteristics they were
a popular raw material for the manufacture of decorative and
ceremonial artefacts — pendants and amulets, which are re-
corded only in Lepenski Vir I–II (Srejović & Babović 1983:
190–194).
Serpentinites
These are generally loose, highly fractured rocks of lesser
hardness (about 4). Although being of low hardness, compact
varieties of this rock could be, due to its toughness, suitable
for tool manufacture, especially when partially silicified.
Most common are green or dark green varieties, but brown
and nearly black varieties also occur. Serpentinites were a
preferred material for the manufacture of small woodworking
tools, such as chisels and small adzes, during the Starčevo
and late Vinča culture periods in Serbia (Antonović 1998: 36,
144). In the early Holocene Iron Gates localities serpentinite
22
axes, adzes and chisels are recorded in the Lepenski Vir IIIb
levels, chisels are recorded at Velesnica, and
grindstones/polishers for awls were found in the Vlasac II
levels (Srejović & Babović 1981: 77).
Other rocks
Fine-grained sedimentary and thermal-contact metamorphic
rocks
These terms cover the main rock types used for ground stone
artefacts in the Neolithic of Serbia. They are fine-grained,
very dense rocks, resistant to percussion, of hardness 6–6.5,
and having an irregular conchoidal fracture, which makes
them suitable for knapping. They are variable in colour —
light to dark greyish-green, grey, brown, green, dark grey and
black, commonly with lighter or darker bands and light spots.
They comprise mainly hornfelses, skarnoids, alevrolithic
metasandstones, and silicified pelites. The hornfelses and
skarnoids are contact metamorphic rocks, whereas the alev-
rolithic metasandstones and pelites are sedimentary rocks
that have undergone slighter changes. Fine-grained variants
of these rocks are recorded in Lepenski Vir I and III, where
they were used for axe, adze and chisel manufacture. They
are also present at Velesnica, where they were used for the
manufacture of axes.
Jadeite/nephrite
Jadeite/nephrite are an interesting occurrence on Neolithic
sites in the Iron Gates. These are actually two different min-
erals — jadeite is pyroxene and nephrite is amphibole — but
cannot be distinguished macroscopically (both are known
under the common name, ‘jade’). For that reason, it is not
possible to say which of the minerals occurs at the Iron Gates
sites. They are very dense and tough minerals, 6–6.5 in
hardness, translucent, and green to olive green in colour.
They are hard to work because they are hard to break, and
thus are suitable for the manufacture of high quality tools.
Jade is a rare mineral in the early Holocene Iron Gates sites.
It is recorded in Lepenski Vir Ib–c, where it was used for
manufacturing chisels. At Velesnica a fragment of a
jadeite/nephrite ring was found.
Malachite and azurite
These minerals are different hydrocarbonates of copper, pro-
duced by weathering of primary copper ores, and thus are
very widespread. Malachite is green, sometimes blackish
green, and azurite is dark blue in colour. Both minerals are of
low hardness (3.5–4), and are easy to work. A rare occurrence
of these minerals is recorded from Lepenski Vir IIIa, in the
form of amulets (Srejović & Babović 1981: 92, 94).
Malachite also occurs at Padina.
Occurrence of raw materials at selected sites
Tables 1–4 summarize the occurrence of raw materials used
for manufacturing ground stone artefacts at four key sites.
Lepenski Vir
The analysis of the material from Lepenski Vir (Table 1) is
based on direct inspection of material in the National
 Dragana Antonović: The ground stone industry in the Serbian Iron Gates

Figure 2. Geological map of the Iron Gates area (after: Geological Map of Yugoslavia 1970; Pavelescu 1961; Kovačević
et al. 1997; Vasković & Matović 1997). A. Holocene, B. Miocene–Pliocene clastic, claystones, limestones, C. Upper and
Lower Cretaceous limestones, dolomites, clastics, D. Cretaceous andesites and pyroclastics, E. Upper and Lower Jurassic
limestones, dolomites, clastics, F. gabbro, diabase, G. serpentinite, H. Permo-Carboniferous grey sandstones with coal, I.
Permo-Triassic red sandstones, J. hornfels, K. Hercynian granites (massifs in eastern Serbia — 1. Brnjica, 2. Neresnica, 3.
Gornjane, 4. Plavna), L. Lower Palaeozoic phyllites, greenschists, amphibolites, clastics, M. Riphean-Cambrian +
Proterozoic schists, gneisses, amphibolites (Romania), N. Riphean-Cambrian greenschists, O. amphibolites, P. Riphean-
Cambrian gneisses, micaceous schists, amphibolites, Q. Proterozoic gneisses, R. corundum, S. quartzites.

Museum in Belgrade, as well as on the field documentation of the material (Padina A and Padina B) was not attempted,
and publications concerning Lepenski Vir. The sample of 119 since the final detailed analysis of excavations at Padina is in
specimens relates to the implements; decorative and progress, which should shed new light on the stratigraphy of
ceremonial–magic artefacts are not included in the total. the site.

Vlasac
The analyses performed and published by the investigators of
this site (Srejović & Letica 1978: 98–100) have been used to
compile the list in Table 2.
Padina
Only part of the stone material from the early Holocene strata
of Sectors I, II and III has been analysed (Table 3), so that the
picture from this site is incomplete. The main source of the
data was the field documentation for Padina; the actual raw
materials were examined in a small sample of artefacts from
all three sectors. A more precise chronological classification
Velesnica
In the analysis of the ground stone industry from Velesnica,
the material from the Early Neolithic layers of block A, ex-
cavated in 1982, was used (see Vasić, this volume). A sample
of 109 specimens was examined (Table 4).
Origin of the raw materials
The choice of rocks for the ground stone implements found in
the Iron Gates sites exhibits considerable variety. The most
common are sandstones, used equally for both art objects

23
 The Iron Gates in Prehistory

Table 1. Lepenski Vir.

Rock Implements Horizons
1. Granites Hammers L. Vir Ie, II
2. Diabase Chisels, axes L. Vir II, IIIa
3. Andesite Axes L. Vir IIIa
4. Basic magmatic rocks Axes, chisels L. Vir Ib–e, IIIa
5. Sandstones Axes, sceptres, weights, grinders/polishers, pounders L. Vir Ic–e, II, III
6. Limestones Axes, adzes, hammers, sceptres, beads, amulets L. Vir Ib–e, II, IIIa
7. Marlstone Sceptres L. Vir I, II
8. Mudstones without calcite Sceptres L. Vir II
9. Schists (slates) Sceptres L. Vir Ia–b, II
10. Phyllite Chisels L. Vir Ic
11. Phyllite–mica schist Axes L. Vir IIIa
12. Granite gneiss Sceptres L. Vir Ib–d
13. Eclogite Chisels L. Vir I
14. Marbles Amulets L. Vir Ib–c, II
15. Serpentinite Axes, adzes, chisels L. Vir IIIb
16. Fine-grained rocks Axes, adzes, chisels L. Vir Ib–e, IIIa
17. Nephrite Chisels L. Vir Ib–c
18. Malachite, azurite Amulets L. Vir IIIa

Table 2. Vlasac.
Rock Implements
1. Granite Grinding/polishing stones
2. Gabbro Grinding/polishing stones
3. Andesite Sling-balls
4. Tuff Tranchets
5. Undefined magmatic rocks Sceptres, hammerstones, working plates, sling balls
6. Sandstones Sceptres, pebble-axes, hammerstones
7. Limestones ?
8. Undefined sedimentary rocks Sceptres
9. Amphibolite Sceptres
10. Schists Sceptres

Table 3. Padina.
Rock Implements
1. Granite Anvils
2. Gabbro Axes
3. Silicified tuff Tranchets
4. Sandstone Grinding/polishing stones, anvils, pounders, mallets, sceptres, querns–altars
5. Schist Sceptres
6. Malachite Pendant?

24
 Dragana Antonović: The ground stone industry in the Serbian Iron Gates
Table 4. Velesnica.
Rock
1. Granite
2. Aplite
3. Diabase
4. Gabbro, gabbro-diabase
5. Ill-defined magmatic rocks
6. Sandstones
7. Silicified limestones
8. Silicified pelite
9. Granite gneiss
10. Gneiss
11. Amphibole–muscovite schist
12. Silicified serpentinite
13. Fine-grained contact-metamorphic and sedsedimentary rocks
14. Jadeite
(sculptures) and various kinds of implements. Next are crys-
talline schists, most commonly used for sceptres, but also for
other implements. This representation of raw materials cor-
responds to the distribution of these rocks in the wider Iron
Gates region (Fig. 2). In this region there are extensive
formations bearing sandstones and metamorphic rocks,
mostly crystalline schists, on both sides of Danube (Fig. 2: H,
F, L–Q). Magmatic rocks are also abundant. Granites are the
most widespread, found equally on both the Romanian and
Serbian sides of the river (in the Serbian part of the Iron Gates
there are four massifs: Brnjica, Neresnica, Gornjane and
Plavna — Fig. 2: K). Andesites, pyroclastic rocks (including
tuffs which are commonly used at Padina for the manufacture
of tranchets and axes), gabbro and diabase occur mainly in
the hilly hinterland of the Padina, Lepenski Vir and Vlasac
localities.
All of the rocks mentioned above, as well as limestones,
marlstones and marbles, have certainly been found in the
form of pebbles in the numerous rivers and streams that form
the tributaries of the Danube in this area. This cannot be said
for hornfels, nephrite/jadeite and malachite. Most probably,
the acquisition of these rocks and minerals was organized.
Only in the case of hornfelses, which occur in the western
part of the Iron Gates region, where the Danube makes con-
tact with the Brnjica massif, is there a possibility that they
could be collected from the surface as pebbles (Fig. 2: J).
Much more numerous deposits of hornfels are found in cent-
ral Serbia, and it may be assumed that some of this material
was imported.2 Jadeite/nephrite deposits have not been re-
corded in the Iron Gates or elsewhere in Serbia; the nearest
source is in Macedonia, to the south of Skopje (Solunska
Glava — Marić 1945: 225). They are also present on the
Kikládhes Islands (Cyclades) in Greece, as well as in the
Piemonte region of western Italy. Therefore, it may be as-
sumed that jade was imported from distant regions. On the
other hand, malachite almost certainly derives from eastern
Serbia, where extensive copper deposits occur.
% Implements
12.84 Hammerstones, querns, mallets
1.84 Fishing weights
1.84 Hammerstones
6.42 Axes, pounders, fishing weights
10.09 Pebble-axes, fishing weights, working plates
42.20 Grinding/polishing stones, whetstones, pounders, querns, hammerstones,
pebble-axes, mallets, fishing weights, working plates, spool
0.92 Axes
0.92 Axes
8.26 Hammerstones, pounders, grinding/polishing stones, mallets, fishing weights
0.92 Pounders
0.92 Axes
0.92 Chisels
8.26 Axes, adzes
0.92 Ring
Typological–functional analysis
There are two basic assemblages of ground stone artefacts
from the Iron Gates sites that date to the early part of the
Holocene. The first comprises implements of the so-called
‘local’ ground stone industry: mallet-sceptres, mallet-
weights, hammerstones, anvils, pounders, grinding/polishing
stones, whetstones, working plates, querns, mortars, sling-
balls, fishing weights, pebble-axes (blunt axes), amulets, and
various ornamental artefacts. To these can be added
tranchets. In the second group are artefacts of the ‘imported’
ground stone industry, which includes ground edge tools
(axes, adzes and chisels), used for woodworking, and ham-
mers made from broken and blunted ground edge tools. The
term ‘imported’ should not be taken literally. Judging by the
raw materials used for their manufacture, these artefacts are
not imported as such, although they are present in smaller
numbers. It is the technology that was imported — developed
elsewhere and adopted in its entirety by the Iron Gates popu-
lation. This is the case at Ajmana, where there are workshops
for the manufacture of ground edge axes (Stalio 1986: 29).
Artefacts of the ‘local’ ground stone industry are produced
from pebbles, naturally shaped for use, which are also trans-
formed into tools with minimal working. The pebbles are of
many different rocks, but sandstones are the most numerous.
These are very abundant in the region and thus are available
in the watercourses. Implements of the ‘imported’ variant are
made most commonly from both fine-grained sedimentary
and thermal contact metamorphic rocks, as well as from
rocks that closely resemble them in their physical-technical
features — gabbro, diabase, andesite, fine-grained basic
magmatic rocks, very fine-grained sandstones, limestones,
eclogite, serpentinite, and jadeite/nephrite. These sorts of
rocks can be worked by chipping because of their fine-
grained texture, and using this technique the implements
were manufactured from large flakes and from the cores
themselves.
25
 The Iron Gates in Prehistory

Figure 3. Mallet-sceptres: A. Lepenski Vir I–II (Srejović & Babović 1983: 187); B. Lepenski Vir I–II (Srejović & Babović
1983: 189); C. Padina, sector III, trench 5, block 1, F.I. 374/70; D. Padina, sector II, F.I. 291/70; E. Vlasac III (Srejović &
Letica 1978: pl. CXVIII); F. Vlasac II (Srejović & Letica 1978: pl. CXVIII).

26
 Dragana Antonović: The ground stone industry in the Serbian Iron Gates

Figure 4. Mallet-weights: A. Lepenski Vir II, block d/II, F.I. 805; B. Velesnica, block A/7, F.I. 417; C. Padina, sector I, profile
1, F.I. 17/68; D. Padina, sector I, profile 1, F.I. 8/68; E. Padina, sector III, profile III.

27
 The Iron Gates in Prehistory

Figure 5. Anvils: A. Padina, sector I, profile I, F.I. 31/68; F. Vlasac II (Srejović & Letica 1978: pl. CXIX); I. Vlasac II
(Srejović & Letica 1978: pl. CXIX). Hammerstones: B. Velesnica, block A/7, layer 26; C. Padina, F.I. 370; D. Velesnica,
block A/9, F.I. 454. Grindstones/polishers: E. Velesnica, block A/3; G. Vlasac I (Srejović & Letica 1978: pl. CXIX); H.
Velesnica, block A/3, F.I. 451.

28
 Dragana Antonović: The ground stone industry in the Serbian Iron Gates

Figure 6. Working plates: A. Velesnica, block A/1, layer 30. Querns: D. Velesnica, block A/1, layer 31; C. Padina, sector
III, profile III, F.I. 360/70. Mortars: B. Lepenski Vir Ib–c (Srejović & Babović 1983: 163). Fishing weights: E. Velesnica,
block A/2-2a, F.I. 469; F. Velesnica, block A/2-2a, F.I. 420; G. Velesnica, block A/3, F.I. 402. Sling-balls: H. Vlasac III
(Srejović & Letica 1978: pl. CXVIII); I–J. Padina, sector III, trench 6, block 1, house floor.

29
 The Iron Gates in Prehistory

Figure 7. Pebble-axes: A. Velesnica, block A/8, F.I. 416; B. Velesnica, block A/7, F.I. 455; C. Vlasac III (Srejović & Letica
1978: pl. CXIX); D. Vlasac II (Srejović & Letica 1978: pl. CXIX).

30
 Dragana Antonović: The ground stone industry in the Serbian Iron Gates

Figure 8. Ornamental and ceremonial artefacts: A. Labret of limestone — Lepenski Vir IIIb (Srejović & Babović 1981: 94);
B. Amulet of marble — Lepenski Vir Ib–c (Srejović & Babović 1983: 190); C. Ring of jadeite — Velesnica, block A/7, F.I.
498; D. Winder of fine-grained sandstone — Velesnica, block A/8, F.I. 422. Tranchets: E. Kula I (Sladić 1986: Fig. 6/25);
F. Padina, sector II, trench 2, block 2a, F.I. 96/70; G. Padina, F.I. 165; H. Padina, sector II, trench 2, block 2a, F.I. 145/70; I.
Vlasac I (Srejović & Letica 1978: pl. CXVII); J. Lepenski Vir III (Kozłowski & Kozłowski 1984: 289). Cutting-edge tools:
K. Padina, sector I, profile I, F.I. 27/68; L. Velesnica, block A/8, F.I. 418; M. Padina, sector I, profile I, F.I. 7/68; N. Velesnica,
block A/3, F.I. 393; O. Padina, sector III, trench 5, block 1, F.I. 281/70; P. Velesnica, block A/7, F.I. 401.

31
 The Iron Gates in Prehistory

Figure 9. Cutting-edge tools: A. Lepenski Vir IIIa, no. 227; 3 B. Lepenski Vir IIIa, no. 230; C. Lepenski Vir III, no. 249; D.
Lepenski Vir Ia, house 20, no. 254; E. Lepenski Vir Id–e, house 26, no. 262; F. Lepenski Vir Ie, house 8, no. 306; G. Lepenski
Vir Ie, house 27, no. 244; H. Lepenski Vir Ib-c, house 7, no. 242, I. Lepenski Vir IIIb, depot 2, no. 250; J. Lepenski Vir IIIb,
depot 2, no. 251; K. Lepenski Vir Ib–c, house 35, no. 253; L. Lepenski Vir Ie, house 21, No. 256.

32
 Dragana Antonović: The ground stone industry in the Serbian Iron Gates
The ‘local’ ground stone industry
Mallet-sceptres (Fig. 3)
These are massive stone artefacts, up to 50 cm long, made
from elongated pebbles of elliptical or circular cross-section.
Final shaping was done by grinding, and some specimens
have ornamentation made by shallow incision. It is likely that
they were used for killing game and large fish, although there
is an opinion that they are sceptres used in sympathetic magic
and fertility rituals (Srejović & Babović 1983: 183). On the
Serbian bank of the Danube, mallet-sceptres are present at
Lepenski Vir (Srejović & Babović 1983: 185–189),
Hajdučka Vodenica (Radovanović 1996: 323), Vlasac
(Srejović & Letica 1978: 99) and Padina. Usually they are
made of crystalline schists, and less commonly from fine-
grained sandstones, marlstones or limestones.
Mallet-weights (Fig. 4) 4
As the name implies, these are massive pebbles used to shat-
ter hard materials such as concreted soil or stone. The natural
shape is altered only in the middle where a groove was made
by hammering. They are recorded at Lepenski Vir and dated
to LV Ie–II (Srejović 1972: 133). They were also found at
Velesnica and Ajmana (Stalio 1986: 29), and in all the early
Holocene strata at Padina and Hajdučka Vodenica
(Radovanović 1996: 323). They are usually interpreted as
fishing weights or as a kind of primitive mallet. The latter is
more plausible given their size, weight and the used end.
They are mainly made from large sandstone pebbles, and oc-
casionally from certain other rocks (granite, granite gneiss).
Hammerstones, anvils, pounders (Fig. 5: A–D, F, I)
These terms refer to artefacts with the same pebble shape and
partially reworked surfaces, but which had different uses.
The hammerstones are ball-shaped, egg-shaped or disc-
shaped, suitable for holding in the hand, and were used for
stone working, as can be seen from scars or minor recesses
(one or more) on their working surfaces. When working stone
and bone implements, the tools could have been held in the
hand (hammerstones; Fig. 5: B–D) or used as a support (anvil
— Fig. 5: F, I). Some of these pebbles were used for seed
grinding (pounders). They have been found in all Mesolithic
horizons at Kula (Kula I–III; Sladić 1986: 432–433), and in
Vlasac I–III (Srejović & Letica 1978: 99–100), Hajdučka
Vodenica (Radovanović 1996: 323), Lepenski Vir (Srejović
1972: 133), Padina and Velesnica. They are made from sand-
stones, magmatic rocks and tuffs.
Grindstones/polishers and whetstones (Fig. 5: E, G, H)
The grindstones/polishers are implements used for grinding
and polishing objects of hard materials (stone, bone or
antler). Therefore, they are manufactured only of rocks with
abrasive characteristics, such as sandstones, and very occa-
sionally magmatic rocks with a high quartz content.
Grinders/polishers for working awls and needles, with one or
more grooves (Fig. 5: G), were found in Lepenski Vir I
(Srejović 1972: 133), Vlasac I–III (Srejović & Letica 1978:
99) and Padina. They are manufactured from sandstones and
magmatic rocks. Immobile grinders for working artefacts of
hard materials (Fig 5: E), were found in Lepenski Vir I, in the
form of large trapezoidal tables made from fine-grained red-
33
dish sandstone, embedded in the floors of the houses
(Srejović 1972: 133), as well as at Ajmana (Stalio 1986: fig.
5) and Velesnica. Whetstones are usually linked with the
sharpening of metal objects, but these implements do occur,
albeit rarely, in the Iron Gates area long before the appear-
ance of metal. They were found at Velesnica, where they
were used for the fine sharpening of bone awls and needles
(Fig. 5: H).
Working plates (Fig. 6: A)
In Vlasac II and III, fragmented plates made of magmatic
rocks were found. These were worked by hammering and
grinding, with numerous circular recesses on the sides. How
these plates were used is not clear; it is possible that they
were utilized as working tables or as artefacts of ceremonial
character (‘altars’; Srejović & Letica 1978: 100). At
Velesnica the plates are made from sandstone and basic
plutonic rocks. They are of irregular shape and have a ground
working surface with circular recesses in it, and are inter-
preted as operating tables for working stone tools (Fig. 6: A).
Querns and mortars (Fig. 6: B–D)
The querns are massive stone objects either with a flat work-
ing surface or with shallow hollows, used for grinding seeds.
Mortars are, like querns, massive stone implements, but with
deeper concave hollows resembling a recipient (vessel) on
one side (see Radovanović 1996: 278). Following
Radovanović (1996: 277–278) some of the ‘altars’ from
Lepenski Vir and Velesnica (Srejović & Babović 1983:
158–172; Vasić 1986, 277, fig. 7) may be interpreted as
querns and mortars. The querns/altars from Padina may also
be included in this category (Fig. 6: B–C). Simple querns
were found in Padina, Sector IV–Babice (Radovanović 1996:
341) and at Velesnica (Fig. 6: D). The querns are made of
sandstone, seldom of granite and other magmatic rocks.
Sling-balls (Fig. 6: H–J)
These artefacts are rare, for the reason that they were hunting
weapons. Sling-balls are not genuine ground stone artefacts,
because it is not clear whether they are naturally ground
pebbles or shaped by hand. They were recorded in the
Mesolithic strata of Vlasac (Srejović & Letica 1978: 100) and
at Padina. Relatively small pebbles (2.5–5 cm diameter) of
magmatic rocks were used.
Fishing weights (Fig. 6: E–G)
These are small tabular pebbles with minor shallow notches
in the lateral edges at the narrower end, made by hammering
and knapping, which were probably used for fixing a line
around the implement. There is an opinion that these are am-
ulets (Vasić 1986: 269). Such weights are the commonest
ground stone artefacts at Velesnica. On the basis of their
abundance, we are inclined to think that they are fishing
weights (for both fishnets and fishhooks). They were also re-
corded at Knjepište and Ušće Kameničkog Potoka
(Stanković 1988: 97).
Pebble-axes (blunt axes; Fig. 7)
These are massive implements of elongated form made of
magmatic rocks, one end of which has a worked cutting edge
 The Iron Gates in Prehistory

Table 5. Lepenski Vir.

%
Artefacts LV I LV II LV III ?5 Total
1. Ground edge tools 17.80 6.78 11.86 30.51 66.95
2. Hammers 1.70 0.85 0 0 2.54
3. Undefined fragments 2.54 0.85 0 3.39 6.78
4. Sceptres 16.95 2.54 0.85 0 20.34
5. Mallets 0.85 0.85 0 1.70 3.39
6. Pebble-axes +6 +
7. Grinders/polishers + +
8. Necklaces, pendants + + + +

Table 6. Vlasac.
%
Artefacts Vlasac I Vlasac II Vlasac III Total
1. Sceptres 6.11 12.21 19.85 38.17
2. Pebble-axes 0 1.53 0.76 2.29
3. Grinders/polishers 0.76 1.53 1.53 3.82
4. Anvils, hammerstones 8.40 25.19 17.56 51.15
5. Sling-balls 0 1.53 3.05 4.58

Table 7. Padina.
%
Artefacts Sector I Sector II Sector III Total
1. Ground edge tools 10.66 1.64 13.93 26.23
2. Mallets 11.48 3.28 13.93 28.69
3. Anvils, pounders, hammerstones 7.38 17.21 5.74 30.33
4. Querns, altars 1.64 0 1.64 3.28
5. Grinders/polishers 0 0.82 4.92 5.74
6. Sceptres 0 0 3.28 3.28
7. Sling-balls 0 0 2.46 2.46

Table 8. Velesnica.

Artefacts %
1. Ground edge tools 8.26
2. Hammerstones, pounders 22.02
3. Grinders/polishers, whetstones 7.34
4. Querns 4.59
5. Mallets 2.75
6. Pebble-axes 4.59
7. Fishing weights 41.28
8. Working plates 0.92
9. Ring 0.92
10. Spool 1.83
11. Undefined fragments 5.51

34
 Dragana Antonović: The ground stone industry in the Serbian Iron Gates
or has been rounded. Such tools have been found at Lepenski
Vir on the floors of the houses in settlements Id-e (houses 1,
21, 27, 32 and 37; Srejović 1972: 133) and in Vlasac II–III
(Srejović & Letica 1978: 99), Hajdučka Vodenica Ib
(Radovanović 1996: 323) and in the earliest strata of
Velesnica. The most developed shape of the pebble-axes oc-
curs at Velesnica (Fig. 7: A–B). These are ellipsoidal pebbles
that, besides having one end transformed into a cutting edge
or rounded, are characterized by hammered regular grooves
at the other end, which was used for fixing the tool on a
handle. Such implements could be used as axes when one end
was worked into a blade, or as hammers when the end had
been rounded. They are most commonly made from
sandstone.
Ornamental artefacts (Fig. 8: A–D)
In the early Holocene Iron Gates sites small numbers of stone
ornamental artefacts were found. They are made of both or-
namental stones and gemstones, such as marble, crystalline
limestone, ordinary limestones of attractive colours, as well
as of malachite, azurite and jade. The ornamental objects
were mostly found at Lepenski Vir, in all horizons of the site.
They comprise pendants of various shapes, made of marble
and crystalline limestones, and necklaces with beads manu-
factured of pink and white limestone, all of them being found
in Lepenski Vir I–II (Srejović & Babović 1983: 190–197).
From the Neolithic horizons of Lepenski Vir (IIIa–b) there
are amulets in form of a labret, made of limestone, as well as
pendants of malachite and azurite (Srejović & Babović 1981:
92, 94). A tabular piece of malachite, found at Padina, could
be interpreted as a half-finished pendant. At Velesnica was
found a jadeite/nephrite ring fragment. Here also can be
mentioned some delicately worked flat winders, from
Velesnica and Knjepište (Stanković 1986b: 448), which may
represent some kind of ornament, but not an object for use in
everyday life.
Tranchets (Fig. 8: F–J)
Although these artefacts can be regarded as products of the
chipped stone industry, and have been studied and published
as such (Radovanović 1981: 50; Kozłowski & Kozłowski
1984: 274; Srejović & Letica 1978: 98) they are mentioned
here, because they are identical in shape to the unfinished
ground edge tools from Neolithic localities in other parts of
Serbia. The tranchets are made only from tuffs that, in their
physical-technical features, resemble hornfelses (which is
the basic material for the manufacture of ground edge tools in
other parts of Serbia). They were used for the extraction of
roots and other ‘agricultural’ activities, and have been found
in Vlasac I, Padina A, Lepenski Vir III and Kula I (Sladić
1986: 432).
The ‘imported’ ground stone industry
Ground edge implements
These comprise axes, adzes and chisels (Fig. 8: K–P; Fig. 9:
A–E, G–L). In terms of shape, axes are implements with a
symmetrical profile, the cutting edge occurring in the plane
of symmetry (Fig. 8: L, P; 9: A–C, G). These are tools whose
basic traits indicate that they were used for cutting trees.
Adzes differ from axes by the asymmetry of the blade — i.e.
35
the blade is not in the plane of symmetry of the tool (Fig. 8:
N; Fig. 9: H, I). They were used only for woodworking (hol-
lowing out and splitting) as visible from striations. Chisels
are put into a special category on the basis of their smaller
dimensions (Fig. 9: D, E, J–L). They were used in the same
manner as adzes, but due to their small size, were also used
for finer woodworking. Ground edge tools occur in Lepenski
Vir I–III, Velesnica, Padina, Ajmana (Stalio 1986: 29), as
well as in the Neolithic horizons of Vlasac (Vlasac IV;
Srejović & Letica 1978: 135), Ušće Kameničkog Potoka
(Stanković 1986a: 468) and Knjepište (Stanković 1986b:
448). They are manufactured from very fine-grained, com-
pact and tough rocks, such as hornfelses, silicified pelites,
limestones, fine-grained sandstones, diabases, andesites,
gabbros, and other poorly defined fine-grained basic mag-
matic rocks, as well as phyllites, phyllite-mica schists, chlor-
ite schists, eclogites and serpentinites.
Hammers (Fig. 9: F)
In Lepenski Vir I and II damaged and blunted ground edge
tools were subsequently used as hammers. So far, these tools
have been recorded only at this site, where they are made of
limestone and granite.
* * * * * *
Tables 5–8 summarize the occurrence of the different types
of ground stone artefacts at the sites of Lepenski Vir, Vlasac,
Padina and Velesnica.
Conclusions
The Iron Gates early Holocene ground stone industry is
composed of two elements: local and ‘imported’. The local
variant developed quite independently, against a background
of sedentary settlement, satisfying the essential needs of the
inhabitants of this region. This led to the production of
mallet-sceptres, mallet-weights with grooves, pebble-axes,
grindstones/polishers and hammerstones. Judging by their
shape and considerable dimensions, these implements were
probably used in game hunting and for killing large fish. On
the other hand, because of the blunt cutting edges of the
pebble axes and the massiveness of the tools themselves, it
may be assumed that they were not used in woodworking or
agriculture. In parallel with this autochthonous ground stone
industry, there occur in the Iron Gates early Holocene sites
tools typical of the Starčevo and Vinča cultures — the ‘im-
ported’ variant. Thus, on the floors of the houses of Lepenski
Vir I were found axes and adzes of Starčevo–Vinča type,
along with mallet-sceptres typical of the Lepenski Vir cul-
ture. These two variants of the ground stone industry are of
quite different character. In essence, the local variant is based
on the pebble, its natural shape being only finished. On the
other hand, the ground edge implements of the ‘imported’
variant are produced by working large flakes of fine-grained
rocks with a characteristic grey and grey-greenish colour,
thus having a distinctive appearance reminiscent of the
Starčevo–Vinča ground stone industry. It is this raw material
that makes the ground stone industry of the entire Serbian
 The Iron Gates in Prehistory

III
I

II

?
Vir

Vir

Vir
i
nsk

i
nsk

i
nsk
pe

pe

pe
Le

Le

Le

Figure 10. The frequencies of particular tool types in the major horizons of Lepenski Vir. III
sac

I
II

I
III

II
la

ca

Vir
Ku

Vir
cto
sni

Vir
r
cto
Vla

cto

i
e

nsk

i
e

Vel

a–S

nsk

i
e

nsk
a–S

a–S

pe

pe
din

pe
din

Le
din

Le

Le
Pa
Pa

Pa

Figure 11. The frequencies of local and ‘imported’ ground stone industries at selected Iron Gates sites.

36
 Dragana Antonović: The ground stone industry in the Serbian Iron Gates
Neolithic definitely recognizable. From these observations,
the occurrence of ground edge implements, made of grey-
greenish rocks (whether hornfelses — the main Neolithic raw
material in Serbia — or local rocks with the same physical-
technical properties) indicates the import of this technology
and, occasionally, the import of the artefacts themselves into
the Iron Gates region. The common appearance of two vari-
ants of the ground stone industry, different in character, is
perhaps further evidence of the parallelism of the Iron Gates
Late Mesolithic with the developed Neolithic of the neigh-
bouring areas, which is otherwise indicated by absolute dates
(Radovanović 1996: 291–292). The implements of the ‘im-
ported’ ground stone industry (ground edge tools — axes,
adzes and chisels) were used only for woodworking, and oc-
cur in greater numbers in settlements showing more intensive
construction activity. Thus, they are well represented at
Lepenski Vir (where the largest number of houses is recor-
ded) and Padina–sector III; although their complete absence
from Vlasac is surprising.
Both the local and ‘imported’ variants occur in parallel in
the Iron Gates sites. In some sites artefacts of the local in-
dustry are more abundant, while in others there are more
‘imported’ artefacts. In yet other settlements the ‘imported’
ground stone variant is absent (Kula I–III, Vlasac I–III). On
the basis of the presence of both local and ‘imported’ ground
stone variants, the following sequence of the Iron Gates early
Holocene settlements can be made: in Vlasac I–III and Kula
I–III only tools of local variant occur; in Velesnica and
Padina the ground stone implements are largely restricted to
the local variant; whereas in Lepenski Vir the local variant is
of greater importance only in the deepest levels, but it disap-
pears in layers II and III (Figs 10 & 11). This is not a chro-
nological sequence, but merely an example illustrating the
extent to which some Iron Gates early Holocene settlements
accepted Neolithic products from outside. In this respect,
Lepenski Vir appears to a higher degree as a neolithized set-
tlement based on hunting and gathering supported by intens-
ive fishing. In contrast, Velesnica is far behind Lepenski Vir
and, according to the ground stone material, was a highly
developed fisher’s settlement.
Here, it should be mentioned that the local variant contin-
ued its development during the Late Neolithic of the Iron
Gates. A considerable number of ‘local variant’ tools
(mallet-weights, sling-balls, hammerstones, pounders,
querns, grindstones) were found, along with ground edge
implements of Starčevo–Vinča type in the Vinča settlement
of Zbradila near Korbovo (Babović 1984: 95–96, 1986: 96).
When attempting to establish analogies and parallels with
the ground stone industries of other regions, it has to be said
that there is a remarkable similarity between the Iron Gates
local variant and the ground stone products from very distant
areas, with which the Iron Gates inhabitants were not in dir-
ect contact. The Mesolithic and Early Neolithic stone imple-
ments of the northwest regions of the European part of Russia
show a close resemblance, in both form and raw materials, to
the material from Lepenski Vir, Padina and Velesnica (Gurina
1961: 190–201, 1973: 46–49). There is a basic similarity also
with the Late Mesolithic stone industries of the Dnestr–Don
area (Levenok 1966: 93; Telegin 1982: 110), and the coastal
1
37
areas of eastern Germany (Gramsch 1973: 19–30) and
Pomerania in Poland (Galinski 1992: 103, 156–157, 162). All
of these cultures developed near large bodies of fresh- or salt
water and fishing was an important occupation, although
none of these cultures in terms of their character or territory
is close to the early Holocene Iron Gates. Thus it is not a
matter of cultural influence, but of a similar pattern of devel-
opment of sedentary living near to a body of water, which
provided a reliable source of food all year round. In all of the
regions mentioned the development of stone implements was
initiated by the need to exploit large fish. This shows once
more that culture is a response to the natural conditions to
which humans were trying to adapt. Therefore, it may be
stated that in the Iron Gates too there was an autochthonous
development of the ground stone industry (only in the form
of the local variant), and this development was directed by
the natural surroundings and living conditions of the Iron
Gates inhabitants. The ‘local variant’ implements were adap-
ted for the successful fishing and hunting economy, suppor-
ted by various kinds of percussive [hammering] implements,
whereas the ‘imported variant’ artefacts can be seen as a later
addition to the quality of life of the Iron Gates population.
Notes
1. The term working plate refers to a large stone with a flat or
slightly concave working surface; usually they are thin tabular
stones, but not exclusively.
2. More detailed and precise analyses of the raw materials used for
the manufacture of the ground stone implements would give a
clearer picture of the possible circulation of raw materials in
these districts.
3. Inventory number in the National Museum, Belgrade.
4. These artefacts are most commonly called weights.
5. Lacking contextual and stratigraphic details. For the majority of
ground edge tools, the field documentation does not give details
of the find circumstances or stratigraphic position, but they were
found along with pottery. This is not an automatic indication that
they are restricted to the Lepenski Vir III level, because a large
number of tools (of both local and imported variants) derive from
the Lepenski Vir I and II levels, and the field documentation re-
cords the presence of advanced Starčevo pottery in those levels
(Starčevo IIa–b in Garašanin’s classification). Such is the case
with the ground edge tools that are registered in the Field
Inventory (F.I.) under numbers 122, 123, 124 and 129, found on
the floor of house 5 (LV I), the axe F.I. no. 369 from house 1 (LV
Id–e), axe F.I. no. 459, found in house 30 (LV Ic), chisel F.I. no.
476 from the floor of house 9 (LV Id), axe F.I. no. 667 from the
floor of house 32 (LV Id–e), chisel F.I. no. 738 from house 47 (LV
Ia). Pottery is found on the floor of houses 1, 32 and 47, but was
not recorded in houses 5, 9 and 30 (Srejović 1972: 133).
6. For some of the artefacts from Lepenski Vir it is known only that
they existed at this site, but precise data on their frequency are
not available.
Acknowledgments
Thanks are due to colleagues who made it possible for me to study
the field documentation from the Iron Gates sites, which has been an
important source of data for this paper — Borislav Jovanović and
Rastko Vasić of the Archaeological Institute in Belgrade, as well as
Ljubinka Babović of the National Museum in Belgrade. Special
thanks are due to the late Professor Dragoslav Srejović, who gave
me access to the important material from Lepenski Vir.
 The Iron Gates in Prehistory
References
Antonović, D. 1992: Predmeti od glačanog kamena iz Vinče.
Belgrade: Filozofski fakultet — Centar za arheološka
istraživanja.
— 1998: Nastanak i razvoj industrije glačanog kamena u neolitu
Srbije. Unpublished PhD dissertation, Department of
Archaeology, Faculty of Philosophy, University of Belgrade.
Babović, Lj. 1984: Zbradila, Korbovo: izveštaj o arheološkim is-
traživanjima u 1980. godini. Đerdapske sveske 2: 93–100.
— 1986: Zbradila — Korbovo: compte-rendu des fouilles en 1981.
Đerdapske sveske 3: 95–115.
Galinski, T. 1992: Mezolit Pomorza. Szczecin: Muzeum narodowe.
Geological Map of Yugoslavia 1970: 1:500,000. Belgrade: Federal
Geological Institute.
Gramsch, B. 1973: Das Mesolithikum im Flachland zwischen Elbe
und Oder, vol. 1. Veröffentlichungen des Museums für Ur- und
Frühgeschichte Potsdam 7. Berlin: Deutscher Verlag der
Wissenschaften.
Gurina, N.N. 1961: Drevnjaja istorija severo-zapada evropeiskoi
časti SSSR. Materialy i issledovanija po arheologii 87.
Moscow–Leningrad: Akademija nauk SSSR.
— 1973: Drevnie pamjatniki Kol’skogo poluostrova. In Gurina,
N.N. (ed.) Etnokul’turnije obštnosti lesnoj i lesostepnoj zony ev-
ropejskoj časti SSSR v epohu neolita. Materialy i issledovanija
po arheologii 172. Leningrad: Akademija nauk SSSR, 45–53.
Kovačević, J., Jovanović, M. & Radošević, B. 1997: Nonmetallic
mineral raw materials in Djerdap area. In Grubić, A. & Berza, T.
(eds) Geology of Djerdap Area. International Symposium.
Belgrade: Geoinstitute / Bucharest: Geological Institute of
Romania, 169–174.
Kozłowski, J.K. & Kozłowski, S.K. 1984: Chipped stone industries
from Lepenski Vir, Yugoslavia. Preistoria Alpina 19: 259–293.
Levenok, V.P. 1966: Mezolit srednorusskogo Dneprovsko-
Donskogo meždureč’ja i ego rol’ v složenii metnoj neolitičeskoj
kul’turi. In Gurina, N.N. (ed.) U istokov drevnih kul’turi (epoha
mezolita). Materialy i issledovanija po arheologii 126.
Moscow–Leningrad: Akademija nauk SSSR, 88–98.
Marić, L. 1945: Sistematska petrografija. Zagreb: Nakladni zavod
Hrvatske.
Pavelescu, L. 1963: Contribution à l’étude du soubassement cristal-
lin et l’aire de distribution des différentes formations cristallo-
38
phylliennes dans la partie centrale et orientale de l’autochtone
danubien. In Association géologique Carpato–Balkanique, Ve
Congrès, 4–19 septembre 1961, Bucharest. Vol. II:
Communications scientifiques. I-ère section: minéralogie-
pétrographie. Bucharest: Association géologique
Carpato–Balkanique, 167–179.
Radovanović, I. 1981. Ranoholocenska kremena industrija sa
lokaliteta Padina u Đerdapu. Belgrade: Arheološki institut.
— 1996: The Iron Gates Mesolithic. Ann Arbor: International
Monographs in Prehistory.
Semenov, S.A. 1976: Prehistoric Technology. London: Moonraker
Press.
Sladić, M. 1986: Kula près de Mihajlovac — un site préhistorique.
Đerdapske sveske 3: 432–442.
Srejović, D. 1972: Europe’s First Monumental Sculpture: New
Discoveries at Lepenski Vir. London: Thames & Hudson.
Srejović, D. & Babović, Lj. 1981: Lepenski Vir: Menschenbilder
einer frühen europäischen Kultur. Mainz: von Zabern.
Srejović, D. & Babović, Lj. 1983: Umetnost Lepenskog Vira.
Belgrade: Jugoslavija.
Srejović, D. & Letica, Z. 1978: Vlasac: A Mesolithic Settlement in
the Iron Gates. Vol. 1, Archaeology. Belgrade: Serbian Academy
of Arts and Sciences.
Stalio, B. 1986: Le site préhistorique Ajmana à Mala Vrbica. Đer-
dapske sveske 3: 27–50.
Stanković, S. 1986a: Emboucher du ruissen Kamenički potok site
du groupe de Starčevo: compte-rendu des fouilles de 1981. Đer-
dapske sveske 3: 467–471.
— 1986b. Localité Knjepište — une station du groupe de Starčevo;
fouilles de 1982–1983. Đerdapske sveske 3: 447–452.
— 1988: Šljivik — Stragari, Trstenik. In Srejović, D. (ed.) The
Neolithic of Serbia. Belgrade: Faculty of Philosophy — Centre
for Archaeological Research, 95–97.
Telegin, D.Ja. 1982: Mezolitični pam’jatki Ukraini (IX–VI tisjačol-
ittja do n. e.). Kiev: Naukova dumka.
Vasić, R. 1986: Compte-rendu des fouilles du site préhistorique à
Velesnica. Đerdapske sveske 3: 264–277.
Vasković, N. & Matović, V. 1997: The Hercynian granitoides of
Djerdap. In Grubić, A. & Berza, T. (eds) Geology of Djerdap
Area. International Symposium, 23–26.IX.1997, Donji
Milanovac–Orsova. Belgrade: Geoinstitute / Bucharest:
Geological Institute of Romania, 129–141.
 Sturgeon fishing in the middle and lower Danube region

László Bartosiewicz, Clive Bonsall & Vasile Şişu

Abstract: Migrating sturgeons were the largest fish in the middle and lower Danube region. Most of these species, however, have been
brought to the brink of extinction by habitat loss and overfishing. This review is a synthesis of sporadic archaeological evidence, zoologic-
al and environmental data as well as ethnohistorical information in two regions: the Iron Gates at the southeast edge of the Carpathian
Basin and the Hungarian section of the Danube within the basin. In addition to ichthyological and taphonomic questions, fishing tech-
niques as well as the varying perceptions of these large fish are summarized in an attempt to draft a multidisciplinary interpretive frame-
work for the archaeological evaluation of future finds.
Key words: sturgeon, Acipenseridae, fishing, weirs, Danube, Iron Gates, seasonality

Introduction Although the disciplinary boundaries between these areas of

research often overlap, this list is intended to provide an in-
Sturgeon has become one of the most elusive animals in terpretive framework for archaeologists interested in any
modern day fishing. Its osseous remains, decimated by ta- period, touching upon the complex interactions between
phonomic loss in archaeological deposits, yield relatively nature and society as reflected in sturgeon fishing.
scarce evidence of their dietary role in past times. Their Our research hypothesis is that it should be possible to
dwindling stocks, brought to the brink of extinction in the outline geographical locations as well as seasons when the
20th century, are only a pale shadow of their economic im- probability of catching various species of sturgeon was in-
portance until the recent past. creased. These parameters should correspond to and comple-
Over-exploitation, habitat loss and pollution have severely ment the scanty archaeological evidence for these important
hit all 27 species in the Acipenserid family worldwide. fish.
Sturgeons, the largest fish in the Danube, were relatively This overview of zooarchaeological, documentary and
common until dams were built in the Iron Gates section ethnohistorical evidence of sturgeon fishing in the middle
where the river forms the border between Romania and Danube region represented by two contrasting sections of the
Serbia. The first dam (Iron Gates I) which became opera- river, in Hungary and the Iron Gates (Fig. 1), is also aimed at
tional in 1971, was built where the Danube leaves the Iron interpreting ichthyoarchaeological data within a broader,
Gates gorge, and effectively marks the divide between the culture–historical context.
middle and lower Danube. The dam was designed in part to
improve navigation through the 130 km long gorge section,
where the Danube (prior to impounding) was characterized
by strong currents, rapids and exposed rocks that were haz-
ardous to shipping. The second dam (Iron Gates II), opera-
tional in 1984, is located 80 km downriver, 875 km from the
river’s mouth. The dams effectively cut off the migration
route of endangered beluga sturgeon and other anadromous
fish in all major sections of the Danube, as reviewed at the
1994 International Conference on Sturgeon Biodiversity and
Conservation in New York (Bacalbaşa-Dobrovici 1997;
Hensel & Holčík 1997).
Against this background, reconstruction of ancient stur-
geon fishing becomes a truly multidisciplinary task:
1. Ichthyoarchaeological finds offer evidence of which spe-
cies were targeted.
2. The palaeohydrological reconstruction of riverine habitats
helps in identifying locations where sturgeon fishing may
have taken place.
3. Familiarity with fish behaviour points to seasons when
migrating sturgeons were most easily caught.
4. Historical accounts describe techniques by which stur-
geons were caught.
5. Ethnohistorical records reveal attitudes toward these great Figure 1. The Danube Valley in Europe and areas dis-
fish as food as well as symbols. cussed in this study (see also Figures 7–8).

39
 The Iron Gates in Prehistory

mains represented in the Schela Cladovei assemblage, dorsal

Figure 2. Lengths of Danubian sturgeon in the historical
record (Hungary) and by prehistoric estimates (Iron Gates).
Ichthyoarchaeological data
This study is built around the osteological evidence for stur-
geon from the Danube. Sturgeon remains recovered from ar-
chaeological sites are discussed in terms of taxonomic iden-
tifiability and taphonomic bias, as well as possibilities of size
reconstruction.
The Romanian–British excavations (1992–1996) at the
Late Mesolithic and Early Neolithic settlement of Schela
Cladovei (Romania) 7 km downriver from the Iron Gates I
dam brought to light 139 sturgeon bones, mostly identifiable
only to family level (Table 1). These remains have been re-
covered by water sieving from several features, located on
one of the river’s terraces, near the present day riverbank.
Although these numbers of sturgeon bones may seem mod-
est, given the sample size of hand-collected bones at this site
(Bartosiewicz et al. 2001: 16, table 1) and others in the region
they may be considered relatively high. Of the skeletal re-
and lateral scutes as well as pectoral fin rays could, to some
extent, be identified to species. Osseous elements from small
but mature individuals could be assigned to sterlet on the
basis of size.
The first (i.e. most cranially located) pectoral fin ray is
well developed and has been used in osteometric analyses of
common sturgeon by Desse-Berset (1994: 84). The me-
diolateral width of this bone in beluga sturgeon
(Bartosiewicz & Takács 1997: 12, fig. 8/1) has also been used
in size estimations in the present paper. Greatest lengths of
the Schela Cladovei sturgeons were estimated as shown in
Table 2.
When plotted together with modern, historical data (Khin
1957), the distribution of these estimated lengths largely cor-
responds to that of the largest specimens from historical
periods (Fig. 2). The latter display a slight positive skew, and
the distribution of the prehistoric specimens is within the
same range, suggesting that randomly caught prehistoric
sturgeons were as large at Schela Cladovei as the largest
modern specimens in Hungary. While, owing to the small
number of cases no significant difference can be observed
between the two groups, the Schela Cladovei sturgeons were
indubitably large. Sporadic records of the amount of meat
some record animals represented (Table 3) indicate that ap-
proximately two-thirds of the live weight estimated for large
prehistoric sturgeons represented edible protein, and other
lines of evidence show that fish were the major source of an-
imal protein in the diet of the Mesolithic inhabitants of
Schela Cladovei (Bonsall et al. 1997, 2000).
Ichthyorachaeological research has shown early signs of
overfishing common sturgeon (Acipenser sturio Linné 1758)
in the southern Baltic region (Benecke 1986: 16, fig. 1) at
archaeological sites in Gdańsk, Poland (10th to 13th century
AD) and Ralswiek, Germany (8th to 12th century AD).
Osteometric data show that these sturgeons were at least lar-
ger than the 1.4 m long modern reference specimen available
to that author (Benecke 1986: 17, fig. 2).
Recent sturgeon sizes in the Danube are also worth con-
sidering. Aside from a random element (fishermens’ luck),
the frequency and actual size of the largest individuals landed

Table 1. Sturgeon bones from Schela Cladovei.

Mesolithic Early Neolithic

Acipenserid Russian Sterlet Acipenserid Sterlet

Sturgeon
frontale 3
parashenoideum 5
praemamillo-maxillare 1
dentale 1
praeoperculare 1 1
operculare 1
cleithrum 10 1 2 1
dorsal scute 1 1
lateral scute 5 2 3 3
pectoral fin ray 7 2 2
dorsal/anal fin ray 2 1
flat bone 75 2 6
Total number 108 2 10 14 5
Find weight (g) 287.5 2.4 8.2 63.6 2.2

40
 László Bartosiewicz et al.: Sturgeon fishing in the middle and lower Danube region

Table 2. Length and weight estimates for prehistoric sturgeons

from Schela Cladovei.

Period GW fin ray Estimated Estimated live

(mm) length (m) weight (kg)
Mesolithic 24.4 1.44 28.2
Mesolithic 36.4 2.15 72.8
Mesolithic 43.3 2.56 110.1
Mesolithic 50.0 2.95 154.0
Neolithic 37.5 2.21 77.7
Neolithic 42.0 2.48 102.1
Neolithic 44.1 2.60 114.2

Table 3. The meat output of some record sturgeons from Hungary.

Year Location Live weight Carcass Live weight

(kg) weight (kg) (kg)
1922 Gemenc 90 70 70.0
1957 Paks 138 100 72.0
1987 Paks 181 100* 55.2*

* Pure meat with spine but without skin

Figure 3. Diachronic temporal decline in gross sturgeon catch in Romania and in the size of record sturgeon in Hungary.

depends on the reproductive capacity and growth character- Sturgeon taxonomy

istics of fish stocks (Miranda et al. 1987: 219). Since 1800, an
estimated 1.84 kg average annual decrease in record sturgeon
body weights was found to be statistically significant
(Bartosiewicz & Takács 1997: 9). This decrease is paralleled
by a decline of the overall weight of sturgeons landed in
Romania (Fig. 3).
The unusually large, 181 kg specimen caught at Paks
(Hungary) in 1987 (Fig. 4), nearly two decades after the
closure of the Iron Gates 1 dam (Pintér 1989: 24), may have
attained this respectable size after having been trapped up-
stream, behind the dam.
In spite of the difficulties of identification at a species level in
this family of fish, the body dimensions as well as skeletal
variability among, and habitat preferences of, Acipenserids
are worth considering. Appraising these parameters is of help
in defining ranges within which archaeological finds can be
better understood.
Sturgeons (Acipenseriformes) are large (80–600 cm), late
maturing (almost 25 years for some females) fish of con-
comitant longevity. They inhabit coastal sea waters, rivers
and lakes, mostly within a latitudinal range from 30–70° N

41
 The Iron Gates in Prehistory

Figure 4. Sturgeon, weighing 181 kg, caught at Paks (Hungary) in 1987. Photo: István Takács.

Table 4. Characteristic dimensions of Danubian sturgeons.

Species Total length (m) Live weight (kg)

Beluga sturgeon Huso huso Linné 1758 2–3 (max. 10) 80–100 (max. 1000)
Common sturgeon Acipenser sturio Linné 1758 1.5–2.5 (max. 3.5) 70–100 (max. 320)
Russian sturgeon Acipenser gueldenstaedti Brandt 1833 2–2.5 80–100
Ship sturgeon Acipenser nudiventris Lovetzky 1828 2 40–50
Stellate sturgeon Acipenser stellatus Pallas 1771 1.5–2 30–50
Sterlet Acipenser ruthenus Linné 1758 1–1.2 15–16 (max.)

(Hankó 1931: 9). They feed on small animals, including
molluscs, crustaceans and small fishes (anchovies, sprats,
gobies) as well as plants, and are anadromous, i.e. migrate
up-river to breed and spawn. Because of this migratory be-
haviour, Luigi Ferdinando Marsigli, an 18th century military
engineer surveying the lower Danube valley, classified be-
luga sturgeons as fluviatiles marini, marine fish that live in
rivers (Marsigli 1726). Systematic work on the complex tax-
onomy of sturgeons began relatively recently (Berg 1904).
As with the archaeological record, most historic refer-
ences to sturgeon are vague as to species identification. This
is related to the fact that some sources not only predate
Linnæan nomenclature, but the living fish species are also
often difficult to tell apart. This is, to some extent, a con-
sequence of natural hybridization. The main dimensions of
species relevant to this study are summarized in Table 4. The
large species central to this study include beluga sturgeon,
Russian sturgeon, ship sturgeon and stellate sturgeon (Fig. 5).
Occurrences of common sturgeon have been reported
from the Danube only relatively recently. According to
Antipa (1905) these fish spawn on the sandy sea bottom be-
fore the Danube estuary. Some distribution maps (Muus &
Dahlström 1965: 63, fig. 26; Terofal 1971; Vuković &
Ivanović 1971: 111; Wheeler 1978: 56; Müller 1983: 123, fig.
28) suggest that it does live in the Danube. Other sources,
however, assert that common sturgeon is a fish of the
Atlantic/Baltic region (Maitland & Linsell 1978: 78; Curry-
Lindahl 1985: 230). This species certainly has not been re-
corded in the Hungarian section of the Danube (Berinkey
1966: 17; Pintér 1989: 24), and a reference to its occurrence
in Transylvania has also been questioned by Hankó (1931: 9).
Although common sturgeon was mentioned in the discussion
of ‘Neolithic’ Padina (Clason 1980: 167), that faunal list only
contains the item ‘Acipenser sp./Huso huso’ (Clason 1980:
149).
Sterlet is the smallest Acipenserid in the region. It is not
anadromous, having adapted to freshwater conditions.
Owing to its smaller size and behaviour it falls beyond the
focus of this paper.
Osteology and taphonomy
The selective survival of different animal remains has a direct
bearing on archaeological interpretations. Studying the post
mortem history of excavated bone is indispensable for the
critical understanding of archaeozoological assemblages.
Taphonomic analyses must be based on familiarity with the
original, complete skeleton. Interspecific comparisons
between morphologically similar Acipenserid species are
made difficult by at least two skeletal characteristics of these
fish:

42
 László Bartosiewicz et al.: Sturgeon fishing in the middle and lower Danube region

1m

Figure 5. Sturgeon species of major importance in the Danube, drawn to scale on the basis of mean lengths in Table 4
(compiled and redrawn to scale after Berinkey 1967 and Pintér 1989).

1. The sheets of dermal bone covering the head of
Acipenserids tend to be numerous and irregular in shape to
such an extent that they may even be strongly asymmetric
within the same individual.
2. While bones from large fish would more likely be re-
covered, their survival is poorest in old Acipenserids (es-
pecially beluga sturgeon), whose skeleton reabsorbs min-
erals with the advancement of age. Thus, the largest bones
tend to be most easily destroyed or eroded beyond recog-
nition in archaeological deposits.
Brinkhuizen (1986) reviewed differences between the usu-
ally resistant and morphologically most characteristic dermal
scutes of beluga and Russian sturgeons from the Iron Gates.
The dorsal scutes of beluga sturgeon are oval in shape with an
elongated, horn-like process. In mature individuals, these
scutes are covered by skin. The lateral scutes of beluga stur-
geon are toothed. In old individuals, they are partially reab-
sorbed and develop a spongy, eroded look.
The dorsal scutes of common and stellate sturgeon display
a more marked morphological difference. These bones are
almond-shaped with roof-like cross-sections. The medial
edge of the ‘roof’ is largely symmetric in common sturgeon,
while it is slightly skewed in a cranial direction in stellate
sturgeon giving it a rose-thorn profile line. A strong radial
pattern is also characteristic of stellate sturgeon. In other
Danubian Acipenserids, even this character looks transitional
between the two morphological extremes.
It is chiefly large and compact dermal scutes with charac-
teristic surface patterning that show up even in hand-
collected assemblages, which otherwise contain few fish re-
mains. These scutes are arranged in dorsal, lateral and ventral
rows along the body of sturgeons and differ both in size and
shape by anatomical location (Casteel 1976: 38, fig. 19).
According to Berinkey (1966: 18–22), Vuković &
Ivanović (1971: 106–112), Pintér (1989: 24–31) and a review
of nine authors by Brinkhuizen (1989: 41), the number of
dermal scutes and fin rays varies between the discussed spe-
cies as shown in Table 5 and Figure 6. As is shown by these
data, dermal scutes not only form continuously changing
rows on individuals of varying sizes, but also their numbers
differ by species which makes their quantitative analysis a
nightmarish enterprise.

43
 The Iron Gates in Prehistory

Table 5. The number of some skeletal elements in Danubian Acipenserids.

Species Fin rays Scutes Branchio-

spinae
dorsal anal dorsal lateral ventral
Beluga sturgeon Huso huso Linné 1758 62–73 28–41 9–17 37–53 9–12 17–31
Common sturgeon Acipenser sturio Linné 1758 31–43 22–27 9–16 24–39 9–14 18–25
Russian sturgeon Acipenser gueldenstaedti Brandt 1833 33–51 21–33 5–19 24–50 14–16 15–31
Ship sturgeon Acipenser nudiventris Lovetzky 1828 45–57 23–37 11–17 50–74 11–17 24–36
Stellate sturgeon Acipenser stellatus Pallas 1771 40–46 24–29 9–16 26–43 9–14 24–26
Sterlet Acipenser ruthenus Linné 1758 37–54 19–31 12–17 57–71 10–19 16–21

Figure 6. Variability in the number of dermal scutes in Danubian acipenserid species.

Habitat reconstruction
The presence/absence of sturgeon bones at archaeological
sites is not only a matter of preservation. It may be presumed
that, at least in prehistoric times, primary butchery of large
sturgeon took place near where the fish was landed. The
Schela Cladovei finds on a low-lying terrace of the Danube
are indicative of this tendency. In later periods, carcass dis-
memberment may have taken place away from where the fish
was pulled ashore, at markets or high status sites of
consumption.
Prehistoric findspots (Fig. 7) as well as historically recor-
ded catch sites (Fig. 8) suggest that understanding habitat
preferences of sturgeon may point to locations where they
could be caught most efficiently — information whose in-
timate knowledge was essential to fisherfolk throughout the
millennia.
Water properties
Anadromous Acipenserids migrate into rivers for spawning:
for this purpose they need deep, well aerated waters with a
hard, preferably sandy or rocky substrate. Geographical
latitude/climate, distance upstream from the river’s mouth,
and variation in discharge rate affect the abundance and dis-
tribution of anadromous fish populations (Schalk 1977).
The quantity of oxygen dissolved in water (y mg/l) is a
function of current speed (x1 m/s) and water temperature
x1 = speed of current: y = 1.953x + 1.984 r = 0.943
x2 = temperature: y = -0.221x + 13.669 r = -0.979
(x2 ºC). On the basis of empirical data published by Pénzes
and Tölg (1977: 327, table 4) as well as Harka (1993) this
relationship was expressed using the following regression
equations by Bartosiewicz & Bonsall (2004: 263):
The regression coefficients express the idea that every addi-
tional 1 m/s in water velocity increases dissolved oxygen
content by almost 2 mg/l in rivers, while a 1 °C rise in tem-
perature would result in a 0.2 mg/l loss. This is why oxygen-
loving sturgeons seek rapid rivers during the spring for
sterlet (Acipenser ruthenus Linné 1758) 3.0–3.5 mg/l
pikeperch (Stizostedion lucioperca Linné 1758) 2.0–3.0 mg/l
tench (Tinca tinca Linné 1758) 0.7 mg/l
spawning, and why water temperature is of decisive import-
ance in the timing of their migration. The minimum require-
ments of dissolved oxygen by three characteristic freshwater

44
 László Bartosiewicz et al.: Sturgeon fishing in the middle and lower Danube region

Figure 7. Archaeological sites and sturgeon bone finds in the Iron Gates gorge section of the Danube.

Figure 8. Archaeological sturgeon finds and the occurrence of 19–20th century record specimens (full circles) in present-
day Hungary. Archaeological site codes: 1= Tiszaföldvár (prehistoric), 2=Tiszaug (prehistoric), 3=Ács (Roman period),
4=Szentendre, 5=Esztergom, 6=Pilisszentkereszt, 7=Visegrád, 8=Vác, 9=Szentendre, 10–12=Buda Castle,
13=Sárszentlőrinc.

45
 The Iron Gates in Prehistory
Figure 9. The relative frequency of sturgeon bones in hand-collected fish assemblages along the Iron Gates section of the
Danube (after Bartosiewicz & Bonsall 2004: 266).
fish species are as follows (Pénzes & Tölg 1977: 327):
Although sterlet is not an anadromous species, it clearly
illustrates the highest requirement of dissolved oxygen of
Acipenserids in this comparison. As the speed of the river is
greater towards its source (and its temperature tends to de-
crease with increasing altitude), the further upstream stur-
geons move, the better the circumstances for spawning.
Aeration parameters were summarized by Harka (1993).
From the tabulated summary of his data it is evident that
foothill and lower foothill river sections with 3.0 to 4.0 mg/l
of dissolved oxygen would provide ideal spawning grounds
for sturgeons (Bartosiewicz & Bonsall 2004: 263, table 4).
Even during the 20th century, beluga sturgeons as heavy
as 900 kg used to be caught by the Donji Milanovac fisher-
men inside the Iron Gates gorges. Naturally, the distribution
of such spots varies with the discharge along a river as de-
termined by climate and topography.
The c. 2300 m 3 /s average discharge of the Danube at
Budapest more than doubles to over 5600 m3/s in the Iron
Gates gorge. Before the construction of the Iron Gates 1 dam,
this c. 130 km long section of the Danube was characterized
by extreme changes in water levels. Minimum discharge was
1400 m3 /s, while 16,000 m 3/s values were also measured
(Bǎncilǎ et al. 1972: 9). Moreover, prior to dam closure, the
riverine environment of the Iron Gates gorges was character-
ized by strong currents, hard substrates, and was rich in nu-
trients, aquatic plants, insects and invertebrates (e.g.
Gammaridae and Corophiidae sp.) that sustained rich and
varied fish resources. Thus, this section of the Danube
provided an ideal habitat for large sturgeons as well as sterlet.
Topography and changes in riverbed gradient
The river was confined to a width of only 170 m in the
Khazan gorge. As is shown by the pre-regulation measure-
ments of the river from 1872 (Bartosiewicz & Bonsall 2004:
265, fig. 6), depths varied tremendously, between 0.5 and
50 m before the river exited to the plain. In the lower part of
the Iron Gates gorge, upstream from Schela Cladovei (near
Turnu-Severin) in Romania, the riverbed has a very steep
gradient, falling 8 m in only 20 km. By contrast, over the
935 km between Schela Cladovei and the Black Sea, the ri-
verbed declines overall by only 34 m (Giurescu 1964: 101).
46
Thus the net gradient is three orders of magnitude different
(40% vs 0.037%) not to mention the considerable differences
in topography and relief between these two sections of the
Danube valley.
Lake sturgeon in Canada spawn in rivers at depths of
c. 0.5–5 m, in areas of swift water or rapids at the foot of low
falls that slow down further migration (Needs-Howarth 1996:
149). Given the mass movement during the spawn run, such
places must have been packed with sturgeons of all sorts in
the Iron Gates gorge as well, making them increasingly vul-
nerable to human predation at these natural traps.
This possibility seems to be supported by the differential
proportion of sturgeon bones to those of other large fish in
hand-collected archaeozoological assemblages from sites at
different locations within the Iron Gates (Fig. 9; Bartosiewicz
1996, 1997). The Late Mesolithic/Early Neolithic site of
Schela Cladovei, located downstream from the rapids, prior
to dam construction marked the exit from the Iron Gates
gorge, and must have been one of the ideal fishing spots
where great numbers of migrating fish could be targeted.
The reported absence of sturgeon bones at the prehistoric
sites of Lepenski Vir and Vlasac raises the question of
author-related bias: the bones were identified by Sándor
Bökönyi, a leading expert in mammalian osteology. Did
some of the large ‘catfish’ bones identified by Bökönyi (1969,
1978) originate from sturgeon? However, if nothing else, the
ornate head bones and unmistakable dermal scutes of stur-
geons would surely have attracted his attention since they can
be most easily recognized in excavated assemblages (Desse
& Desse-Berset 1992); Bökönyi himself identified sturgeon
remains at the Neolithic site of Mihajlovac–Knjepiste
(Bökönyi 1992: 79). More recent work suggests that
Acipenserid bones did occur on sites upstream from Schela
Cladovei, including sites within the Iron Gates gorges. Borić
& Dimitrijević (2005) have reported sturgeon (beluga) bones
from the floors of some of the trapezoidal buildings at
Lepenski Vir. In addition to sterlet, commonly represented at
prehistoric sites along the Danube in Romania, remains of
great sturgeon (beluga) were also reported from the
Epipalaeolithic sites of Ostrovul Banului and Icoana. Bones
of Epipalaeolithic Russian sturgeon were also identified at
Cuina Turcului (Pǎunescu 2000: 342). In the time of the
 László Bartosiewicz et al.: Sturgeon fishing in the middle and lower Danube region
Figure 10. Schela Cladovei and major sturgeon catch sites along the Danube.
Boian and Gumelniţa cultures, anadromous sturgeons were
also caught at sites along the lower Danube, notably
Borduşani, Hârsova and Isaccea (Radu 1997, 2003).
When some historically renowned sturgeon fishing
grounds are plotted along the entire section of the Danube
(Fig. 10), many of them pinpoint locations downstream from
reaches with a steep gradient in the riverbed (large sturgeon
were regularly caught at Tulln, upstream from Vienna, and a
1692 record is known from Bavaria).
In addition to the river’s gradient, changes along its course
also created opportunities for sturgeon fishing. Until the
mid-19th century, the Danube meandered through flood-
plains and wetlands, shifting its course with each spring
flood. Fish turned such marshes, e.g. around Szentendre,
north of Budapest, into rich hatcheries.
In the plains, the variable course of the river resulted in
underwater shoals, fords, smaller and larger islands that all
influenced currents and created spots where sturgeons could
be caught more easily. In 1690, “50–100 sturgeons were
caught and butchered daily” at the island of Ada-Kaleh,
downstream from Orşova in the Iron Gates (Marsigli 1726).
Plotting the catch sites of 19th–20th century record speci-
mens on the map of present-day Hungary (cf. Fig. 8), the ho-
rizontal patterning appears more homogeneous, although
confluences with tributaries and major river bends seem to be
indicative of good sturgeon fishing grounds.
Sturgeon were regularly caught in many of the Danube’s
tributaries, including the Váh, Maros and Tisza rivers (Hankó
1931: 9). In the latter, sporadic prehistoric bone finds as well
as modern records (Bartosiewicz 1999) confirm Hankó’s
statement (cf. Fig. 8). In 1518, following a long medieval
tradition, the city of Komárom in northern Hungary was giv-
en the rank of Royal Sturgeon Fishing Grounds (Herman
1980: 267). This strategically important point is located at the
confluence of two branches of the Danube, downstream from
47
Europe’s largest inland river delta where sturgeon were
already caught in Roman times (Bartosiewicz 1989: 611).
The monopoly of Komárom fisherfolk can still be detected in
the 18th century, when they let sturgeon fishing rights from as
far as the Tisza river between Tiszacsege and Tiszafüred
(Bencsik 1970: 98).
The mouth of a small left bank tributary of the Danube,
across from the northern tip of Margaret Island in Budapest,
is called ‘Sturgeon Catcher’, another indication that such
natural topographic features were exploited in sturgeon
fishing.
Sturgeon shoals thinned out as the river was regulated and
walled off from the floodplain (Woodard 2000). Large-scale
commercial navigation along the Danube also had a negative
impact as it became necessary to keep the riverbed clean of
sand and gravel deposits. By the mid-20th century, beluga
sturgeon seldom swam upstream beyond the Iron Gates.
Sturgeon behaviour
Sturgeons may live for up to a 100 years. Male beluga stur-
geons become sexually mature by the age of 12 years.
Females usually start spawning at 18 years (Deckert 1967:
66). Prior to sexual maturity, Acipenserids approach the
coastal areas of the Black Sea. Following a short period of
adaptation, they start moving upstream into the rivers. From
that point onwards, patterns of migration and seasonality
directly influenced the success of sturgeon fishing inland.
Some stocks already begin to move into the Danube delta
in September–October. Others begin the spawning run in the
sea in March–April. By late summer, young sturgeons move
into the lower Danube. They reach sexual maturity by next
spring (Table 6). The arrival of these immature individuals,
as well as their prolonged presence in rivers seems to indicate
 The Iron Gates in Prehistory

Table 6. Spawning parameters of various sturgeon species in the Danube (Berinkey 1966).

Species Spawning time Water temperature

Beluga sturgeon Huso huso Linné 1758 March–May 9 ºC

Common sturgeon Acipenser sturio Linné 1758 April–May 8–18 ºC
Russian sturgeon Acipenser gueldenstaedti Brandt 1833 April–May 10–17 ºC
Ship sturgeon Acipenser nudiventris Lovetzky 1828 April–June, —
Stellate sturgeon Acipenser stellatus Pallas 1771 April–June 10–17 ºC
Sterlet Acipenser ruthenus Linné 1758 April–June 12–17 ºC

Figure 11. Seasonal variations in river discharge in the Iron Gates in ‘characteristic years’ prior to dam construction.
Inset: mean annual values. (after Bǎncilǎ et al. 1972).

that they are keen on adapting to fresh water. Groups of mi-
grating sturgeon, therefore, tend to include individuals of
different ages and sizes, as was observed in North America
(Roussow 1957).
Beluga sturgeon is the first to move into fresh water, as
soon as icy floods are over (Hankó 1931: 9). They are fol-
lowed by Russian and stellate sturgeon swimming up to the
Iron Gates. All three species are actively targeted by fisher-
men between January to June as well as October to
December. These two periods largely correspond to the
spring and autumn migrations in the Iron Gates gorge. Most
notably, they also coincide with seasons of high discharge in
this region (Fig. 11; Bǎncilǎ et al. 1972: 19), when low water
temperatures and high water velocity favour spawning.
Sturgeon fishing between June and September only took
place opportunistically in the Iron Gates when water temper-
atures were high and discharge low. Mátyás Bél (1764: 39),
an 18th century Hungarian naturalist, documented the same
two seasons in Hungary: spring fishing began in March and
continued uninterrupted until June (cf. Table 7). The autumn
season lasted from August until December, unless winter
began early. These data contradict somewhat 19th century
records that in Hungary only 7–15% of the yearly sturgeon
catch were landed between February and April, while
41–57% were caught between September to November. The
rest were caught opportunistically throughout the year (Khin
1957). Identifying the season of death may be attempted by
thin-sectioning pectoral fin rays from archaeological sites.
The low precision of this method, however, makes the iden-
tification of fish caught during the spawn run contradictory
(Desse-Berset 1994; Needs-Howarth 1996). Seasonal differ-
ences in fish size are also difficult to establish. Some
19th–20th century record specimens of known date are listed
by season in Table 7.
Owing to the random nature of the catch, no significant
seasonal weight or length difference could be detected
between record sturgeons. These data reconfirm, however,
that equally large sturgeons could be caught both during the
spring and in the late autumn/winter.
Interspecific differences in spawning temperatures, as
well as the presence of off-season stragglers meant that stur-
geons were available year round. The use of their bones as
seasonal indicators in archaeological assemblages, therefore,
is limited to probabilistic interpretations. It would seem lo-
gical that sturgeons could be most successfully targeted when
they were rushing upstream along the Danube in great num-
bers. However, this is contradicted by the 41–57% September
to November catch statistics given previously.

48
 László Bartosiewicz et al.: Sturgeon fishing in the middle and lower Danube region

Table 7. Measurements of sturgeon caught at known seasonal dates in the Hungarian

section of the Danube.

Year Date Location Weight (kg) Length (cm)

1957 March 2 Paks 138 273

1922 March 3 Gemenc 90 220
1950 March 8 Százhalombatta 132 300
1936 March 18 Dunapataj 63 215
1927 ‘Good Friday’ Dunapentele 86 235
1987 May 16 Paks 181 300
1957 May Baja 134 280
1954 May 27 Paks 50 203
Mean value, spring 109.3 253.3
Standard deviation 44.2 39.5
Winter
1911 early September Orsova (Iron Gates) 102 252
1956 October 6 Ercsi 117 263
1870 ‘winter catch’ Ásvány-Győr 250 365
1894 ‘winter catch’ Pozsony 100 200
1955 January 31 Ercsi 117 263
1953 February 1 Ercsi 63 213
Mean value, spring 109.3 253.3
Standard deviation 44.2 39.5
P-value of Student’s t-test 0.600 0.819

Historical accounts of sturgeon in the Danube
While large Acipenserid bones commonly occur in some
Mesolithic and Neolithic archaeological assemblages, they
seem to become rare by the late Middle Ages. In part, this
may be explained by a shift in the focus of archaeological
research: catch sites were more likely to coincide with the
sites of consumption during prehistory. By the Middle Ages,
however, sturgeon remains are best known from high status
settlements in the Danube Bend gorge in Hungary (Fig. 8).
This valuable fish is known to have been traded over both
short and long distances. From an archaeological point of
view, however, most relevant historical information relates to
methods of catching these animals. Given the large size of
beluga sturgeon especially, landing it posed a special chal-
lenge to fisherfolk in all periods. According to Masen (1681:
898), “the beluga is similar to sturgeon and as strong as a
tuna, frequent in the Danube”. While there is no direct evid-
ence of prehistoric sturgeon fishing techniques along the
Danube, elaborate systems for sturgeon fishing have been
documented since the Middle Ages.
Sturgeon weirs
In accordance with the large size and strength of these anim-
als, traps or enclosures placed in rivers to catch sturgeon, of-
ten included massive timber structures, sometimes in com-
bination with large-holed netting. The placement of weirs
was of strategic importance. Familiarity with sturgeon habit-
ats, i.e. features of riverbed and adjacent topography, as well
as the knowledge of sturgeon behaviour were, thus, instru-
mental in minimizing labour expenditure required for the
construction of weirs.
The German naturalist Johann Georg Gmelin became a
professor of chemistry and natural sciences in St. Petersburg,
Russia in 1731. He documented a sturgeon weir shaped like
a maze and made from timber fencing which stretched across
the Volga river. Sturgeons swimming upstream were trapped
in the four, curvaceous ‘pockets’ of this complex structure
(Khin 1957: 12, fig. 7).
Nineteenth century sturgeon traps in the Iron Gates gorge
were described by Jókai (1872: 7) as follows: “Between the
islets, the narrow branches of the Danube are disrupted [by]
double post structures made from robust timber, arranged in
a V-shape, opening downstream... Once the sturgeons enter, it
is not their habit to turn downstream. As they proceed in the
ever-narrowing funnel, they wind up in the ‘death chamber’
at the end...”.
Mixed-media weirs included the use of strong nets as well.
Khin (1957: 14–15) refers to a 16th century description by
Miklós Oláh. Wooden posts were staked across the Danube in
November, before the waters grew icy. Fishermen in boats
stretched nets between the evenly placed posts, then cannons
were fired to scare sturgeons into the traps.
Medieval documents suggest that full closure of rivers
would have been considered an unfriendly or even illegal
move. A 1528 court case in Hungary between the city of Vác
and the royal capital, Buda (located downstream), clearly
shows the competing interests of their sturgeon fishing com-
munities (Szilágyi 1995: 114). Weirs, therefore, were often
placed between the bank and smaller islets, thus, exploiting
only one of the river’s branches.
One of the most instructive documents on sturgeon fishing
in the Iron Gates gorge is found in Volume IV of Marsigli’s
1726 work. The frontispiece (Fig. 12) portrays numerous
technical details. It shows a tunnel of three A-shaped gates
built between a small island and the right bank of the Danube
(an arrow clearly indicates the downstream direction). In the
lower half of the picture the narrow gorge is flanked by steep,
misty cliffs. Dense ripples indicate a more rapid current here.
Sturgeon are being caught at the downstream end of this sec-

49
 The Iron Gates in Prehistory

Figure 12. Sturgeon fishing at the exit from the Iron Gates gorge (from Marsigli’s 1726 book).

tion, where the hills and the plain meet. Fishermen in small
boats haul sturgeons onto the bank, where primary butcher-
ing takes place on a makeshift table.
According to a 1702 description by Bél (1764: 39), willow
withe fencing was combined with strong hemp nets at a slight
elevation in the riverbed at Földvár (Tolna county, Hungary).
The soft substrate mentioned in his text (onto which stur-
geons were driven and caught) may be a reference to a sub-
merged sandbank.
The construction of weirs, even at the best loci, required
tremendous investments of labour. In the 16th century along
the Tisza river entire villages of serfs were enlisted to build
weirs using oak logs, under the direction of the magister
clausurae. Aside from having an allowance of fish during
construction work, these serfs had the right to half of the
catch in the new weir — with the exception of sturgeons
(Maksay 1959: 703).
Building and maintaining sturgeon weirs not only required
labour. In agricultural areas, shipments of appropriate oak
logs from distant forests also had to be organized. According
to a 1493–1495 price list, for example, 220 such ‘weir’ logs
were bought by the Eger episcopate in Hungary (Kandra
1887: 378). Although the sporadic nature of the written data
do not permit us to estimate the costs of sturgeon weirs, their
construction must have represented a major investment, af-
fordable only by well-organized estates in medieval Hungary.
The lack of archaeological evidence for weirs does not
exclude the possibility that such complex structures could
have been erected in ancient times. The possibility of such
cooperative efforts is clearly illustrated by the construction in
AD 103–105 of Trajan’s bridge across the Danube at Drobeta
(modern-day Turnu-Severin) a few kilometers downriver

50
 László Bartosiewicz et al.: Sturgeon fishing in the middle and lower Danube region
from the prehistoric site of Schela Cladovei.
Other equipment
By the 19th–20th century hardly any references to sturgeon
weirs occur in Hungary (Szilágyi 1995: 108). Large and
strong, so-called pipola nets, were made especially for stur-
geon fishing out of hemp. However, by the second half of the
19th century large sturgeons became scarce (cf. Fig. 3), so
that these special nets were used but rarely (Herman 1887:
281). Sturgeons could also be caught with gill nets, suspen-
ded vertically in the water to trap fish by their gills, as the
strong pectoral fins of sturgeons easily became hooked in its
meshes (von Brandt 1964: 170).
Sturgeons were also caught using sharp hooks strung on a
strong rope and stretched across rivers. In contrast to similar
methods of catfish (Silurus glanis Linné 1758) fishing, such
hooks did not have to be baited, since curious sturgeons often
‘played’ with these glossy pieces of metal and were caught by
chance (Khin 1957: 16). By the late 19th century, such hooks
became the primary means of sturgeon fishing (Herman
1887: 368).
Lake sturgeons in North America could be speared even
from the shore in shallow waters (Needs-Howarth 1996:
149). Fish, injured and dazed, were then dragged near the
bank. According to Marsigli (1726), fishermen enveloped
such fish in a large net in the Danube and provoked it further
onto the bank by ‘titillating’ it until it became stranded.
Captured sturgeons were sometimes kept tethered to trees or
strong poles, before being towed upstream by boat to the
nearest market. The capitals of Buda and Vienna were sup-
plied with live sturgeon in this way (Bél 1764: 41).
Aside from hefty hooks and harpoons, grapnels must have
been important tools in landing these large fish at all times.
By the 18th century, the great sturgeons were even ‘hunted’
with firearms (Marsigli 1726, in Deák 2004: 74).
Cognitive aspects of sturgeon fishing
Sturgeons were the largest fish in the Danube. Their sheer
size and powerful movement must have impressed people
throughout history. The renowned abundance of sturgeons
during the spawning run must have further enhanced their
awesome perception. Thus, it is likely that these great beasts
had dualistic symbolic meanings related to both life and
death in the spiritual life of peoples in the region.
Positive perceptions
Isotopic studies of human remains by Bonsall et al. (1997,
2000, 2004) have confirmed the results of previous faunal
analyses (e.g. Bökönyi 1978, Bartosiewicz et al. 1995), in-
dicating that Mesolithic populations in the Iron Gates con-
sumed considerable amounts of fish (aquatic resources were
still exploited during the Early Neolithic, but the dietary role
of terrestrial resources increased). Even if sturgeon was
available on a seasonal basis, owing to the quantities of meat
even single individuals yielded, its contribution to the diet
cannot be underestimated, and probably determined the pre-
historic perception of these animals.
Highly valued sturgeon meat won the name ‘Royal Fish’
51
for beluga sturgeon in Medieval Hungary. Aside from the
impressive quantity of food provided by a single specimen,
the quality of sturgeon flesh was also appreciated, regardless
of size. Of the species that resembled great sturgeon, Marsigli
(1726) devoted most attention to the small sterlet whose meat
was said to be simply ‘the best’. A quarter of a millennium
later, this view was confirmed during ethnographic inter-
views conducted by Vasile Şişu among the fishermen of
Dubova district (Iron Gates gorge, Romania).
According to the Torah, only fish with scales and fins are
kosher. Accordingly, the bones of sturgeon and similarly
scaleless catfish occurred only in the Christian layers above
deposits associated with the 13th century Jewish community
in a well excavated in the Buda Royal Castle in Hungary. The
latter contained only remains of kosher carp-like fish
(Cyprinidae) and high status pike (Esox lucius L.;
Bartosiewicz 2003). The ganoid type (Lagler et al. 1977: 108,
fig. 4.2) scutes of sturgeon are not considered scales in this
religious context, since the skin is often torn during de-
scaling that renders the fish tref. Sturgeon meat and caviar
therefore were avoided by Ashkenazi Jews who settled in
Medieval Eastern Europe. ‘Legalizing’ sturgeon meat with
reference to its ganoid scales, however, became an important
halachic issue in the early 19th century Jewish religious re-
form in Hungary (Frojimovics et al. 1995: 293).
Needs-Howarth (1996: 153) concluded that concentra-
tions of lake sturgeon bones at certain archaeological sites in
Canada indicate that these animals had mythical or religious
meaning for Iroquoian people. In their artistic renditions of
Lake Superior native American mythology, Longfellow
(1855) and Kohl (1859 II: 143) indeed equate the ‘King of
Fishes’ with sturgeon.
It is not difficult to imagine that the much-awaited spring
arrival of masses of sturgeon amounted to some sort of a feast
during the Mesolithic and Early Neolithic of the Iron Gates.
According to a structuralist interpretation by Radovanović
(1997: 88–89), the upstream movement of beluga sturgeon
may have symbolized life to the prehistoric inhabitants of the
gorge, possibly counterweighted by the downstream orienta-
tion of the deceased in Mesolithic burials along the
riverbank.1
Negative perceptions
The perception of animals tends to be dualistic, ranging
between extremes (Bartosiewicz 1998: 69). At the prehistoric
site of Lepenski Vir in the Iron Gates gorge, the stone statue
dubbed Danubius by the excavator (Srejović 1972: fig. 52)
has been tentatively identified with sturgeon, having even a
crest of ‘dermal scutes’ carved along its back (Radovanović
(1997: 93, figs. 1–2). Although the exaggerated, even fright-
ening, facial features (such as the large, bulging eyes) are not
at all reminiscent of the modern perception of sturgeons —
sturgeon have very small eyes (cf. Fig. 4), this hypothesis is
not implausible if, as suggested by Srejović (1972: 111), the
sculpture was conceived as an apotropaic representation with
the power to avert evil or catastrophe. Bonsall et al. (2002)
suggested that the Lepenski Vir sculptures were intended to
protect against unpredictable and catastrophic floods. If so,
the ability of sturgeon to swim against the flow even at times
of high discharge, overcoming the power of the river, may
 The Iron Gates in Prehistory
explain their use in imagery.
Other, more negative associations must also be reckoned
with. Although ancient fisherfolk were perfectly able to dis-
tinguish between species based on to their intimate first-hand
knowledge, the image of powerful sturgeons may have been
distorted by beliefs in water monsters. For example, in an
Ojibway myth from North America, a giant fish threatens to
swallow a young boy (Williams 1956: 168). However, in
Lake Superior, the only fish that large is lake sturgeon, a
harmless bottom feeder.
A comparably large Danubian species is European catfish
(also known as sheathfish or wels). Large specimens of this
fish species reach the average size of sturgeons (c. 2.5 m,
120 kg; Pintér 1989: 135) and may grow up to 5 m (330 kg)
in the Dniepr (Curry-Lindahl 1985: 259). In contrast to stur-
geon, however, catfish are indeed ferocious carnivores with
fearful reputations. One of their vernacular synonyms, para-
sztfaló, bluntly means ‘peasant gobbler ’ in Hungarian
(Gozmány 1979: 957). The English synonym, sheathfish,
seems to relate this species with sturgeon, whose Indo-
European names were associated with ‘greave’ (a piece of
armor worn to protect the shin) by von Sadovszky (1995:
17–19). Moreover, the Hungarian name for Acipenserids,
tokhal, could be best translated as ‘case fish’ or, actually,
‘sheathfish’. Although the linguistic evaluation of apparent
similarities is beyond the scope of this paper, it may indicate
how the perceptions of these two large fish merged in popular
mythology.
Both in the Csallóköz region (the aforementioned inland
delta, located upstream from Komárom), and in the Iron
Gates, sturgeons following boats were considered a bad
omen. As a 19th century fisherman put it: “See, how that in-
fernal monster swims ceaselessly alongside our boat. It’s an
old sturgeon, almost half a ton. It always means trouble,
when evil beasts are racing with the boat like that” (Jókai
1872: 94).
Danubian sturgeons of all species have four barbels in
front of the mouth equipped with taste buds (Lagler et al.
1977: 350, fig. 11.9). According to Kriesch (1876: 68), rare
specimens without such ‘whiskers’ were called ‘calamity
fish’ in Hungary, an omen of disaster, threatening the region
or even the entire country. They must have occurred among
the young of stellate sturgeon especially. This means that the
negative perception of sturgeon was not necessarily linked
with large size and the concomitant challenge posed by stur-
geon fishing.
Overcoming sturgeons
In spite of all these benign and fearsome connotations, it was
only the landing of sturgeons that was indubitably trying and
dangerous. Overcoming such a formidable adversary re-
quired fishing skills, strength and probably involved a major
element of machismo at all times. According to 18th century
Hungarian fishermen, the tail of a landed great sturgeon
should not be approached because “even the most able-
bodied lad can be killed by a single slap from it” (Bél 1764:
40). The hard work of sturgeon fishing probably also served
as an outlet for social aggression.
Taphonomic loss and problems of seasonal dating make it
impossible to tell whether prehistoric sturgeon landings
52
reached the frenzy of the much-publicized, traditional
Grindadrap on the Faeroe Islands, during which pilot whales
(mostly the so-called Atlantic blackfish, Globicephala
melaena Traill 1809), are driven ashore and massacred,
mostly by men. A remarkable similarity between these small
cetaceans and sturgeons in size and, to some extent, shape
could be a functional basis of such hypotheses. The late 1st
century BC Greek author, Strabo (Geographica Lib. VII.3.18,
cited by Bél 1764: 35) actually compared sturgeons to dol-
phins in terms of size.
While it is possible that some prehistoric sturgeon remains
from the Iron Gates were the result of opportunistic fishing,
historical data on the scale and intensity of sturgeon exploit-
ation make the Grindadrap parallel sound reasonable. The
vivid late Baroque description of sturgeon fishing by Bél
(1764: 40–41) is as reminiscent of a gracious rite as of ar-
chetypal human predation.
Conclusions
While it cannot be demonstrated that sturgeons always
played a key role in the lives of all peoples living along the
Danube, their significance should not be neglected at sites
where their bones have been found. This multidisciplinary
review of sturgeon fishing should provide archaeologists
with insights into the possible ways to interpret sturgeon
bones encountered in ichthyoarchaeological assemblages.
1. Owing to their poor resistance to taphonomic loss stur-
geon bones tend to be underrepresented at most sites. On
the other hand, their large size guarantees the recovery of
the surviving remains even in hand-collected as-
semblages. Although species-level identification is limited
to special elements, large Acipenserids must have played
comparable roles in most cultures
2. Animals with such a large body mass must have been a
desirable prey for subsistence fishers. Their capture,
however, required special skills that included a thorough
knowledge of both habitats and fish behaviour, which de-
termined the probability of taking the best catch.
3. Aside from some degree of specialization, sturgeon fish-
ing could only succeed as a fundamentally cooperative
enterprise throughout history. It took planning and co-
ordinated action by fishing communities, which only in-
creased in significance as fishing techniques became ever
more complex. As a group activity, carried out by men in
historical times, sturgeon hunting may have had strong
associations with male identity.
4. Developing infrastructure for large-scale sturgeon fishing
required investments, which in turn increased the com-
mercial value of these animals. Medieval trade in stur-
geons shows the consumer’s end of this chain; some stur-
geon remains were found at sites, spatially separated from
catch sites, which were probably the places of primary
butchering.
5. Sporadic archaeological data cannot be used in studying
the depletion of stocks. Decline by the 19th century,
however, is reflected both in fish sizes and in the ethno-
historical records. Over a century ago Kriesch (1876:
12–17) recognized that deforestation, river regulation
 László Bartosiewicz et al.: Sturgeon fishing in the middle and lower Danube region
works, damming, water pollution, steamboats and timber
floating were directly responsible for habitat degradation.
Owing to their low reproductive rate and high visibility,
sturgeons were particularly hard hit by what he termed
‘human greed’, i.e. overfishing.
6. While the often-anecdotal ethnohistorical and literary ex-
amples cannot be used directly in the interpretation of ar-
chaeological sturgeon finds, they show the complexity of
attitudes (mostly related to the immense size and particu-
lar appearance of sturgeons) that are unlikely to have de-
veloped only in recent centuries.
Note
1. Some Final Mesolithic burials at Lepenski Vir (cf. Bonsall et al.
2004, this volume) are oriented parallel to the Danube with their
heads pointing downstream, but this is not a consistent feature of
Mesolithic burials from the Iron Gates.
Acknowledgements
A preliminary version of this paper was presented in September
2001 at the 12th Meeting of the Fish Remains Working Group
(International Council for Archaeozoology) in Paihía Bay, New
Zealand, organized by Drs Foss Leach and Janet Davidson. Figure
12, from Marsigli’s original work, was kindly provided by Andrea
Kreutzer, Librarian in the Museum of Military History in Budapest.
Dr Antal András Deák kindly gave permission to quote from his
manuscript on Marsigli prior to publication. Thanks are due to Dr
Alice M. Choyke for useful comments on the text as well as Dr Erika
Gál and László Daróczi-Szabó, zooarchaeologists, who translated
texts from Romanian.
References
Antipa, G. 1905: Die Störe und ihre Wanderungen in den europäis-
chen Gewässern mit besonderer Berücksichtigung der Störe der
Donau und des Schwarzen Meeres. Wien: Sitzungberichte der
internationale Fischerei Kongress.
Bacalbaşa-Dobrovici, N. 1997: Endangered migratory sturgeons of
the lower Danube River and its delta. Environmental Biology of
Fishes 48: 201–207.
Băncilă, I., Diakon, A., Djordjević, V., Hvoj, J., Jakovlević, P.
Neimarević, K., Nouresku, A., Sandul, G., Vasiliju, A. & Verčon,
M. (eds) 1972: Système hydro-énergétique et navigable Ðerdap
(Les Portes de Fer). Belgrade: Export-Press.
Bartosiewicz, L. 1989: Animal remains from the fort. In Gabler, D.
(ed.) The Roman Fort of Ács-Vaspuszta (Hungary) on the
Danubian Limes. Part ii. BAR International Series 531. Oxford:
British Archaeological Reports, 600–623.
— 1996: Early Medieval faunal remains from Pontes (Iron Gates
gorge, eastern Serbia). Acta Archaeologica Academiae
Scientiarum Hungaricae 48: 281–315.
— 1997: Őskori vizahalászat a Duna vaskapui szakaszán
(Prehistoric sturgeon fishing in the Danube). Halászatfejlesztés
20: 92–103.
— 1998: Attitudes to pets in the ethnolinguistic record. In Anreiter,
P., Bartosiewicz, L., Jerem, E. & Meid, W. (eds) Man and the
Animal World. Studies in Memoriam Sándor Bökönyi. Budapest:
Archaeolingua Kiadó, 65–78.
— 1999: ‘Bronze Age fish remains in the Tisza River region,
Hungary’. Paper delivered at the 10th meeting of the ICAZ Fish
Remains Working Group, New York City.
— 2003: ‘Eat not this fish — a matter of scaling’. Paper delivered at
the 12th meeting of the ICAZ Fish Remains Working Group,
1
53
Guadalajara, Jalisco State, México.
Bartosiewicz, L., Bonsall, C., Boroneanţ, V. & Stallibrass, S. 1995:
Schela Cladovei: a preliminary review of the prehistoric fauna.
Mesolithic Miscellany 16(2): 2–19.
Bartosiewicz, L., Boroneanţ, V., Bonsall, C. & Stallibrass, S. 2001:
New data on the prehistoric fauna of the Iron Gates: a case study
fromSchela Cladovei, Romania. In Kertész, R. & Makkay, J.
(eds) From the Mesolithic to the Neolithic. Budapest:
Archaeolingua, 15–22.
Bartosiewicz, L. & Bonsall, C. 2004. Prehistoric Fishing along the
Danube. Antæus 27: 253–272.
Bartosiewicz, L. & Takács, I. 1997: Osteomorphological studies on
the great sturgeon (Huso huso Brandt). Archaeofauna 6: 9–16.
Bél, M. 1764 [1984]: Tractatus de Rustica Hungarorum: A mag-
yarországi halakról és azok halászatáról (Hungarian Country
Life: The Fish of Hungary and their Fishing). Hungarian trans-
lation of the 1764 copy: Antal András Deák, 1984. Budapest:
Vízügyi Történeti Füzetek.
Bencsik, J. 1970: Egy jobbágyközség gazdasági, társadalmi élete az
úrbérrendezéstől a jobbágyfelszabadításig (The economic and
social life of a serfs’ community between the urbarium bill and
emancipation). Acta Universitatis de Ludovico Kossuth
Nominatae Series Historica 10: 49–91.
Benecke, N. 1986: Some remarks on sturgeon fishing in the south-
ern Baltic region in Medieval times. In Brinkhuizen, D.C. &
Clason, A.T. (eds) Fish and Archaeology. Studies in Osteometry,
Taphonomy, Seasonality and Fishing Methods. BAR
International Series 294. Oxford: British Archaeological
Reports, 9–17.
Berg, L.S. 1904: Zur Systematik der Acipenseriden. Zoologischer
Anzeiger 27: 665–667.
Berinkey, L. 1966: Halak – Pisces. Budapest: Akadémiai Kiadó.
Bonsall, C., Lennon, R., McSweeney, K., Stewart, C., Harkness, D.,
Boroneanţ, V., Bartosiewicz, L., Payton, R. & Chapman, J. 1997:
Mesolithic and Early Neolithic in the Iron Gates: a palaeodietary
perspective. Journal of European Archaeology 5: 50–92.
Bonsall, C., Cook, G., Lennon, R., Harkness, D., Scott, M.,
Bartosiewicz, L. & McSweeney, K. 2000: Stable isotopes, ra-
diocarbon and the Mesolithic–Neolithic transition in the Iron
Gates. Documenta Praehistorica 27: 119–132.
Bonsall, C., Macklin, M., Payton, R. & Boroneanţ, A. 2002:
Climate, floods and river gods: environmental change and the
Meso–Neolithic transition in southeast Europe. Before Farming:
the archaeology of Old World hunter-gatherers 3_4(2): 1–15.
Bonsall, C., Cook, G., Hedges, R., Higham, T., Pickard, C. &
Radovanović, I. 2004. Radiocarbon and stable isotope evidence
of dietary change from the Mesolithic to the Middle Ages in the
Iron Gates: new results from Lepenski Vir. Radiocarbon 46:
293–300.
Borić, D. & Dimitrijević, V. 2005: Continuity of foraging strategies
in Mesolithic–Neolithic transformations: dating faunal patterns
at Lepenski Vir (Serbia). Atti della Società per la Preistoria e
Protostoria della Regione Friuli-Venezia Giulia 15 (2004–05):
33–80.
Bökönyi, S. 1969: Kičmenjaci (prethodni izveštaj). In Srejović, D.,
Lepenski Vir: Nova praistorijska kultura u Podunavlju.
Belgrade: Srpska književna zadruga, 224–228.
— 1978: The vertebrate fauna of Vlasac. In Srejović, D. & Letica,
Z. (eds) Vlasac: Mezolitsko naselje u Djerdapu, vol. 2. Belgrade:
Serbian Academy of Science and Arts, 35–65.
— 1992: Animal remains of Mihajlovac–Knjepište, an early
Neolithic settlement of the Iron Gate gorge. (Hommage à Nikola
Tasić à l’occasion de ses soixante ans). Balcanica 23: 77–87.
Brandt, A. von 1964: Fish Catching Methods of the World.
Farnham: Fishing News Books.
Brinkhuizen, D.C. 1986: Features observed on the skeletons of
 The Iron Gates in Prehistory
some recent European Acipenseridae: their importance for the
study of excavated remains of sturgeon. In Brinkhuizen, D.C. &
Clason, A.T. (eds) Fish and Archaeology. Studies in Osteometry,
Taphonomy, Seasonality and Fishing Methods. BAR
International Series 294. Oxford: British Archaeological
Reports, 18–33.
— 1989: Ichthyo-archeologisch onderzoek: methoden en toep-
assing aan de hand van romeins vismateriaal uit Velsen
(Nederland). Unpublished PhD thesis. Rijksuniversiteit
Groningen, Groningen.
Casteel, R.W. 1976: Fish Remains in Archaeology and Palaeo-
environmental Studies. London: Academic Press.
Clason, A.T. 1980. Padina and Starčevo: game, fish and cattle.
Palaeohistoria 22: 141–173.
Curry-Lindahl, K. 1985. Våra fiskar. Havs- och sötvattenfiskar i
Norden och övriga Europa. Stockholm: Norstedt & Söners.
Deák, A.A. 2004: A Duna fölfedezése (The Discovery of the
Danube). Budapest: Vízügyi Múzeum, Levéltár és
Könyvgyűjtemény.
Deckert, K. 1967: Urania-Tierreich: Band 4: Fische, Lurche,
Kriechtiere. Leipzig: Urania Verlag.
Desse, J. & Desse-Berset, N. 1992: Pêches locales, côtières ou
lointaines: le poisson au menu des Parisiens du Grand Louvre, du
14e au 18e siècle. Anthropozoologica 16: 119–126.
Desse-Berset, N. 1994: Sturgeons of the Rhône in Arles (6th–2nd
century BC). In Van Neer, W. (ed.) Fish Exploitation in the Past.
Tervuren: Koninklijk Museum voor Midden-Afrika, 81–90.
Frojimovics, K., Komoróczy, G., Pusztai, V. & Strbik, A. 1995: A
zsidó Budapest. Emlékek, szertartások, történelem I–II. (Jewish
Budapest. Monuments, Rituals, History I–II). Hungaria Judaica
7. Budapest: MTA Judaisztikai Kutatócsoport.
Giurescu, C.C. 1964: Istoria pescuitului şi a pisciculturii în
Romania (The History of Fishing and Pisciculture in Romania).
Bucureşti: Editoriul Academiei.
Gozmány, L. 1979: Vocabularium nominum animalium Europae
septem linguis redactum. Budapest: Akadémiai Kiadó.
Hankó, B. 1931: Magyarország halainak eredete és elterjedése (The
origins and distribution of fishes in Hungary). A Debreceni Tisza
István Tudományegyetem Állattani Intézetéből 10: 3–34.
Harka, Á. 1993: A folyóvizek halrégiói (The fish regions of rivers).
A természet 44(5): 85–87.
Hensel, K. & Holčík, J. 1997: Past and current status of sturgeons in
the upper and middle Danube river. Environmental Biology of
Fishes 48: 185–200.
Herman, O. 1887: A magyar halászat könyve I–II (The Book of
Hungarian Fishing). Budapest: Kir. Magyar Természet-
tudományi Társulat.
— 1980: Halászat és pásztorélet (Fishing and Pastoralism).
Budapest: Gondolat Kiadó.
Jókai, M. 1872 [2001]: Aranyember I (The Golden Man I).
Budapest: Talentum Kiadó.
Kandra, K. 1887: Bakócs-codex vagy Bakócs Tamás egri püspök
udvartartási számadókönyve 1493–6. évekből (The Bakócs
Codex, the court accounting books of Bishop Tamás Bakócs of
Eger, 1493–6). Adatok az Egri Egyházmegye történelméhez 2:
333–452.
Khin, A. 1957: A magyar vizák története (The History of Hungarian
Sturgeons). Budapest: Mezőgazdasági Múzeum Füzetei 2.
Kohl, J.G. 1859: Kitschi-Gami oder Erzählungen vom Oberen See
I–II. Bremen: C. Schünermann.
Kriesch, J. 1876: Hasznos és kártékony állatainkról. Kézikönyv II.
rész. Halak (Handbook of Useful Animals and Pests. Vol. II.
Fish). Budapest: Szent István Társulat.
Lagler, K.F., Bardach, J.E., Miller, R.R. & May Passino, D.R. 1977:
Ichthyology. New York: Wiley.
Longfellow, H.W. 1855 [1964]: The Song of Hiawatha. New York:
1
54
Bounty Books.
Maitland, P.S. & Linsell, K. 1978: Europas sötvattenfiskar.
Stockholm: Albert Bonniers.
Maksay, F. (ed.) 1959: Urbáriumok XVI–XVII. század (16th–17th
Century Urbaria). Budapest: Akadémiai Kiadó.
Marsigli, L.F. 1726: Danubius Pannonico-Mysicus, observa-
tionibus geographicis, astronomicis, hydrographicis, historicis,
physicis perlustratus. Vol. VI. Amsterdam–The Hague.
Masen, J.S.J. 1681: Speculum Imaginum Veritatis Occultae. Ed.
tertia prioribus correctior. Köln: J.A. Kinckius.
Miranda, L.E., Wingo, M.W.M., Muncy, R.J. & Bates, T.D. 1987:
Bias in growth estimates derived from fish collected by anglers.
In Summerfelt, R.C. & Hall, G.E. (eds) Age and Growth of Fish.
Ames, IA: Iowa State University Press, 211–219.
Müller, H. 1983: Fische Europas. Leipzig: Neumann.
Muus, B. & Preben, D. 1965: Meeresfische der Ostsee, der Nordsee,
des Atlantiks: Biologie, Fang, wirtschaftliche Bedeutung.
München: BLV-Verlagsgesellschaft.
Needs-Howarth, S. 1996: Lake sturgeon fishing at prehistoric
Iroquoian sites near Lake Simcoe, Ontario. Archaeofauna 5:
147–154.
Păunescu, A. 2000: Paleoliticul şi Mezoliticul din spaţial cuprins
între Carpaţi şi Dunăre. Bucureşti: Agir.
Pénzes, B. & Tölg, I. 1977: Halbiológia horgászoknak (Fish
Biology for Anglers). Budapest: Natura–MOHOSZ.
Pintér, K. 1989: Magyarország halai (The Fish of Hungary).
Budapest: Akadémiai Kiadó.
Radovanović, I. 1997: The culture of Lepenski Vir: a contribution to
the interpretation of its ideological aspects. In Lazić, M. (ed.)
Antidoron Dragoslavo Srejović completis LXV annis ad amicis,
collegis, discipulis oblatum. Belgrade: Centre for Archaeological
Research, University of Belgrade, 87–93.
Radu, V. 1997: Archaeological researches at Borduşani-Popină
(Ialomiţa County). Preliminary report 1993–1994. Pisces.
Cercetări arheologice, 10: 96–105.
— 2003: Exploitation des ressources aquatiques dans les cultures
néolithiques et chalcolithiques de la Roumanie méridionale.
Thèse de doctorat de l’Université Aix–Marseille I, Préhistoire,
Archéologie, Histoire et Civilisation de l’Antiquité et du
Moyen-Age, Aix-en-Provence, France.
Roussow, G. 1957: Some considerations concerning sturgeon
spawning periodicity. Journal of the Fisheries Research Board of
Canada 14: 553–572.
Sadovszky, O. von 1995: Fish, Symbol and Myth. Budapest:
Akadémiai Kiadó.
Schalk, R.F. 1977: The structure of an anadromous fish resource. In
Binford, L.R. (ed.) For Theory Building in Archaeology: Essays
on Faunal Remains, Aquatic Resources, Spatial Analysis, and
Systemic Modelling. New York: Academic Press, 207–249.
Srejović, D. 1972: Europe’s First Monumental Sculpture: New
Discoveries at Lepenski Vir. London: Thames & Hudson.
Szilágyi, M. 1995: A Tiszai halászat. Az eszközök és fogási módok
történeti változásai (Fishing in the Tisza. Historical Changes in
Equipment and Methods). Budapest: Akadémiai Kiadó.
Terofal, F. 1971: Kwastsnoeken, steuren en verwanten. In Grzimek,
B. (ed.) Het leven der dieren IV, Vissen I. Nijmegen: Thieme &
Cie, 148–146.
Vuković, T. & Ivanović, B. 1971: Slatkovodne ribe Jugoslavije
(Freshwater Fish of Yugoslavia). Sarajevo: Zemaljski Muzej
Sarajevo.
Wheeler, A. 1978: Key to the Fishes of Northern Europe. London:
Frederick Warne.
Williams, M.L. (ed.) 1956: Schoolcraft’s Indian Legends. East
Lansing: Michigan State University Press.
Woodard, C. 2000: Singing the Danube blues. Christian Science
Monitor, May 11, 2000.
 The Mesolithic–Neolithic transition in the Ðerdap
as evidenced by non-metric anatomical variants

Mirjana Roksandic

Abstract: In order to examine the interactions of communities with different modes of subsistence — the foraging communities of Lepenski
Vir and the farming groups of the central Balkans — the non-metric anatomical variants were examined in four Ðerdap sites. This paper
argues for local continuity within the region, with high a degree of initial heterogeneity, and temporal ordering as the most likely explanation
for the pattern of change. A more pronounced difference at the time of contact with Neolithic populations argues for a limited ‘seeping in’
of a non-local population that did not result in a change of either economic base or ideology. There is no evidence of an incoming population
at the time of change to a farming economy and integration of the sites into the cultural circle of the Balkan Neolithic, but rather of a local
population accepting the new way of life.
Key words: Mesolithic, Neolithic, transition, Đerdap, non-metric anatomical traits

Introduction ber of communities in the Holocene that continue with es-

This article seeks to promote a better understanding of the
population dynamics of the Lower Danube as reflected in the
transition from the Mesolithic ‘Lepenski Vir culture’ to the
Neolithic ‘Starčevo complex’ in the Đerdap (Serbia). Since a
long period of coexistence of the Mesolithic and Neolithic
ways of life in this region has been proposed (Radovanović
1995, 1996b), this paper will attempt to reconstruct the extent
and mode of interactions between farmers and foragers
through the examination of skeletal material from four of the
most important sites excavated in the region: Lepenski Vir,
Hajdučka Vodenica, Vlasac and Padina.
The focus of this article being the Mesolithic population of
the Đerdap and its presumed contact with Neolithic peoples,
and since the debate over the meaning of this term has his-
torically played an important role in discussions between
principal investigators of the Đerdap sites (see Srejović 1971,
1979; Jovanović 1972; Srejović & Letica 1978; Boroneanţ et
al. 1999), it is important to provide clear definitions of both
Mesolithic and Neolithic as they are used here.
Mesolithic
Current definitions can be classified into two major groups:
one typological and the other chronological. For the pro-
ponents of a typological definition the characteristics of the
flint industry (such as the appearance of microliths) have the
most decisive value. For example, Orliac (1988: 686) pro-
poses that those industries situated between Palaeolithic and
Neolithic that possess “characters sufficiently different from
those of the industries of the two periods” should be determ-
ined as Mesolithic. While possibly appropriate for Western
Europe, typological definitions fall short in other areas of the
world. For proponents of a chronological definition all
hunter-gatherers of the Holocene are regarded as Mesolithic,
regardless of whether they show differences from
Palaeolithic hunter-gatherers. Since it would include a num-
sentially the same mode of food procurement and mobility
patterns as their Pleistocene counterparts as well as signific-
antly different groups, the need to introduce ‘Mesolithic’ as a
different term is not obvious.
Introduction of economic parameters has, in a number of
cases, led to economic determinism, with an open or subdued
notion of evolutionism. Distinction is made between the
Epipaleolithic — in which the Holocene adaptation does not
produce any changes in way of life and lithic technology
(Leroi-Gourhan 1965; Kozłowski & Kozłowski 1986) — and
the Mesolithic with its substantial changes in economy, eco-
logy, and material culture (Kozłowski & Kozłowski 1986).
The latter would be found only in innovation zones leading to
food production (Leroi-Gourhan 1965), or enabling change
from food collection to food production (Clark 1980). This
definition supposes an unidirectional evolution toward food
production and contradicts the data from large areas of the
world where substantial changes in economy, ecology and
material culture did not lead to the introduction of agriculture
(e.g. the Pacific Northwest Coast).
In order to overcome chronological and typological ambi-
guity, as well as economic determinism characterized by an
implicit evolutionist basis, Radovanović (1996a: 14) argues
that a qualitatively different phenomenon can be found in the
appearance of formal disposal areas for burial of the dead.
Formal disposal areas need not be a phenomenon separate
from the habitation site, as that would exclude all western and
Central European sites (Meiklejohn, pers. comm.), with the
exception of the newly excavated Mesolithic necropolae in
France (Duday & Courtaud 1998) and Belgium (Cauwe
1998). They are defined as “...areas of continuous, ceremo-
nial, mortuary disposal” (Radovanović 1996a: 14). Further,
they are an archaeologically visible phenomenon that is in-
terpreted as arising from the need to lay claim to territory by
its ideological integration (Chapman 1981).1 The need to
claim territory arises from a combination of linear rather than
hexagonal arrangement of units within one hunter-gatherer
group, or higher than usual population densities (Gamble

55
 The Iron Gates in Prehistory
1986: 52–53), based on intensive exploitation of a vital re-
source, or a greater variety of resources in the vicinity, with
semi-sedentism or sedentism. This would result in a structur-
al complexity of the social unit (Srejović 1979) usually ex-
pressed through developed ancestral and mortuary rituals, a
clear sign of changing times in prehistory associated with
changes in the social arena (Chapman 1993). Although ar-
chaeologically visible, the ambiguity in determining what
constitutes a formal disposal area makes them less valid for
defining the Mesolithic.
On the other hand, economic parameters and mobility
patterns combined, once evolutionist connotations are re-
moved, have the potential to make the definition of
Mesolithic more straightforward and appropriate even for
regions where the typological distinction of Mesolithic is not
as easy as in France (for example, Africa where microliths
appear as early as 70,000 BP), or where ceramics (tradition-
ally associated with the Neolithic) appear earlier than agri-
culture, as for example in the Jomon culture of Japan
(Imamura 1996) or in Scandinavia (Werbart 1998: 37, and
quoted literature).
This change in adaptation is usually linked with intensi-
fication of exploitation of one or more abundant resources
(r-selected resources) as opposed to the exploitation of K-
selected resources (see Gamble 1986: 41) that characterizes
mobile hunter gatherers. The availability of an abundant and
stable resource that can be exploited in the relative vicinity of
the camp has been linked to reduced mobility. The combina-
tion seems to be determinative of Mesolithic settlements.
Structural complexity, seen as a segmentation or an in-
crease in the parts that make up the whole (Kent 1989: 10),
could arise from changes in mobility patterns and increased
sedentism resulting in population increase and the need for an
arbiter in settling disputes (Lee 1972a, 1972b). However, it
cannot be taken for granted and must be demonstrated by in-
dependent data in many areas of the world where it has been
assumed (Brinch-Petersen & Meiklejohn, forthcoming).
Settlement distribution and mobility patterning, interact-
ing with more intensive exploitation of r-selected resources
(plants or fish), and aggregation necessary for tasks demand-
ing cooperative effort, can be regarded as both determinative
of the Mesolithic and as providing sufficient archaeological
visibility. Within the Đerdap, the Mesolithic is best defined as
primarily based on intensified exploitation of food resources
on a limited territory with reduced mobility. Regardless of
whether fish played a key role as staple (Bonsall et al. 1997)
or as a vital resource (Radovanovic 1996a), its availability is
the sine qua non of reduced mobility (sensu Kent 1989) and
the change in its exploitation provides clear evidence of
Mesolithic economy and social organization.
Neolithic
An economic definition of Neolithic, in which “the shift in
mode of subsistence to agro-pastoral farming remains the
only process that is relatively clearly defined, geographically
widespread and sufficiently archaeologically detectable” to
act as a signature of the Neolithic (Zvelebil 1996b: 625) is
widely accepted and both ethnologically and archaeologic-
56
ally traceable. Preceding or subsequent changes in social
structure, ideology or any other aspects of life need to be ex-
amined on a regional basis. The other two questions, that of
how much evidence of plant manipulation and animal hus-
bandry constitutes enough evidence (sensu Harris 1996, and
quoted literature), and whether a horticultural stage of farm-
ing economy — characterized by lack of impact on the en-
vironment (Willis & Bennet 1994) — can be perceived as
agriculture, also need to be regarded on a local scale. In
Southeast Europe, the introduction of cultigens and domestic
animals from the Near East seem to solve this problem, as
these are not found in the wild. Their introduction indicates a
shift in subsistence strategy and reliance (at least partial) on
imported cultigens.
The Đerdap context
Crucial to our understanding of the Đerdap
Mesolithic/Neolithic transition is the period of coexistence of
these two modes of life in the immediate vicinity. It is evid-
enced by 14C dates, Starčevo ceramics, and flint blades within
Mesolithic strata of the Lepenski Vir culture. This period
witnesses the coexistence, communication, trade and inter-
action between Lepenski Vir hunter-gatherers and Starčevo
farmers. It is in this light that the semantic discussion of the
two terms becomes increasingly important for understanding
archaeological data.
Interpretations of the Lepenski Vir–Schela Cladovei cul-
ture range from Epipaleolithic (Boroneanţ 1969, Letica 1971,
Lazarovici 1979), Protoneolithic (Srejović 1968),
Epipaleolithic in its early and Protoneolithic in its late phase
(Srejović 1979), Mesolithic (Prinz 1987, Srejović 1989,
Voytek & Tringham 1989), to Mesolithic and Early Neolithic
(Jovanović 1972, 1974). This variety stems to a great degree
from the above-mentioned definitions: one chronological, in
which Holocene hunter-gatherers are differently viewed as
Mesolithic or Epipaleolithic; and the other based on material
culture where microliths are taken as a tell-tale sign of the
Mesolithic, while ceramics, polished stone axes and adzes
are used as markers of Neolithic (for discussion and appreci-
ation of the theoretical positions in these different ap-
proaches, see Radovanović 1996a). The littoral distribution
of the sites in the Đerdap that resulted in reduced mobility or
possibly even sedentism, increased social complexity evid-
enced by specialization of ritual vs domestic activities
(Srejović 1979; but see Chapman 1993) and increased popu-
lation aggregation (Jackes et al., this volume) are all very
prominent features of Mesolithic components in the studied
sites. Neolithic in the region is characterized by the intro-
duction and reliance, at least partial, on Near Eastern culti-
gens and livestock. Although hunting, fishing and gathering
remain important, the proportion of domesticates vs wild
fauna and flora is sufficient to argue for the introduction of a
Neolithic economy.
Therefore, if intensification of exploitation of food re-
sources, on a limited territory with reduced mobility, charac-
terizes the Lepenski Vir Mesolithic, then all strata within
these sites that do not have evidence of food production
and/or introduced domesticates (above 5% as suggested by
 Mirjana Roksandic: The Mesolithic–Neolithic transition in the Ðerdap

Table 1. Synchronisation of the sequences identified at four key sites in the Iron Gates gorge:
shaded areas represent the appearance of ceramics in the stratum: light shading — sporadic
appearance, darker shading — ceramics are common as well as Pre-Balkan Plateau flint,
‘Montbani type’ chipped stone blades, along with geometric microliths (based on
Radovanović 1996a: 289, 1996b, 1996c; Radovanović & Voytek 1997).

Phase – millennium BC Padina L. Vir Vlasac H. Vodenica

6 - mid 6th B(III) II/IIIa — Ib

5 - 7th/6th B(II) I(3) — Ia

4 - 7th (2nd half) B(I) I(2) III Ia

3 - 7th (1st half) A-B I(1) Ib-II

2 - 8th/7th A/A-B Proto-LV Ia-b Ia

1 - 8th A — Ia —

Zvelebil 1996a) should be regarded as Mesolithic. If we ac-
cept that the Neolithic in the region is characterized by intro-
duced cultigens and livestock, then evidence for food pro-
duction and appearance of these cultigens in the strata should
be regarded as crucial for identifying them as Neolithic. In
this perspective, the sporadic appearance of Starčevo-type
ceramics and Starčevo blades in Mesolithic strata, i.e.
Mesolithic-type houses with ceramics in situ (Jovanović
1984b), can be explained by a porous agricultural frontier
with transfer of knowledge, material items, and individuals
across the board (Zvelebil 1996a), whilst recognizing that
this exchange could have involved both different direction
and different form over this long period of coexistence.
The presence of domestic animals should also be con-
sidered in this context. Dog was domesticated in the Đerdap
without doubt and is found in the Mesolithic, Transitional
and Neolithic strata of Lepenski Vir, Vlasac, Padina and
Hajdučka Vodenica. At Icoana selective hunting of wild pig
(very young and very old animals) and possible domestica-
tion of dog point in the same direction (Bolomey 1973). The
analyses from Padina (Clason 1980) and Hajdučka Vodenica
(Greenfield 1984, this volume) are not as conclusive as, in
both cases, the material was studied initially without respect
to chronological units. The scarcity of fish bones at Hajdučka
Vodenica could account for the elevated percentage of do-
mestic animals (c. 20% of identifiable specimens).
Considering only the domestic species, the vast majority be-
long to incipient domestic pig and dog, and only c. 15%
(c. 3% of the total sample) belong to Bos taurus (imported
species).
All of the other units, except for the Lepenski Vir III
(Neolithic) strata present less than 5% of domestic animals
which coincides with Zvelebil’s (1996a) explanation of a
porous agricultural frontier, and serve as evidence of contact
(trade or raiding) rather than of a Neolithic economy. A high
proportion of wild animals and fish in Lepenski Vir III points
to a strong local tradition among the Early and Middle
Neolithic settlers of the site, or to the possibility that
Lepenski Vir was an atypical Neolithic site, a non-sedentary
station for hunting and gathering. Either way, the knowledge
of the region and the know-how of the Mesolithic hunters
was already acquired. Again, it would be of great value to be
able to distinguish between Early (LV IIIa) and Middle (LV
IIIb) Neolithic settlements as the relative importance of
Ovis/Capra and Bos taurus are expected to have changed in
the region from one sub-period to the other.
Bonsall et al. (1997: 56–57) argue that sampling bias
could have played a role in the lack of domestic animals in
Lepenski Vir I and II, which both Chapman (1993) and
Whittle (1996) consider to be synchronous with Neolithic
Starčevo. Although faunal analyses by Bökönyi (1969, 1972,
1974, 1978), performed within the framework of the 1960s
and 1970s, are lacking in detail (see Lyman 1994, for new
approaches to MNI and skeletal elements proportions), they
are consistent throughout the Đerdap material. The very re-
stricted numbers of identifiable specimens in LV I–II (less
than 500) compared to LV III (over 2000) could account for
some bias in species representation (cf. Bonsall et al. 1997),
but not for the total lack of selected exploited species. If
sieving can account for a greater share of fish bones in the
recent excavations at Schela Cladovei (Bartosiewicz et al.
2001), there is no reason to suppose that the overall propor-
tions of mammal skeletons would have been significantly
altered.
Pottery, long held as a tell-tale sign of the introduction of
the Neolithic to Europe, appears in many of the sites within
the Đerdap and has been a major source of discussion. On the
right bank the stratigraphic position of ceramic bearing hori-
zons is reasonably clear in the lower gorge, Ključ and down-
stream from Ključ (Hajdučka Vodenica, Kula, Ajmana,
Velesnica). Most problematic is the situation of Lepenski Vir
and Padina, both of which are located in the upper gorge
(Radovanović 1996a: 282). Since both sites have complex
vertical and horizontal stratigraphy and evidence of other
imported material, but no evidence of change in the econom-
ic structure and ideological world of the local inhabitants
(until the LV III stratum), the appearance of pottery (Table 1)
is recognized as evidence of contact between local foragers

57
 The Iron Gates in Prehistory
and pottery-bearing Neolithic farmers. Pottery appears in all
horizons of Padina B at all sectors, while there is no evidence
of its appearance in Padina A, or A/B. Similarly, Proto-
Lepenski Vir and Lepenski Vir Ia–b (Radovanović phase I-1)
did not contain pottery, while it starts appearing in Lepenski
Vir Ic-e (Srejović 1969), or Radovanović phase I-2 and I-3.
Lepenski Vir II did not contain any pottery, and it appears
again with the Neolithic economy of Lepenski Vir IIIa and
IIIb. Vlasac I–III, akin to the lower gorge settlements, did not
contain any pottery until the fully Neolithic Vlasac IV strat-
um. The appearance of pottery coincides with the distribution
of Pre-Balkan Plateau flint, and argues for greater importance
of trade.
While assuming that pottery is necessarily a Neolithic in-
vention throughout Europe is inherently problematic, there is
no reasonable doubt that pottery was brought into the Đerdap
Mesolithic communities by neighbouring Neolithic people
since it fits well within the Gura Baciului and Starčevo tradi-
tion (Jovanović 1984a, 1987). In terms of the newly proposed
periodization by Tasić (1997, 1998), the ceramics found at
the sites of Lepenski Vir and Padina fall well within the Early
and Middle Neolithic of the central Balkans with no particu-
lar developments that would suggest local invention.
Therefore, the appearance of pottery on these sites can serve
as a marker of contact between farmers and foragers, inde-
pendent of absolute chronology and uncertainties of 14 C
dates. The rationale behind the use of pottery as an inde-
pendent marker of contact is found in its non-local origin that
supposes either trade, transfer of knowledge, or transfer of
people with this particular knowledge into the Đerdap com-
munities. All of these imply the availability of contact, even
where there is no firm evidence of contact itself.
Conceptually, three periods of the Đerdap sequence are
distinguished in the present study:
1. Mesolithic proper, based on a hunting-gathering economy
with increased sedentism.
2. Mesolithic/Neolithic or Contact period, based on the same
economic pattern but with either evidence or possibility of
contact with Neolithic peoples — the ‘availability phase’
of Zvelebil.
3. Neolithic proper, marked by change from a predominantly
foraging economy to a predominantly food producing one.
No unidirectional evolution of economic pattern is assumed.
Regardless of the economic pattern of a particular site, or
phase within the site, once contact with Neolithic peoples in
the region becomes possible, it is no longer regarded as
purely Mesolithic but falls within the Mesolithic/Neolithic
group, signifying the availability of the contact. The eco-
nomic behaviour at any particular site will further determine
whether it is Mesolithic/Neolithic (with little or no change in
the economic domain) or Neolithic (implying increased im-
portance of domesticates).
This approach justifies the preference of relative over ab-
solute chronology. Since our determination of any of the
strata in the four settlements is based on economic patterns,
if burials can be associated reasonably accurately with any of
these occupations then absolute dates do not provide useful
additional information.
The synchronization for the four sites in question (Table 1)
summarizes the chronology and outlines the period when
58
contact with ceramic-producing farming communities in the
region becomes established. Although the appearance of
ceramics and Pre-Balkan Plateau flint does not necessarily
imply the ‘invasion’ or even ‘moving in’ of farming com-
munities into the region, it is evidence of the availability of
contact between Đerdap foragers and Balkan farmers.
Anthropological research
Most of the previous anthropological research in the Iron
Gates gorge was based on comparisons of metric data for the
two sites that had yielded larger numbers of measurable cra-
nial remains, namely Vlasac and Lepenski Vir. One of the
first syntheses came from Nemeskéri and his colleagues
(Nemeskéri 1969, 1978; Nemeskéri & Lengyel 1978;
Nemeskéri & Szathmary 1978a, 1978b, 1978c, 1978d,
1978e). Nemeskéri’s research was very influential and re-
mains one of the most comprehensive studies of the Vlasac
material, albeit within the paradigm of ‘anthropotypology’.
Given recent understanding of problems associated with ‘ra-
cial taxonomy’ (Jacobs et al. 1996), these conclusions could
be either rejected or incorporated into a different under-
standing of population genetics and micro-evolutionary
changes. If his findings are regarded without the taxonomic
labels, several characteristics of the population can be
identified:
1. Great heterogeneity within the local Mesolithic
population.
2. Greater heterogeneity within the male group and greater
homogeneity within the female group.
3. A temporal trend toward gracilization.
4. A trend toward homogenization of the population in later
phases, which Nemeskéri attributes to greater inter-group
gene flow.
Concerning the Lepenski Vir material, in a preliminary
report Nemeskéri (1969) concludes that the Mesolithic strata
contain the finds of Cro-Magnon type, while the Neolithic
strata show most probably three distinct types belonging to
the Mediterranean taxon. It is important to note that the
Mediterranean taxon does not imply Mediterranean origin,
but only a skeletal ‘type’.
The first synthesis of the Đerdap material by Mikić (1980,
1981a, 1981b) follows the general division into A and B types
outlined by Nemeskéri and Szathmary. Mikić developed an
explanatory scheme that accounts for possible processes that
could have induced the change within the series (Mikić
1981a: 104, fig. 1). In his first synthesis of the material,
strongly influenced by Srejović’s hypothesis of indigenous
Lepenski Vir cultivation and domestication, Mikić proposed
that micro-evolutionary trends at Lepenski Vir could account
for gracilization as a consequence of neolithization. He in-
troduced another set of ‘types’ all based on a generalized
‘Mediterranean’ morphology, all of which were derived
through micro-evolutionary processes from the ‘Cro-
Magnon’ type. This micro-evolution occurs within layer II of
Lepenski Vir, and subsequent changes in both shape and size
of the skeletons occur without interruption into the Neolithic,
eventually producing ‘gracile’ and ‘generalized’
Mediterranean types. As with Nemeskéri, the ‘Mediter-
 Mirjana Roksandic: The Mesolithic–Neolithic transition in the Ðerdap
ranean’ in the label does not imply the movement of people
from the Mediterranean into the region. Abandoning the ty-
pological classification in his later works (Mikić 1988),
Mikić argues for local continuity and isolation stressing the
morphological similarities between the Palaeolithic Climente
specimen and Late Mesolithic and even Neolithic individuals
from Lepenski Vir (Mikić 1992: 40).
Padina and Hajdučka Vodenica were very summarily
treated by Živanović and most of the conclusions were based
on archaeological interpretation of the sites. Thus, on the
basis of morphological examination, he singled out seven
skeletons belonging to the Lepenski Vir culture at the site of
Hajdučka Vodenica, attributing all others to the Iron Age
stratum (Živanović 1976). Although Živanović claimed to-
have recognized a number of more robust skeletons belong-
ing to the ‘Padina racial sub-group’ and substantially more
gracile ones belonging to a much later population (Živanović
1976: 124), I was unable to make any such distinction once
the sex was accounted for. This ‘Padina racial sub-group’ is
viewed as autochthonous, different from all other known
groups and best described as the ‘Obercassel type of the
Dinaric race’ (Živanović 1975a, b). Živanović ascribes to
Nemeskéri the conclusion that Lepenski Vir belonged to the
same ‘Proto-Dinaric’ population. However, Nemeskéri dis-
claimed this quote (Nemeskéri & Szathmary 1978b: 180).
All authors agreed on one important aspect of the material
— essential continuity within the region. All of the changes
were attributed to micro-evolution towards more gracile
forms with intensification of contacts and admixture at the
time of the Vlasac II/Lepenski Vir I phases. None of the au-
thors perceived any abrupt change in the Neolithic popula-
tions of the region. Schwidetzky & Mikić (1988) argued that
gracilization cannot be assumed to coincide with Neolithic
adaptation. They reach the conclusion that the high rates of
change support the micro-evolutionary processes in the Iron
Gates rather than abrupt change in population (Schwidetzky
& Mikić 1988: 117). Quite a different conclusion was
reached by Menk (Menk & Nemeskéri 1989). While he also
claimed a sharp decrease in robusticity between the Terminal
Mesolithic and Early Neolithic of the series (see Roksandic
1999, for a critique) as well as considerable change in shape,
the author concluded that the change could not be explained
by local evolution, but rather by a progressive replacement of
the population. Menk applied PCA to cranial and postcranial
measurements provided by Nemeskéri. As Menk notes, the
Lepenski Vir Neolithic sample “fractions into three parts with
a remarkable gap in the central part of its area” (Menk &
Nemeskéri 1989: 534). In itself, coupled with a small sample
size and a problematic pooling of the sexes, this finding can
invalidate the analysis, since the central value of the
Neolithic Lepenski Vir population is derived from the strong
dissimilarity of the individuals of which it is made.
A look at the PCA graphic output for components 1, 2 and
3 offer by Menk & Nemeskéri (1989) shows strong differen-
tiation along axis 1 (corresponding to size) for Lepenski Vir
Mesolithic and Lepenski Vir Neolithic. But, the same is true
for the distance between the Vlasac I and I? and Lepenski Vir
Mesolithic. However, the distance between the two on the 2,
3 (measuring some form of shape differentiation) axis is
small. It is, in effect, much smaller than between Vlasac I and
59
Vlasac I?. Furthermore, the Starčevo-Criş Neolithic sample
seems to be less removed on the size axis from Mesolithic
Lepenski Vir and practically identical with it on the shape
axis (axis 2, 3). The only actual difference between different
subsamples on the ‘shape’ axis is between Vlasac I and
Vlasac II, III that show as much difference on the size axis as
the Lepenski Vir Mesolithic and Neolithic samples. The con-
clusion, although not necessarily wrong, cannot be demon-
strated on the basis of the published results; however, the
article offers an interesting view on Mesolithic heterogeneity
that is in accordance with all other published results.
Material and methods
Since basic demographic data are given elsewhere
(Roksandic 1999; Jackes et al., this volume), only the period
designation for each of the burials is presented here (Table 2).
Padina
According to the chronology of Padina and the division of the
site strata into Mesolithic and Mesolithic/Neolithic contact,
skeletal remains were assigned to either of the two periods
according to the site documentation, superpositioning of cer-
tain features, and Radovanović’s (1996a) analysis of the
burials. It was not possible to assign four individuals from the
disturbed unidentified graves to either period with acertainty.
Lepenski Vir
According to the presented chronology of the site and divi-
sion of strata into the Mesolithic, Mesolithic/Neolithic con-
tact and Neolithic periods, the skeletal remains were assigned
to particular phases according to Radovanović’s analysis of
Mesolithic burials (Radovanović 1996a: 174–189) and site
documentation (kindly provided by Prof. Srejović, for the
Neolithic burials (Antunović 1987).
Vlasac
According to the building horizons, Srejović & Letica (1978)
identified three phases of the Mesolithic settlement (Vlasac I,
II and III) and one of the Neolithic (Vlasac IV). Since no hu-
man skeletal remains were associated with the latter, it will
not be discussed in detail. Radovanović has observed signi-
ficant changes in burial practices over time and has proposed
a division of formal disposal areas into two chronological
phases. Based on the published material, as well as field
documentation (kindly provided by the principal investigat-
or, the late Prof. D. Srejović) she was able to distinguish an
important change in burial practices that began to occur at the
end of Vlasac I but was certainly present in the Vlasac II
phase (Radovanović 1996a: 187–218, appendix II).
However, both Vlasac I and II would belong to the Mesolithic
pre-contact period and only Vlasac III would chronologically
belong to the period when contact with the Neolithic popula-
tions was possible. All of the cases where Radovanović could
not distinguish between the Vlasac II and III burials are
treated as Mesolithic/Neolithic contact. However, a separate
test was run with these individuals included in the Mesolithic
group since contacts between Lepenski Vir and the sur-
rounding farming population(s) is least attested in Vlasac of
 The Iron Gates in Prehistory

Table 2: Classification of burials from Padina, Lepenski Vir, Vlasac, and Hajdučka Vodenica.

PADINA
Mesolithic:
single burials: 1; 18b; 21; 22; 39.
double burials: 12; 12(1); 14; 14(1); 17; 17(1); 23; 23(1).
multiple burials: 15(15-16a); 15-16(15-16a); 16(15-16a); 16a(15-16a); 16(1)(15-16a).
Mesolithic/Neolithic Contact:
single burials: 1a; 3; 6; 7; 8; 9; 10; 11; 13; 18; 19; 24; 26; 26a; 27; 28; 29; 30.
double burials: 2, 2(1); 6a, 6a(1); 25, 25(1).
multiple burials 4/4+5+5a/, 5/4+5+5a/, 5a/4+5+5a/, 5a(1)/4+5+5a/; 20, 20(1), 20(2).

LEPENSKI VIR
Mesolithic:
single burials: 3; 21; 22; 46; 60; 61; 67; 69; 110; 111; 112; 113; 117; 118; 119; 120; 121; 132; 133.
double burials: 50, 50(1); 64, 64(1); 99, 99(1); 102, 102(1); 109,109a.
multiple burials: 101, 101(1), 101(2).
Mesolithic/Neolithic Contact:
single burials: 11; 12; 15; 16; 17; 23; 26; 28; 46; 68; 70; 90; 91; 92; 94; 95; 96; 97; 100; 103; 104; 105; 115; 116; 122; 126; 127; 128; 129; 130;
131; 134.
double burials: 7a,7b; 13, 13(1); 14, 14(1); 45a, 45b; 63, 63(1); 74, 74(1); 93, 93(1); 98, 98a; 99, 99(1); 106, 106(1); 107, 107(1); 108, 108(1);
114, 114(1); 123, 123(1); 124, 124(1); 125, 125(1).
multiple burials 54a, 54b, 54c, 54d, 54d(1), 54e.
Neolithic:
single burials: 1; 4; 5; 6; 8; 9; 20; 31a; 33; 35; 37; 38; 39; 42a; 43; 44; 48; 51; 53; 56; 57; 59; 66; 71; 88.
double burials: 18, 18(1); 19, 19a; 32a, 32c; 52, 52a; 55a, 55b; 73, 73(1).
multiple burials: 83a, 83a(1), 83b.
Period uncertain:
single burials: 2; 10; 24; 25; 29; 36; 40; 41; 49; 58; 62; 65; 72; 75; 76; 78; 86.
double burials: 77, 77(1); 80; 81; 82; 84, 84(1).
multiple burials: 27a (27a+e), 27(27b), 27 (27C), 27 (27d), 27(1), 27 (27f), 27(2); 34a, 34b, 34c; 79a, 79b, 79c; 85, 85a, 85b/85(1)/, 85b; 87,
87(1), 87(2), 87(3), 87(4); 89a, 89b, 89b(1).

VLASAC
Mesolithic:
single burials: 7; 8; 10, 11; 13, 20; 25; 28, 30; 31; 32; 33; 34; 37; 38; 39; 40; 41; 44; 59; 61; 63; 68; 72; 79; 81.
double burials: 9, 9(1); 12a, 12b; 19, 19a; 35, 35a; 42a, 42b; 47, 47a; 48, 48(1); 53, 53(1); 56, 56(1); 57, 57(1); 60, 60(1); 62, 62(1); 66a, 66a(1);
80, 80a; 84, 84(1).
multiple burials: 4a, 4b, 4b(1); 5, 5(1), 5(2); 6, 6a+6(1), 6(2); 18a, 18b, 18c; 21, 21(1), 21(2); 36, 36(1), 36(2); 45, 45(1), 45a+45(1); 49, 49(1),
49(2); 50, 50(1)+50a(1), 50a, 50a(2)+50b(1), 50a(3), 50b; 51, 51a, 51b, 51+51a+51b(1), 51+51a+51b(2); 52, 52(1), 52(2), 52(3);
54, 54(1), 54(2); 55, 55(1), 55(2), 55(3); 58, 58a, 58b; 64, 64a, 64b; 65, 65(1), 65a; 67, 67(1), 67(2), 67(3); 82, 82(1)+ 82b,
82(2)+82c, 82(3), 82(4)+82a; 83, 83a+83(1), 83(2), 83(3).
Mesolithic/Neolithic Contact:
single burials: 2; 14; 16; 17; 22; 23; 24; 43; 46; 75; 76.
double burials: 15, 15(1); 27, 27(1); 70, 70(1); 71, 71(1); 73, 73(1); 77, 77(1); 78, 78a.
multiple burials 29, 29(1), 29a; 69, 69(2), 69a, 69(1)+69a(1); 74, 74(1), 74(2).
Period uncertain:
single burials: 1; 3.
double burials: 26, 26(1).

HAJDUČKA VODENICA
Mesolithic/Neolithic Contact:
missing burials: 1; 2; 3; 4; 5; 6; 7; 9; 10; 12.
single burials: 8; 11; 16; 21; 22; 30; 31.
double burials: 14, 14(1); 29, 29(1); 33, 33(1).
multiple burials: 15 m, 15 s, 15s(1); 17/17-20(3)/, 18/17-20/, 19/17-20(2)/, 20/17-20(4)/, 17-20(1), 20, 20(1); 23+24+25(1), 23+24+25(2),
23+24+25(3), 23+24+25(4), 23+24+25(5); 26+27+28(1), 26+27+28(2), 26+27+28(3), 26+27+28(4); 32, 32(1), 32(2).

60
 Mirjana Roksandic: The Mesolithic–Neolithic transition in the Ðerdap
all of the sites; no pottery was found in these layers, and all
of the Pre-Balkan Plateau flint was explained as intrusive
(Srejović & Letica 1978; Kozłowski & Kozłowski 1982).
Hajdučka Vodenica
Within the proposed framework, all of the burials from
Hajdučka Vodenica would fit within the Mesolithic/Neolithic
contact period. Unlike the other sites, the graves show re-
markable uniformity in the burial position and even orienta-
tion. This uniformity, as well as the restricted and well-
respected space for the burials, is well in accord with the
(comparative to other sites) short time-span of the necropolis
and of the chamber tomb.
Analyses
Several reasons have guided preference towards non-metric
traits: 1) They have been understudied in the present series.
2) They are potentially more appropriate for material of un-
equal preservation, as we are discussing population frequen-
cies and not individual characteristics. This would be in-
creasingly important in case of differential treatment for
different groups. 3) Although there is an ongoing discussion
among anthropologists about the ability of different analyses
to establish relationships among skeletal samples, a paired
study of analyses of diverse non-metric traits and cranial
measurements demonstrates that the former are more power-
ful in this respect (Jackes et al. 1997).
Non-metric traits do not follow simple Mendelian patterns
of inheritance. Hauser & De Stefano (1989: 5–10) accept the
model of ‘threshold character’ proposed by Falconer (1965)
in relation to pathological conditions, as the underlying the-
oretical basis for all of the characters. The liability to develop
a trait is normally distributed, and depending on the position
of the individual’s inherited tendency to develop the charac-
ter relative to the threshold, the character may or may not be
expressed. The genes involved are multiple genes with small
additive frequencies. Threshold models permit a number of
other environmental and developmental factors to be in-
cluded in the determination of the trait’s expression and allow
for the observed gradients in some of the traits. An individual
situated just below the threshold in one environment may be
pushed over it in another (Hauser & De Stefano 1989: 7)
which reinforces the population specific character of the fre-
quencies of trait expressions. The proportion of total variance
attributed to the additive effects of genes, known as the her-
itability of the trait, was calculated from the study of the fre-
quency of the condition in a series of related individuals of
known sex and age (Sjøvold 1984) and was shown to be sig-
nificant although low. However, any attempt to relate indi-
viduals within a series to one another failed to perform, be-
cause of this underlying complex genetic basis of the traits
(e.g. Crubezy 1991).
Choice of characters
The choice is based on a number of characters for which low
environmental influences were suggested by Saunders (1989)
61
and Buikstra & Ubelaker (1994) as well as some additional
characters observed during the first field season on the Đer-
dap material itself (sutura squamo-mastoidea and tubercu-
lum marginale).
The original list comprised 29 paired (that could be recor-
ded bilaterally) traits and eight axial (that had only sagittal
expression) traits for the skull, and 21 paired and one sagittal
character for the postcranial skeleton. Procedures for record-
ing followed Standards (Buikstra & Ubelaker 1994) where
applicable, and Hauser & De Stefano (1989) in all other
cases. Most of the traits were recorded on a scale rather than
present or absent in order to allow more flexibility in the final
analysis. However, they are treated as discrete in the statist-
ical analysis. Since Buikstra & Ubelaker (1994) offer very
little in terms of postcranial non-metric traits, a list of post-
cranial traits adapted from Czarnetski (1972b), Czarnetski et
al. (1985) and Saunders (1978) was added.
Only adult skeletons from all of sites were taken into con-
sideration since the occurrence of non-metric traits in
subadults can be ambiguous. This has reduced the total num-
ber of individuals examined from 438 (MNI) from all four
sites to 259 (MNI) adult individuals.
Since lack of intercorrelation and correlation with age and
sex cannot be assumed, trait frequencies were compared with
regards to age, sex and each other. Since the chance of purely
random significant correlation occurring on the tested
samples becomes greater with the number of correlation tests
performed, the first step in the procedure was to remove all of
the variables that could not be observed (both as absent or
present) on at least 10% of the examined adult sample. This
resulted in the elimination of seven cranial and nine postcra-
nial traits. As, depending on the sample size, a small absolute
number of occurrences (less than 5) can produce biased res-
ults, six cranial and four postcranial variables were excluded
before their frequencies within subpopulations were ex-
amined. Of the remaining variables another group of charac-
ters, those with low overall frequencies, were checked
against chronological and spatial subpopulations in order to
assess their overall variability. (Sjøvold 1977; Molto 1983:
113). Sjøvold (1977) recognizes two types of low variability
traits: those that have reached fixation in every sample stud-
ied, and others that have very low uniform incidence in any
set of population samples. Rather than using the χ2 or Fisher’s
exact test between samples to exclude the variables for which
the significant difference is not obtained in at least one pair-
wise comparison (as suggested by Sjøvold 1977), the empir-
ical results of Molto (1983: 114) for an Ontario Iroquois
sample were applied. In Molto’s study (1983: 115) the largest
range of frequencies among the traits that had low variability
was 7.1 (for example, 0.0% in one sample to 7.1% in anoth-
er). Molto has excluded these traits from further considera-
tion and kept those with minimum range in any of the
samples equal or greater than 10% (e.g. 21% in one and 31%
in another). By using this observation as a rule of thumb in
the present study, rather than increasing the possibility of
finding statistical significance (where there might be none)
through a large number of tests performed, three paired and
one sagittal traits were determined as having low overall
variability and excluded.
Three variables were excluded because of the high inter-
 The Iron Gates in Prehistory

Table 3. Traits used in various combinations in the final analyses. Sides pooled.

Trait name – common Latin Code Trait no.

Marginal tubercle tuberculum marginale (TZ) 1

Squamomastoid suture sutura squamomastoidea (SSM) 2
Supraorbital notch incisura supraorbitalis (SN) 3
Supraorbital foramen supraorbital foramen (SF) 4
Zygomatico facial foramen foramen zygomatico-faciale (ZFF) 5
Parietal foramen foramen parietale presens (PF) 6
Coronal ossicle (COR) 7
Lambda ossicle (LAM) 8
Auditory torus torus auditivus (AUDT) 10
Mental foramen foramne mentale (MEF) 11
Maxillary torus torus maxilaris (MAX) 12
Mylohyoid bridge ponticulus mylohyoideus (MHBD) 13
Septal aperture perforatio fossae olecranii (PFO) 15
Third trochanter trochanter tertius (TT) 17
Apical bone (APIC) 23
Inca bone os inca (INCA) 24
Palatine torus torus palatinus (PAL) 26

or intra-observer error calculated from the observations re- Assumption 1

corded in 1996 and those recorded in the 1998 field season on
a randomly chosen subsample.
The Fisher exact test of significance was performed on the
rest of the traits in order to check for possible correlation with
sex. Only one variable pair was found to be potentially cor-
related with sex: the mandibular torus. In the published liter-
ature, there is no definite pattern of preference according to
sex, but the general trend of predominance in females is re-
ported (Hauser & De Stefano 1989, and quoted literature).
The trait was excluded from further consideration.
Tests of side correlation were performed on all of the
paired traits. The ones that showed correlation were ex-
cluded. In doing so the risk of increasing the probability of
false correlation was ignored, as potential benefits in in-
creasing the number of observations by pooling the sides
outweighed the concerns.
Traits that had less than 9 observations on left and right
side combined in any of the subsamples (see further discus-
sion on the sides recording of the traits) were also excluded.
Only 17 traits were used in different combination in the ana-
lyses (Table 3). Since a more restricted number of appropri-
ately chosen variables can distinguish better between popu-
lations (Molto 1983; Krenzer 1996) this reduction was
deemed beneficial. Trait description and scoring procedures
are detailed in Roksandic (1999).
Three assumptions were made: 1) Non-metric traits have a
strong genetic basis. 2) The sample is representative of the
populations we are trying to compare. 3) The outlier is a
substantially removed population studied by the same
methods.
Although the influence of environment (occupation, habitat,
nutrition) cannot be excluded for most traits, assumption is
reasonably well founded in the research on the genetics of
non-metric traits. We are examining the population structure
and not the genetic make-up of the individuals, and therefore
even if the influence of environment on the expression of
traits (due to their threshold character) cannot be disregarded,
the validity of the population comparisons is not reduced.
Assumption 2
The assumption of the representativeness of the sample that
is examined can rarely be proven in archaeological samples.
Burial samples are often (if not always) biased as not every-
one gets a formal burial, and mode of deposition is strongly
dependent on the social persona (Masset 1993; Roksandic
2002, and quoted literature). Furthermore, the direction of
bias can be discerned only rarely, after a thorough study of all
of the social, biological and taphonomic aspects of the skel-
eton. Although we can not assume the representativeness of
the sample for the purpose of studying mortuary ritual and its
social implications, there is a hope that the populational bio-
logy (or the genetic make-up) of a changing population will
still be represented adequately. Only under the circumstances
of a completely different burial ritual for the local and the
supposed incoming population, that would obliterate one or
the other from our sample, the assumption of the represent-
ativeness could not be sustained. Since burial ritual in the
Mesolithic varies greatly and becomes more or less canon-
ized only in the Lepenski Vir IIIb period which belongs to the
Middle Neolithic (Antunović 1990), and since inhumation is
a demonstrated pattern for both of the periods, there is no

62
 Mirjana Roksandic: The Mesolithic–Neolithic transition in the Ðerdap
reason to suspect total obliteration of one group or another in
the current sample.
Assumption 3
Inclusion of Franzhausen I, a Bronze Age population from
Austria (Wiltschke-Schrotta 1992), was based on a number
of variables that were recorded in both studies, and the sys-
tem of recording that followed the same general procedures
(Czarnetzki 1972a, 1972b, 1972c; Hauser & De Stefano
1989; Buikstra & Ubelaker 1994). Also, the site is suffi-
ciently removed chrono-spatially, but still within the same
general geographical area. Importantly, Wiltschke-Schrotta
has recorded sides separately and has presented the side data
in a manner that made it possible to add them up without
problems and obtain a methodologically comparable sample.
Statistical analysis of non-metric traits
Following the publication of Berry & Berry’s (1967) article
on population studies based on cranial non-metric traits, the
Smith–Grewal statistic was commonly used for calculation
of average distances between sample populations (Saunders
1989: 98). Smith’s Mean Measure of Divergence (MMD) has
further been investigated and developed by Sjøvold (1977)
and serves, with minor modifications, as the major statistic
used for examining the inter-sample distance. Stated simply,
the Mean Measure of Divergence (MMD) is the summed di-
vergence between two samples, divided by the number of
traits included in the analysis.
For this type of analysis, proportions of the sample exhib-
iting a trait are given as theta (θ) values symmetrical around
0, such that the incidence of 50% equals a theta of zero
(Jackes et al. 1997: 645). Sjøvold has determined that the
Anscombe formula is the best modification for calculation of
θ, most accurately transforming the incidences of traits and
stabilizing the variance well, except in cases when incidences
are extremely high or low (Sjøvold 1977). If the sample sizes
are small and incidences are accordingly low, the
Freeman–Tukey transformation is judged to provide some-
what better variance stabilization than Anscombe (Jackes et
al. 1997: 645).
The actual formulae used in this study were taken from
Jackes et al. (1997). The programming as well as the running
of some of the data sets was done by Mary Jackes on the
QuattroPro spreadsheet program at the University of Alberta
in Edmonton. Others were run by myself on the Microsoft
Excel program provided by Jackes.
In analyzing the distance between the populations, it is
possible to use MMD, Z and ‘stand’, all of which are repres-
ented on the tables. The choice here is based on the fact that
Z is a way of standardizing the MMD in case of unequal
sample sizes and is more explicitly correlated with both
MMD and DI, the latter being the measure of significance of
the distance (Jackes & Gao, forthcoming). Therefore Z
matrices are reproduced with the results.
Since dendrograms can link the samples in only one dir-
ection, a spatial distance between different samples can be
better appreciated through multidimensional scaling (MDS)
plots. To produce MDS plots, the same distance matrices as
63
for dendrograms were used. Monotonic scaling, Kruskal
Stress (measuring how well the curve fits all the points), and
two dimensions because of the small number of points plot-
ted were found the most appropriate. In each of the MDS plot
figures, captions include the scores for the two dimensions,
the Kruskal stress of final configuration — which should be
less than 0.1 in a ‘good fit’ (Wilkinson et al. 1996: 667) —
and proportion of variance expressed.
Of further methodlogical concern is the treatment of sides.
I have opted for adding sides and treating the material by
elements and not by individuals. This method is preferred by
Ossenberg (1981). She argues that the observed correlation
between the intensity of trait incidence and the proportion of
bilateral expression reflects genetic factors since an indi-
vidual expressing a trait bilaterally has a stronger dose of trait
positive alleles than an individual with unilateral expression
of the trait. Therefore, computing the frequency of a discrete
trait on the basis of pooled sides quantifies the genetic po-
tential in the population better than does the individual count.
This way of recording has the benefit of expressing the un-
derlying threshold character of the traits as it takes into ac-
count the total genetic potential for the trait expression within
a population. It also increases the sample sizes in many cases
without violating the biological bases of the trait expression.
Accordingly, sides were added in the following manner:
k/N L + k/N R = k/N or (2/5 + 3/8 = 5/13)
where k is the number of instances in which the trait was re-
corded as present, while N is the total number of possible
observations.
Subsamples based on sites with Franzhausen I
The results of these analyses are presented in Tables 4–6 and
Figure 1.
A glance at Table 6 shows that Hajdučka Vodenica and
Padina are virtually identical, while all other sites seem to be
significantly different from each other. Franzhausen, as ex-
pected, is the most removed from other sites, while Padina
and Lepenski Vir and Padina and Vlasac show the next most
significant difference.
The multidimensional scaling plot also shows the clear
grouping of the Iron Gates gorge sites against the more re-
moved Franzhausen I site. The general pattern is that of het-
erogeneity (Fig. 1).
Subsamples based on chronology
The results of these analyses are presented in Tables 7–9 and
Figure 2.
As described above, the basis for distinguishing the units
is provided by the evidence of economic behaviour and
evidence of contact with peoples with different economic
patterns. Mesolithic refers to the strata within any of the sites
where there is no evidence of contact. The
Mesolithic/Neolithic is the period when contact with farming
communities in the region becomes possible, while Neolithic
is, primarily, characterized by greater importance of domest-
icates in the economic base (>5%). Evidence of adoption of
 The Iron Gates in Prehistory

Table 4. ‘k’ and ‘N’ values used in the analysis of sites with Franzhausen I.

Trait 1 HV 2 L. Vir 3 Padina 4 Vlasac 5 FR.I

no. k N k N k N k N k
2 4 9 14 46 13 26 4 62 8 588
6 7 10 31 60 23 28 44 77 211 425
10 7 22 5 48 17 33 27 76 130 638
11 1 21 2 60 3 40 6 82 46 530
12 1 9 6 41 1 10 15 59 8 365
13 2 14 12 18 4 25 16 32 28 446
17 0 10 12 29 4 13 13 35 116 318

Table 5. The output of the statistical analysis of sites with Franzhausen I.

Site 1 Site 2 Site 1 Site 2 mmd sd Stand Total Z di S formula

name name FT FT FT N FT FT FT
1 2 HVmn LV 0.2954 0.0590 5.0063 57 3.4171 0.1774 24.6591 ft
1 3 HVmn P 0.0233 0.0724 0.3214 39 0.2896 -0.1216 7.5861 ft
1 4 HVmn V 0.2678 0.0547 4.8985 74 3.0388 0.1585 22.0737 ft
1 5 HVmn FRI 0.3548 0.0460 7.7127 486 4.2742 0.2628 31.0451 ft
2 3 LV P 0.2900 0.0429 6.7659 68 5.3277 0.2043 39.9015 ft
2 4 LV V 0.0946 0.0263 3.5981 104 3.5686 0.0420 25.7342 ft
2 5 LV FRI 0.4219 0.0169 24.9836 516 9.5513 0.3881 86.5518 ft
3 4 P V 0.2290 0.0385 5.9491 85 4.9147 0.1520 36.2973 ft
3 5 P FRI 0.3769 0.0299 12.5989 498 8.7299 0.3171 76.0812 ft
4 5 V FRI 0.2554 0.0118 21.6578 533 8.9086 0.2319 78.3023 ft

Table 6. Matrix of Z values for sites with Franzhausen I. Significant relationships are outlined in bold.

Z(ft) 1 2 3 4 5
matrix H. Vodenica L. Vir Padina Vlasac FR.I

1 0
2 3.4171 0
3 0.2896 5.3277 0
4
3.0388 3.5686 4.9147 0
5
4.2742 9.5513 8.7299 8.9086 0

Dimensions 1 2
HV 0.16 -0.14
LV 1.00 0.39
PA 0.12 -1.00
VL 0.23 0.69
FR -1.51 0.05

Kruskal Stress of final configuration: 0.0270

Proportion of variance: 0.9941

Figure 1. Multidimensional scaling for the sites with Franzhausen I. Based on dissimilarity matrix.

64
 Mirjana Roksandic: The Mesolithic–Neolithic transition in the Ðerdap

Table 7. ‘k’ and ‘N’ values for traits used in the analysis of chronological units.

Trait Mesolithic 1 Meso/Neo 2 Neolithic 3

no. k N k N k N
1 30 33 20 45 1 8
2 10 55 16 66 8 14
3 40 60 43 77 6 14
4 9 66 14 65 2 12
5 23 42 29 43 4 9
6 42 67 48 76 8 18
7 0 62 4 70 0 18
8 16 62 23 73 5 16
10 30 50 26 85 0 15
11 6 81 4 82 1 18
12 12 46 7 51 2 13
23 8 33 3 32 0 10

Table 8. The output of the analysis of chronological units.

Site 1 Site 2 Site 1 Site 2 mmd sd Stand Total Z di S formula

name name FT FT FT N FT FT FT
1 2 Meso MN 0.1298 0.0155 8.3884 119 4.8968 0.0988 46.9740 ft
1 3 Meso LVn 0.4667 0.0398 11.7260 69 6.4826 0.3871 63.6011 ft
2 3 MN LVn 0.1017 0.0388 2.6230 78 2.4289 0.0242 26.0986 ft

Table 9. Matrix of Z values for chronological units. Significant relationships are outlined in bold.

Z(ft) 1 2 3
matrix Mesolithic Meso/Neo Neolithic

1 0
2 4.8968 0
3 6.4826 2.4289 0

Dimensions 1 2
Meso -1.34 0.00
Contact 0.27 0.00
Neo 1.07 0.00

Kruskal Stress of final configuration: 0.00

Proportion of variance: 1.00

Figure 2. a) Dendrogram showing internal relationship of chronological units.Derived from dissimilarity matrix, Euclidean
distance, complete linkage. b) MDS plot of chronological units based on dissimilarity matrix.

65
 The Iron Gates in Prehistory
cultural elements of the surrounding farmers of the
Balkano–Anatolian and Balkano–Carpathian basin (ENCB
and MNCB of Tasić 1998) although considered, was not
taken as sufficient for determining the find as Neolithic.
Clustering of the Mesolithic/Neolithic (Contact) period
with the Neolithic period contradicts the wave of advance
model for the neolithization of this region. In order to
demonstrate the spread of Neolithic farmers themselves, and
not only their domesticates and/or knowledge, the result
should show a slight to non-existent change in the Contact
period (as some exchange of genes could be expected) and an
abrupt change with the advent of the Neolithic. This pattern
would argue for an insurgence of people with different ge-
netic make-up who brought about the change in the economic
base (as proposed by Ammerman & Cavalli Sforza 1971;
1984; Cavalli Sforza 1996).
All of the relationships are significant according to the ‘di’
value (Table 7). It is noteworthy that the distance between
Mesolithic and Contact is more than twice the distance
between Contact and Neolithic. This is even more suggest-
ively shown by the multidimensional scaling plot.
If there is, indeed, an exchange of genes as well as goods
at the time of the first contact, it does not destabilize the
Mesolithic society and ideology. Even more importantly, as
the basis of subsistence remains hunting, gathering and fish-
ing, this supposed exchange of genes does not bring about a
fully developed farming economy.
It is important to note that a certain amount of change in
the genetic make-up, as evidenced in the non-metric traits,
would be expected due to secular trends. However, for secu-
lar trends to be the only source of change, the distances
between different periods would need to be approximately
the same.
The diagram in Figure 2b strongly suggests that, apart
from the obvious secular trend reflected in the alignment of
the units, a greater amount of change happens between
Mesolithic and Contact periods. The introduction of an out-
lier in the next analysis is aimed at clarifying how important
this difference was in the amount of genetic change.
Subsamples based on chronology with
Franzhausen I
These results are presented in Tables 10–12 and Figure 3. As
is evident from Table 10, the difference between Contact and
Neolithic periods ceases to be significant when an outlier is
introduced (shown by a negative ‘di’ value). Furthermore,
distances between Mesolithic and Contact, and Contact and
Neolithic, according to the Z statistic, become almost equal.
The introduction of Franzhausen shows that the distances
on the local scale become less obvious and that in general
they follow the secular trend. While the Contact period is al-
most equidistant from the Mesolithic and Neolithic periods,
the two are differently positioned with respect to
Franzhausen (Bronze Age). In other words, the Contact,
Neolithic and Franzhausen are to be found on the same axis,
while the Mesolithic period forms a different pattern and is
situated on a different axis together with the Contact period.
66
Subsamples based on combined chronology
and sites
Subsamples that respected both sites and chronological de-
termination were analyzed in order to provide a more fine-
grained understanding of the relationship among them
(Tables 13–18; Figs 4–5). The Padina sample, when divided
into Mesolithic and Mesolithic/Neolithic, made comparisons
almost impossible because of the small numbers of observa-
tions in almost all variables. Since Padina clusters consist-
ently with Hajdučka Vodenica, and since most of the indi-
viduals from Padina belonged to the same chronological unit
(Mesolithic/Neolithic) they were assigned to the Hajdučka
Vodenica subsample and thus form the HVPmn (Hajdučka
Vodenica–Padina Mesolithic/Neolithic). The remaining indi-
viduals from Padina that belong to the Mesolithic were as-
signed to the Lepenski Vir Mesolithic subsample (forming
LVPm — Lepenski Vir–Padina Mesolithic). This was done in
order to strengthen the Mesolithic sample of Lepenski Vir
after a careful examination of frequencies. Although the fre-
quencies do not show substantial differences, this should be
kept in mind in the analyses and interpretation.
According to the distance matrix produced here, the most
similar are the Hajdučka Vodenica/Padina Contact group and
the Lepenski Vir Neolithic. This points to a strong continuity
between the two periods and is important because this con-
tinuity is not site specific. Along these lines is the similarity
between Lepenski Vir Contact and Lepenski Vir Neolithic.
But the Hajdučka Vodenica/Padina group also shows little
difference from the Lepenski Vir Mesolithic subsample. At
Lepenski Vir itself, the change is pronounced at the time of
Contact and very restricted between Contact and Neolithic.
It is interesting to note that Lepenski Vir Mesolithic is
most different from Vlasac Mesolithic and Lepenski Vir
Contact and less, but still significantly, different from
Lepenski Vir Neolithic. Lepenski Vir Neolithic shows little
difference from Vlasac Mesolithic and somewhat more from
Vlasac Contact.
Vlasac Mesolithic and Mesolithic/Neolithic appear to be
virtually identical along the second dimension and different
along the first dimension where they are pooled by similarity
to Lepenski Vir Neolithic. The general outline argues for
temporal trend and continuity within the sample with greater
variability in the Contact period.
According to the ‘di’ values (Table 14), several distances
are non-significant: Lepenski Vir Contact shows little dis-
tance from the Neolithic period at the same site, and there is
a clear continuity between Mesolithic and Contact at Vlasac.
Such a strong association raises doubts that the chronological
determination based on stratigraphy and stylistic analysis
(Radovanović 1996a) could be incorrect. In order to check if
this grouping is indeed evidence of continuity, and not a
consequence of unreliable separation into different chrono-
logical groups, a different chronological division of the
sample, based on Srejović’s (Srejović & Letica 1978) de-
terminations was run through the same procedures. None of
the significant relationships changed and, more importantly,
Vlasac retains practically the same non-difference for its two
chronological subsamples. Lepenski Vir Neolithic is
equidistant from Vlasac Mesolithic and Vlasac Contact.
 Mirjana Roksandic: The Mesolithic–Neolithic transition in the Ðerdap

Table 10. ‘k’ and ‘N’ values for traits. Analysis of chronological units with Franzhausen I.

Trait Mesolithic 1 Meso/Neo 2 Neolithic 3 Franz.I 4

no. k N k N k N k N
2 10 55 16 66 8 14 8 588
5 23 42 29 43 4 9 90 451
6 42 67 48 76 8 18 211 425
8 16 62 23 73 5 16 27 190
10 30 50 26 85 0 15 130 638
11 6 81 4 82 1 18 46 530
12 12 46 7 51 2 13 8 365
13 16 39 15 42 1 5 28 446
23 8 33 3 32 0 10 16 194

Table 11. The output of the analysis of chronological units with Franzhausen I.

Site 1 Site 2 Site 1 Site 2 mmd sdFT Stand Total Z di S formula

name name FT FT N FT FT FT
1 2 Meso MN 0.0411 0.0187 2.1905 114 2.0405 0.0036 18.9949 ft
1 3 Meso N 0.3244 0.0530 6.1157 66 5.1131 0.2183 42.6539 ft
1 4 Meso FrI 0.3561 0.0111 31.9687 478 12.7567 0.3339 142.4636 ft
2 3 MN N 0.1024 0.0522 1.9597 74 2.4522 -0.0021 21.6171 ft
2 4 MN FrI 0.2874 0.0103 27.8134 486 12.0167 0.2668 130.2460 ft
4 3 FrI N 0.3327 0.0448 7.4201 438 5.8392 0.2430 49.6241 ft

Table 12. Matrix of Z values for chronological units with Franzhausen I. Significant
relationships are outlined in bold.
Z(ft) 1 2 3 4
matrix Mesolithic Meso/Neo Neolithic Franz.I

1 0.0000
2 2.0405 0.0000
3 5.1131 2.4522 0.0000
4 12.7567 12.0167 5.8392 0.0000

Dimensions 1 2
Meso -0.88 0.48
Contact 0.75 -0.35
Neo 0.15 -0.27
Franz. I 1.48 0.14

Kruskal Stress of final configuration: 0.00

Proportion of variance: 1.00

Figure 3. Multidimensional scaling plots for chonological units with Franzhausen I. Based on dissimilarity matrix.

67
 The Iron Gates in Prehistory

Table 13. ‘k’ and ‘N’ values for traits used in the analysis of subsamples based on site/chronology combination.

Trait HVP(mn) 1 LVP(m) 2 LV(mn) 3 LV(n) 4 V(m) 5 V(mn) 6

no. k N k N k N k N k N k N
1 6 11 12 15 2 11 1 8 18 28 12 23
2 11 24 7 19 4 15 8 14 3 36 1 26
3 16 26 20 22 8 19 6 14 20 48 19 32
4 4 19 2 20 5 18 2 12 7 46 5 28
5 7 12 10 12 5 9 4 9 13 30 17 22
6 17 24 19 22 12 22 8 18 23 45 19 30
7 0 17 0 20 2 18 0 18 0 42 2 25
8 4 17 2 18 4 18 5 16 14 44 15 28
11 4 34 0 27 0 20 1 18 6 54 0 28
13 5 24 4 19 6 6 1 5 12 20 4 26
23 1 9 2 9 0 9 0 10 6 24 2 14
24 1 11 1 10 2 11 0 9 0 28 0 15

Table 14. The output of the analysis for site/chronology combination.

Site 1 Site 2 mmd sd Stand Total Z di S formula

FT FT FT N FT FT FT
1 2 HVPmn LVPm 0.0205 0.0520 0.3935 37 1.0880 -0.0836 17.3097 ft
1 3 HVPmn LVmn 0.2698 0.0584 4.6233 34 3.0129 0.1531 30.4878 ft
1 4 HVPmn LVn -0.0104 0.0630 -0.1647 32 -0.1692 -0.1364 10.7029 ft
1 5 HVPmn Vm 0.0959 0.0381 2.5196 56 2.6350 0.0198 27.6085 ft
1 6 HVPmn Vmn 0.0875 0.0437 2.0041 44 2.5068 0.0002 26.6643 ft
2 3 LVPm LVmn 0.5018 0.0586 8.5607 32 5.3937 0.3846 51.9135 ft
4 2 LVn LVPm 0.3109 0.0632 4.9183 30 4.0290 0.1845 38.9387 ft
2 5 LVPm Vm 0.2715 0.0382 7.1145 55 5.6937 0.1952 55.0150 ft
2 6 LVPm Vmn 0.1599 0.0438 3.6504 43 3.5494 0.0723 34.8213 ft
3 4 LVmn LVn 0.1905 0.0718 2.6525 27 1.2510 0.0469 18.2819 ft
3 5 LVmn Vm 0.2242 0.0459 4.8892 52 3.3118 0.1325 32.8670 ft
3 6 LVmn Vmn 0.3885 0.0505 7.6860 39 4.1027 0.2874 39.5918 ft
4 5 LVn Vm 0.1661 0.0507 3.2744 50 2.4544 0.0647 26.2831 ft
4 6 LVn Vmn 0.1712 0.0553 3.0968 37 3.0302 0.0606 30.6230 ft
6 5 Vmn Vm 0.1217 0.0300 4.0608 62 3.2961 0.0617 32.7397 ft

Table 15. Matrix of Z values for site/chronology combination Significant relationships are outlined in
bold.
Z(ft) 1 2 3 4 5 6
matrix HVP(mn) LVP(mn) LV(mn) LV(n) V(m) V(mn)

1 0
2 1.0880 0
3 3.0129 5.3937 0
4 -0.1692 4.0290 1.2510 0
5 2.6350 5.6937 3.3118 2.4544 0
6 2.5068 3.5494 4.1027 3.0302 3.2961 0

68
 Mirjana Roksandic: The Mesolithic–Neolithic transition in the Ðerdap
Figure 4. Multidimensional scaling plot showing internal relationship between the site/chronology units in the Iron Gates
gorge. Based on dissimilarity matrix.
The relationships between different site/chronological
units become more complex. Lepenski Vir and Padina in the
Mesolithic resemble the Contact period at Hajdučka
Vodenica and Padina. This could be due to the fact that
Padina is present in both components. However, while the
Padina Contact sample is almost the same size as Hajdučka
Vodenica in the same period, the Padina Mesolithic sample is
very small and comparable in frequencies to Lepenski Vir
Mesolithic. Therefore it is unlikely that the analysis would
pool these two sites together were they different. As stated, a
strong case of continuity is present between Vlasac
Mesolithic and Vlasac Contact, as well as Lepenski Vir
Contact and Neolithic groups. Some shifting and moving of
population within the region could explain the similarities
between the Contact subsamples at Lepenski Vir, Vlasac,
Hajdučka Vodenica and Padina. This would coincide with
Radovanović’s phase of greater territorial integrity and more
ideological integration in the region (Radovanović 1995,
1996a, 1996b, 1996d). Availability of contact with Neolithic
farmers in the region could have emphasized the need for
ideological and conceptual unity among the foragers.
Franzhausen, as expected, is far removed from the rest of
the sample and is a definite outlier. The sites examined form
a pattern similar to the ‘horseshoe’ shape typical of chrono-
logical series (Greenacre 1984). However, several features
contradict an interpretation of the pattern as reflecting merely
change over time. According to the temporal change explan-
ation, the Mesolithic sites should be on one end,
Mesolithic/Neolithic at the bottom and Neolithic on the other
end of the ‘horseshoe’ diagram. Although the pattern ob-
served reflects this situation, Hajdučka Vodenica and Padina
Contact and Vlasac Mesolithic do not fit perfectly.
Contribution of traits
Theta values obtained in the Site/Chronology analysis
(above) were submitted to Principal Component Analysis.
69
Dimensions 1 2
HVPmn -0.39 0.20
LVPm 1.33 -0.24
LVmn 0.97 0.79
LVn 0.27 0.34
Vm 0.89 0.63
Vmn -0.41 -0.95
Kruskal Stress of final configuration: 0.027
Proportion of variance: 0.993
This, somewhat unorthodox approach was suggested by M.
Jackes (pers. comm.) based on a published analysis by
Christensen (1997). Simply stated, ‘θ’ values are treated as
ordinal values and submitted to the PCA in order to ascertain
which of the traits contributed the most to the observed pat-
tern. The site chronology analysis being the one on which the
interpretation is mostly based, it was deemed unnecessary to
subject the results of other analyses to the same procedures.
The output in Table 19 shows that trait 1 (marginal
tubercle 0.90), trait 3 (supraorbital notch 0.92), trait 6 (pari-
etal foramen 0.89) and trait 23 (apical bone 0.80), contribute
the most to the first dimension. In real ordinal data this di-
mension represents size; here it shows the traits with high
frequencies. Trait 4 exhibits a strong negative association
with dimension one (supraorbital foramen – 0.89), while trait
no 13 (mylohyoid bridge – 0.59) shows negative association
of a lesser extent. On the second dimension, trait 24 (Inca
bone 0.89) has a strong positive association, while trait no 11
(double mental foramen – 0.69) has a strong but negative as-
sociation. The two components explain 66.6% of the total
variation within the sample.
In Varimax rotation (Table 20), applied to reduce the
number of variables on the ‘size’ axis, Variables 1 and 23
show even stronger positive association (0.97 and 0.98, re-
spectively) while variable 4 shows strong negative associ-
ation. On the second component, variable 11 shows even
stronger negative association, while variable 5 shows the
strongest positive association with this component. These
two components explain 55.7% of the variation in the sample.
All of the analyses (with or without Franzhausen) show a
strong temporal trend. A significant amount of change within
the examined population may be due to non-directional
micro-evolution that is expected for a series covering a time-
span of 1500 years. However, the position of Hajdučka
Vodenica and Padina and the MDS of chronological units
show significantly more change occurring between
Mesolithic and Contact period (due to the availability of
 The Iron Gates in Prehistory

Table 16. ‘k’ and ‘N’ values for traits used in the analysis of subsamples based on site/chronology combination with Franzhausen I.

Trait HVP(mn) 1 LVP(m) 2 LV(mn) 3 LV(n) 4 V(m) 5 V(mn) 6 FR.I 7

no. k N k N k N k N k N k N k N
2 11 24 7 19 4 15 8 14 3 36 1 26 8 588
5 7 12 10 12 5 9 4 9 13 30 17 22 90 451
6 17 24 19 22 12 22 8 18 23 45 19 30 211 425
8 4 17 2 18 4 18 5 16 14 44 15 28 27 190
10 16 38 12 23 1 18 0 15 18 47 9 29 130 638
11 4 34 0 27 0 20 1 18 6 54 0 28 46 530
12 1 16 3 13 0 9 2 13 9 33 6 26 8 365
23 1 9 2 9 0 9 0 10 6 24 2 14 16 194

Table 17. The output of the analysis for site/chronology combination with Franzhausen I.

Site 1 Site 2 mmd sd Stand Total Z di S formula

FT FT FT N FT FT FT
1 2 HVPmn LVPm 0.0236 0.0625 0.3779 40 0.8608 -0.1014 11.2045 ft
1 3 HVPmn LVmn 0.0638 0.0683 0.9335 37 1.8817 -0.0729 16.5584 ft
1 4 HVPmn LVn 0.1346 0.0668 2.0161 36 2.5822 0.0011 20.8349 ft
1 5 HVPmn Vm 0.0891 0.0439 2.0302 61 2.1528 0.0013 18.1553 ft
1 6 HVPmn Vmn 0.1679 0.0510 3.2912 47 3.3415 0.0659 26.0245 ft
1 7 HVPmn FRI 0.2940 0.0317 9.2615 444 7.1037 0.2305 60.2433 ft
2 3 LVPm LVmn 0.2480 0.0708 3.5028 33 3.1803 0.1064 24.8741 ft
2 4 LVPm LVn 0.4032 0.0692 5.8225 32 4.5975 0.2647 35.8746 ft
2 5 LVPm Vm 0.1918 0.0462 4.1505 57 3.7504 0.0994 29.0578 ft
2 6 LVPm Vmn 0.1510 0.0534 2.8303 43 2.9377 0.0443 23.1927 ft
2 7 LVPm FRI 0.5434 0.0339 16.0297 441 8.7306 0.4756 79.4253 ft
3 4 LVmn LVn -0.0231 0.0754 -0.3070 29 -0.1041 -0.1740 7.1021 ft
3 5 LVmn Vm 0.1954 0.0523 3.7372 54 2.9894 0.0908 23.5464 ft
3 6 LVmn Vmn 0.1621 0.0593 2.7350 40 2.4154 0.0436 19.7722 ft
3 7 LVmn FRI 0.1425 0.0400 3.5631 438 3.0453 0.0625 23.9311 ft
4 5 LVn Vm 0.2797 0.0509 5.4949 53 4.0364 0.1779 31.2792 ft
4 6 LVn Vmn 0.3378 0.0580 5.8255 40 4.2705 0.2218 33.1581 ft
4 7 LVn FRI 0.3628 0.0385 9.4119 437 5.9376 0.2857 48.1238 ft
5 6 Vm Vmn 0.0671 0.0349 1.9210 65 1.8855 -0.0028 16.5800 ft
5 7 Vm FRI 0.1764 0.0153 11.5023 462 6.3152 0.1457 51.8994 ft
6 7 Vmn FRI 0.3445 0.0226 15.2427 448 8.1575 0.2993 72.3665 ft

Table 18. Matrix based of Z values for site/chronology combination with Franzhausen I. Significant relationships are
outlined in bold.

Z(ft) 1 2 3 4 5 6 7
matrix HVP(mn) LVP(m) LV(mn) LV(n) V(m) V(mn) FR.I

1 0
2 0.8608 0
3 1.8817 3.1803 0
4 2.5822 4.5975 -0.1041 0
5 2.1528 3.7504 2.9894 4.0364 0
6 3.3415 2.9377 2.4154 4.2705 1.8855 0
7 7.1037 8.7306 3.0453 5.9376 6.3152 8.1575 0

70
 Mirjana Roksandic: The Mesolithic–Neolithic transition in the Ðerdap
Figure 5. Multidimensional Scaling plot showing internal relationships between site/chronology units with the introduction
of Franzhausen. Based on dissimilarity matrix.
contact with a different population?) than between Contact
and Neolithic. This can be interpreted as showing a high de-
gree of population heterogeneity during the Mesolithic
and/or availability of contact with some influx (but not on a
large scale) of new genes from a different population(s). The
degree of difference (as well as economic behaviour) of these
populations with the indigenous foragers remains, of course,
impossible to assess from the current study.
Combining the lines of interpretation
In view of the proposed porous frontier between Mesolithic
and Neolithic cultures in the study region, can we presume
interactions between bearers of these respective cultures?
What forms did these interactions take? Was neolithization
their ultimate consequence?
Interactions, in the broadest sense, involving any amount
of change within a population and resulting from the avail-
ability of contact or presence of another population, can be
assumed even without any specific explanatory mechanism.
It is improbable that two populations existing in a relatively
restricted geographic area would never interfere or interact
with one another. Beyond assumptions, the contact between
groups with distinct material culture, which in the case of
Đerdap archaeology correspond well to subsistence groups,
is evidenced on many of the sites in the region through the
exchange of trade items. The question is therefore centred
more on the nature and consequences of this contact than on
its existence.
First it is important to stress that this contact need not be
uniform and could have been site specific. For example,
while there is no evidence for ceramics in the Contact period
at Vlasac, Hajdučka Vodenica is rich in potsherds, and
ceramics were found in situ in Padina houses. Is the close
clustering of the two periods at Vlasac indicative of the
71
Dimensions 1 2
HVPmn 0.46 0.42
LVPm 1.10 0.31
LVmn -0.25 0.14
LVn -0.42 0.77
Vm 0.15 -0.77
Vmn 0.67 -0.69
Franz. I -1.70 -0.19
Kruskal Stress of final configuration: 0.064
Proportion of variance: 0.97
greater isolation of Vlasac as a specific locality?
In terms of anthropological change within the period,
some regrouping of the population is evident. Vlasac seems
to be very closed and little change occurs at the time of
availability of contact. Similarities between Padina,
Hajdučka Vodenica and Lepenski Vir seem to point towards
greater mobility within the group as a result of possible pres-
sure from the outside. This seems to confirm Radovanović’s
interpretation of this period as a phase of consolidation of the
Lepenski Vir culture, with greater ideological integration
among the previously dispersed and distinct sites, reflected
archaeologically in increased evidence of art and ritual
(Radovanović 1996b, 1996c, 1996d; Radovanović & Voytek
1997). Can the observed difference between Mesolithic and
Contact populations be attributed to admixture between the
existing local subpopulations of the Mesolithic Iron Gates
gorge, or does it provide evidence for the influx of other,
more remote, genes?
The observed differences between the chronological units
examined seems to be largely due to a secular trend. At the
time of first contact with the Neolithic population, a slightly
more marked change in the genetic profile of the population
occurs. This indicates higher levels of admixture with a non-
local population. It could have been brought about by an in-
flux of non-local foragers, by an influx of surrounding farm-
ers, or both. In order to answer this question with certainty, a
better understanding of local Neolithic populations as well as
a wider base of the Mesolithic Iron Gates populations (both
from the Romanian side of the Danube as well as from sites
situated further away) would be needed. The Neolithic site of
Velesnica contained only three female skeletons, while
Ajmana (with 17 individuals) was not available for study at
the time of this research. The published report by
Radosavljević-Krunić (1986) does not give enough informa-
tion for its inclusion in any of the analyses undertaken. On the
Romanian side, only Schela Cladovei has yielded a signific-
 The Iron Gates in Prehistory

Table 19. PCA output for Theta values of traits analyzed in the text.

Latent Roots (Eigenvalues)

1 2 3 4 5 6 7 8 9
5.0401 2.9539 2.3817 1.1441 0.4803 0.0000 0.0000 0.0000 0.0000
Component loadings
1 2 3 4
V1 0.9061 -0.1902 0.0914 0.3483
V2 -0.0838 0.2467 -0.8662 -0.4264
V3 0.9282 0.3265 -0.0888 -0.1543
V4 -0.8026 0.4243 0.0552 0.0716
V5 0.7571 0.4489 0.4066 -0.2393
V6 0.8946 0.3982 -0.0725 0.0127
V7 -0.4109 0.6277 0.6539 -0.0315
V8 -0.3348 -0.4634 0.7356 -0.3070
V11 -0.2278 -0.6903 -0.5614 0.1728
V13 -0.5891 0.4175 0.0267 0.6585
V23 0.8072 -0.3803 0.1785 0.4010
V24 0.0366 0.8978 -0.3536 0.1874

Table 20. PCA output for Theta values of traits used in the text.
Varimax rotation.

Rotated Loading Matrix ( VARIMAX, Gamma = 1.0000)

1 2 3 4
V1 0.9748 0.1010 0.0652 0.1492
V2 -0.4468 -0.2620 0.7440 0.4220
V3 0.5848 0.4893 0.4623 0.4524
V4 -0.7235 0.0737 0.0293 -0.5501
V5 0.4276 0.8425 0.1079 0.3048
V6 0.6245 0.4965 0.5019 0.2757
V7 -0.4395 0.7166 -0.2264 -0.4832
V8 -0.2219 0.0959 -0.9464 0.0893
V11 0.0161 -0.9302 -0.0160 0.0872
V13 -0.2559 -0.0350 0.1576 -0.9295
V23 0.9802 -0.0225 -0.1318 0.1019
V24 -0.1509 0.3787 0.8303 -0.3345
‘Variance’ Explained by Rotated Components
1 2 3 4
3.8798 2.8133 2.7148 2.1118
Percent of Total Variance Explained
1 2 3 4
32.3321 23.4440 22.6232 17.5984

ant number of individuals (100+: C. Bonsall, pers. comm.) Based on the data presented here, the distances between
which are still under study, while a survey beyond the banks the Mesolithic components of Vlasac, Lepenski Vir and
of the Danube on both the Romanian and Serbian sides is yet Padina seem to be important. The great heterogeneity of the
to be undertaken. population observed by other researchers also supports this

72
 Mirjana Roksandic: The Mesolithic–Neolithic transition in the Ðerdap
finding. However, a simple trend towards homogenization in
the Contact period, would have resulted in the
Mesolithic/Neolithic subsamples being positioned close to
one another and equidistant from the Mesolithic subsamples.
This is not the case. As stated earlier, Vlasac seems to remain
the most isolated while a marked degree of similarity is ob-
served between the Lepenski Vir Mesolithic and the
Hajdučka Vodenica and Padina Contact periods. Some intro-
duction of new genes is possible. Ascertaining that they come
either from surrounding Neolithic people, or other foragers
moving as a consequence of neolithization of the surrounding
region would be over-interpreting the scant evidence.
If some influx of new genes is indicated in the Contact
period, the Neolithic in the region (in terms of population
biology) represents a continuation of the local Mesolithic.
This is evident in both the non-metric traits where the
Neolithic group helps make the ‘horseshoe-shaped’ curve
typical of temporal ordering, and in metrics where there is
practically no significant difference between the Contact and
the Neolithic groups (Roksandic 1999).
Conclusions
In conclusion, no large-scale population admixture can be
demonstrated from the above data. Some ‘seeping in’ of the
population suggested by Menk (Menk & Nemeskéri 1989:
531), but without the successive replacement that he argued
for, can be proposed on the basis of current research. This
‘seeping in’ happens more perceptibly at the beginning of the
contact, rather than at the time of change in subsistence. Even
when the change in subsistence finally occurs, fishing and
hunting still account for the major portion of the animal as-
semblage in both Neolithic sub-phases at Lepenski Vir.
Although the reasons for the change in subsistence are bey-
ond the scope of this research, it can be stated on the basis of
the anthropological information that it is not brought about
by an incoming population. It must be regarded as a con-
sequence of cultural and social factors operating within the
Mesolithic of Lepenski Vir itself, which brought about its
disintegration.
The Mesolithic Lepenski Vir culture is based on the rich
riverine environment. Such environments tend to support
more affluent societies, and these are not ‘among the first to
make the transition to food production. Rather they appear to
be late lasting in historic terms’ (Brinch Petersen &
Meiklejohn, in press). The Lepenski Vir Mesolithic success-
fully paralleled local Neolithic developments over a long
period of time. Contacts with the Neolithic population in the
region seem to have helped to form an ideological unity of
sites and thus brought into full expression the artistic
achievements of an already affluent society. Ideological in-
tegration evidenced at the time of possible Contact could
have resulted from the growing wealth of the sites based on
trade in salt-preserved foodstuffs (fish from Iron Gates gorge
and wild game from Gura Baciului) as proposed by Tasić
(1998). Internal conflicts, over-exploitation of the environ-
ment and innumerable other factors may have played a role in
the disintegration of the Lepenski Vir tradition. The biolo-
gical descendants of the Lepenski Vir culture remained in
73
theAcknowledgements
My research was made possible by Wenner-Gren Foundation
for Anthropological Research Grant no. 6250, an Ada and
Isabelle Steele Memorial Scholarship (SFU), and a Graduate
Travel Award (SFU). For making the material available for
the present research, I am indebted to Dr Borislav Jovanović
and late Professor Dragoslav Srejović, principal investigators
of the Iron Gates gorge sites studied. My thanks also go to
Chris Mieklejohn and Mary Jackes for their invaluable
support.
locality, in smaller groups, and adopted a different material
culture and architecture, but retained the same burial prac-
tices, and to a large extent the hunting and fishing economic
base. The greater percentage of domestic animals and definite
use of domesticated varieties of cereals classifies them as a
Neolithic group, but in many respects this population re-
mained unchanged. Only within the fully developed Starčevo
phase (Lepenski Vir IIIb) and with a change in burial ritual
towards more canonized forms (Antunović 1990) did this
population finally integrate itself into a larger Neolithic
community.
Note
1. Although this monocausal explanation, which was applied to all
the mortuary monuments in Neolithic Europe, is overly
simplistic and reveals more about the preoccupations of
modern-day western scholars than the prehistoric inhabitants of
Europe (cf. Cullen 1995: 286), it remains one of the possible, and
even plausible, reasons but cannot be perceived as the only cause
of the growing importance of mortuary ritual in the period
(Masset 1993).
Acknowledgements
My research was made possible by Wenner-Gren Foundation for
Anthropological Research Grant no. 6250, an Ada and Isabelle
Steele Memorial Scholarship (SFU), and a Graduate Travel Award
(SFU). For making the material available for the present research, I
am indebted to Dr Borislav Jovanović and late Professor Dragoslav
Srejović, principal investigators of the Iron Gates gorge sites stud-
ied. My thanks also go to Chris Mieklejohn and Mary Jackes for
their invaluable support.
References
Ammerman, A.J. & Cavalli-Sforza, L.L. 1971: Measuring the rate
and spread of early farming in Europe. Man 6: 674–688.
Ammerman, A.J. & Cavalli-Sforza, L.L. 1984: The Neolithic
Transition and the Genetics of Populations in Europe. Princeton,
New Jersey: Princeton University Press.
Antunović, M. 1987: Populacija i način sahranjivanja na
Centralnom Balkanu tokom ranog I srednjeg Neolita.
Unpublished BA Honours paper, Department of Archaeology,
University of Belgrade.
— 1990: Anthropological and archaeological survey concerning
mortuary practices in the central area of Balkan Peninsula during
Early and Middle Neolithic. In Cahen, D. & Otte, M. (eds)
Rubané et Cardial, Actes du colloque international de Liège
(11–13 décembre 1988). E.R.A.U.L. 39. Liège: University of
Liège, 39–50.
Bartosiewicz, L., Boroneanţ, V., Bonsall, C. & Stallibrass, S. 2001:
New data on the prehistoric fauna of the Iron Gates: a case study
1
from Schela Cladovei, Romania. In Kertész, R. & Makkay, J.
 The Iron Gates in Prehistory
(eds) From the Mesolithic to the Neolithic. Budapest:
Archaeolingua (Main Series), 15–22.
Berry, A.C. & Berry, R.J. 1967: Epigenetic variation in the human
cranium. Journal of Anatomy 101: 361–379.
Bolomey, A. 1973: An outline of the late Epipalaeolithic economy at
the ‘Iron Gates’: the evidence of bones. Dacia n.s. 27: 41–52.
Bökönyi, S. 1969: Kičmenjaci (prethodni izveštaj). In Srejović, D.,
Lepenski Vir, Nova praistorijska kultura u Podunavlju. Belgrade:
Srpska književna zadruga, 224–228.
— 1972: The vertebrate fauna. In Srejović, D., Europe’s First
Monumental Sculpture: New Discoveries at Lepenski Vir.
London: Thames & Hudson.
— 1974: History of Domestic Mammals in Central and Eastern
Europe. Budapest: Akadémiai Kiadó.
— 1978: The vertebrate fauna of Vlasac. In Srejović, D. & Letica,
Z. (eds) Vlasac: A Mesolithic Settlement in the Iron Gates, vol. 2.
Belgrade: Serbian Academy of Science and Arts, 35–65.
Bonsall, C., Lennon, R., MCSweeney, K., Stewart, C., Harkness, D.,
Boroneanţ, V., Bartosiewicz, L., Payton, R. & Chapman, J. 1997:
Mesolithic and Early Neolithic in the Iron Gates: a palaeodietary
perspective. Journal of European Archaeology 5: 50–92.
Boroneanţ, V. 1969. Découverte d’objets d’art épipaleolithique dans
la zone des Portes de Fer du Danube. Rivista de Science
Prehistoriche 24: 283–298.
Boroneanţ, V., Bonsall, C., MCSweeney, K., Payton, R. & Macklin,
M. 1999: A Mesolithic burial area at Schela Cladovei, Romania.
In Thévenin, A. (ed.) L’Europe des Derniers Chasseurs:
Épipaléolithique et Mésolithique. (Actes du 5e colloque interna-
tional UISPP, commission XII, Grenoble, 18–23 septembre
1995). Paris: Éditions du Comité des Travaux Historiques et
Scientifiques, 385–390.
Brinch Petersen, E. & Meiklejohn, C. Forthcoming. Historical con-
text of the term ‘complexity’ in the Scandinavian Mesolithic. In
Janik, L., Kaner, S., Matsui, A. & Rowley-Conwy, P. (eds) From
the Jomon to Star Carr. Oxford: Archaeopress.
Buikstra, J.E. & Ubelaker, D.H. (eds) 1994: Standards for Data
Collection from Human Skeletal Remains. Fayetteville: Arkansas
Archaeological Survey Research Series 44.
Cauwe, N. 1998: Sépultures collectives du mésolithique au néolith-
ique. In Guilaine, J. (ed.) Sépultures d’Occident et génèse des
mégalithismes (9000–3500 avant notre ère). Paris: Editions
Errance, 11–25.
Cavalli-Sforza, L.L. 1996: The spread of agriculture and nomadic
pastoralism: insights from genetics, linguistics and archaeology.
In Harris, D.R. (ed.) The Origins and Spread of Agriculture and
Pastoralism in Eurasia. London: UCL Press, 51–69.
Chapman, J. 1993: Social power in the Iron Gates Mesolithic. In
Chapman, J. & Dolukhanov, P. (eds) Cultural Transformations
and Interactions in Eastern Europe. Aldershot: Avebury, 71–121.
Chapman, R. 1981: The emergence of formal disposal areas and the
‘problem’ of megalithic tombs in prehistoric Europe. In
Chapman, R., Kinnes, I. & Randsborg, K. (eds) The Archaeology
of Death. Cambridge: Cambridge University Press, 71–81.
Christensen, A.F. 1997: Cranial non-metric variation in north and
central Mexico. Anthropologischer Anzeiger 55:15–32.
Clark, J.G.D. 1980: Mesolithic Prelude. The Paleolithic–Neolithic
Transition in Old World Prehistory. Edinburgh: Edinburgh
University Press.
Clason, A.T. 1980: Padina and Starčevo: game, fish and cattle.
Palaeohistoria 22:141–173.
Crubézy, E. 1991: Caractères Discrets et Evolution. Example d’une
Population Nubienne: Missiminia (Soudan). Thèse de Sciences.
Université de Bordeaux I.
Cullen, T. 1995: Mesolithic mortuary ritual at Franchthi Cave,
Greece. Antiquity 69: 270–289.
Czarnetzki, A. 1972a: Epigenetische Skelettmerkmale im
1
74
Populationsverleich. I. Rechts-links-Unterschiede bilateral
angelegter Merkmale. Zeitschrift für Morphologie und
Anthropologie 63: 238–254.
— 1972b: Epigenetische Skelettmerkmale im Populationsverleich.
II. Frequenzunterschiede zwischen den Geschlechtern.
Zeitschrift für Morphologie und Anthropologie 63: 341–350.
— 1972c: Epigenetische Skelettmerkmale im Populationsverleich.
III. Zur frage der Korrelation zwischen den Große des epigenet-
ischen Abstandes und dem Grad der Allopatrie. Zeitschrift für
Morphologie und Anthropologie 64: 145–158.
Duday, H. & Courtaud, P. 1998: La nécropole mésolithique de La
Vergne (Charente-Maritime). In Guilaine, J. (ed.) Sépultures
d’Occident et génèse des mégalithismes (9000–3500 avant notre
ère). Paris: Editions Errance, 25–39.
Falconer, D.S. 1965: The inheritance of liability to certain disease,
estimated from the incidence among relatives. Annals of Human
Genetics 29: 51–76.
Gamble, C. 1986: The Paleolithic Settlement of Europe. Cambridge:
Cambridge University Press.
Greenacre, M. J. 1984. Theory and Application of Correspondence
Analysis. London: Academic Press.
Greenfield, H.J. 1984: Analysis of the vertebrate fauna from
Hajdučka Vodenica: an Early Neolithic site in the Iron Gates,
Yugoslavia. Unpublished manuscript submitted to Borislav
Jovanović, Archaeological Institute, Belgrade.
Harris, D.R. 1996. Introduction: themes and concepts in the study of
early agriculture. In Harris, D.R. (ed.) The Origins and Spread of
Agriculture and Pastoralism in Eurasia. London: UCL Press,
1–9.
Hauser, G. & De Stefano, G.F. 1989: Epigenetic Variants of the
Human Skull. Stuttgart: Schweizerbart.
Imamura, K. 1996. Prehistoric Japan. New Perspectives on Insular
East Asia. London: UCL Press.
Jacobs, K., Wyman, J. & Meiklejohn, C. 1996. Pitfalls in the search
for ethnic origins: a cautionary tale regarding the construction of
‘anthropological types’ in pre-Indo-European northeast Europe.
In Jones-Bley, K. & Huld, M.E. (eds) The Indo-Europeanization
of Northern Europe. Journal of Indo-European Studies 17.
Washington D.C.: Institute for the Study of Man, 285–305.
Jackes, M. & Gao, Q. Forthcoming. Jiangzhai and BanPo (Shaanxi,
PRC): new ideas from old bones. In Janik, L., Kaner, S., Matsui,
A. & Rowley-Conwy, P. (eds) From the Jomon to Star Carr.
Oxford: Archaeopress.
Jackes, M., Lubell, D. & Meiklejohn, C. 1997: Healthy but mortal:
human biology and the first farmers of Western Europe.
Antiquity 71: 639–658 (supplementary material at
http://intarch.ac.uk/antiquity).
Jovanović, B. 1972: The autochthonous and the migrational com-
ponents of the Early Neolithic in the Iron Gates. Balcanica 3:
49–58.
— 1974: Praistorija Gornjeg Đerdapa. Starinar n.s. 22: 1–22.
— 1984a: Hajdučka vodenica, praistorijska nekropola. Starinar n.s.
23–24:305–313.
— 1984b: Padina, naselje mezolita i starijeg neolita. Starinar n.s.
23–24: 159–167.
— 1987: Die Architektur und Keramik der Siedlung Padina B am
Eisener Tor, Jugoslawien. Germania 65: 1–16.
Kent, S. 1989: Cross-cultural perceptions of farmers as hunters and
the value of meat. In Kent, S. (ed.) Farmers as Hunters: The
Implications of Sedentism. Cambridge: Cambridge University
Press, 2–17.
Kozłowski, J.K. & Kozłowski, S.K. 1982: Lithic industries from the
multi-layer Mesolithic site Vlasac in Yugoslavia. In Kozłowski,
J.K. (ed.) Origin of the Chipped Stone Industries of the Early
Farming Cultures in Balkans. Prace archeologiczne 33.
Warszawa–Kraków: Państwowe Wydawnictwo Naukowe,
 Mirjana Roksandic: The Mesolithic–Neolithic transition in the Ðerdap
11–109.
Kozłowski, J.K. & Kozłowski, S.K. 1986: Foragers of Central
Europe and their acculturation. In Zvelebil, M. (ed.) Hunters in
Transition: Mesolithic Societies of Temperate Eurasia and their
Transition to Agriculture. Cambridge: Cambridge University
Press, 95–108.
Krenzer, U. 1996: Anthropologischer Beitrag zur
mesolitisch–neolithischen Transition. Homo 47: 223–247.
Lazarovici, G. 1979: Neoliticul Banatului. Cluj-Napoca: Muzeul de
Istorie al Transilvaniei.
Lee, R.B. 1972a: The intensification of social life among the !Kung
Bushmen. In Spooner, B. (ed.) Population Growth:
Anthropological Implications. Cambridge, MA: M.I.T. Press,
343–350.
— 1972b: Population growth and the beginnings of sedentary life
among the !Kung Bushmen. In Spooner, B. (ed.) Population
Growth: Anthropological Implications. Cambridge, MA: M.I.T.
Press, 329–343.
Leroi-Gourhan, A. (ed.) 1965: La Préhistoire. Paris: Press
Universitaire Française.
Letica, Z. 1971: Vlasac — nouvel habitat de la culture Lepenski Vir
à Djerdap. Archaeologia Iugoslavica 10: 7–12.
Lyman, R.L. 1994: Vertebrate Taphonomy. Cambridge: Cambridge
University Press.
Masset, C. 1993: Les dolmens. Sociétés néolithiques et pratiques
funéraires. Paris: Editions Errance.
Menk, R. & Nemeskéri, J. 1989: The transition from Mesolithic to
Early Neolithic in Southeastern and Eastern Europe: an anthro-
pological outline. In Hershkovitz, I. (ed.) People and Culture in
Change. Proceedings of the Second Symposium on Upper
Palaeolithic, Mesolithic and Neolithic Populations of Europe
and the Mediterranean Basin. BAR International Series 508.
Oxford: British Archaeological Reports, 531–539.
Mikić, Ž. 1980: Anthropologische Typen der Đerdap (Eisernen-Tor)
— Serie. In Kozłowski, J.K. & Machnik, J. (eds) Problèmes de
la néolithisation dans certaines régions de l’Europe: actes du
colloque international. Prace Komisiji Archeologicznej 21.
Kraków: Polska Akademia Nauk, 151–162.
— 1981a: Đerdapska serija kao antropološki model neolitizacije.
Glasnik Etnografskog instituta SANU 30: 103–130.
— 1981b: Stanje i problemi fizičke antropologije u Jugoslaviji —
Praistorijski period. Posebna izdanja 53. Sarajevo: Centar za
balkanološka ispitivanja, knjiga 9.
— 1988: Die Anthropologischen Funde vom Eiserner Tor als
Neolithisationsmodell. Berytus: Archaeological Studies 26:
45–52.
— 1992: The Mesolithic population of the Iron Gates region.
Balcanica 23: 33–45.
Molto, J.E. 1983: Biological Relationships of Southern Ontario
Woodland Peoples: The Evidence of Discontinuous Cranial
Morphology. Ottawa: National Museums of Canada.
Nemeskéri, J. 1969: Stanovnistvo Lepenskog Vira (Prethodni
izvestaj). In Srejović, D. Lepenski Vir. Nova praistorijska kultura
u Podunavlju. Belgrade: Srpska Književna Zadruga, 239–262.
— 1978: Demographic structure of the Vlasac Epipaleolithic popu-
lation. In Srejović, D. & Letica, Z. (eds) Vlasac: A Mesolithic
Settlement in the Iron Gates, vol. 2. Belgrade: Serbian Academy
of Sciences and Arts, 97–134.
Nemeskéri, J. & Lengyel, I. 1978: The results of paleopathological
examinations. In Srejović, D. & Letica, Z. (eds) Vlasac: A
Mesolithic Settlement in the Iron Gates, vol. 2. Belgrade: Serbian
Academy of Sciences and Arts, 231–260.
Nemeskéri, J. & Szathmary, L. 1978a: Analysis of the variation of
quantitative traits. In Srejović, D. & Letica, Z. (eds) Vlasac: A
Mesolithic Settlement in the Iron Gates. Vol. 2,
Geology–Biology–Anthropology. Belgrade: Serbian Academy of
1
75
Sciences and Arts, 157–176.
Nemeskéri, J. & Szathmary, L. 1978b: Anthroposcopic and epigen-
etic variation. In Srejović, D. & Letica, Z. (eds) Vlasac: A
Mesolithic Settlement in the Iron Gates, vol. 2. Belgrade: Serbian
Academy of Sciences and Arts, 135–156.
Nemeskéri, J. & Szathmary, L. 1978c: Individual data on the Vlasac
anthropological series. In Srejović, D. & Letica, Z. (eds) Vlasac:
A Mesolithic Settlement in the Iron Gates, vol. 2. Belgrade:
Serbian Academy of Sciences and Arts, 285–426.
Nemeskéri, J. & Szathmary, L. 1978d: Sex and sexualisation. In
Srejović, D. & Letica, Z. (eds) Vlasac: A Mesolithic Settlement in
the Iron Gates, vol. 2. Belgrade: Serbian Academy of Sciences
and Arts, 77–96.
Nemeskéri, J. & Szathmary, L. 1978e: Taxonomical structure of the
Vlasac Mesolithic population. In Srejović, D. & Letica, Z. (eds)
Vlasac: A Mesolithic Settlement in the Iron Gates, vol. 2.
Belgrade: Serbian Academy of Sciences and Arts, 177–229.
Orliac, M. 1988: Mésolithique. In Leroi-Gourhan, A. (ed.)
Dictionnaire de la Préhistoire. Paris: Presses Univérsitaires de
France, 686.
Ossenberg, N.S. 1981: An argument for the use of total side fre-
quencies of bilateral nonmetric skeletal traits in population dis-
tance analysis: the regression of symmetry on incidence.
American Journal of Physical Anthropology 54: 471–479.
Prinz, B. 1987: Mesolithic Adaptations on the Lower Danube:
Vlasac and the Iron Gates Gorge. BAR International Series 330.
Oxford: B.A.R.
Radosavljević-Krunić, S. 1986: Resultats de l’étude anthropolo-
gique des squélettes provenant du site Ajmana. Đerdapske sveske
3: 51–58.
Radovanović, I. 1995: Mesolithic and Early Neolithic in the Iron
Gates region — settlements, subsistence and chronology. In
Thévenin, A. (ed.) L’Europe des Derniers Chasseurs:
Épipaléolithique et Mésolithique. (Actes du 5e colloque interna-
tional UISPP, commission XII, Grenoble, 18–23 septembre
1995). Paris: Éditions du Comité des Travaux Historiques et
Scientifiques, 379–383.
— 1996a: The Iron Gates Mesolithic. Ann Arbor: International
Monographs in Prehistory.
— 1996b: Kulturni identitet đerdapskog mezolita. Zbornik radova
Narodnog muzeja 26(1): 39–47.
— 1996c: The Lepenski Vir culture: a contribution to interpretation
of its ideological aspects. In Lazić, M. (ed.) Antidoron completis
LXV annis Dragoslavo Srejović ab amicis collegis discipulis ob-
latum. Belgrade: Center for Archaeological Research, Faculty of
Philosophy, University of Belgrade, 87–93.
— 1996d: Mesolithic/Neolithic contacts: a case of the Iron Gates
region. Poročilo o raziskovanju paleolitika, neolitika in eneo-
litika v Sloveniji 23: 39–48.
Radovanović, I. & Voytek, B. 1997: Hunters, fishers or farmers:
sedentism, subsistence and social complexity in the Djerdap
Mesolithic. Analecta Praehistorica Leidensia 29: 19–31.
Roksandic, M. 1999: Transition from Mesolithic to Neolithic in the
Iron Gates Gorge: Physical Anthropology Perspective.
Unpublished PhD thesis, Simon Fraser University, Burnaby, B.C.
— 2002: Position of Skeletal Remains as Key to Understanding
Mortuary Behaviour. In Haglund, W.G. & Sorg, M.H. (eds)
Advances in Forensic Taphonomy. Boca Raton, FL: CRC Press,
95–113.
Saunders, S.R. 1978: The Development and Distribution of
Discontinuous Morphological Variation of the Human
Infracranial Skeleton. Archaeological Survey of Canada Paper
no. 81.Ottawa: National Museums of Canada.
Saunders, S.R. 1989: Nonmetric skeletal variation. In Iscan, M.Y. &
Kennedy, K.A.R. (eds) Reconstruction of Life from the Skeleton.
New York: Liss, 95–108.
 The Iron Gates in Prehistory
Schwidetzky, I. & Mikić, Ž. 1988: Lepenski Vir und das
Grazilisationsproblem in der Anthropologie. Godišnjak Centra
za Balkanološka ispitivanja (ANUBiH) 24: 113–123.
Sjøvold, T. 1977: Non-metrical divergence between skeletal popu-
lations. Ossa 4 (supplement I): xii, 133.
— 1984: A report on the heritability of some cranial measurements
and non-metric traits. In van Vark, G.N. & Howells, W.W. (eds)
Multivariate Statistical Methods in Physical Anthropology: a
Review of Recent Advances and Current Developments.
Dordrecht–Boston: Reidel, 223–246.
Srejović, D. 1968: Lepenski vir (Boljetin) — predneolitska i neolit-
ska naselja. Arheološki Pregled 10: 85–87.
— 1969: Lepenski Vir, Nova praistorijska kultura u Podunavlju.
Belgrade: Srpska književna zadruga.
— 1971: Predneolitske i neolitske kulture u Đerdapu. Materijali 6
(VIII kongresa arheologa Jugoslavije): 19–25.
— 1979: Protoneolit – kultura Lepenskog Vira. In Praistorija
Jugoslovenskih, zemalja II: Neolitsko Doba. Sarajevo: Svjetlost,
33–76.
— 1989: The Mesolithic of Serbia and Montenegro. In Bonsall, C.
(ed.) The Mesolithic in Europe. Edinburgh: John Donald,
481–491.
Srejović, D. & Letica, Z. 1978: Vlasac: A Mesolithic Settlement in
the Iron Gates, 2 volumes. Belgrade: Serbian Academy of
Sciences and Arts.
Tasić, N. 1997: Apsolutna hronologija neolita centralnog Balkana.
Unpublished PhD thesis, Department of Archaeology, University
of Belgrade.
— 1998: U potrazi za solju. IQ 3: 12–38.
Voytek, B.A. & Trinham, R. 1989: Rethinking the Mesolithic: the
case of southeast Europe. In Bonsall, C. (ed.) The Mesolithic in
Europe. Edinburgh: John Donald, 492–499.
76
Werbart, B. 1998: Subneolithic: what is it? — ‘Subneolithic’ soci-
eties and the conservative economies of the circum-Baltic re-
gion. In Zvelebil, M., Dennell, R. & Dománska, L. (eds)
Harvesting the Sea, Farming the Forest: The Emergence of
Neolithic Societies in the Baltic Region. Sheffield: Sheffield
Academic Press, 37–44.
Whittle, A. 1996: Europe in the Neolithic: The Creation of New
Worlds. Cambridge: Cambridge University Press.
Wilkinson, L., Blank, G. & Gruber, C. 1996: Desktop Data Analysis
with SYSTAT. Upper Saddle River, NJ: Prentice Hall.
Willis, K.J. & Bennet, K.D. 1994: The Neolithic transition — fact or
fiction? Palaeoecological evidence from the Balkans. The
Holocene 4: 326–330.
Wiltschke-Schrotta, K. 1992. The Early Bronze Age graveyard of
Franzhausen I, lower Austria. 3. Study of epigenetic character-
istics. Anthropologischer Anzeiger 50(1-2): 27–49.
Zvelebil, M. 1996a: The agricultural frontier and the transition to
farming in the circum-Baltic region. In Harris, D.R. (ed.) The
Origins and Spread of Agriculture and Pastoralism in Eurasia.
London: UCL Press, 323–345.
— 1996b: What’s in the name: the Mesolithic, the Neolithic and
social change at the Mesolithic–Neolithic transition. In
Edmonds, M. & Richards, C. (eds) Understanding the Neolithic
of North-Western Europe. Glasgow: Cruithne Press, 1–35.
Živanović, S. 1975a: Mesolithic population in Đerdap region.
Balcanica 6: 1–7.
— 1975b: A note on the anthropological characteristics of the
Padina population. Zeitschrift für Morphologie und
Anthropologie 66: 161–175.
— 1976: Ostaci ljudskih skeleta iz praistorijskog nalazišta na
Hajdučkoj vodenici. Starinar n.s. 26: 124–129.
 Demography of the Đerdap Mesolithic–Neolithic
transition

Mary Jackes, Mirjana Roksandic & Christopher Meiklejohn

Abstract: Models for the Mesolithic–Neolithic transition are critically dependent on demographic variables. A major issue has been whether
Mesolithic populations were stationary or increasing in demographic terms. The development of methods allowing for palaeodemographic
comparison among archaeological skeletal samples provides the opportunity to test assumptions about the adequacy of samples and about
changes in fertility during the European Late Mesolithic and Early Neolithic. Previous work has suggested that the Mesolithic was a period of
stationary population in the far west and north of Europe, while the Neolithic brought with it a significant increase in fertility. We apply our
standardized palaeodemographic methods to data from Ðerdap skeletal samples to examine whether period assignments need reassessment,
and whether the Neolithic brought increased fertility to the Iron Gates region. Our conclusions indicate that continuing reassessment of the
assignment of Lepenski Vir skeletons to archaeological periods is needed, that Mesolithic populations were stationary, and that there is a
high probability for increased fertility rates in the Neolithic at Lepenski Vir.
Key words: Đerdap, demography, Mesolithic, Neolithic, transition, Lepenski Vir

Introduction A key element of osteological work relates to the accurate

The shift to food production and the Neolithic is one of the
great transitions in human history. In this paper we will ex-
amine the evidence provided by skeletons excavated from the
Đerdap sites in order to test whether, prior to the introduction
of agriculture, the Đerdap Mesolithic population was in-
creasing or was stable and stationary.
Demographic factors are central to how we view the
transition. Arguments about the social or environmental basis
of the shift return repeatedly to the issue of the relative size
of Mesolithic and Neolithic populations in Europe. The ana-
lyses underlying the ‘demic diffusion’ model are dependent
upon the relative population densities of ‘indigenous’ and
‘colonizing’ groups. For Greece and the Balkans other than in
the Đerdap much is made of the invisibility of the Mesolithic
population. As a result, a key question is whether the appar-
ent peak and concentration of population in the early
Holocene Đerdap is real, especially in the Mesolithic. This is
a major problem both for archaeology and demography.
General palaeodemographical methodology:
the questions that can be answered by
demographic data
Demography deals with variables such as birth rates, death
rates, and their mutual relationship in figures such as growth
rates. Demographers studying modern populations have the
advantage of a relatively complete data base. Errors are
minor and hidden within the statistical power afforded by
large samples. Those who study prehistoric populations are
less fortunate. Reconstructions must be based on a single set
of connected variables, related to death, and extended with an
unknowable degree of accuracy to breakdown by sex and
age. As a result, studies of prehistoric demography focus on
the distribution of ages at death.
determination of demographic variables. Problems involved
in the determination of skeletal age, especially in adults, are
important here (Jackes 1992, 1994, 2000). Whereas markers
of sex in adult skeletons are relatively consistent throughout
the life span, age markers are dependent on the process of
ageing itself. Since this process is gradual and dependent
upon factors such as health and physical condition, the de-
termination is problematic. Individuals who are both healthy
and in good physical condition show slower rates of change.
As a result, ageing methods tend to provide ‘physiological’
age as opposed to ‘chronological’ results. The exception to
this is in subadults, where the nature and rapidity of age-
dependent changes allows for quite accurate determination.
Parameters used in this analysis and the relationships
among them
Because of variability among adults in the expression of
age-dependent characteristics, there is debate over the de-
termination of prehistoric demographic variables that require
the accurate estimation of skeletal age. There have been pro-
posals that archaeological settings require an approach dif-
ferent from the full adult age profiles used in demographic
analysis. Jackes (1986, 1992) has suggested the conjoint use
of two values that may provide insight into overall population
structure, the juvenile to adult ratio (J:A) and mean childhood
mortality (MCM). The J:A (the Indice de Juvenilité, or IJ,
introduced by Bocquet & Masset 1977) is the ratio of chil-
dren 5 to 14 years of age to adults over 20 years, both readily
determined from skeletal samples. MCM (Jackes 1986) is the
mean mortality for ages 5 to 19: the mean of the life table
mortality quotients is a summary value of the probability of
death before reaching adulthood among those surviving to
age 5. It is calculated from the q values across the three
five-year age categories, 5–9, 10–14 and 15–19 years.
J:A and MCM can be determined with some accuracy, and
seem to bear a consistent relationship to each other unless

77
 The Iron Gates in Prehistory

Figure 1 Total fertility (TF) estimates derived from mean childhood mortality (MCM) and juvenile:adult ratio (J:A). Plot
of 170 points: reference points 1. Dobe !Kung, 2. Hutterites dying between 1941 and 1950, 3. Coale’s Index of Marital
Fertility; grey circles 132 archaeological samples (excludes the Đerdap material); solid triangles three archaeological
samples from Dickson Mounds; open triangles 31 historical data sets.

perturbed by special circumstances, such as migration or
sample bias. Model demographic data (Coale & Demeny
1983) provide the basis for estimates of population fertility.
For each of a large number of pooled sex West model tables,
increasing, decreasing and stationary, we calculate J:A,
MCM and the total fertility rate (TF). These three variables
are very highly correlated and allow us to estimate the TF
values for archaeological populations by regression: J:A and
MCM, which express the relative numbers of subadults
among the dead, provide a reflection of population fertility.
While there is a perception that human fertility without con-
traception would lead to an average woman who survives to
menopause having 15 to 20 children, this does not seem to be
true. This is reflected in the fertility values we use here,
which derive from West model tables 1 to 8, at levels of in-
crease up to 5% for West 1.1
We can demonstrate the validity of the approach by testing
the values derived from the Coale and Demeny model data
against modern populations that show high and low fertility.
Since fertility has biological and behavioural constraints, it is
clear that normal unbiased samples must have J:A and MCM
values which have their own limits and a biologically de-
termined relationship among the variables can be assumed. If
shown to be valid, the estimates will provide us with a meth-
od of examining past population demography.
Is the use of West model tables valid?
It may be argued that the West model tables are not appro-
priate bases for conjectures on the fertility of past popula-
tions; the fertility estimated from West model tables could be
too low because the West tables specifically exclude data
with high infant and early childhood mortality, and are de-
rived from 129 sets of data from, in general, industrialized
countries, mostly Northwest Europe or countries colonized
from the British Isles. The West tables might therefore largely
represent a part of the world with low fertility because of
factors like late marriage and a high incidence of unmarried
people (Hajnal 1982).
Are the fertility estimates used here plausible? To test this,
we compare the Coale and Demeny model fertility predic-
tions (incorporating also United Nations model data: United
Nations 1982) with data for American Hutterite women,
considered the best example of ‘natural fertility’ under almost
ideal conditions. The total fertility (TF) 2 value for Hutterite
married women aged 15 to 49 from 1921 to 1930 was estim-
ated as 12.44 live-born children. This figure is normally giv-

78
 Mary Jackes et al: Demography of the Đerdap Mesolithic–Neolithic transition
en as the value of Coale’s Index of Marital Fertility, by which
the Hutterites of this period are taken to represent the highest
overall level of childbearing, exceeded only under rare con-
ditions (e.g. Weeks 1996). Figure 1 demonstrates that the
samples used for reference in this study have an estimated
total fertility below this limit (marked by reference point 3).
A TF of 12.44 is markedly above the TF of other historical
data sets from Europe, Asia, North and South America cov-
ering the period 1650 to 1950.
While reference point 3 marks maximum Hutterite fertil-
ity in the 1920’s, reference point 2 on Figure 1 marks the es-
timated level of fertility derived from the 309 Hutterite
deaths between1941 to 1950 (age at death reordered from
Eaton & Mayer 1953: table 16). The calculated J:A of 0.278
for this decade gives an estimated total fertility of 8.9 live-
born children for women aged 15 to 44; TF would be 7.5
when estimated from MCM. The higher estimated TF value
for Hutterites dying between1941 and 1950 may be more
correct, because the summed age-specific fertility rates of all
Hutterite women of reproductive age between 1936 and 1940
gives a TF rate of 9.4 (see also Jackes & Meiklejohn 2004).
A more conservative estimate would be influenced by the 8.1
value for Hutterite TF from 1946 to 1950 (Eaton & Mayer
1953). This was obviously a period of change, one disturbed
by World War II, as is probably evidenced by the lack of ac-
cord between the J:A and MCM estimates of TF.
Nevertheless, it is clear that the estimated values give a sense
of Hutterite fertility even for a time of transition.
Bocquet-Appel (pers. comm. 2000) has suggested that
Hutterite fertility is lower than it might be, because infant
mortality is low; when infant mortality is high, fertility is
high. The relationship between infant mortality and fertility
is, in fact, complex (see, e.g., Montgomery & Cohen 1998).
Infant mortality may lead to shorter birth intervals, but this in
turn is associated with an increased risk of adverse effects
perinatally (Zhu et al. 1999) and maternal mortality
(Andersson et al. 2000; Conde-Agudelo & Belizán 2000).
If we move to the opposite extreme, low fertility, then ex-
cluding such recent contraceptive regimes as twentieth cen-
tury USA, the Dobe !Kung lie at the bottom of the scale of
known fertility levels as shown on Figure 1. The location of
the Dobe !Kung (reference point 1) marks the total fertility as
estimated from the age at death distribution. The Dobe !Kung
fertility is estimated by the J:A fit as 3.9, a reasonable ap-
proximation of the Dobe !Kung TF for the 1963–1973 period
of 4.3 (Howell 1979) given the small sample size and un-
stable conditions.
Constraints on fertility
Obviously there are behavioural restraints on high fertility,
even in non-contracepting societies, and we must add biolo-
gical factors to any discussion on fertility: higher risks of
foetal loss or stillbirth with increasing age and parity; the ad-
verse effects on mother and foetus of short birth intervals; the
possibility of longer postpartum amenorrhoea with high par-
ity or age (Larsen & Vaupel 1993).
While 35 years of continuous exposure to sexual inter-
course could theoretically lead to 26 children per woman,
each newborn enjoying 6 months of breast feeding, this
model is not plausible, even if there were no disease, death,
79
famine, marital separation, behavioural constraints, no steril-
ity, and no age specific variation in fertility. In fact, foetal
wastage due to chromosomal abnormalities is high, and thus
the chance of a live birth resulting from intercourse is sur-
prisingly low. Holman et al. (2000) show that 50% of a
20-year-old woman’s pregnancies result in foetal loss and
that this increases with age. The general proposition is that
over 70% of conceptions result in lost pregnancies. As a res-
ult, both this foetal loss (detected and undetected) and, of
great importance for small archaeological populations, the
effects of inbreeding (Dorsten et al. 1999; Ober et al. 1999)
must be taken into consideration.
While acknowledging that our fertility estimates for past
populations could be conservative, we point out that they
seem very realistic. Even in a situation encouraging rapid
population growth, such as the southern region of North
America during the first half of the eighteenth century, total
marital fertility was 8.1. Yet here high fertility could be ex-
pected, especially because of the possibility of wet nurses
reducing the period of lactational amenorrhoea (Houdaille
1995).
Fertility in archaeological populations
We cannot know details of the fertility of archaeological
populations: the type of evidence available for the seven-
teenth century Huron (Jackes 1994) is surely almost unique.
But that evidence does indicate low fertility. Even Colyton,
the seventeenth century English parish studied for over 30
years, from written records, and determined to have ‘natural
fertility’, is still under discussion (Vann 1999); the evidence
suggests that age at marriage determined the period of
highest fertility, and that there was family limitation towards
the end of a woman’s reproductive period. Archaeological
demography will, then, always present us with questions, but
translating the J:A and MCM values into fertility estimates
using quadratic regression provides a method of deriving a
basic demographic parameter from skeletal remains, and ac-
knowledges and circumvents the central problem of our in-
ability to provide correct ages for adult skeletons. We may, in
this way, recognize archaeological demographic trends.
The database collected by Jackes, together with selected
data from the database of Steckel et al. (2002; McCaa 2002),
demonstrates that under normal conditions J:A and MCM
values for historical and archaeological samples, of reason-
able size and without obvious bias, will not fall beyond cer-
tain limits. We therefore consider that there are normal bio-
logical limits which are reached at slightly under J:A = 0.4
and MCM = 0.14, and that at this point TF must be below 14.
Only seven problematical samples of the 142 archaeological
samples in our database (which is taken to exclude the Đer-
dap material for the purpose of this discussion) fall beyond
this limit on both axes.
When the estimates derived from J:A and MCM are plot-
ted against each other, we have a method of identifying
samples which are probably flawed by errors and biases in
such a way that an adequate TF estimate cannot be made.
Some examples of problematic data appear on Figure 1,
which illustrates the fertility estimations derived from ar-
chaeological skeletal age distributions. The two extreme out-
liers are one of the Dickson Mounds samples discussed in
 The Iron Gates in Prehistory

Figure 2 Total fertility (TF) estimates derived from mean childhood mortality (MCM) and juvenile:adult ratio (J:A).
General archaeological samples (excluding the Đerdap material) and the three reference points provide the background
context for estimates for Portuguese Mesolithic and Neolithic samples, suggesting fertility increase in the Neolithic.

Jackes (1993) and a sample of 170 individuals from Irene
Mound (Steckel et al. 2002). The method demonstrates that
the Irene Mound sample, representing c. 61% of the original
excavated, must be considered inadequate.
Thus, while we acknowledge that biases inherent in meth-
ods of adult age assessment may weaken the value of pa-
laeodemography as an instrument of interpretation in bioar-
chaeology, we propose that the use of the J:A and MCM will
provide a method of estimating population fertility and allow
us to determine whether a sample is unsuitable for analysis
(Jackes 1993), because of biases which may derive from in-
complete excavation or reporting, taphonomic factors or se-
lective burial.
Demography and the Mesolithic–Neolithic
transition
What was not clear, however, was the actual relationship
between the two transitions: agricultural and demographic.
Deevey assumed the appearance of agriculture to come first,
with demographic change being the dependent variable.
However, by the late 1960s and 1970s, based on the demon-
stration that sedentary settlement preceded agriculture in the
Near East, several scenarios (e.g. Binford 1968; Smith &
Young 1972; Cohen 1977) argued that the prime mover was
the demographic transition, and Meiklejohn argued forcibly
over many years for a ‘population pressure’ driver to the ag-
ricultural transition (Meiklejohn 1978, 1979; Meiklejohn et
al. 1984). It now seems that one of the key assumptions be-
hind the population pressure model, the presence of high fer-
tility levels prior to the transition, may be demonstrably false
(Jackes 1988; Meiklejohn et al. 1997). This paper will dis-
cuss this possibility and the place of the Đerdap samples in
the debate.

Previous assumptions Analysis of Portuguese data — the Mesolithic is not a

It is forty years since Deevey (1960) outlined the concept of high growth population
demographic transition and applied it to the appearance of In the 1980s, Mesolithic skeletal samples excavated in the
agriculture, postulating that the scale of modern overall pop- nineteenth and earlier twentieth centuries from the shell
ulation size could be traced to the agricultural transition. midden sites at Muge in central Portugal were re-examined

80
 Mary Jackes et al: Demography of the Đerdap Mesolithic–Neolithic transition

Figure 3 Total fertility (TF) estimates derived from mean childhood mortality (MCM) and juvenile:adult ratio (J:A).
General archaeological samples (excluding the Đerdap material) and the three reference points provide the background for
estimates for northern European Mesolithic samples (Skateholm in Scania, Olenii Ostrov in Russian Karelia and combined
data from sites on the Danish island of Sjælland) indicating a stationary population.

(Jackes 1988; Jackes & Lubell 1999a, 1999b; Lubell et al.
1994; Jackes 1992; Meiklejohn et al. 1997; Jackes et al.
1997a, 1997b). As part of that study, there was an attempt to
understand the demographic structure of the samples from
the Mesolithic sites Cabeço da Arruda and Moita do
Sebastião and the Neolithic site Casa da Moura. While the
analysis concentrated on mortality profiles, showing that the
Mesolithic samples had lower mortality than the Neolithic
sample, the clear implication was that mortality also reflects
fertility (as with Figure 2). The Mesolithic samples did not
show marked signs of population growth. 3 while the
Neolithic sample did. Our work does not support ‘population
pressure’ as the prime impetus for the agricultural transition
in Portugal, nor do we see evidence that immigration fuelled
the Neolithic population growth (see Jackes et al. 2001),
leading to the conclusion that fertility increased during
and/or after the period of transition.
Analysis of north European Mesolithic data in accord
with Portuguese results
The ideas in Jackes’ (1988) paper were not tested until almost
a decade later, when Meiklejohn was asked to discuss his
work on Danish Mesolithic samples (Meiklejohn et al. 1997).
The Danish material was excavated, mostly under the
Vedbæk Project, on the island of Sjælland. Because the
Danish sample was small and therefore had to be pooled,
other North European Mesolithic material (primarily
Skateholm in Scania and Olenii Ostrov in Russian Karelia)
was compared with the Portuguese results. The comparison
demonstrated that, despite the problems of mixed samples
and the absence of juveniles in both the Danish and
Skateholm groups, the results were reasonably consistent
with those found in Portugal. None of the Mesolithic samples
was in the high mortality, high fertility grouping. Though of-
ten treated as a population whose ‘complexity’ prefigured the
Neolithic (e.g. Price & Brown 1985; Tilley 1996), the
Mesolithic population of southern Scandinavia could be in-
terpreted as “...a stationary population, with both low mor-
tality and fertility” (Meiklejohn et al. 1997: 320). This con-
clusion still seems valid (Fig. 3), indicating a stationary
population in Mesolithic Northern Europe.

81
 The Iron Gates in Prehistory
The Đerdap samples — their importance to the
question
Questions arising from the above analyses
Confirmation of the pattern of lower Mesolithic fertility and
higher Neolithic fertility requires examination of further
large Mesolithic samples. Work on material from four Đer-
dap sites (Roksandic 1999, 2000) provides the opportunity.
Meiklejohn et al. (1997) concluded that the data from
Nemeskéri’s (1978) study of the Vlasac sample were closely
comparable to Olenii Ostrov, suggesting that the Đerdap
samples would conform to the pattern. Furthermore,
Meiklejohn & Zvelebil (1991), based on previously pub-
lished data, suggested that the general health of the Đerdap
population was similar to that found in southern Scandinavia.
Roksandic (1999, 2000) studied four Đerdap samples:
Hajdučka Vodenica, Lepenski Vir, Padina and Vlasac. She
demonstrates that the data used by Nemeskéri and others are
incomplete. We present a new set of demographic profiles
here, and also examine whether the revised Đerdap data sup-
port our model of low mortality and fertility in Mesolithic
populations and higher fertility in Neolithic populations.
The basic demographic profiles: methods specific to the
Đerdap site analysis
The burial practices of Mesolithic peoples are characterized
by great variability, and this is certainly true for the Đerdap
Mesolithic. The burial practices include cremation, primary
inhumation and secondary interment including removal and
re-organization of body parts, with re-burial of skulls and
fragmentary remains.
The Đerdap sites were excavated as a rescue operation,
and re-analysis of the excavation records continues. Further
study of burial practices planned by Roksandic will include
detailed consideration of taphonomic factors. This will no
doubt alter the period assignments used here, especially in
light of new analyzes already published (Bonsall et al. 2000;
Radovanović 2000) and others in progress.
In this paper we use the previously accepted archaeolo-
gical assignment of the burials (Radovanović 1996a;
Roksandic 2000). Our method of demographic analysis
provides a contribution to the on-going discussion about the
validity of the attribution of burials to archaeological units.
Within any single burial, the assessment of the MNI fol-
lowed the common procedures of establishing recognizable
osteological elements that were doubled, as well as those that
presented incompatibility of age and sex markers. Pairing of
bones on the basis of age and general robusticity was accep-
ted only in cases of good preservation and obvious similarit-
ies. Since the burials usually comprised one or a few indi-
viduals already recognized as separate entities in the field,
and since there was some mixing of the smaller elements, all
burials were treated as units. For Lepenski Vir, assessment
was limited because field documentation was not available,
but for Vlasac we could rely on published drawings (Srejović
& Letica 1978) and for Padina and Hajdučka Vodenica on
unpublished documentation provided by B. Jovanović. Since
the MNI in any single burial unit did not exceed eight indi-
viduals, it was not necessary to use the procedures appropri-
ate for ossuaries.
82
While ‘extra individuals’ within any of the burials could
be the result of the inclusion in the grave of earth from dis-
turbed burials, these ‘extra bones’ are so common in Đerdap
burials that their patterning requires more detailed analysis in
order to allow their incorporation into the palaeodemograph-
ic reconstruction (Roksandic, in preparation).
A further set of problems results from the loose human
remains, bones and fragments of bone found in the archae-
ological deposits without any evidence of burial, a circum-
stance noted as a common occurrence in the Mesolithic else-
where (Meiklejohn & Denston 1987). Theoretically such
bones could belong to any of the buried individuals; verific-
ation of this would have required all individuals in the series
to be checked (an impossible task given time constraints), so
they were not included in the sample as separate individuals.
The situation is especially complex for Lepenski Vir where
the quantity of loose bones can sometimes exceed the quant-
ity of bones present in a recognized grave. In order to see
whether this situation created a bias, repeated analyses were
run, both including and excluding these ‘individuals’ from
the Lepenski Vir sample.
Sex determination was based on the pelvic bones
whenever possible and included standard procedures
(Phenice 1969; Workshop of European Anthropologists
1980; Buikstra & Ubelaker 1994). It is noteworthy that the
preauricular sulcus was present in almost all of the examined
pelves that showed female morphology. As discussed by
Roksandic (1999, 2000), the degree of sexual dimorphism is
remarkable and the secondary skeletal markers of sex on
postcranial bones could be used with great reliability where
the pelvic remains were missing. A different pattern is ob-
served with skull remains, which could account for discrep-
ancies between present determinations and those of
Nemeskéri (1978), Zoffmann (1983) and Živanović (1975).
Age determination presented more problems. In order to
avoid point age estimates in adults, since they are highly de-
pendent on the reference population (Bocquet-Appel &
Masset 1982) and unreliable in building mortality profiles
(Müller & Love 1999), adult ages were assigned to two large
categories, namely, ‘young’ and ‘old’. For the present study,
all individuals older than 25 were grouped (cf. Jackes 1992).
The problem created by individuals represented by a single
bone or a bone fragment could be only partially circumvented
in this way. Although for most of these partial skeletons it
was possible to establish whether they were adults over 25 or
not, some had to remain in the undetermined group compris-
ing all from 15 to 80+ years of age.
Age determination for children up to 12 years of age was
based on observation of tooth formation and eruption and
long bone epiphyseal union, when available. Age was as-
signed by reference to tables in Buikstra & Ubelaker (1994).
In other cases, the general aspect of bones was used to estab-
lish that the skeleton belonged to any of the subadult groups.
The precision with which the age in subadults was assessed
depended greatly on preservation and the representation of
different body parts, therefore some of the individuals were
assigned to quite a wide age range (2–15 years, for example).
All these problems had to be circumvented in the statistical
analysis.
Archaeological considerations are of greatest importance.
 Mary Jackes et al: Demography of the Đerdap Mesolithic–Neolithic transition

Table 1. Lepenski Vir.

Age category Mesolithic Meso–Neo M & M–N Neolithic M–N & N All
0 20.6 34.4 55.0 5.3 39.7 60.3
5 2.2 4.2 6.4 6.0 10.2 12.4
10 2.8 2.7 5.5 3.9 6.6 9.4
15 0.0 1.5 1.5 1.5 3.0 3.0
20 1.3 2.5 3.8 1.3 3.8 5.1
25 13.6 45.5 59.1 40.9 86.4 100.0
Total 40.5 90.8 131.3 58.9 149.7 190.2
TF estimated from MCM 7.1 4.2 4.8 5.9 4.9 5.2
TF estimated from J:A 10.8 5.1 6.2 7.6 6.2 6.8

Table 2. Vlasac.
Age category Mesolithic Meso–Neo Neolithic All
0 30.6 0.3 0 30.9
5 9.1 0.3 0 9.4
10 2.3 0.3 0 2.6
15 5.0 0 5.0
20 0.0 3.0 0 3.0
25 81.9 31.1 0 113.0
Total 128.9 35.0 0 163.9
TF estimated from MCM 4.6 4.2 3.8
TF estimated from J:A 4.9 5.1 4.1

Table 3. Padina.
Age category Mesolithic Meso–Neo Neolithic All
0 1 2 0 3
5 1 1 0 2
10 0 1 0 1
15 0 0 0 0
20 0 1 0 1
25 15 26 0 41
Total 17 31 0 48
TF estimated from MCM 3.2
TF estimated from J:A 3.4

Table 4. Hajdučka Vodenica.

Age category Mesolithic Meso–Neo Neolithic All
0 0 0 0 0
5 0 2 0 2
10 0 2 0 2
15 0 2 0 2
20 0 1 0 1
25 0 29 0 29
Total 0 36 0 36

83
 The Iron Gates in Prehistory

Table 5. Đerdap sites grouped.

Age category Mesolithic Meso–Neo Neolithic All
0 52.3 36.7 5.3 94.2
5 12.3 7.5 6.0 25.8
10 5.1 6.0 3.9 14.9
15 5.0 3.5 1.5 10.0
20 1.3 7.5 1.3 10.1
25 110.5 131.6 40.9 283.0
Total 186.5 192.8 58.9 438.2
TF estimated from MCM 4.6 3.5 5.9 4.2
TF estimated from J:A 5.4 3.9 7.6 4.9

Table 6. Neolithic samples.

Age category Neo LV Velesnica Ajmana All Neo
0 5.3 2 4 11.3
5 6.0 2 3 11.0
10 3.9 3 6.9
15 1.5 2 3.5
20 1.3 1.3
25 40.9 3 5 48.9
Total 58.9 7 17 82.9
TF estimated from MCM 5.9 9.7
TF estimated from J:A 7.6 11.6

Table 7. Site and period groupings.

Age category Vlasac, Padina & Hajdučka Vodenica All
0 33.9 42.0
5 13.4 13.5
10 5.6 9.9
15 7.0 5.0
20 5.0 8.8
25 183.0 172.5
Total 247.9 251.7
TF estimated from MCM 3.7 3.9
TF estimated from J:A 4.0 4.7

For comparison of Mesolithic and Neolithic mortality and
fertility patterns, it was necessary to assign individuals to
different periods within the series. The details of period
definition and chronological ordering of the skeletons can be
found in Roksandic (1999, 2000), which is based on
Radovanović (1996a) as well as re-analysis of the docu-
mentation from Padina (Jovanović, pers. comm. to MR,
1998). To strengthen the numbers of Neolithic individuals,
Velesnica (Roksandic, this volume) and Ajmana
(Radosavljević-Krunić 1986) were included in the analysis.
Analysis of the data
Because sample sizes are small, the data were reworked so
that any unknown age or unknown period individuals were
included in the analysis. The unknowns were redistributed
proportionately into the known cells.4 It was immediately
evident that no simple conclusions were to be derived from
the Đerdap samples.
The Lepenski Vir and Vlasac samples are summarized in
Tables 1 and 2. The unreliability of the Lepenski Vir
Mesolithic subsample can be judged from the fact that TF

84
 Mary Jackes et al: Demography of the Đerdap Mesolithic–Neolithic transition

Figure 4 Total fertility (TF) estimates derived from mean childhood mortality (MCM) and juvenile:adult ratio (J:A).
General archaeological samples (excluding the Đerdap material) and the three reference points provide the background for
estimates derived from pooled Đerdap data: Padina (Table 3); Vlasac (Table 2); ALL (Table 5 plus Velesnica, Ajmana from
Table 7); Lepenski Vir (Table 1). The distribution shown indicates low Mesolithic fertility and increased fertility in the
Neolithic.

estimates range from 7 to 11 – such a broad range is a good
indication of a problematic sample. To say that the rich riv-
erine environment of the Lepenski Vir Mesolithic resulted in
population increase would push the evidence beyond accept-
able limits, because the Vlasac Mesolithic estimate, which is
clearly robust, suggests that a Mesolithic woman’s completed
family size would be around 4 children (Table 2). The Vlasac
sample is interpreted on other evidence as representing a
closed and conservative population (Roksandic 2000).
Initial arguments for continuity in subsistence practices at
Lepenski Vir (Radovanović 1996a, 1996b) are being recon-
sidered (Bonsall et al. 2000; Radovanović 2000), and the
Mesolithic– Neolithic transition at Lepenski Vir is dated to a
period of unstable climate (von Grafenstein et al. 1998;
Barber et al. 1999). While it might be argued that there is a
dramatic drop in fertility during the transition, so that the
Lepenski Vir Mesolithic–Neolithic sample came to have a TF
of between 4 and 5, it is more likely that the Mesolithic es-
timate is unacceptable, either because of the small sample
size, or because of incorrect period assignment. Thus, the
suggestion of a stable and generally stationary Mesolithic TF
of 4 to 5 children seems most reasonable.
The sample from Padina (Table 3) is too small to give a
reasonable assessment of its demography; we can simply in-
dicate probable low fertility. The paucity of children under
age five suggests sample bias, and this is even more obvious
for Hajdučka Vodenica (Table 4). While the method of es-
timating fertility used here is specifically designed to cir-
cumvent the frequent problem of infant under-representation
in archaeological sites, it is important to note instances in
which childhood under-representation, beyond five years of
age, may also occur as at Hajdučka Vodenica (Table 4).
Given inadequate samples, grouping sites might provide a
method to get better demographic estimates by period (Table
5). The Neolithic sample is clearly the weakest, and data from
two other sites are added in Table 6.
Two further groupings may be made in order to attempt to
provide appropriate sample sizes. In Table 7 we examine
firstly all sites and all periods pooled, but exclude Lepenski
Vir because the Mesolithic data for that site suggest special

85
 The Iron Gates in Prehistory
circumstances. We then group all the data for the
Mesolithic–Neolithic with those for the Neolithic, in order to
see whether high fertility could be considered a characteristic
of contact with and development of agricultural practices.
Both these groupings of samples again indicate a TF of
around 4.
Overall, whatever method of pooling is used, the data
suggest a rather low fertility, one unlikely to lead beyond a
stationary population unless circumstances were exceptional.
There is, however, one set of data that is unexpected, the
Lepenski Vir Mesolithic. The sample is small and the data
give an indication of some bias. A special use of Lepenski Vir
for the preferential burial of subadults is one possibility. On
the other hand, the anomalous nature of the Mesolithic
sample would be diminished by the addition of further adults,
as suggested by the stable C and N isotopic values (Bonsall
et al. 1997: table 5).
One other factor should be considered in Đerdap demo-
graphy: the transition period has a slight over-representation
of adults among the dead. This might occur, not because of
low fertility, preservation bias against the young, or selective
burial of adults, but as a result of immigration of young
adults. The Mesolithic–Neolithic sample could indicate a fall
in fertility consequent upon a period of instability associated
with cultural change and an influx of adults from outside
leading to an apparent over-representation of adults. Such an
influx could result in a drop in fertility: the drop could be ac-
tual, as a result of the changing and unstable conditions, or
perceived, resulting from an unbalanced sex ratio among the
migrants (an excess of males).
However, it is worth pointing out that several of the
Mesolithic–Neolithic sample adults might be considered
Mesolithic on stable isotope values for the 33 Mesolithic and
Neolithic skeletons from Lepenski Vir first analyzed (Bonsall
et al. 1997), and that, until full details are published on the
entire sample of 68 Lepenski Vir individuals analyzed for
stable isotopes by Bonsall et al. (2000), we cannot speculate
on the transition period demography.
Nevertheless, a calibration of 14C dates for the Black Sea 5
suggests there may have been a change in Danube aquatic
resources just before 8000 years ago. Thus, a drop in fertility
might result from altered circumstances.
Figure 4 provides a summary restatement of what we can
derive from the Đerdap demographic data. Firstly, the Vlasac
and Padina data indicate a stationary population for the
Mesolithic, just as for the North European Mesolithic. Thus,
if the contact period did have a fall in fertility, the result of
resource instability, immigration or climate change, there
would have been little buffering in the event of crises.
Thirdly, while data for Neolithic samples are shown to be in-
adequate, nevertheless there is an indication of a marked in-
crease in fertility such that population increase would occur.
This can be said on the basis of Lepenski Vir data. The pres-
ence at Lepenski Vir of Neolithic material has clearly raised
the TF above the previous levels. In fact, the Neolithic TF
must be 7 as a minimum estimate. Overall, the 462 individu-
als represented in this analysis (the unlabelled triangle in
Figure 4) provide a reasonable sample that will be of great
value to palaeodemography once questions of period assign-
ments are clarified.
86
Conclusions
The evidence is imperfect, because of inadequate sample
sizes and the complication of apparent differences among
sites, perhaps resulting from the difficulty of differentiating
among periods. Our method has allowed us to identify those
samples that must be considered as problematic. Despite un-
certainties, the evidence suggests a stable and stationary
Mesolithic population in the Đerdap. While a possible ‘seep-
ing in’ of immigrants at the Mesolithic–Neolithic contact
period, indicated by the non-metric traits in Đerdap samples
(Roksandic 1999, 2000, this volume), could be indicated by
a drop in fertility and/or in-migration of adults, such low fer-
tility could not be maintained for any length of time. The
Neolithic appears to be a time of population growth. In gen-
eral, then, the pattern of stationary Mesolithic and increasing
Neolithic population structures discerned in the far west and
north of Europe is repeated and confirmed by the analysis of
samples from the Đerdap.
Notes
1. West model tables used in the regression analysis comprise 15
decreasing tables with a mean total fertility (TF) of 3.5 (SD =
0.614); 35 increasing tables with a mean TF of 8.9 (SD = 5.967);
11 stationary tables, mean TF 3.9 and SD of 0.782. TF was cal-
culated for 30 years of childbearing from the Cx column of the
life tables (the sex ratio for these 30 years is derived from the
appropriate model tables and the excess of males over females
varies between 0.01 and 0.04). TF estimates are predicted values
derived from quadratic model curve estimation regression (SPSS
v.12).
2. The total fertility rate is the number of live births a woman might
have were she to live to menopause and bear children according
to the age-specific fertility rates for the population. Note that live
births, not pregnancies, are counted. As Terrisse (1986) points
out, this theoretical figure may be shown to overestimate fertility
when tested against actual historical data.
3. The demography was based on mandibular dentition MNI es-
timates from collections in Lisbon and Porto. For discussion on
methodology and sites, see Jackes & Meiklejohn (2004) and
Jackes & Alvim (2006). The present paper was written before a
complete assessment of Arruda history and MNI: new informa-
tion is provided in Jackes & Meiklejohn (2004). We have also
now done the same for Moita, during which process the tech-
nique of estimating fertility has been tested and confirmed as re-
liable, in conjunction with a complete reassessment of the
demography (Jackes & Meiklejohn in press).
4. This explains why the figures for age distribution of individuals
in Tables 1–2 and 5–7 are not whole numbers.
5. Data from Ballard et al. (2000) calibrated with OxCal 3.5 (Bronk
Ramsay 1998) using R = 67±26 (Siani et al. 2000) and the mar-
ine curve (Stuiver & Braziunas 1993). While there are serious
questions regarding the interpretation of the Black Sea data (e.g.
Aksu et al. 2002), the reality of a period of reduced temperature
and precipitation over several hundred years around 8200 BP is
now well established (see, e.g., Barber et al. 1999; Mayewski et
al. 2004).
Acknowledgements
MR thanks Dr B. Jovanović, Prof. Ž. Mikić, and the late Prof. D.
Srejović for permission to study the Đerdap material and is grateful
to the Wenner-Gren Foundation (Grant No. 6250) and Simon Fraser
University for providing funding. Research on Portuguese materials
was funded by operating grants 410-84-0030 and 410-86-2017 from
 Mary Jackes et al: Demography of the Đerdap Mesolithic–Neolithic transition
the Social Sciences and Humanities Research Council of Canada to
D. Lubell, MJ and CM who thank Dr. M.M. Ramalho, Serviços
Geológicos de Portugal, for permission to study collections in his
care. MJ thanks D. Lubell for his help, Dr Robert McCaa for ob-
taining permission for the use of the databases collected by Steckel
and Rose, and Dr J-P. Bocquet-Appel for helpful comments. CM
thanks Dr. D.C. Merrett for a very close reading of the manuscript
prior to final submission. Two anonymous reviewers provided use-
ful suggestions.
References
Aksu, A.E., Hiscott, R.N., Kaminski, M.A., Mudie, P.J., Gillespie,
H., Abrajano, T. & Yaşar, D. 2002: Last Glacial-Holocene pa-
leoceanography of the Black Sea and Marmara Sea: stable iso-
topic, foraminiferal and coccolith evidence Marine Geology 190:
119–149
Andersson, T., Bergstrom, S., Hogberg, S. & Hogberg, U. 2000:
Swedish maternal mortality in the nineteenth century by different
definitions: previous stillbirths but not multiparity risk factor for
maternal death. Acta Obstetrica Gynecologica Scandinavica 79:
679–686.
Ballard, R., Coleman, D. & Rosenberg, G. 2000: Further evidence
of abrupt Holocene drowning of the Black Sea shelf. Marine
Geology 170: 253–261.
Barber, D.C., Dyke, A., Hillaire, M.C., Jennings, A.E., Andrews,
J.T., Kerwin, M.W., Bilodeau, G., McNeely, R., Southon, J.,
Morehead, M.D. & Gagnon, J.M. 1999: Forcing of the cold event
of 8,200 years ago by catastrophic drainage of Laurentide lakes.
Nature 400: 344–348.
Binford, L.R. 1968: Post-Pleistocene adaptations. In Binford, S. &
Binford, L.R. (eds) New Perspectives in Archaeology. Chicago:
Aldine, 313–341.
Bonsall, C., Lennon, R., McSweeney, K., Stewart, C., Harkness, D.,
Boroneanţ, V., Bartosiewicz, L., Payton, R. & Chapman, J. 1997:
Mesolithic and Early Neolithic in the Iron Gates: a palaeodietary
perspective. Journal of European Archaeology 5: 50–92.
Bonsall, C., Cook, G., Lennon, R., Harkness, D., Scott, M.,
Bartosiewicz, L. & McSweeney, K. 2000: Stable isotopes, ra-
diocarbon and the Mesolithic–Neolithic transition in the Iron
Gates. Documenta Praehistorica 27: 119–132.
Bocquet, J-P. & Masset, C. 1977: Estimateurs en paléodémographie.
L’Homme 17: 65–90.
Bocquet, J-P. & Masset, C. 1982: Farewell to palaeodemography.
Journal of Human Evolution 11: 321–333.
Bronk Ramsey, C. 1998: Probability and dating. Radiocarbon 40:
461–474.
Buikstra, J.E. & Ubelaker, D.H. (eds) 1994: Standards for Data
Collection from Human Skeletal Remains. Fayetteville: Arkansas
Archaeological Survey Research Series 44.
Coale, A.J. & Demeny, P., with B. Vaughan 1983: Regional Model
Life Tables and Stable Populations, 2nd edition. New York:
Academic Press.
Cohen, M.N. 1977: The Food Crisis in Prehistory. New Haven: Yale.
Conde-Agudelo, A & Belizán, J.M. 2000: Maternal morbidity and
mortality associated with interpregnancy interval: cross sectional
study. British Medical Journal 321: 1255–1259.
Deevey, E.S. Jr. 1960: The human population. Scientific American
203: 195–204.
Dorsten, L.E., Hotchkiss, L. & King, T.M. 1999: The effect of in-
breeding on early childhood mortality: twelve generations of an
Amish settlement. Demography 36: 263–271.
Eaton, J.W. & Mayer, A.J. 1953: The social biology of very high
fertility among the Hutterites: the demography of a unique pop-
ulation. Human Biology 25: 206–264.
87
Hajnal, J. 1982: Two kinds of preindustrial household formation
system. Population and Development Review 8: 449–494.
Holman, D.J., Wood, J.W. & Campbell, K.L. 2000: Age-dependent
decline of female fecundity is caused by early fetal loss. In te
Velde, E.R., Broekmans, F. & Pearson, P. (eds) Female
Reproductive Ageing. Studies in Profertility Series, vol 9.
Carnforth UK: Parthenon Publishing Group, 123–136.
Howell, N. 1979: Demography of the Dobe !Kung, Population and
Social Structure. New York: Academic Press.
Houdaille, J. 1995: Puritans and Virginians. Population: an English
Selection 7: 204–210.
Jackes, M. 1986: The mortality of Ontario archaeological popula-
tions. Canadian Journal of Anthropology 5: 33–48.
— 1988: Demographic change at the Mesolithic–Neolithic trans-
ition: evidence from Portugal. Rivista di Antropologia (supple-
ment) 66: 141–158.
— 1992: Paleodemography: problems and techniques. In Saunders,
S.R. & Katzenberg, M.A. (eds) Skeletal Biology of Past Peoples:
Research Methods. New York: Wiley-Liss, 189–224.
— 1993: On paradox and osteology. Current Anthropology 34:
434–439.
— 1994: Birth rates and bones. In Herring, A. & Chan, L. (eds)
Strength in Diversity: a Reader in Physical Anthropology.
Toronto: Canadian Scholar’s Press, 155–185.
— 2000: Building the bases for paleodemographic analysis: adult
age determination. In Katzenberg, M.A. & Saunders, S.R. (eds)
Biological Anthropology of the Human Skeleton. New York:
John Wiley and Sons, 407–456.
Jackes, M. & Alvim, P. 2006: Reconstructing Moita do Sebastião,
the first step. In Bicho, N. & Veríssimo, N.H. (eds) Do
Epipaleolítico ao Calcolítico na Península Ibérica. Faro: Centro
de Estudos de Património, 13–25.
Jackes, M. & Lubell, D. 1999a: Human skeletal biology and the
Mesolithic–Neolithic transition in Portugal. In Thévenin, A. (ed.)
L’Europe des derniers chasseurs: Épipaléolithique et
Mésolithique. Actes du 5e colloque international UISPP, com-
mission XII, Grenoble, 18–23 septembre 1995. Paris: Éditions du
CTHS, 59–64.
Jackes, M. & Lubell, D. 1999b: Human biological variability in the
Portuguese Mesolithic. Arqueologia 24: 25–42.
Jackes, M., Lubell, D. & Meiklejohn, C. 1997a: Healthy but mortal:
human biology and the first farmers of western Europe. Antiquity
71: 639–658 (supplementary material at http://intarch.ac.uk/
antiquity).
Jackes, M., Lubell, D. & Meiklejohn, C. 1997b: On physical an-
thropological aspects of the Mesolithic–Neolithic transition in
the Iberian Peninsula. Current Anthropology 35: 839–846.
Jackes, M. & Meiklejohn, C. 2004: Building a method for the study
of the Mesolithic–Neolithic transition in Portugal. Documenta
Praehistorica 31: 89–111.
Jackes, M. & Meiklejohn, C. In press: The palaeodemography of
central Portugal and the Mesolithic–Neolithic transition. In
Bouquet-Appel, J.-P. (ed.) Recent Advances in Paleodemog-
raphy: Data, Techniques, Patterns. New York: Springer Verlag.
Jackes, M., Silva, A.M. & Irish, J. 2001: Dental morphology – a
valuable contribution to our understanding of prehistory. Journal
of Iberian Archaeology 3: 97–119.
Larsen, U. & Vaupel, J.W. 1993: Hutterite fecundability by age and
parity: strategies for frailty modeling of event histories.
Demography 30: 81–102.
Lubell, D., Jackes, M., Schwarcz, H., Knyf, M. & Meiklejohn, C.
1994: The Mesolithic–Neolithic transition in Portugal: isotopic
and dental evidence of diet. Journal of Archaeological Science
21: 201–216.
McCaa, R. 2002: Paleodemography of the Americas from ancient
times to colonialism and beyond. In Steckel, R.H. & Rose, J.C.
 The Iron Gates in Prehistory
(eds) The Backbone of History: Health and Nutrition in the
Western Hemisphere. New York: Cambridge University Press,
94–124.
Mayewski, P.A., Rohling, E. E., Stager, J.C., Karlén, W., Maasch,
K.A., Meeker, L. D., Meyerson, E.A., Gasse, F., van Kreveld, S.,
Holmgren, K., Lee-Thorp, J., Rosqvist, G., Rack, F.,
Staubwasser, M., Schneider, R.R. & Steig, E. J. 2004: Holocene
climate variability. Quaternary Research 62: 243–255.
Meiklejohn, C. 1978: Review of M.N. Cohen, ‘The Food Crisis in
Prehistory: Overpopulation and the Origins of Agriculture’. Yale
University Press [1977]. American Journal of Physical
Anthropology 49: 286–287.
— 1979: Ecological aspects of population size and growth in Late-
Glacial and early Postglacial north-western Europe. In Mellars,
P.A. (ed.) The Early Postglacial Settlement of Northern Europe:
an Ecological Perspective. London: Duckworth, 65–79.
Meiklejohn, C. & Denston, B. 1987: The human skeletal material:
inventory and initial interpretation. In Mellars, P.A. (ed.)
Excavations on Oronsay: Prehistoric Human Ecology on a Small
Island. Edinburgh: Edinburgh University Press, 290–300.
Meiklejohn, C. & Zvelebil, M. 1991: Health status of European
populations at the agricultural transition and the implication for
the adoption of farming. In Bush, H. & Zvelebil, M. (eds) Health
in Past Societies: Biocultural Interpretations of Human Skeletal
Remains in Archaeological Contexts. BAR International Series
567. Oxford: Tempus Reparatum, 129–145.
Meiklejohn, C., Schentag, C.T., Venema, A. & Key, P. 1984:
Socioeconomic change and patterns of pathology and variation
in the Mesolithic and Neolithic of western Europe: some sug-
gestions. In Cohen, M.N. & Armelagos, G.J. (eds)
Paleopathology at the Origins of Agriculture. Orlando:
Academic Press, 75–100.
Meiklejohn, C., Wyman, J.M., Jacobs, K. & Jackes, M.K. 1997:
Issues in the archaeological demography of the agricultural
transition in western and northern Europe: a view from the
Mesolithic. In Paine, R.R. (ed.) Integrating Archaeological
Demography: Multidisciplinary Approaches to Prehistoric
Population. Southern Illinois University, Center for
Archaeological Investigations, Occasional Paper no. 24.
Carbondale: Southern Illinois University, 311–326.
Montgomery, M.R & Cohen, B. (eds) 1998: From Death to Birth:
Mortality Decline and Reproductive Change. Washington, D.C.:
National Academy Press.
Müller, H-G. & Love, B. 1999: ‘A semi-parametric model for es-
timating a lifetime distribution from paleodemographic data’.
Paper presented at the Workshop Mathematic Modeling in
Paleodemography: Coming to Consensus, Rostock, Germany,
9–12 June 1999.
Nemeskéri, J. 1978: Demographic structure of the Vlasac
Epipaleolithic population. In Srejović, D. & Letica, Z. (eds)
Vlasac: A Mesolithic Settlement in the Iron Gates, vol. 2.
Belgrade: Serbian Academy of Sciences and Arts, 97–134.
Ober, C, Hyslop, T. & Hauck, W.W. 1999: Inbreeding effects on
fertility in humans: evidence for reproductive compensation.
American Journal of Human Genetics 64: 225–231.
Phenice, T.W. 1969: A newly developed visual method of sexing the
Os pubis. American Journal of Physical Anthropology 30:
297–302.
Price, T.D. & Brown, J.A. (eds) 1985: Prehistoric Hunter-
Gatherers: the Emergence of Cultural Complexity. Orlando:
Academic Press.
Radosavljević-Krunić, S. 1986: Resultats de l’étude anthropolo-
gique des squélettes provenant du site Ajmana. Ðerdapske sveske
3: 51–58.
88
Radovanović, I. 1996a: The Iron Gates Mesolithic. Ann Arbor:
International Monographs in Prehistory.
— 1996b: Mesolithic–Neolithic contacts: a case of the Iron Gates
region. Poročilo o raziskovanju paleolitika, neolitika i eneolitika
v Sloveniji 23: 39–48.
— 2000: Houses and burials at Lepenski Vir. European Journal of
Archaeology 3: 330–349.
Roksandic, M. 1999: Transition from Mesolithic to Neolithic in the
Iron Gates Gorge: Physical Anthropology Perspective.
Unpublished PhD thesis, Department of Anthropology, Simon
Fraser University, Burnaby, B.C.
— 2000: Transition from Mesolithic to Neolithic in the Iron Gates
gorge: physical anthropology perspective. Documenta
Praehistorica 27: 1–100.
— In preparation. ‘Extra individuals at Lepenski Vir: burial infill or
grave inclusions?’.
Siani, G., Paterne, M., Arnold, M., Bard, E., Métivier, B., Tisnerat,
N. & Bassinot, F. 2000: Radiocarbon reservoir ages in the
Mediterranean Sea and Black Sea. Radiocarbon 42: 271–280.
Smith, P.E.L. & Young, T.C. Jr. 1972: The evolution of agriculture
and early culture in greater Mesopotamia: a trial model. In
Spooner, B. (ed.) Population Growth: Anthropological
Implications. Cambridge, MA: M.I.T. Press, 1–59.
Srejović, D. & Letica, Z. 1978: Vlasac: A Mesolithic Settlement in
the Iron Gates, 2 volumes. Belgrade: Serbian Academy of
Sciences and Arts.
Steckel, R.H, Sciulli, P.W. & Rose, J.C. 2002: A health index from
skeletal remains. In Steckel, R.H. & Rose, J.C. (eds) The
Backbone of History: Health and Nutrition in the Western
Hemisphere. New York: Cambridge University Press, 61–93.
Stuiver, M. & Braziunas, T. 1993: Modelling atmospheric 14C in-
fluences and 14 C ages of marine samples to 10,000 BC.
Radiocarbon 35: 137–189.
Terrisse, M. 1986: Reconstitution des familles en Scandinavie.
Annales de démographie historique: 325–352.
Tilley, C.Y. 1996: An Ethnography of the Neolithic. Cambridge:
Cambridge University Press.
United Nations 1982: Model Life Tables for Developing Countries.
Department of International Economic and Social Affairs,
Population Studies no. 77. New York: United Nations.
Vann, R.T. 1999: Unnatural infertility, or, whatever happened in
Colyton? Some reflections on ‘English population history from
family reconstitution 1580–1837’. Continuity and Change 14:
91–104.
Von Grafenstein, U., Erlenkeuser, H., Müller, J., Jouzel, J. &
Johnsen, S. 1998: The cold event 8200 years ago documented in
oxygen isotope records of precipitation in Europe and
Greenland. Climate Dynamics 14: 73–81.
Weeks, J.R. 1996: Population: an Introduction to Concepts and
Issues. 6th edition. Belmont (California): Wadsworth Publishing
Company.
Workshop of European Anthropologists 1980: Recommendations
for age and sex diagnosis of skeletons. Journal of Human
Evolution 9: 517–549.
Zoffmann, Zs. 1983: Prehistoric skeletal remains from Lepenski Vir
(Iron Gates, Yugoslavia). Homo 34: 129–148.
Zhu, B.P., Rolfs, R.T., Nangle, B.E. & Horan, J.M. 1999: Effect of
the interval between pregnancies on perinatal outcomes. New
England Journal of Medicine 340: 589–594.
Živanović, S. 1975: A note on the anthropological characteristics of
the Padina population. Zeitschrift für Morphologie und
Anthropologie 66: 161–175.
 Approaches to Starčevo culture chronology

Joni L. Manson

Abstract: The Early Neolithic period in the central and northern Balkans is represented by the cultures of the Starčevo-Körös-Criş complex.
Pottery is considered one of the hallmarks of this culture complex, and the dating of Early Neolithic sites in the region has typically relied
on typological sequences based on analyses of pottery decoration. Radiocarbon dating got off to a slow start in the Balkans partly because
of the long-held emphasis on seriation and partly because of the poor preservation of organic material at many sites. A third dating technique,
based on archaeomagnetic intensity analyses, holds great promise for dating these pottery-rich sites. The results of archaeomagnetic intensity
analyses on potsherds from twelve Starčevo sites are described. By combining the typological sequences, radiocarbon dates, and archaeo-
magnetic intensity data it is possible to propose a more refined chronology of the Starčevo culture.
Key words: Starčevo culture, chronology, Neolithic, pottery, radiocarbon, archaeomagnetic intensity

Introduction
The Starčevo culture was named for the archaeological site
Starčevo-Grad, located just west of the small village of
Starčevo, some 20 km east-northeast of Beograd, Serbia.
Starčevo-Grad lies on the northern (left) bank of the Danube,
practically opposite the later Neolithic site, Vinča. M. Grbić
conducted brief test excavations at Starčevo-Grad in 1928. A
joint Yugoslav–American team carried out larger-scale ex-
cavations in 1931 and 1932. The concept of an early
Neolithic ‘Starčevo culture’ was defined as a result of this
work (Fewkes et al. 1933; Ehrich 1977; Benac et al. 1979a).
Starčevo sites are concentrated in the central Balkan and
southern Pannonian region, particularly Serbia, including
Vojvodina and Kosovo, as well as portions of Croatia,
Bosnia-Hercegovina, and Macedonia (Fig. 1). The Starčevo
culture shares many similarities with roughly contemporan-
eous neighbouring cultures, especially the Anzabegovo-
Vršnik culture (Macedonia), the Karanovo culture (southern
Bulgaria), the Kremikovci culture (western Bulgaria), the
Criş culture (Romania), and the Körös culture (Hungary). For
detailed information on these culture groups, see Tringham
(1971), Girić (1974), Brukner (1979), Dimitrijević (1979),
Garašanin (1979), Radišić (1984), Manson (1990), Whittle
(1996), and Bailey (2000). Although the territory of many of
these groups roughly corresponds to modern political
boundaries, it is possible that they may also reflect slight
variations in adapting to the many diverse microenviron-
ments of the Balkan peninsula (Kaiser 1984). Typically, these
cultures are considered part of the Early Neolithic period in
this area (Fig. 2). Many Iron Gates sites (e.g. Lepenski Vir,
Schela Cladovei, Padina, and Hajdučka Vodenica) have Early
Neolithic Starčevo and Starčevo-Criş occupations.
The term ‘Neolithic’ (as it is used in the central Balkans)
implies an agricultural and stockraising economy as well as
the production of polished stone tools and pottery (e.g. Benac
et al. 1979a). However, organic materials tend to be poorly
preserved at many Starčevo sites. Combined with a slow ac-
ceptance of the practices required for the systematic retrieval
of the floral and faunal remains necessary for detailed eco-
nomic and ecological analyses, this has meant that the inter-
pretation of many sites as ‘Neolithic’ has depended almost
entirely on the presence of a single artefact type — pottery.
Starčevo pottery and typological sequences
Potsherds are by far the most commonly recovered artefacts
from Starčevo sites, accounting for almost 95% of the total
assemblage from some sites (Dimitrijević 1979). The pottery
is generally described in terms of its fabric, form, and decor-
ation. Strong similarities exist in the pottery found at
Starčevo, Körös, Anzabegovo-Vršnik, and Veluško-Porodin.
Starčevo ceramics include coarse, medium, and fine fab-
rics, or wares. Paste texture and surface decoration are the
major factors involved in the designation of wares. Coarse
ware is predominant at all Starčevo sites. In coarse ware
ceramics, the clay often appears to have been poorly cleaned.
At Starčevo-Grad, the native loess underlying the site was
used for pottery production (Fewkes et al. 1933; see also
Horton 1938). The clay itself was relatively fine-textured, but
the pottery also contained chaff, mica, sand, and occasionally
small pebbles. It is assumed that the chaff would have been
purposely added as a tempering material; some grain im-
pressions on pottery have been identified as cultivated cer-
eals. The sand, mica, and pebbles may have been natural ac-
cessories in the raw clay or they may have been intentionally
added. Coarse ware vessels were often large and thick-
walled. This pottery usually appears to be low-fired, with a
dark central core. Exteriors may be light brown to red in col-
our; interior surfaces are usually darker and may be finely
polished, possibly to reduce vessel wall permeability
(Garašanin 1984; Manson 1990).
Medium ware closely resembles the coarser ware in clay
texture and tempering material. Vessel walls may be thinner,
but they have the same dark core. Both surfaces are usually
light brown, although hints of red or yellow may be seen.
Fine ware contains little or no organic tempering, but fine
sand or mica is usually present. The vessel walls are thin and
usually even-coloured throughout. Surface colours cover a

89
 The Iron Gates in Prehistory
Figure 1. Geopolitical map of the central and northern Balkans.
wide range from yellow to buff to red to brown or grey.
Starčevo pottery occurs in a limited range of forms. The
basic shape is a spherical or hemispherical pot, regardless of
fabric or size. Vessels with flaring walls, and shouldered pots
with cylindrical necks of various heights, are found in smal-
ler numbers. Straight or slightly thickened bases appear on
vessels of every fabric but are more common on coarse ware
pots. Low ring bases are also associated with all fabric types,
but higher ring bases and true pedestal bases are more com-
mon on fine ware vessels. Some larger vessels had handles,
usually in the form of protuberances (some with holes to fa-
cilitate hanging or carrying), but band or strap handles are
also found.
Most Starčevo ceramic analyses have directed their atten-
tion to pottery decoration. In very broad terms, coarse ware is
associated with barbotine surfaces, medium ware with im-
pressed or incised decoration, and fine ware with painted
decoration. The term ‘barbotine’ is credited by Fewkes
(1938) to M. Vasić, who first used the term to describe the
coarse pottery found in the lowest levels at Vinča. Fewkes
considered coarse barbotine ceramics to be the ‘chief do-
mestic ware’ of the Starčevo culture.
Two major variants of barbotine decoration have been
defined. In the first, small clumps of clay appear to have been
spattered in an irregular or regular manner on the exterior
surface of a vessel. In the second, a rough coating of clay was
applied and then streaked over with sticks or fingers, giving
the appearance of a barbotine coating that was arranged or
organized into strips or ridges. In cross-section, the barbotine
90
coating may appear to be a different colour than the vessel
wall, but the coating consists of the same material as the wall.
It has the same fabric texture and contains the same inclu-
sions (Horton 1938). Although barbotine is generally con-
sidered a decorative technique, the roughened surfaces would
also have made the vessel less slippery to handle (Kutzián
1947) and may have increased the vessel’s resistance to
thermal shock by increasing the exterior surface area (Kaiser
1984).
Incised and impressed ornamentation appears on both
coarse and (more commonly) on medium ware pots. Incising
was less common, and generally took the form of a net-like
pattern across the body of the vessel. Impressed designs were
most often executed by means of fingertips or fingernails,
either in definite patterns or unevenly spaced. Finger and fin-
gernail impressions also appear on appliqué bands or
rosettes. Occasionally impressions were made with an in-
strument (such as a reed) which left circular designs in the
clay. A number of medium ware pots were simply smoothed,
or sometimes burnished, and left undecorated.
Although painted pottery is considered a diagnostic trait of
the Early Neolithic Balkan cultures, painted ceramics make
up a very small fraction of the total ceramic inventory at most
of these sites. Pots were covered with a wash (interior and
exterior), painted, burnished, and then fired (Fewkes et al.
1933). Three types of painting occur: white painting on a
reddish background, dark painting on a reddish background,
and polychrome (usually white outlining dark) painting on a
reddish background. The dark paint colours were probably
 Joni Manson: Approaches to Starčevo culture chronology
Figure 2. Neolithic cultures of the Balkans.
derived from iron oxides, while the white paint is thought to
have an organic origin (Dimitrijević 1979). Painting is asso-
ciated with fine wares, although some fine ceramics were
burnished and left unpainted (‘monochrome’ ceramics).
Common motifs of white painted ceramics include gar-
lands, zigzags, and crosshatch designs near the rim. Groups
of vertical lines, joined on the lower portion of the pot’s body,
are found, as well as some curvilinear designs and (rarely) the
spiral motif. Dark painted motifs include the above, plus tri-
angles, groups of vertical bands filled with a crosshatch
design, several types of spirals (including ones that end in
‘claws’), and meanders. Sometimes a rim will have a dark
band painted around the top or slightly inside. Polychrome
painting is relatively rare. In most cases, dark designs are
outlined with white. Most motifs are curvilinear (including
spirals) but some linear designs also occur in polychrome.
The earliest published chronological sequence of pottery
from Starčevo sites was devised by Vladimir Milojčić
(1950). By comparing pottery from various sites, he estab-
lished a four-part ceramic sequence and relative chronology,
Starčevo I–IV (Fig. 3). The next serious work on a Starčevo
seriation was carried out by Draga Aranđelović-Garašanin
(1954). Her sequence closely followed that of Milojčić but
recognized closer similarities between the middle two peri-
ods. Thus, the stages were labelled Starčevo I, IIa, IIb, and
III. Aranđelović-Garašanin based much of her analysis on
about 50,000 sherds recovered during the early excavations
at Starčevo-Grad, particularly material from pit 5A, which
she considered a closed, stratified feature.
91
According to Aranđelović-Garašanin’s (1954) sequence,
painted ceramics are absent from Starčevo I; coarse wares,
particularly barbotine ceramics, predominate but fine burn-
ished monochrome ceramics are also fairly common. Incised,
impressed, and appliquéd pottery is present but not common.
Garašanin (1979) includes materials from Lepenski Vir III,
Donja Branjevina, Gura Baciului, Cîrcea, Grivac, and
Divostin I in the Starčevo I period (Fig. 4).
Starčevo II is considered the beginning of ‘classical’
Starčevo. In Starčevo IIa, coarse barbotine ceramics pre-
dominate, but fine ware makes up a slightly higher percent-
age of the total ceramic assemblage. The proportion of ap-
pliquéd, impressed, and incised sherds also increases slightly.
White and dark painted linear and curvilinear designs (in-
cluding spirals) make an appearance. In Starčevo IIb, white
painted designs disappear, except for an occasional poly-
chrome piece, and the frequency of all styles of painted
ceramics declined.
During Starčevo III, the percentage of fine ware increased
again. Coarse barbotine is still predominant, but it was now
more often applied in a regular, organized pattern. There is a
somewhat higher frequency of appliquéd, impressed, and in-
cised ceramics compared to the preceding period. The same
dark painted designs continued, with spirals (especially in
polychrome) becoming more common. The frequency of
biconical forms, and high pedestals also increased.
Aranđelović-Garašanin based the Starčevo III period on finds
from other sites because material from this period was not
collected at Starčevo-Grad until the 1969–70 excavations.
 The Iron Gates in Prehistory
Figure 3. Typological sequences of Starčevo ceramics.
Criticism of the Aranđelović-Garašanin (1954) typologic-
al sequence has centred on the reliability of the stratigraphy
of pit 5A at Starčevo-Grad and questions regarding the use of
painted pottery as a major distinguishing trait of the se-
quence. Typically, painted sherds make up less than 2% of the
total ceramic assemblage from Starčevo sites. See Korošec
(1973) and Ehrich (1977) for discussions on the stratigraphy
of Starčevo-Grad and a statistical re-examination of
Aranđelović-Garašanin’s typology. Srejović (1972) proposed
combining Aranđelović-Garašanin’s Starčevo IIa, IIb, and III
into a Classical Starčevo period. He saw little evidence of a
Starčevo I period, and preferred to call materials from
Lepenski Vir III, ‘Proto-Starčevo’ (see also Vetnić 1986).
An alternative ceramic typology was offered by Stojan
Dimitrijević (1974, 1979). His sequence begins with a two-
stage ‘Preclassic’ Starčevo, characterized by the absence of
barbotine ceramics. The first part of this Preclassic is called
the Monochrome Phase, the second is the White Linear
(Linear A) Phase. This is followed by a two-stage ‘Classic’
Starčevo, which includes the Dark Linear (Linear B) Phase
and the Garlandoid Phase. The major feature of
Dimitrijević’s Classic Starčevo is the very large proportion of
barbotine pottery. The Dark Linear Phase is characterized by
such motifs as vertical bands, zigzags, triangles, and
crosshatching executed in dark paint. In the Garlandoid
Phase the major motif is a garland, executed in either dark or
white paint. Dimitrijević’s Classic Starčevo corresponds to
Aranđelović-Garašanin’s Starčevo II period.
Dimitrijević’s ‘Late Classic’ Starčevo also consists of two
phases: Spiraloid A and Spiraloid B. The spiral motif first
appears in Spiraloid A and increases in frequency in Spiraloid
B. A ‘Final’ Starčevo period follows the Late Classic. This
period appears to be restricted to peripheral sites and the
pottery exhibits much more similarity to Körös forms than to
Starčevo ones. A majority of the pottery consists of finger-
streaked or spattered barbotine and appliquéd specimens.
Dimitrijević (1979) suggested that this Final Starčevo period
may have served as the basis of the Linear Banded Pottery
92
culture (LBK) of Central Europe (cf. Milisauskas & Kruk
1989; Bogucki & Grygiel 1993; Bogucki 1995; Gronenborn
1999).
The Dimitrijević typology seems to have more relevance
for the northern and western Starčevo sites (in Croatia) than
for the majority of the Starčevo sites, and has not been readily
accepted by archaeologists working in the region. In spite of
the criticisms levelled against it, the ceramic typology of
Aranđelović-Garašanin (1954) is now deeply embedded in
the literature of this region and is still relied on for ceramic
sequencing. The system does appear to be supported at the
few Starčevo sites that have vertical stratigraphy, especially
Rudnik at Metohija (Garašanin 1979). In fact, this pottery
typology often serves as the only source of chronological
control for the Starčevo culture because vertically stratified
sites and radiocarbon dates are both scarce (e.g. Brukner
1979; Garašanin 1979).
The chronological relationship between Körös and
Starčevo sites also remains somewhat problematic. In many
cases, the internal chronology of the Körös culture was based
on a correlation with Starčevo material. Most of this research
associated Early Körös with Starčevo II and Late Körös with
Starčevo III (Brukner 1968, 1979; Brukner et al. 1974;
Radišic 1984). Brukner (1979) added that the Late Körös
period may have continued up to the beginning of the Vinča-
Tordoš (B1) period.
Radiocarbon dates
Many of the archaeologists specializing in Starčevo studies
were slow to accept the radiocarbon dating technique (Benac
et al. 1979b; see also Evans & Rasson 1984), and relatively
few 14C dates have been published for Starčevo sites although
the situation is improving (e.g. Whittle et al. 2002).
Throughout this article, radiocarbon dates are given as years
BP; calibrated dates are cited as cal BC. As some referenced
sources gave radiocarbon dates and some gave calibrated
 Joni Manson: Approaches to Starčevo culture chronology
Figure 4. Selected Neolithic sites in the Balkans.
dates, conversions were performed with the CALIB 5 pro-
gram distributed by the Quaternary Isotope Laboratory of the
University of Washington (Stuiver & Reimer 1993; Stuiver et
al. 2005) using the IntCal04 curve (Reimer et al. 2004).
In Macedonia, the eleven 14C dates for Anzabegovo I in-
dicate a temporal span from about 7250 to 7100 BP
(c. 6100–6000 cal BC). Eight dates for Anzabegovo II span
the period of about 7050 to 6850 BP (c. 5950–5750 cal BC).
And four dates for Anzabegovo III indicate a span from about
6850 to 6500 BP (c. 5750–5500 cal BC) (Gimbutas 1976).
Gimbutas correlates Anzabegovo I with the Gura Baciului
culture, Anzabegovo II with Starčevo IIa, and Anzabegovo
III with Starčevo IIb and III.
A radiocarbon date for the earliest level of the stratified
site of Veluška Tumba, also in Macedonia, shows that
Veluška Tumba I dates to about 6950 BP (c. 5800 cal BC),
while Veluška Tumba II was dated to about 6700 BP (5600
cal BC) (Todorović et al. 1977). Garašanin (1979) correlated
Veluška Tumba I and II to the Gura Baciului culture, Veluška
Tumba III to Anzabegovo II (therefore Starčevo IIa), and
Veluška Tumba IV to Anzabegovo III (therefore Starčevo IIb
and probably III).
A single 14C date for the Starčevo settlement at Gornja
Tuzla in Bosnia-Hercegovina indicates an occupation around
6650 BP (c. 5600 cal BC). The site was correlated with
Starčevo III (Tasić & Tomić 1969).
Bone samples from the later excavations at Starčevo-
Grad, Serbia, yielded five radiocarbon dates between 6850
and 6650 BP (c. 5750–5600 cal BC) (Ehrich 1977). Seven
14C dates from Divostin I span the period from about 7200 to
93
6900 BP (c. 6050–5750 cal BC). The Starčevo occupation at
Grivac dates to about 7250 BP (c. 6100 cal BC), that at Banja
to about 7050 BP (c. 5950 cal BC) (McPherron 1988).
Garašanin (1979) believed Divostin I could be correlated
with Starčevo I, while Gimbutas (1974) fitted Divostin I into
Starčevo IIa.
Radiocarbon dates from Padina indicate a Starčevo occu-
pation from about 7100 to 6550 BP (c. 5950–5500 cal BC)
(Clason 1980). Similar dates have been obtained from char-
coal samples from the houses at Lepenski Vir that contained
Starčevo pottery. A series of 14C dates from these houses in-
dicates a Starčevo occupation around 7150 to 6750 BP
(c. 6000–5650 cal BC) (Budja 1999; Bonsall et al. 2000).
Recently two studies focusing on the origins of the LBK
culture of Central Europe looked at 14C dates from Starčevo
and Körös sites (Gläser 1991; Lenneis & Stadler 1995).
Gläser proposed dates of 5950 to 5470 cal BC for the dura-
tion of the Starčevo culture and 5700 to 5260 cal BC for the
Körös culture. Lenneis & Stadler’s study dated the Starčevo
culture to 6050 to 5450 cal BC, with dates of 5760 to 5280 cal
BC for the Körös culture. Research reported by Hertelendi et
al. (1995) dated Early Körös sites to 5950–5400 cal BC, and
Late Körös sites to 5770–5230 cal BC.
Archaeomagnetic intensity dating
The plasticity of clay and the nature of pottery-making result
in products well suited to stylistic analyses and seriation.
However, the physical characteristics of the clay also allow
 The Iron Gates in Prehistory
us to date pottery by means of archaeomagnetic intensity
analysis.
Clay often contains magnetic minerals such as magnetite
and haematite as impurities. When clay is heated to a tem-
perature higher than the ‘blocking temperature’ of these
minerals, the magnetic particles can record the direction and
strength of the earth’s magnetic field in that location at that
particular time. Blocking temperatures fall just below a
magnetic mineral’s Curie point (the temperature above which
a substance loses its ferromagnetic properties). The blocking
temperatures of magnetite and haematite lie in the range of
500° C to 700° C.
The primary remanent magnetization acquired by heating
clay above the blocking temperature of its magnetic minerals
is called thermoremanent magnetization (TRM). Since the
geomagnetic field changes direction and strength over time
and space, it is possible to determine when an artefact of
baked clay was last subject to temperatures high enough to
allow acquisition of TRM by comparing its magnetic para-
meters with the known geomagnetic record for a particular
region. Although the directional data is not particularly useful
when dealing with portable objects such as pottery or figur-
ines, these artefacts can still provide important information in
the form of intensity data.
Most of the techniques now being used to determine mag-
netic intensities of baked clay artefacts are variations of the
method devised by Thellier & Thellier (1959). This method
of determining ancient geomagnetic intensity involves the
use of double heatings at a series of temperature steps. The
Thellier–Thellier technique depends on the principle that
magnetic grains which have acquired a remanent magnetiza-
tion after cooling through the blocking temperature will also
lose that remanent magnetization at the same temperature if
reheated. A sample of baked clay usually has such a variety of
magnetic grains that the blocking temperatures are continu-
ously distributed from the Curie temperature downward
(Aitken 1983: 19). Therefore, a series of laboratory reheat-
ings to increasingly higher temperatures up to the Curie point
will yield a series of partial TRMs which can be plotted
against the natural remanent magnetization (NRM) being re-
moved from the sample. The NRM is generally considered to
be the magnetism a sample has before it is subjected to ex-
perimental procedures in the laboratory. The progressive de-
crease of NRM plotted against the progressive increase of
TRM should result in a straight line with a slope proportional
to the ancient field intensity. This relationship can be ex-
pressed as: FA = FL x Δ NRM / Δ TRM, where FA is the an-
cient magnetic field intensity, FL is the laboratory field in-
tensity, and Δ NRM / Δ TRM is the absolute value of the
slope of the line.
Pottery, which has been described as pyrophilic in nature,
is generally well suited to the multiple heatings of the
Thellier–Thellier method (Carmichael 1977). The first step in
this procedure is to measure the NRM of the sample. Then the
sample is heated to a temperature, T1, and allowed to cool in a
known magnetic field, FL, with a known orientation. The re-
manent magnetism, J(T1, 0), resulting from this step is meas-
ured. The sample is rotated 180° about a horizontal axis, re-
heated to T 1 , cooled in F L again, and the remanent
magnetism, J(T1, 180), is measured. The vector sum of J(T1,
94
0) and J(T1, 180) equals twice the PTRM (partial thermore-
manent magnetization) that the sample acquired in the an-
cient field when it cooled from above its Curie point, Tc , to
T1 . The vector difference of J(T1, 0) and J(T1, 180) equals
twice the PTRM acquired in field FL between T1 and room
temperature (Collinson 1983: 427–428). This double heating
process is repeated at progressively higher temperatures up to
the Curie point. One curve can then be generated from the
sums of the measurements obtained from the initial and re-
versed positions for each temperature, while a second curve
can be derived from the differences between these measure-
ments (Michels 1973: 144). A third curve, the Arai diagram
(Nagata et al. 1963), or NRM–TRM diagram, generally de-
picts NRM lost on a y-axis and TRM gained on an x-axis.
The Thellier–Thellier method has features built into it that
allow researchers to check the reliability of the results and to
discard those which cannot pass these checks. In fact, it is not
unusual for the rate of failure to be relatively high — often
more than one-third of the samples must be rejected on the
basis of a lack of internal consistency (e.g. Sasajima 1965;
Boyd 1986; Wei et al. 1987). A number of explanations have
been offered to account for deviations from ‘ideal’ behaviour,
which often show up in the form of non-linear curves on the
Arai plots. Many researchers have suggested guidelines to
follow when determining which results can be considered
reliable (Table 1).
Manson (1990) subjected 101 samples of Starčevo pottery
to archaeomagnetic intensity analysis (Table 2). The standard
Thellier–Thellier method was used throughout the laboratory
procedure, which was carried out at the Paleomagnetism Lab
at the University of Pittsburgh, under the direction of Victor
Schmidt. The NRM (at 20° C) of each sample was first es-
tablished using a cryogenic magnetometer from
Superconducting Technology, Inc. (SCT). The first four
samples (from Divostin, Vinča, and Starčevo) were pro-
cessed using a spinner magnetometer from Princeton Applied
Research. All of the other specimens were processed with the
cryogenic, or SQUID (Superconducting Quantum
Interference Device), magnetometer. The heating and cool-
ing of the specimens was carried out in a Schonstedt thermal
specimen demagnetizer.
Of the 101 samples processed, 65 yielded results con-
sidered to be acceptable. Samples that lacked at least three
points on a straight-line segment of the NRM–TRM plots
were immediately rejected. The remaining samples were then
checked for consistency on Zijderveld diagrams, which
combine intensity and directional changes on the same plot
(Fig. 5). Regularity on these diagrams ensures that loss of
TRM and gain of NRM are both univectoral, indicating that
there are no additional components complicating the picture.
Samples with irregular Zijderveld diagrams were also rejec-
ted from the study. Of the samples considered acceptable, the
number of points on the straight-line segment of the
NRM–TRM plot averaged 5 or 6. Ordinarily, the mid-range
of temperature points formed the linear part of the plot.
Furthermore, the linear portion of the plots covered an aver-
age span of 60% of the total extrapolated NRM.
The slopes of the lines were calculated using SAS pro-
grams. In order to facilitate comparisons with the published
master curves derived from other archaeomagnetic intensity
 Joni Manson: Approaches to Starčevo culture chronology

Table 1. Determining the reliability of NRM–TRM diagrams.

Source Minimum # of Temperature (°C) Additional Comments

Temperature Points Range to Use

Dubois & Watanabe (1965) —— 100–400 ——

Sasajima (1965) —— —— Don’t use non-linear plots (this eliminated >40% of his sample).

Weaver (1966) —— 150–450 ——

Bucha (1967) 4 —— Points should cover 2/3 of the distance on the line between the Y-
and X-axes.

Bucha (1967) 3 —— Standard deviation from the line should not exceed ± 10%.

Coe (1967) 6 —— Standard deviation from the line should not exceed ± 5%.

Bucha et al. (1970) 4 —— ——

Rusakov & Zagniy (1973) 6 —— ——

Barbetti et al. (1977) —— —— Don’t use high temperature points of concave-up lines.

Levi (1977) —— —— Linear plot is a necessary but insufficient condition for judging
reliability.

Coe et al. (1978) 4 —— Points must span at least 15% of the total extrapolated NRM.

Kovacheva (1980) —— mid-range ——

Tanaka (1982) —— —— Need at least 3 samples per site.

Arbour & Schwartz (1982) 3 —— ——

Kovacheva (1983) 6 mid-range ——

Thomas (1983) —— —— Portion of plots used is highly subjective.

Walton (1983) —— mid-range Linearity is no guarantee that mineral alteration hasn’t occurred.

Aitken (1983) —— —— Need regular NRM demagnetization as well as linear plots.

Aitken et al. (1983)
—— —— Linear portion should cover 200°C and 60% of the demagnetization
range (for Class A); or 150°C and 50% of the range (for Class B+);
TRM capacity must not change more than 10% (Class A) or 20%
(Class B+) between 200°C and the upper point of the linear portion;
unless VRM is present, the 150°C point should be no more than
10% (Class A) or 20% (Class B+) away from an extension of the
line; the 550°C point should be no more than 10% (Class A) or 40%
(Class B+) away from an extension of the line.

Sakai & Hirooka (1985) —— —— The more points used, the better the reliability (in their sample, 172
out of 300 specimens gave ‘good’ results).

Burakov & Nachasova (1985) —— up to and incl. 450°C ——

Kovacheva & Zagniy (1985) —— mid-range Use small temperature steps (25°C) to insure having enough points
on a line.

Wei et al. (1987) —— —— Standard deviation should not exceed ± 6%.

Kovacheva (1995) 4 —— Remanence direction must remain stable; allow <10% change in
PTRM capacity; precision increases as sample number increases.

95
 The Iron Gates in Prehistory

Table 2. Sites (with geographical coordinates), numbers of samples, and modern magnetic
intensity (F0 = intensity at time of collection).

Site Name # Samples # Samples Latitude (°N) Longitude (°E) F0 (oe.)

Tried Accepted

Ludaš-Budžak 26 15 46° 10´ 19° 50´ 0.46910

At 7 2 45° 08´ 21° 15´ 0.47160

Nosa 4 3 46° 10´ 19° 50´ 0.46690

Kaonik-Gradina 5 2 43° 30´ 21° 30´ 0.46730

Pančevo-Nadela 33 22 44° 52´ 20° 45´ 0.47000

Tečić 7 5 43° 50´ 21° 10´ 0.46210

Kozluk 2 2 45° 07´ 21° 18´ 0.46390

Vrtište 6 5 43° 20´ 21° 45´ 0.46080

Aradac-Leje 7 5 45° 22´ 20° 18´ 0.46970

Divostin 1 1 44° 01´ 20° 52´ 0.47470

Vinča 2 2 44° 46´ 20° 37´ 0.46720

Starčevo 1 1 44° 48´ 20° 42´ 0.46740

Table 3. Intensity values of Starčevo sites.

Sample % of # of Temp. (°C) FA FA / F0 ± S.E. Site Average

# Extr. Temp. Range Used (S.E. = Standard FA / F0 ± S.E.
NRM Points Error of Estimate)

AT
A001 44 4 100–350 0.2900 0.61 ± 0.00 0.70 ± 0.04
A008 67 5 350–500 0.3718 0.79 ± 0.07

LUDAŠ–BUDŽAK
B001 41 3 200–400 0.3443 0.73 ± 0.05
B003 90 6 450–575 0.3170 0.68 ± 0.04
B004 44 3 200–350 0.2486 0.53 ± 0.03
B005 50 8 300–550 0.3751 0.80 ± 0.04
B006 37 5 350–500 0.2590 0.55 ± 0.01
B007 47 4 200–400 0.2342 0.50 ± 0.09
B014 38 4 200–400 0.2973 0.63 ± 0.08
B016 42 5 350–500 0.3890 0.83 ± 0.02 0.69 ± 0.04
B019 90 6 350–550 0.2897 0.62 ± 0.02
B021 67 6 200–475 0.2296 0.49 ± 0.02
B025 65 4 300–450 0.4188 0.89 ± 0.02
B026 99 9 350–600 0.3655 0.78 ± 0.05
B027 77 9 100–525 0.3844 0.82 ± 0.04
B028 57 9 200–550 0.4052 0.86 ± 0.04
B029 77 5 300–475 0.3171 0.68 ± 0.03

KOZLUK
K001 53 5 200–450 0.4714 1.02 ± 0.09 0.93 ± 0.07
K002 41 5 300–475 0.3885 0.82 ± 0.05

KAONIK–GRADINA
KG001 42 6 300–500 0.3997 0.86 ± 0.09 0.86 ± 0.09
KG003 54 5 200–450 0.3973 0.85 ± 0.09

ARADAC–LEJE
L001 69 5 100–400 0.4049 0.86 ± 0.04
L002 32 3 200–350 0.4430 0.94 ± 0.02
L003 47 4 200–400 0.4754 1.01 ± 0.04 1.02 ± 0.04
L004 66 7 200–500 0.5322 1.13 ± 0.06
L007 51 7 200–500 0.5397 1.15 ± 0.06

96
 Joni Manson: Approaches to Starčevo culture chronology

Table 3 (cont). Intensity values of Starčevo sites.

Sample % of # of Temp. (°C) FA FA / F0 ± S.E. Site Average

# Extr. Temp. Range Used (S.E. = Standard FA / F0 ± S.E.
NRM Points Error of Estimate)

NOSA
N001 68 6 200–475 0.4097 0.88 ± 0.06
N003 47 4 200–450 0.3984 0.85 ± 0.07 0.85 ± 0.07
N004 67 5 100–400 0.3889 0.83 ± 0.08

PANČEVO–NADELA
PN001 32 5 300–475 0.1994 0.42 ± 0.02
PN003 58 5 300–475 0.3241 0.69 ± 0.03
PN004 65 7 200–500 0.4512 0.96 ± 0.07
PN005 65 5 300–475 0.5117 1.09 ± 0.06
PN006 61 6 350–525 0.3351 0.71 ± 0.02
PN007 64 8 300–550 0.4723 1.00 ± 0.07
PN014 64 4 100–350 0.6396 1.36 ± 0.05
PN015 47 5 300–475 0.5139 1.09 ± 0.04
PN016 32 3 300–400 0.5380 1.14 ± 0.09
PN018 92 6 200–475 0.1388 0.30 ± 0.02
PN021 58 5 200–450 0.3635 0.77 ± 0.06 0.87 ± 0.05
PN023 50 5 200–450 0.2943 0.63 ± 0.05
PN024 48 5 300–475 0.3876 0.82 ± 0.06
PN026 92 5 100–400 0.4767 1.01 ± 0.08
PN027 59 6 300–500 0.4318 0.92 ± 0.05
PN029 68 6 100–450 0.5401 1.15 ± 0.06
PN030 74 9 350–625 0.4152 0.88 ± 0.01
PN031 79 9 200–550 0.4189 0.89 ± 0.05
PN033 65 7 300–575 0.3978 0.85 ± 0.05
PN035 51 5 350–500 0.4191 0.89 ± 0.08
PN035b 49 5 350–500 0.3449 0.73 ± 0.05
PN035c 57 5 350–500 0.4001 0.85 ± 0.04

TEČIĆ
T002 87 7 200–500 0.5309 1.15 ± 0.06
T003 27 4 300–450 0.4315 0.93 ± 0.11
T004 55 4 200–400 0.3777 0.82 ± 0.02 0.93 ± 0.06
T005 69 9 200–575 0.4318 0.93 ± 0.07
T006 66 9 350–600 0.3817 0.83 ± 0.04

VRTIŠTE
V002 84 5 200–450 0.4380 0.95 ± 0.04
V004 67 6 350–525 0.2591 0.56 ± 0.02
V005 76 9 350–600 0.2971 0.64 ± 0.02 0.74 ± 0.03
V006 59 8 200–525 0.3807 0.83 ± 0.05
V007 68 9 300–575 0.3303 0.72 ± 0.04

DIVOSTIN
YU.4 64 5 200–500 0.265 0.57 0.57

VINČA
YU.1 63 4 400–550 0.285 0.61 0.81
YU.3 82 6 200–550 0.465 1.00

STARČEVO
YU.2 46 3 300–450 0.500 1.07 1.07

experiments in Southeast Europe (Kovacheva & Veljovich
1977, 1985; Kovacheva 1977, 1980, 1983, 1995), the results
were reported as the ratio of ancient field intensity to modern
field intensity at the site (FA / F0). ‘Modern’ field strength ac-
tually refers to the field intensity at the site at the time the
sample was collected. Modern field strengths were determ-
ined through the use of the program GEOMAG (U.S.
Department of Commerce n.d.). The program uses a site’s
latitude, longitude, elevation, and collection date to calculate
the appropriate geomagnetic data for a particular location and
time (Manson & Schmidt 1991).
Acceptable sample intensity ratios were then averaged to
obtain average site intensity values. The relatively short life
span and single component quality of most Starčevo sites
(Manson 1990) lent support to this decision. Table 3 lists the
number of temperature points and the temperature range
used, the percentage of total extrapolated NRM included, the
experimentally derived field intensity (FA), and the ratio of
ancient to modern field intensity (FA / F0) for each sample, as
well as the average FA / F0 for each site.
Since magnetic field intensity does not vary unidirection-
ally through time, any intensity value may correspond to

97
 The Iron Gates in Prehistory

The future of Starčevo chronological studies

The dates proposed in Table 4 for the Starčevo and Körös

cultures of the central Balkans cover a slightly broader range
than the dates suggested by Gläser (1991), Hertelendi et al.
(1995) and Lenneis & Stadler (1995). The latter three studies
all derived their chronologies from available radiocarbon
dates. It is likely that a combination of dating techniques will
offer the greatest potential for creating absolute chronologies
and for assessing the validity of the ceramic typologies cur-
rently in place (Sternberg 1997; Bonsall et al. 2002).
It is also obvious that clear communication between the
archaeologists who recover the objects, and the geophysicists
who often perform the chronometric analyses on the samples,
is essential for achieving reliable results (Sternberg 1990;
Kovacheva 1995). The archaeologist is responsible for
providing the contextual data about site stratigraphy and
stylistic seriations and the geophysicist must be able to ex-
plain the basic theory, methodology and limitations of the
techniques, as well as the results of the analysis. Finally, the
archaeologist must be able to interpret that information and
incorporate it into new research designs.
Many of the major research questions concerning the
Starčevo culture require a well-grounded chronological
framework. Questions surrounding the introduction and ad-
Figure 5. NRM–TRM (Arai) and Zijderveld diagrams. option of domesticated plant and animal species, relation-
ships to neighbouring (contemporaneous?) groups, relation-
ships to earlier and later populations in the same area,
more than one possible date. The Aranđelović-Garašanin changes in technology (including pottery production), settle-
(1954) typology was used to determine a relative temporal ment strategies, and social complexity can best be addressed
association for each site and its most plausible position on the when a temporal framework has been established (e.g.
archaeomagnetic intensity master curve. The master curve Greenfield 1986; Tringham et al. 1992; Bailey et al. 1998;
for the region (see especially Kovacheva & Veljovich 1985; Manson 1990, 1995; Gronenborn 1999; Bonsall et al. 2000).
Kovacheva 1995) was based on materials from radiocarbon- The goal, after all, is not merely to create more refined
dated sites in Bulgaria and Serbia, so it has absolute dates chronological frameworks, but to use those frameworks to
assigned to portions of the curve. After determining the posi- focus attention on other aspects of past cultures. The estab-
tions of the intensity values on the curve, it was possible to lishment of temporal frameworks is a vital step in archae-
1 ology, but it can no longer be considered the final objective of
the discipline. A more-refined chronology for the Starčevo
Table 4. Starčevo and Starčevo-Körös chronology. culture should be viewed as an important tool for under-
standing the life and times of these Neolithic Balkan peoples.
Phase 14C age range Calibrated age range
Acknowledgements
Starčevo I At least 7250–7050 BP c. 6100–5900 cal BC
I gratefully acknowledge the financial assistance of the Institute for
Starčevo IIa c. 7050 to 6800 BP c. 5900–5700 cal BC International Education (Fulbright Scholarship for Dissertation
Research in Yugoslavia), Southern Illinois University at Carbondale
Starčevo IIb c. 6800 to 6450 BP c. 5700–5400 cal BC (Doctoral Research Award), and the United States Department of
Education (Jacob K. Javits Fellowship). The analyses reported here
Starčevo III c. 6450–?6150 BP c. 5400–5100 cal BC could not have been completed without the help and cooperation of
many people in Beograd, Pančevo, Zrenjanin, Vršac, Priština, Niš,
Starčevo I = Gura Baciului Kikinda, Bitola, Subotica, Zagreb, and Svetozarevo, as well as in the
Starčevo IIa + IIb = Classical Starčevo = Early Körös
United States. The archaeomagnetic analyses were performed at the
Starčevo III = Late Körös
Paleomagnetism Laboratory of the Department of Geology and
Planetary Science at the University of Pittsburgh, under the direc-
assign probable absolute dates to each of the Starčevo culture tion of Victor Schmidt. Alan McPherron, Department of
Anthropology, University of Pittsburgh, arranged for my first field
sites in the study. Furthermore, by combining the typological
visit to Yugoslavia (Kragujevac). Carroll Riley, Department of
sequences with the archaeomagnetic intensity data and the Anthropology, Southern Illinois University at Carbondale, cheer-
relatively few 14C dates available from Starčevo culture sites, fully (for the most part) provided much-appreciated advice at many
it is possible to propose a correlation between the stages of the project. I also wish to thank my husband, David
Aranđelović-Garašanin ceramic sequence and absolute dates Snyder, for his encouragement and assistance, especially where
(Table 4). computers are concerned.

98
 Joni Manson: Approaches to Starčevo culture chronology
References
Aitken, M.J. 1983: Basic techniques for archaeointensity determin-
ation. In Creer, K.M. et al. (eds) Geomagnetism of Baked Clays
and Recent Sediments. Amsterdam: Elsevier, 78–87.
Aitken, M.J., Alcock, P.A., Bussell, G.D. & Shaw, C.J. 1983:
Paleointensity studies on archaeological material from the Near
East. In Creer, K.M. et al. (eds) Geomagnetism of Baked Clays
and Recent Sediments. Amsterdam: Elsevier, 122–127.
Aranđelović-Garašanin, D. 1954: Starčevačka kultura. Ljubljana.
[see also Garašanin, D.]
Arbour, G. & Schwartz, E.J. 1982: Archeomagnetic intensity of
Indian potsherds from Quebec, Canada. Journal of
Geomagnetism and Geoelectricity 34: 129–136.
Bailey, D.W. 2000: Balkan Prehistory. Exclusion, Incorporation
and Identity. London: Routledge.
Bailey, D.W., Tringham, R., Bass, J., Stevanović, M., Hamilton, M.,
Neumann, H., Angelova, I. & Raduncheva, A. 1998: Expanding
dimensions of early agricultural tells: The Podgoritsa
Archaeological Project, Bulgaria. Journal of Field Archaeology
25: 373–396.
Barbetti, M.F., MCElhinny, M.W., Edwards, D.J. & Schmidt, P.W.
1977: Weathering processes in baked sediments and their effects
on archaeomagnetic field-intensity measurements. Physics of the
Earth and Planetary Interiors 13: 346–354.
Benac, A., Garašanin, M. & Srejović, D. 1979a: Uvod. Praistorija
Jugoslavenskih zemalja II: 11–31. Sarajevo: Akademija Nauka i
Umjetnosti Bosne i Hercegovine.
Benac, A., Garašanin, M. & Srejović, D. 1979b: Zaključna razmat-
ranja. Praistorija Jugoslavenskih zemalja II: 635–667. Sarajevo:
Akademija Nauka i Umjetnosti Bosne i Hercegovine.
Bogucki, P. 1995: The Linear Pottery culture of central Europe:
conservative colonists? In Barnett, W.K. & Hoopes, J.W. (eds)
The Emergence of Pottery. Washington D.C.: Smithsonian
Institution Press, 89–97.
Bogucki, P. & Grygiel, R. 1993: The first farmers of central Europe:
a survey article. Journal of Field Archaeology 20: 399–426.
Bonsall, C., Cook, G., Lennon, R., Harkness, D., Scott, M.,
Bartosiewicz, L. & MCSweeney, K. 2000: Stable isotopes, ra-
diocarbon and the Mesolithic–Neolithic transition in the Iron
Gates. Documenta Praehistorica 27: 119–132.
Bonsall, C., Cook, G., Manson, J. & Sanderson, D. 2002: Direct
dating of Neolithic pottery: progress and prospects. Documenta
Praehistorica 29: 47–59.
Boyd, M. 1986: A new method for measuring palaeomagnetic in-
tensities. Nature 319: 208–209.
Brukner, B. 1968: Neolit u Vojvodini. Belgrade: Arheološko Društvo
Jugoslavije, Dissertationes V.
— 1979: Körös-grupa. Praistorija Jugoslavenskih zemalja II:
213–226. Sarajevo: Akademija Nauka i Umjetnosti Bosne i
Hercegovine.
Brukner, B., Jovanović, B. & Tasić, N. 1974: Praistorija Vojvodine.
Novi Sad: Institut za Izučavanje Istorije Vojvodine i Savez
Arheoloških Društava Jugoslavije.
Bucha, V. 1967: Intensity of the earth’s magnetic field during ar-
chaeological times in Czechoslovakia. Archaeometry 10: 12–22.
Bucha, V., Taylor, R.E., Berger, R. & Haury, E.W. 1970:
Geomagnetic intensity: changes during the past 3000 years in the
western hemisphere. Science 168: 111–114.
Budja, M. 1999: The transition to farming in Mediterranean Europe
— an indigenous response. Documenta Praehistorica 26:
119–141.
Burakov, K.S. & Nachasova, I.E. 1985: Correcting for chemical
change during heating in archaeomagnetic determinations of the
ancient geomagnetic field intensity. Physics of the Solid Earth
21(10): 801–803. Izvestiya, Akademiya Nauk, SSSR.
1
99
Carmichael, C.M. 1977: Introduction to ‘Paleomagnetic field in-
tensity, its measurement in theory and practice’. Special issue of
Physics of the Earth and Planetary Interiors 13(4): v–vii.
Clason, A.T. 1980: Padina and Starčevo: game, fish, and cattle.
Palaeohistoria 22: 141–173.
Coe, R.S. 1967: The determination of paleo-intensities of the earth’s
magnetic field with emphasis on mechanisms which could cause
non-ideal behavior in Thellier ’s method. Journal of
Geomagnetism and Geoelectricity 19: 157–179.
Coe, R.S., Grommé, S. & Mankinen, E.A. 1978: geomagnetic pa-
leointensities from radiocarbon-dated lava flows on Hawaii and
the question of the Pacific nondipole low. Journal of
Geophysical Research 83: 1740–1756.
Collinson, D.W. 1983: Methods in Rock Magnetism and
Palaeomagnetism. Techniques and Instrumentation. New York:
Chapman & Hall.
Dimitrijević, S. 1974: Problem stupnjevanja Starčevačke kulture s
posebnim obzirom na doprinos južnopanonskih nalazišta
rješavanju ovog problema. In Tasić, N. (ed.) Počeci Ranih
Zemljoradničkih Kultura u Vojvodini i Srpskom Podunavlju.
Beograd: Srpsko Arheološko Društvo / Subotica: Gradski Muzej.
Materijali 10: 59–83.
— 1979: Sjeverna zona. Praistorija Jugoslavenskih zemalja II:
229–360. Sarajevo: Akademija Nauka i Umjetnosti Bosne i
Hercegovine.
Dubois, R.L. & Watanabe, N. 1965: Preliminary results of investig-
ations made to study the use of Indian pottery to determine the
paleointensity of the geomagnetic field for the United States
600–1400 AD. Journal of Geomagnetism and Geoelectricity 17:
417–423.
Ehrich, R.W. 1977: Starčevo revisited. In Markotić, V. (ed.) Ancient
Europe and the Mediterranean. Warminster: Aris & Phillips,
59–67.
Evans, R.K. & Rasson, J.A. 1984: Ex Balcanus Lux? Recent devel-
opments in Neolithic and Chalcolithic research in southeast
Europe. American Antiquity 49: 713–741.
Fewkes, V.J. 1938: Neolithic sites in the Yugoslavian portion of the
lower Danubian valley. Proceedings of the American
Philosophical Society 78: 329–402.
Fewkes, V., Goldman, H. & Ehrich, R. 1933: Excavations at
Starčevo, Yugoslavia, seasons 1931 and 1932. Bulletin of the
American School of Prehistoric Research 9: 33–54.
Garašanin, D. 1984: Naselje Starčevačke Kulture. In Vinča u
Praistoriji i Srednjem Veku. Beograd: Galerija Srpska Akademije
Nauka i Umjetnosti, 13–21.
Garašanin, M. 1979: Centralnobalkanska zona. Praistorija
Jugoslavenskih zemalja II: 79–212. Sarajevo: Akademija Nauka
i Umjetnosti Bosne i Hercegovine.
Gimbutas, M. 1974: Anza, ca. 6500–5000 B.C.: a cultural yardstick
for the study of Neolithic southeast Europe. Journal of Field
Archaeology 1: 26–66.
— 1976: Neolithic Macedonia as Reflected by Excavation at Anza,
Southeast Yugoslavia. Los Angeles: Institute of Archaeology,
University of California.
Girić, M. 1974: Körös–Starčevo nalazišta u severnom Banatu. In
Tasić, N. (ed.) Počeci Ranih Zemloradničkih Kultura u Vojvodini
i Srpskom Podunavlju. Beograd: Srpsko Arheološko Društvo /
Subotica: Gradski Muzej. Materijali 10: 169–187.
Gläser, R. 1991: Bemerkungen zur absoluten datierung des beginns
der westlichen Linienbandkeramik. Banatica 11: 53–64.
Greenfield, H.J. 1986: The Paleoeconomy of the Central Balkans
(Serbia): A Zooarchaeological Perspective on the Late Neolithic
and Bronze Age (4500–1000 BC). BAR International Series 304.
Oxford: British Archaeological Reports.
Gronenborn, D. 1999: A variation on a basic theme: the transition to
farming in south-central Europe. Journal of World Prehistory 13:
 The Iron Gates in Prehistory
123–210.
Hertelendi, E., Kalicz, N., Raczky, P., Horváth, F., Veres, M.,
Svingor, É., Futó, I. & Bartosiewicz, L. 1995: Re-evaluation of
the Neolithic in eastern Hungary based on calibrated radiocarbon
dates. Radiocarbon 37: 239–244.
Horton, D. 1938: Note on a microscopic study of a sample group of
barbotine sherds with positive appliqué from ‘Grad’, Starčevo.
Proceedings of the American Philosophical Society 78: 397–402.
Kaiser, T. 1984: Vinča Ceramics: Economic and Technological
Aspects of Late Neolithic Pottery Production in Southeast
Europe. Unpublished PhD Dissertation, University of California,
Berkeley. (University Microfilms, Ann Arbor).
Korošec, B. 1973: Application de méthodes d’analyse statistique au
problème de la chronologie de Starčevo. Arheološki Vestnik –
Acta Archaeologica 24: 271–302.
Kovacheva, M. 1977: Archaeomagnetic investigations in Bulgaria:
field intensity determinations. Physics of the Earth and
Planetary Interiors 13: 355–359.
— 1980: Summarized results of the archaeomagnetic investigation
of the geomagnetic field variation for the last 8000 yr in south-
eastern Europe. Geophysical Journal of the Royal Astronomical
Society 61: 57–64.
— 1983: Archaeomagnetic data from Bulgaria and south eastern
Yugoslavia. In Creer, K.M. et al. (eds) Geomagnetism of Baked
Clays and Recent Sediments. Amsterdam: Elsevier, 106–110.
— 1995: Bulgarian archaeomagnetic studies. In Bailey, D.W. &
Panayotov, I. (eds) Prehistoric Bulgaria. Madison, WI:
Prehistory Press, 209–224.
Kovacheva, M. & Veljovich, D. 1977: Geomagnetic field variations
in southeastern Europe between 6500 and 100 years ago. Earth
and Planetary Science Letters 37: 131–138.
Kovacheva, M. & Veljovich, D. 1985: Arheomagnitno izsledvane na
praistoricheski obekti ot NRB i SFRYU. Arheologia 4(2): 21–26.
Kovacheva, M. & Zagniy, G. 1985: Archaeomagnetic results from
some prehistoric sites in Bulgaria. Archaeometry 27: 179–184.
Kutzián, I. 1947: The Körös Culture. Budapest: Dissertationes
Pannonicae.
Lenneis, E. & Stadler, P. 1995: Zur absolutchronologie der
Linearbandkeramik aufgrund von 14 C-Daten. Archäologie
Österreichs 6: 4–13.
Levi, S. 1977: The effect of magnetite particle size on paleointensity
determinations of the geomagnetic field. Physics of the Earth
and Planetary Interiors 13: 245–259.
McPherron, A. 1988: Radiocarbon determinations. In McPherron,
A. & Srejović, D. (eds) Divostin and the Neolithic of Central
Serbia. Pittsburgh: University of Pittsburgh Ethnology
Monograph 10, 379–383.
Manson, J.L. 1990: A Reanalysis of Starčevo Culture Ceramics:
Implications for Neolithic Development in the Balkans.
Unpublished Ph.D. dissertation, Southern Illinois University,
Carbondale (University Microfilms, Ann Arbor).
— 1995: Starčevo pottery and Neolithic development in the central
Balkans. In Barnett, W.K. & Hoopes, J.W. (eds) The Emergence
of Pottery. Washington D.C.: Smithsonian Institution Press,
65–77.
Manson, J.L. & Schmidt, V.A. 1991: Archaeomagnetic dating of
Neolithic Yugoslavian potsherds. Paper presented at the 56th
Annual Meeting of the Society for American Archaeology, New
Orleans.
Michels, J.W. 1973: Dating Methods in Archaeology. New York:
Seminar Press.
Milisauskas, S. & Kruk, J. 1989: Neolithic economy in central
Europe. Journal of World Prehistory 3: 403–446.
Milojčić, V. 1950: Körös–Starčevo–Vinča. Mainz: Reinecke
Festschrift.
Nagata, T., Arai, Y. & Momose, K. 1963: Secular variation of the
1
100
geomagnetic total force during the last 5000 years. Journal of
Geophysical Research 68(18): 5277–5281.
Radišić, R. 1984: Starčevačka kultura i Kereška kultura. In
Praistorijske Kulture na Tlu Vojvodine. Novi Sad: Vojvoðanski
Muzej, 9–13.
Reimer, P.J., Baillie, M.G.L., Bard, E., Bayliss, A., Beck, J.W.,
Bertrand, C., Blackwell, P.G., Buck, C.E., Burr, G., Cutler, K.B.,
Damon, P.E., Edwards, R.L., Fairbanks, R.G., Friedrich, M.,
Guilderson, T.P., Hughen, K.A., Kromer, B., MCCormac, F.G.,
Manning, S., Bronk Ramsey, C., Reimer, R.W., Remmele, S.,
Southon, J.R., Stuiver, M., Talamo, S., Taylor, F.W., van der
Plicht, J. & Weyhenmeyer, C.E. 2004: IntCal04 terrestrial ra-
diocarbon age calibration, 0–26 cal kyr BP. Radiocarbon 46:
1029–1058.
Rusakov, O.M. & Zagniy, G.F. 1973: Intensity of the geomagnetic
field in the Ukraine and Moldavia during the past 6000 years.
Archaeometry 15: 275–285.
Sakai, H. & Hirooka, K. 1985: Variation of the geomagnetic intens-
ity deduced from archaeological objects in Japan. Rock
Magnetism and Paleogeophysics 12: 12–19.
Sasajima, S. 1965: Geomagnetic secular variation revealed in the
baked earths in west Japan (part 2). Change of the field intensity.
Journal of Geomagnetism and Geoelectricity 17: 413–416.
Srejović, D. 1972: Europe’s First Monumental Sculpture: New
Discoveries at Lepenski Vir. New York: Stein & Day.
Sternberg, R.S. 1990: The geophysical basis of archaeomagnetic
dating. In Eighmy, J.L. & Sternberg, R.S. (eds) Archaeomagnetic
Dating. Tucson: University of Arizona Press, 5–28.
— 1997: Archaeomagnetic dating. In Taylor, R.E & Aitken, M.J.
(eds) Chronometric Dating in Archaeology. London: Plenum
Press, 323–356.
Stuiver, M. & Reimer, P.J. 1993: Extended 14C database and revised
CALIB radiocarbon calibration program. Radiocarbon 35:
215–230.
Stuiver, M., Reimer, P.J. & Reimer, R.W. 2005: CALIB 5.0. [WWW
program and documentation].
Tanaka, H. 1982: Geomagnetic paleointensities for the period 6000
to 3000 years BP determined from lavas and pyroclastic flows in
Japan. Journal of Geomagnetism and Geoelectricity 34:
601–617.
Tasić, N. & Tomić, E. 1969: Crnolačka Bara: Naselje Starčevačke
i Vinčanske Kulture. Kruševac: Narodni Muzej / Beograd:
Arheološko Društvo Jugoslavije. Dissertationes 8.
Thellier, E. & Thellier, O. 1959: The intensity of the earth’s mag-
netic field in the historical and geological past. Izvestiya.
Geophysical Series 7: 929–949. Akademiya Nauk SSSR.
Thomas, R.C. 1983: Mineralogical problems. In Creer, K.M. et al.
(eds) Geomagnetism of Baked Clays and Recent Sediments.
Amsterdam: Elsevier, 98–104.
Todorović, J., Simoska, D. & Kitanoski, B. 1977: The settlement of
Pešterica and the problem of the Early Neolithic in Pelagonia.
Archaeologia Iugoslavica 18: 1–8.
Tringham, R. 1971: Hunters, Fishers and Farmers of Eastern
Europe 6000–3000 B.C. London: Hutchinson.
Tringham, R., Brukner, B., Kaiser, T., Borojević, K., Bukvić, L.,
Šteli, P., Russell, N., Stevanović, M. & Voytek, B. 1992:
Excavations at Opovo, 1985–1987. Socioeconomic change in the
Balkan Neolithic. Journal of Field Archaeology 19: 351–386.
U.S. Department of Commerce n.d.: Program #609 (GEOMAG).
Environmental Science Services Administration, Coast and
Geodetic Survey, Geomagnetism Div., Washington Science
Center, Rockville, MD. Converted to VAX/VMS and MacIntosh
by Victor A. Schmidt, University of Pittsburgh, Pittsburgh, PA.
Vetnić, S. 1986: Elementi Protostarčeva-Starčeva I na neseljima tipa
Rajkinac kod Svetozarevu u gornjem Pomoravlju. Starinar n.s.
37: 89–93.
 Joni Manson: Approaches to Starčevo culture chronology

Walton, D. 1983: The reliability of ancient intensities obtained by
thermal demagnetization. In Creer, K.M. et al. (eds) Geomag-
netism of Baked Clays and Recent Sediments. Amsterdam:
Elsevier, 88–97.
Weaver, G.H. 1966: Measurement of the past intensity of the earth’s
magnetic field. Archaeometry 9: 174–186.
Wei, Q., Zhang, W. & Li, D. 1987: Geomagnetic intensity as evalu-
ated from ancient Chinese pottery. Nature 328: 330–333.
Whittle, A. 1996: Europe in the Neolithic. The Creation of New
Worlds. Cambridge: Cambridge University Press.
Whittle, A., Bartosiewicz, L., Borić, D., Pettitt, P. & Richards, M.
2002: In the beginning: new radiocarbon dates for the Early
Neolithic in northern Serbia and south-east Hungary. Antæus 25:
63–117.

101
 Faunal assemblages from the Early Neolithic
of the central Balkans: methodological issues in the
reconstruction of subsistence and land use

Haskel J. Greenfield

Abstract: The earliest food producing societies in Europe north of the Aegean littoral appear in the central Balkans during the Early
Neolithic. As a result, they are frequently included in discussions of the evolution of animal domestication in Europe. However, relatively
little zooarchaeological research has been undertaken in the central Balkans. This paper has two goals. First, it introduces and discusses the
nature, quality and problems pertaining to each of the major Early Neolithic zooarchaeological samples from the central Balkans to
demonstrate the difficulty of inter-assemblage comparison. Second, it explores the potential of these assemblages for reconstructing animal
exploitation strategies and land use patterns during the Early Neolithic, which are essential for understanding the economic processes in-
volved in the evolution of early food producing societies in Southeast Europe. It is shown that it would be misleading to uncritically accept
these assemblages as high quality sources of information for reconstructing the origins of animal domestication in Southeast Europe and
animal exploitation strategies and land use patterns during the Early Neolithic. These assemblages have a very limited potential for under-
standing the economic processes involved in the evolution of early food producing societies in Southeast Europe.
Key words: fauna, Neolithic, southeast Europe, subsistence, land use patterns

The earliest food producing societies in temperate Southeast
Europe north of the Mediterranean littoral appear in the
central-north part of the Balkan peninsula (hereafter referred
to as the central Balkans — Fig. 1) during the earlier half of
the Neolithic. The Early Neolithic (Starčevo culture) of the
central Balkans, c. 7300–6400 BP (6150–5350 cal BC), wit-
nessed the earliest spread of farming adaptations in Europe
north of the Mediterranean littoral (Tringham 1971; Barker
1985). It is within this region that food production strategies
based essentially upon an eastern Mediterranean complex of
plants and animals are modified before spreading to the tem-
perate climatic zones of Central and Northern Europe
(Champion et al. 1984; Barker 1985).
Many scholars have proposed explanatory models, often
including economic or subsistence components (with a
paucity of supporting data), for the spread of the Early
Neolithic food producing cultures of temperate Southeast
Europe (e.g. Ammerman & Cavalli-Sforza 1971; Garašanin
1973, 1983; Bökönyi 1974a; Barker 1985; Whittle 1985).
While the Early Neolithic of the central Balkans is con-
sidered one of the most crucial periods in European prehis-
tory, it nonetheless remains one of the most poorly investig-
ated in terms of prehistoric economics. But, if economic
processes are recognized as fundamental to the emergence of
food producing society, economic data must be collected and
used to test models for their evolution.
The appearance of early food producing societies in tem-
perate Southeast Europe is often assumed by archaeologists
to be the result of the migration of previously established
food producing communities from the south (e.g. Childe
1929, 1958; Ammerman & Cavalli-Sforza 1971; Tringham
1971: 70ff.; Champion et al. 1984: 100, 120; cf. Dennell
1985: 153ff.; Whittle 1985: 54). Two often unstated assump-
tions are used to support this view of the spread of food pro-
ducing societies into temperate Southeast Europe:
1. If domestic fauna are found to be dominant in Early
Neolithic sites in the region, these sites belong to intrusive
colonists colonizing an essentially open nearly uninhab-
ited environment; and
2. Since terminal Mesolithic sites are poorly represented in
the region (with the exception of agriculturally marginal
areas such as the Danubian Iron Gates or Montenegrin
highlands), significant indigenous Mesolithic hunting-
gathering populations survived only in such areas, and
slowly adopted domestic economies.
Therefore, any sites with domesticated plants and animals
found outside of these refugia were assumed to belong to in-
trusive agricultural populations. The first Early Neolithic
zooarchaeological studies from the region demonstrated a
preponderance of domestic fauna in Early Neolithic sites
outside of the Iron Gates (e.g. Bökönyi 1974b, 1976, 1988;
Clason 1980). These were used in support of this circular
reasoning. Recent studies indicate a more complex picture.
As will be shown in this paper, the frequency of domestic
fauna in sites cannot be used to argue for the presence or ab-
sence of indigenous populations in and outside of agricultur-
ally marginal areas. Rather, the frequencies of faunal remains
in sites are a reflection of differential resource exploitation,
recovery methodology, and bone assemblage attrition.
The Early Neolithic of the central Balkans has been ex-
tensively investigated because of the interest in the origins of
food production in Europe. But, little effort has been placed
upon the systematic recovery of faunal remains from such
sites. This has resulted in a dramatically biased zooarchae-
ological database, and one that has been uncritically accepted
in the literature to explain the origins of food production in
Southeast Europe (e.g. Murray 1970; Barker 1985; Whittle
1985; Halstead 1988). Neither the few systematic zooar-
chaeological studies nor the secondary discussions employ-
ing such data have paid attention to the taphonomic history of

103
 The Iron Gates in Prehistory

Figure 1. Map of the central Balkans, showing Early Neolithic sites with analyzed faunal assemblages mentioned in text.
Site name abbreviations used in all figures and tables are: A=Anza; B=Blagotin; BC=Bukovačka Česma; D=Divostin;
F=Foeni-Salaş; G=Golukot; HV=Hajdučka Vodenica; LV=Lepenski Vir; LB=Ludoš Budžak; M=Madžari; N=Nosa;
O=Obre; P=Padina; RB=Rug Bair; S=Starčevo; V=Vlasac.

samples or to the methodological problems connected with
the faunal samples from this period and region (cf. Gifford
1981; Greenfield 1991, 1993). In the region, there is a dearth
of published systematic analyses of Early Neolithic fauna.
Most published accounts consist of species-frequency lists or
preliminary reports, usually lacking detailed supporting data.
These shortcomings in the database are significant influences
during reconstruction of prehistoric economies on the basis
of inter-site patterning in bone assemblages. It is only after
methodological and taphonomic issues are addressed that
animal exploitation strategies can be discussed. Some of the
more important Early Neolithic faunal samples from the
central Balkans are discussed here in order to demonstrate
some of the difficulties involved in inter-assemblage com-
parison and their potential for reconstructing animal exploit-
ation strategies and land use patterns.
The region, archaeological ‘culture’ and sites
The region
The central Balkans encompasses the eastern half of the
former Yugoslavia. It includes the countries presently known
as Serbia, Bosnia, and Macedonia. The term Balkan was ori-
ginally a Turkish word, meaning a chain of (forested) moun-
tains (Naval Intelligence Division 1944; Klaić 1982). The

104
 Haskel Greenfield: Early Neolithic faunal assemblages from the central Balkans
Balkan Peninsula is a topographically complex environment
with several interconnected mountain systems coursing
through the area. Highlands and lowlands are juxtaposed,
with more than 37% of the land above 500m (Turnock 1989:
8). This variation in landforms within a small area has a
strong influence on local climate. The regional climate is
transitional between that of temperate Central Europe and the
more arid Mediterranean basin. Climate and plant and animal
communities take on different characteristics not only in a
general north–south gradient, but also with increasing alti-
tude. Neighbouring valleys often exhibit very different com-
binations of regional environmental variables, yet retain the
general pattern of environmental diversity within the area as
a whole (Pounds 1969).
The Early Neolithic of the central Balkans
The advent of the Neolithic in this region is generally con-
nected to the appearance of food producing technology and
adaptations, such as settled communities, ceramics, polished
stone tools, domestic plants and animals. These are found in
large quantities for the first time with the Starčevo culture.
The Early Neolithic of the central Balkans coincides for the
most part with the spatial extent of the Starčevo ‘culture’
c. 7300–6400 BP (6150–5350 cal BC — cf. Tringham 1971;
Gimbutas 1976; Garašanin 1983).1 Geographical variants of
the Starčevo culture are known from Romania (Criş) and
Hungary (Körös). The Starčevo culture has been temporally
subdivided into several sub-phases. Several competing chro-
nological systems have been proposed on the basis of,
primarily, the seriation of ceramic materials found at strati-
fied and unstratified sites throughout the region (e.g. Milojčić
1949; Aranđelović-Garašanin 1954; Gimbutas 1976;
Garašanin 1979, 1983; Srejović 1988). It is considered to be
the archaeological manifestation of the earliest Neolithic
populations in the region.
Large-scale excavations have demonstrated that Starčevo
settlements are not internally differentiated into functionally
distinct areas (such as domestic houses, specialized work
areas, cemeteries). There is little evidence for socio-
economic differentiation among houses or burials that would
imply significant status distinctions within communities.
Remains of structures include both semi-subterranean and
ground-level dwellings, often with associated hearths and
refuse/storage pits. Artefacts include coarse and fine ceram-
ics, small ceramic altars, amulets, chipped and ground stone
tools, bone and antler tools, and unmodified animal bones
and shells. The ceramics are thick-walled and decorated with
a variety of typical Starčevo decorative motifs, including
mono- and polychrome painting, interior burnishing, pinch-
ing, impressing, channelling, and barbotine surface decora-
tions (Tringham 1971; Garašanin 1979, 1983; Srejović 1979;
Leković 1985; Bogdanović 1988; McPherron & Srejović
1988; Chapman 1989; Greenfield 2000; Jongsma &
Greenfield 2001).
The sites
Most of our knowledge of Early Neolithic subsistence de-
rives from four groups of sites. Each is located in a different
environmental context. The nature of the subsistence data
differs between each of these environments.
105
1. The first are found along the banks of the Danube in the
gorge known as the Iron Gates (Lepenski Vir: Bökönyi
1971; Padina: Clason 1980; Hajdučka Vodenica:
Greenfield, this volume). The sites in the gorge share ac-
cess to a similar range of resources and have similar faunal
proportions.
2. The second group is located in the northern end of the
Mediterranean environmental zone. They are also at the
southern end of the Starčevo range. Sites include Anza and
Rug Bair in a small Macedonian upland basin (Bökönyi
1976; Schwartz 1976).
3. The third group includes those found in the rolling hills
and river valleys of central Serbia and Bosnia, such as
Divostin and Blagotin nestled among the rolling hills of
central Serbia, Bukovačka Česma in the Morava river
valley, and Obre I in a valley deep in the mountains of
northern Bosnia (Bökönyi 1974b, 1988; Greenfield 1994,
Greenfield & Jongsma n.d.). This is the transition zone
between the Mediterranean and the Central European
temperate climatic system. River valleys, such as the
Morava, Bosna and Vardar were important routes for the
spread of Early Neolithic adaptations from the
Mediterranean to Central Europe and a centre for Starčevo
settlement (cf. Barker 1975; Srejović 1979). As such, they
represent an important link in understanding the spread of
early farming adaptations out of the southern Balkans into
the rest of Europe.
4. The fourth group is found in the Pannonian Plain (a non-
political term for the great lowlands encompassing
Hungary, eastern Romania, and the northern part of the
former Yugoslavia). These sites are at the northern end of
the Starčevo range. Sites from this group include the type
site of Starčevo on the Danube along the south edge of the
plain and those further north and east, in the interior of the
plain, such as Foeni-Salaş, Ludoš-Budžak, Nosa, and
Golukot (Bökönyi 1974a; Clason 1980; Lazić 1988;
Greenfield et al. n.d.). They are found close to the
Hungarian or Romanian borders and are located in a more
Central European climatic system (Pounds 1969).
Environmental variability in site location
Based on vegetation, climate and topography, the region can
be divided into four major zones:
1. A Mediterranean zone in the southern half of the Balkan
Peninsula, with low annual precipitation, high summer
temperatures, and semi-steppe vegetation (Anza and Rug
Bair).
2. A temperate transition zone in the hills of Serbia and
Bosnia, transitional in many of its environmental charac-
teristics between the Mediterranean and Central European
systems. This zone is characterized by high, year-round
precipitation, deciduous forests and strong soil develop-
ment. The sites in this zone are located on hills overlook-
ing river floodplains and surrounded by agriculturally fer-
tile soils (Blagotin, Obre I, Divostin, Bukovačka Česma).
3. A temperate highland zone in the Iron Gates gorge of the
Danube, surrounded by the mountains of eastern Serbia,
thick deciduous forests, soils with poor agricultural po-
 The Iron Gates in Prehistory
tential, and overlooking the Danube and its rich aquatic
resources (Hajdučka Vodenica, Lepenski Vir, Padina); and
4. A temperate Central European lowland zone in the plains
of Pannonia (Starčevo, Golukot, Foeni-Salaş, Nosa, and
Ludoš-Budžak). These sites overlook the floodplains of
rivers, streams and marshes, with flora and fauna adapted
to soils with high moisture levels.
Some methodological issues in inter-assemblage
comparison
Although consideration of the taphonomic history of each
sample is a necessary precondition to the reconstruction of
prehistoric subsistence and land use, most of the samples
were not excavated, analyzed or published in a manner that
allows a systematic reconsideration of their taphonomic his-
tory. Consideration of the methodological problems underly-
ing the faunal studies can be used, however, to enhance our
understanding of some of the inter-assemblage variability.
Eventually, we may be able to differentiate patterning in the
samples attributable to methodology from that of prehistoric
human behaviour. Some of the more important methodolo-
gical issues in inter-assemblage comparison are discussed
next. The issues chosen for discussion are those that can be
re-evaluated given the limited nature of most faunal
publications.
Recovery methodology
It has long been a truism of faunal analysis that recovery
methodology dramatically affects the outcome conclusions
of the analysis (Payne 1972; Casteel 1973). As a result, col-
lection methodology must be considered before any conclu-
sions concerning human behaviour can be made. In general,
bone collection from this region has been unsystematic. All
of the samples were unsieved except for Anza, Blagotin,
Foeni-Salaş and Rug Bair. Only Foeni-Salaş and Rug Bair’s
fauna were entirely sieved. The sieved and unsieved samples
from Anza and Blagotin were not separated during analysis.
Hand recovery affects the results of all the analyses in the
following manner:
1. The bones of larger animals will be more completely col-
lected than smaller animals. This will result in over-
representation of larger animals in assemblages. For ex-
ample, cattle remains were collected more frequently than
sheep remains, over-representing them in the assemblage
relative to their original frequencies. Such is the situation
at Obre (Bökönyi 1974b; Clason & Prummel 1977;
Greenfield 1986, 1991).
2. The size of the bone is also an important variable. The
larger the bone of the same species, the more likely it is
that it will be collected. Large bones are more frequently
collected than the smaller bones of the same taxon. Thus,
reconstruction of preferential selection of body parts for
use and/or disposal is compromised. If excavators prefer-
entially collect large unbroken long bones (e.g. humerus)
and ignore the smaller bones (e.g. phalanges, tarsals and
carpals), then one may mistakenly assume that the prehis-
toric occupants disposed of the missing bones elsewhere.
However, the reason that the smaller bones are missing
from the assemblage is due more to the behaviour of the
106
archaeologists conducting the excavation than to the be-
haviour of the prehistoric occupants (Lyman 1994).
While most faunal studies from the region do not list the
type and frequency of various element categories, some
studies do (Clason 1980; Greenfield 1994, n.d. a, b, this
volume). There is a systematic recovery bias against small
bone elements. When the distribution of element categories
are compared in the unsieved samples, there is a total or near
absence of small elements (phalanges, carpals and tarsals) for
the medium- and small-sized mammals (e.g. ovicaprines,
pigs, dogs, beaver). They are relatively more common among
large mammals, both domestic and wild. In the sieved
samples, the small element categories are more common.
It is also possible to note the effect of recovery technology
upon overall sample size by comparing the Iron Gates
samples (Lepenski Vir, Padina, Vlasac, and Hajdučka
Vodenica — Bökönyi 1971, 1978; Clason 1980; Greenfield,
this volume). The excavators of Lepenski Vir, Padina, and
Hajdučka Vodenica openly acknowledge that bone recovery
was limited and biased toward what was considered to be
tools or otherwise modified bones (Greenfield, this volume;
B. Jovanović, pers. comm.). The result was extremely low
recovery rates when compared to Vlasac (Srejović & Letica
1978) where efforts were made to avoid the mistakes of the
earlier excavations in the gorge. The other assemblages are
dominated by large mammal taxa (red deer, cattle). At
Vlasac, contrary to the other excavations in the gorge, the
remains of fish and various small- and medium-sized taxa
dominate the assemblage in all phases.
Quantification
Only two methods have been used to quantify species rep-
resentation from Early Neolithic faunal assemblages in the
central Balkans: number of individual specimens (NISP) and
minimum number of individuals (MNI) (see Bökönyi 1970;
Greenfield 1986, 1991 — for reviews of procedures in the
Balkans). In this study, only NISP counts are employed for
several reasons. First, the problems with MNI seem to be
more severe than with NISP (Grayson 1984). For example,
both MNI and NISP appear to be equally predictive of spe-
cies abundance in large samples (>10,000 fragments). Some
of the samples, however, are relatively small, and NISP is
more useful for the analysis of small samples, especially
those with unequal species frequency ratios (Gilbert et al.
1982). Also, MNI counts are not given for all of the samples,
being available only for the total remains of each species
from six of the sites (Divostin, Lepenski Vir, Ludoš-Budžak,
Nosa, Obre and Anza I–III). In each instance, the values were
calculated in the same manner (Bökönyi 1970). Finally, NISP
counts are available for all of the sites, making the samples
more comparable. In the NISP counts from Blagotin, Foeni-
Salaş, Hajdučka Vodenica and Bukovačka Česma, whole and
partially articulated skeletons and limbs were not double-
counted. Articulated specimens were counted only once.
Antler and horn fragments were few and mostly attached to
cranial elements (Greenfield 1986).
It is not known if the NISP counts from the other sites were
calculated in exactly the same way since there is no discus-
sion of NISP methodology in their publication. Even though
the question of individual groups of articulated elements was
 Haskel Greenfield: Early Neolithic faunal assemblages from the central Balkans

Table 1. Frequency distribution of burnt and unburnt bones from Early Neolithic levels of each site.

Site Context Burnt Unburnt Total

N % N % N
Blagotin Exterior 20 2.27% 861 97.73% 881
Pit house - central 319 1.82% 17228 98.18% 17547
Pit house - peripheral 482 3.58% 12982 96.42% 13464
Plough zone above pithouse 2 1.38% 143 98.62% 145
Subtotal 823 2.57% 31214 97.43% 32037
Bukovačka Česma All 1 0.18% 543 99.82% 544
Foeni-Salaş If grey bones are deducted 497 6.06% 7700 93.94% 8197
Hajdučka Vodenica All 2 1.79% 110 98.21% 112

Table 2. Frequency distribution of weathering stages from Early Neolithic levels of each site.

Weathering Stages
Site Context type Light Medium Heavy Total
Blagotin Exterior in cultural horizon 50 14.49% 133 38.55% 162 46.96% 345
Exterior - in ploughzone above ZM02 4 2.17% 135 73.37% 45 24.46% 184
Subtotal - Exterior 54 10.21% 268 50.66% 207 39.13% 529
Interior (Pit house - central) 10761 61.27% 5893 33.56% 908 5.17% 17562
Interior (Pit house - peripheral) 2194 16.30% 9994 74.27% 1268 9.42% 13456
Subtotal - Interior 12955 41.77% 15887 51.22% 2176 7.02% 31018
Grand total 13009 41.24% 16155 51.21% 2383 7.55% 31547
Bukovačka Česma Mixed 544 100.00% 0 0.00% 0 0.00% 544
Foeni-Salaş Exterior in cultural horizon 485 14.52% 2592 77.58% 264 7.90% 3341
Interior (Pit house) 632 12.20% 4375 84.46% 183 3.53% 5190
Grand total 1117 13.09% 6967 81.67% 447 5.24% 8531
Hajdučka Vodenica Mixed 27 24.11% 29 25.89% 56 50.00% 112

not directly addressed in the previously published analyses of
any of the sites except those analyzed by the author, the ar-
ticulated remains of whole skeletons were not found at Anza,
Obre, Divostin, Hajdučka Vodenica and Bukovačka Česma.
Only the analyses conducted by the author have recognized
the importance of documenting the presence of articulated
limbs in order to modify NISP or MNIs. Articulated limbs
were found at Blagotin, Bukovačka Česma and Foeni-Salaş
(Bökönyi 1974b, 1976, 1988; Greenfield 1994, this volume,
n.d. a, b).
Data publication
Each of the published assemblages has been published in
ways that limit reanalysis. Even though the samples from
Anza, Divostin and Obre were excavated by both natural and
artificial stratigraphic units, the Early and Late Neolithic
samples were not distinguished for publication except for
initial summary statistics. For example, while species pro-
portions over time are noted, changes in age and sex propor-
tions are not reported. As a result, it is impossible to identify
shifts in culling strategies.
Curation
Saving and storing bone collections in safe and accessible
facilities is fundamental to the pursuit of the discipline.
Advances made in one generation can be improved upon if
the next generation can access the materials. However, the
central Balkans is plagued by poor storage facilities and an
inadequately developed ethos in relation to the storage of
zooarchaeological materials. The faunal remains from most
of the largest collections were discarded immediately after
analysis (Obre, Lepenski Vir, Divostin — personal commu-
nications from Sandor Bökönyi, Dragoslav Srejović,
Svetozar Stanković). Only the bone tools were kept since
they were considered to be artefacts, in addition to small
samples of other bones. The faunal remains from Anza and
Rug Bair were kept for several years in a storage shed outside
of the local museum. By the time I tried to re-examine the
Anza remains in 1982, rats had gotten into the shed. All of the
bags and tags were destroyed, and the faunal remains sub-
sequently discarded by the curators. The remains from
Hajdučka Vodenica and Padina are stored at the central stor-
age facility for the Serbian Iron Gates sites at Karataš. It is
extremely difficult to access material in this facility since
crates of material are stored one on top of the other, without
any central cataloguing system. The remains from Starčevo
and Blagotin are stored in Belgrade (at the National Museum
and the University of Belgrade, respectively), while those
from Foeni-Salaş are in Timişoara, Romania (Muzeul
Banatului). Selected samples from Hajdučka Vodenica,
Blagotin and Foeni-Salaş are in Winnipeg, Canada
(University of Manitoba). The fate of the other collections is

107
 The Iron Gates in Prehistory

Table 3. Faunal remains from central Balkan Early Neolithic sites. The ‘number of taxa’ column includes all bones iden-
tified to a class or finer taxonomic level. Unidentified ovicaprines (i.e. those that are not identified to either Ovis or Capra)
are not counted if Ovis or Capra are present. If Ovis or Capra are not present, then unidentified ovicaprines would be
counted only once. All invertebrates, bird, and fish remains, respectively, are each treated as a single taxon owing to the
low level of analysis accorded their remains (Bökönyi 1970, 1974, 1976, 1988; Clason 1980; Lazić 1988; Moskalewska
1986; Schwartz 1976; Greenfield n.d. a, b).

Site Figure Number Wild Domestic NISP Notes

abbreviations of taxa
Anza I–III A 14 138 4.25% 3112 95.75% 3250 a
Blagotin B 18 719 8.29% 7959 91.71% 8678 a
Bukovačka Česma BC 11 166 61.48% 104 38.52% 270
Divostin I D 14 203 8.45% 2198 91.55% 2401
Foeni-Salaş F 17 416 19.87% 1678 80.13% 2094
Golukot G 10 893 68.38% 413 31.62% 1306
Hajdučka Vodenica H 11 211 84.40% 39 15.60% 250
Lepenski Vir III L 21 1765 74.50% 640 27.02% 2369
Ludoš-Budžak LB 18 572 20.91% 2163 79.09% 2735
Madžari M 17 185 6.44% 2689 93.56% 2874
Nosa N 16 736 71.88% 229 22.36% 1024
Obre I O 19 1700 20.75% 6491 79.25% 8191
Padina B P 19 2656 92.64% 211 7.36% 2867
Rug Bair R 9 13 1.88% 680 98.12% 693 b
Starčevo S 19 1029 25.91% 2943 74.09% 3972

a. Sieved and unsieved samples were not separated during analysis.

b. Sieved fauna from the American excavations.

unknown at present, but I suspect that they have been dis-
carded. The various analysts also had an influence on cura-
tion. In general, Sandor Bökönyi advised his Yugoslavian
hosts that they could discard the remains after he finished his
analysis. In contrast, all other analysts encouraged that their
collections be curated. However, since the advent of the
Balkan wars of the 1990s, the condition of any of the collec-
tions is uncertain.
Physico-chemical assemblage attrition
Most analyses ignore the issue of assemblage attrition
through weathering and other destructive forces. The contro-
versy surrounding the ‘schlepp effect’ is a good example
(Lyman 1994). Without a discussion of taphonomic vari-
ables, it is impossible to judge the value of assemblages for
subsistence reconstructions.
Burning
Most Early Neolithic faunal reports do not mention burnt
bones at all. Only a very small fraction of burnt bones were
found at Blagotin (2.57%), Bukovačka Česma (0.18%),
Foeni-Salaş (6%) and Hajdučka Vodenica (1.8% — Table 1).
The highest values come from Foeni-Salaş, which is two to
three times higher than the values found in all the rest.
Similarly low frequencies of burnt bones were found in Late
Neolithic and post-Neolithic assemblages from the region
(Greenfield 1986, 1991).
Why are so few bones burnt? There are several possible
reasons. Most bones probably were still covered with meat
when exposed to fire, were boiled in a stew-like concoction,
or had their meat removed prior to cooking. In each case,
there would have been limited contact between the bone and
the flame. Also, the burnt bone may have simply disinteg-
rated in the soil, since burnt bone breaks up into small frag-
ments more readily than uncooked bone (Bonfield & Li
1966). Another reason that few pieces of burnt bone are
found is differential disposal patterns. Cooking areas
(hearths) and nearby middens are the depositional contexts in
which burnt bone most likely would have been discarded.
Other than in the Iron Gates sites, few hearths have been
found. At Bukovačka Česma, almost all of the bones around
the two excavated hearths (trench 5, N=40; trench 8, N=11)
were not burnt. At Anza, most debris was acknowledged to
derive from exterior middens (Bökönyi 1976), where the
burnt bone would have broken up and disintegrated relatively
quickly. However, due to the nature of excavation methodo-
logy at most of the sites that gathered together disparate
pieces of bone from large-scale excavation units, no spatial
analyses of bone distributions can be conducted.
Why were so many more burnt bones found at Foeni-Salaş
than at the other sites? Given the absence of any evidence for
large-scale burnt destruction of the settlement, and the fact
that most of the burnt bone remains were from small uniden-
tifiable fragments, it would appear most likely that the ex-
tensive dry and wet sieving operation would have been re-
sponsible for the differences.
Bone weathering and depositional context
The degree of weathering of samples can also profoundly af-
fect the level of reconstruction. Analyses that avoid this issue

108
 Haskel Greenfield: Early Neolithic faunal assemblages from the central Balkans
Figure 2. Semi-logarithmic scale scatter-plot showing re-
lationship between number of taxa identified and identified
taxa assemblage size in each assemblage.
are in danger of reconstructing the patterns of preservation
rather than any prehistoric behaviour. There are profound
differences in preservation between the samples analyzed by
the author.
One major source of bone weathering is soil pH. Soil pH
levels, however, were not measured for most sites, since they
were excavated prior to an interest in taphonomy in the re-
gion. However, pH was measured at Blagotin and Foeni-
Salaş. At both, the Early Neolithic deposits ranged from pH
5.0 to 7.0, depending upon the context — relatively normal
values for the region. While pH may have had less of an in-
fluence, depositional context probably had a great influence.
There appear to be regional variations in bone preserva-
tion. The material from the Iron Gates seems to be much
more poorly preserved than the other regions. Most of the
Hajdučka Vodenica sample was heavily eroded (50% —
Table 2), with characteristic weathering cracks, pitted bone
Figure 3. Normal scale scatter-plot showing relationship
between number of taxa identified and identified taxa as-
semblage size in each assemblage.
109
surfaces, and the loss of delicate features. Based on a study of
some of the bone tools from Lepenski Vir and Vlasac curated
at the University of Belgrade, the state of preservation of the
Hajdučka Vodenica sample is probably characteristic of the
other Iron Gates samples.
In the central Serbian sites, bones tend to be much better
preserved. The Bukovačka Česma sample is remarkable for
the preservation of even delicate morphological features
(Greenfield 1994). All of the bones were very well preserved.
At Divostin, bones were well preserved when buried in the
ashy and alkaline soil of pits. Bones from the cultural horizon
(which is composed of Smonica-type soils, with a heavy clay
content and high water retention), however, were so fragile
and soft that they fragmented or disintegrated during excav-
ation (Lyneis 1988: 301).2 At nearby Blagotin, there were
similar patterns of preservation to those found at Divostin
(Greenfield 2000, n.d. a; Greenfield & Jongsma n.d.). Most of
the bones were lightly (41%) or moderately (51%) weathered
(Table 2). However, when the remains are divided between
those found within features (such as pits) and those that lay in
the open, the degree of preservation is very different. In the
exterior deposits, whether in the plough zone or in the cul-
tural deposits, most of the remains were either moderately
(51%) or heavily (39%) weathered. Very few were lightly
weathered. In the interior deposits, most of the remains were
moderately (51%) or lightly (41%) weathered. There are
some differences between specific deposits, but the effect of
depositional context on preservation remains (Meadow
1978).
Only one sample had information from Pannonia —
Foeni-Salaş (Greenfield n.d. b; Greenfield et al. n.d.). It had
a very different pattern of preservation. Most of the bones
(81.6% — Table 2) were only moderately preserved, regard-
less of whether bones were recovered in pits or in the exterior
deposits. This is a reflection of the sandy loam deposits of the
site. It is difficult to gauge the pattern of bone preservation at
tell sites (which had deeply stratified deposits that were ad-
vantageous for bone preservation — Obre I, Anzabegovo)
since there is no recorded information on these sites.
In general, preservation is worst in the Iron Gates (where
they were heavily eroded), those from Pannonia were better,
and those from central Serbia were the best. Bukovačka
Česma had the best preserved assemblage of all. Little is
known about preservation at the other sites in the region.
Discussion
Despite all of the above issues, can anything be said about
subsistence practices? Before this can be accomplished, two
other issues must be examined.
Sample size and taxonomic diversity
The samples under consideration range quite widely in size.
While large samples are best for reconstruction of subsist-
ence practices, even small assemblages should be con-
sidered. They can contribute information otherwise unob-
tainable from other sources. Even though the number of
samples and the diversity of environmental settings of
Starčevo sites with analyzed fauna is extremely small, each
 The Iron Gates in Prehistory
Figure 4. Normal scale scatter-plot showing relationship
between percentage of wild animals in the assemblage and
identified taxa assemblage size.
adds a new dimension to our understanding of Starčevo
adaptations.
Species counts are available by separate Starčevo phases
only for Anza. At Obre, it is difficult to correlate the species
frequencies with major occupational phases or strata (cf.
Bökönyi 1974b; Greenfield 1991). The faunal samples from
most sites were not published by separate levels, horizons, or
phases within the Starčevo layer(s) (except at Starčevo, it-
self). As a result, it would be impossible to re-analyze the
material, if and when the sub-phasing of the Starčevo culture,
in general, or excavation units at particular sites, becomes
more refined. The total fragment count of unidentified speci-
mens (Table 1) is unknown from half of the sites (Anza,
Divostin, Lepenski Vir, Obre). Since total fragment count
cannot be used to examine the impact of total number of
fragments upon taxonomic diversity, a proxy measure must
be used instead — i.e. NISP. Only Obre and Blagotin have
substantial samples of identified taxa (8000+ NISP). The rest
have medium (2000–4000 bones) or small samples (c. 1000
or less).
When the number of identified taxa are plotted against the
number of bones identified to a genus or species in the sample
on a semi-logarithmic scale, all the sites fall into a single
large cluster (Fig. 2). A semi-logarithmic scale was chosen
because the sample sizes are so disparate. It would appear
that a curvilinear relationship exists between the two vari-
ables (y = 4005.6 Ln (x) – 7917.9). Statistically, when a cur-
vilinear relationship appears and there is a low correlation
(R2 = 0.1951), it is possible that the nature of the scale was
incorrect.
When the two variables are plotted on a normal scale, a
different pattern emerges. Two separate distributions appear
(Fig. 3). In the lower half of the graph, there is a strong pos-
itive correlation between identified sample size and species
110
diversity (R2 = 0.73). In other words, the greater the size of
the identified sample, the greater is the number of identified
taxa. The small number of identified taxa in small as-
semblages is, therefore, a reflection of sample size. Rare
species may appear in small data sets, but less common forms
appear regularly only when the sample reaches certain min-
imal size levels (usually several thousand identified frag-
ments: Casteel 1973; Grayson 1984). Although the absence
of rare species in small samples does not dramatically affect
the overall analysis of subsistence systems, they are neces-
sary for other types of analysis (e.g. ecological reconstruc-
tion: Hesse & Wapnish 1985). Several factors, such as re-
covery and preservation, may intervene to affect the normally
linear relationship between the sample size of identified
mammal bones and the number of mammal species.
In the upper half of the graph, a different pattern appears,
with the data from Blagotin and Obre. Major increases in
taxonomic diversity can occur even when there are only
modest increases in sample sizes. In these two cases, in-
creases in sample size did not significantly alter taxonomic
diversity. It would appear that once sample sizes reach a cer-
tain level (above 5000 identified specimens), other variables
affect taxonomic diversity. In these cases, location would
have limited the range of taxa available for exploitation. Obre
is found overlooking a major river, while Blagotin is found in
the highlands and away from any major water source. Hence,
location and available surrounding resources can be a com-
pounding issue.
Domestic:wild proportions
Is there a variable governing the relative frequency of do-
mestic and wild animals in assemblages? Initial comparison
of the domestic:wild species percentages from Starčevo sites
reveals that half of the assemblages have relatively high and
the other half relatively low percentages of wild animals.
There is no overlap between the two groups. All three of the
Iron Gates sites have very high wild species frequencies
(>61%). But three of the sites outside of the Iron Gates
(Ludoš-Budžak and Golukot in the Pannonian plain, and
Bukovačka Česma in central Serbia) also have very high wild
species counts. The remaining eight sites outside of the Iron
Gates have a preponderance of domestic animals (>74%).
How can these distributions be interpreted? In the past, sites
outside of the Iron Gates were assumed to be those of intrus-
ive food producers because most of the fauna was domestic.
In contrast, sites in the Iron Gates were assumed to be from
indigenous hunter-gatherers because:
1. Even when food production appeared, it was never a
mainstay of the local economy (cf. Bökönyi 1971, 1978),
and
2. The human skeletal record has been consistently
interpreted to represent population continuity in the Iron
Gates from the Terminal Palaeolithic through the
Mesolithic and into the Early Neolithic (cf. Živanović
1975, 1976, 1979).3
In fact, the reality is more complex; sites with high and low
frequencies of wild fauna are found outside the Iron Gates. If
we continue to use the logic of most researchers on the Early
Neolithic of the region, does this mean that even sites outside
of the Iron Gates belong to hunter-gatherers? This is a point
 Haskel Greenfield: Early Neolithic faunal assemblages from the central Balkans
to be returned to later.
When the number of identified fauna from each of the sites
is plotted against the percentage of wild animals identified in
the assemblage, three clusters appear (Fig. 4). Using the x-
axis, the samples can be divided on the basis of percentage of
wild animals — one group with a high percentage of wild
animals and two groups with low percentages. Each of the
major groups can be subdivided on the basis of sample size:
1. The high wild percentage group contains sites with both
low and medium sample sizes. This group comes from a
variety of environmental settings. It includes the sites
from the Iron Gates. The other sites in this group are found
in the region beyond the narrow seclusion of the Iron
Gates, in a variety of environmental contexts. The implic-
ations of these distributions appear fairly obvious. The
high percentage of wild fauna found in the Iron Gates sites
reflect the isolation, topographic ruggedness and abund-
ance of wild fauna, and the lower energy returns of pas-
toralism versus hunting-fishing. The topography and en-
vironment of the gorge with its steep slopes and thick
forests make pastoralism difficult and less profitable as a
subsistence strategy when compared with the hunting of
wild ungulates. The high percentage of wild fauna in the
sites outside the Iron Gates (Bukovačka Česma, Golukot,
and Nosa) is clearly influenced by the location of the sites
near or on major riverine environments. These sites are
more similar to those from the Iron Gates. This is a very
different pattern than in the other sites found outside of the
Iron Gates sites rather than the generally accepted pattern
for sites outside of the Iron Gates.
2. The low wild percentage group can be divided into two
sub-groups: (i) one sub-group is composed of sites with
small and medium sample sizes; (ii) the other sub-group is
composed of sites with large sample sizes.
On the surface, it would appear that the difference in
wild:domestic ratios may be affected by sample size. The
smaller samples tend to have higher percentages of wild an-
imals. The situation is more complex. One variable is recov-
ery method (sieving or hand recovery). The partially sieved
samples include both those with large (Blagotin) and medium
sample sizes (Anza) among the low wild group. Sieving does
not appear to have much of an effect on wild: domestic ratios.
The preponderance of wild fauna cannot be attributed en-
tirely to recovery method since wild animals are generally
not larger than domestic animals. If anything, many wild taxa
are smaller (e.g. beaver and fish) and would be more difficult
to recover by hand. Hand collection would under-represent
the wild component of the assemblage. Fish remains, in par-
ticular, are poorly collected by hand and therefore could be
predicted to have been present in larger quantities before ex-
cavation. For example, they already represent 10% of the
Bukovačka Česma sample, a hand collected site located
above the floodplain of the Morava river. They were probably
present in even greater frequencies before excavation.
Another possibility is the surrounding environment and its
relationship to the subsistence economy. Most of the sites
with high frequencies of wild fauna are found in a limited set
of environments — in the temperate environmental zone, and
close to rivers, streams and floodplains that even today
abound with wildlife. The sites with smaller sample sizes also
111
happen to be in such environments. Therefore, their high wild
frequencies may not be a reflection of sample size, but of
their exploitation of the rich wild resources in the surround-
ing wetlands. The faunal remains of Bukovačka Česma,
Golukot and Nosa are more similar to those from the Iron
Gates than from other sites spread through the hill country of
Serbia, Bosnia or Macedonia because they also overlook
resource-rich aquatic environments. Much of the source of
variation between samples can therefore be attributed to a
limited set of variables: recovery methodology, sample size
and environmental location. Therefore, the percentage of
wild or domestic fauna may still be used as a general reflec-
tion of overall environment. Recovery can alter the details
significantly, nonetheless.
Much of the source of variation between assemblages can
therefore be attributed to a limited set of variables: recovery
methodology, sample size and environmental location. The
quantity of wild remains in sites outside of the Iron Gates can
be relatively high indicating that traditional assumptions
concerning the nature and rate of spread of the transition to a
food producing economy rate in Southeast Europe need to be
re-evaluated. The spread of food production may be a more
complex process than previously envisioned.
Some thoughts on the transition to food production in
Southeast Europe
If we accept the assumption that sites with wild resources in
the Iron Gates are the remains of indigenous foragers, then it
may even be that some of the sites outside of the Iron Gates
with a predominance of wild resources may also be the re-
mains of acculturating foragers. In fact, we know too little
about the relationship between Mesolithic and Early
Neolithic populations in this region to be able to make such
blanket assumptions. While Early Neolithic sites are wide-
spread across the region, so much is assumed in Mesolithic
studies on the basis of a few highly concentrated localities
(Iron Gates, Montengro).
For much of the Early Neolithic of Central and Northern
Europe, indigenous Mesolithic foragers probably coexisted
with initial and intrusive food producers. All we need to do is
to look at the evidence from other better-studied regions for
comparison. For the first phase of the Early Neolithic of
Central and Northern Europe, indigenous Mesolithic for-
agers coexisted with initial and intrusive food producers (e.g.
Bogucki 1988). There is also evidence for acculturation by
indigenous foragers to an agricultural lifestyle (Price 1987).
In contrast, most studies of Southeast European prehistory
implicitly assume that either the indigenous population im-
mediately adopted food production upon its arrival or the
area was not occupied by foragers at the advent of the
Neolithic. The Mesolithic forager sites are not visible in the
areas of intensive Early Neolithic settlement (in the great
river valleys, beyond the Iron Gates and Montenegrin high-
lands). The widespread presence of Mesolithic sites (in areas
that have been systematically surveyed, i.e. Iron Gates —
Whittle 1985; or in regions with little vegetative cover, e.g.
Montenegro — Djuričić 1991) and the demonstrable con-
tinuity between Mesolithic and Early Neolithic human popu-
lations in the more mountainous and less accessible areas of
the peninsula (i.e. the Iron Gates — Živanović 1975, 1976,
 The Iron Gates in Prehistory
1979; y’Edynak 1978) indicate that it is likely that foragers
were also present in the lowlands at the advent of the
Neolithic.
If indigenous foragers were present at the advent of the
Neolithic, it might be more realistic to assume that they co-
existed with early food producers for a period of time, only
slowly abandoning foraging subsistence economies? During
this transition, they may have retained much of their com-
mitment to foraging while at the same time incorporating
some domestic food producing strategies and technologies
into their subsistence round (as occurred in Southwest,
Northwest and Northeast Europe — Price 1987; Rimantiene
1992), such as domestic animals, pottery production, or other
hallmarks of Starčevo sites. If this is the case, we might ex-
pect foraging communities to retain their mobility, while ad-
opting and incorporating domestic animals into their mobile
subsistence system before adopting other Early Neolithic ac-
coutrements (such as pottery) or vice versa as in northern and
Western Europe (cf. Zvelebil & Dolukhanov 1991). If this
was the case, many of the sites defined as Mesolithic on the
basis of lithic typology and the absence of ceramics may, in
fact, date from the Early Neolithic. The paucity of radiocar-
bon dates from ‘Mesolithic’-type sites makes this a difficult
hypothesis to test immediately. However, there are no cases
where domestic animals are found in clear association with a
Mesolithic-type stone tool technology (and where Early
Neolithic pottery is absent) outside of the Iron Gates. The few
early Holocene sites without pottery, but with a Mesolithic
stone assemblage, do not contain any domestic animals. But
there are several sites outside of the Iron Gates with Early
Neolithic material culture, but with assemblages dominated
by wild animals (Greenfield, this volume).
The question then arises — to whom do these settlements
actually belong?
1. Are they settlements of indigenous foragers slowly accul-
turating to a food producing economy by selectively in-
corporating components of typical Starčevo material
culture?
2. Are they a specialized wild food collection site for a larger
community committed to domestic economies?
3. Do they represent new settlements established by recently
arrived food producers who have yet to build up the agri-
cultural component of their local economy?
4. Or, could they be settlements of food producers who have
moved into an agriculturally marginal area?
Each alternative has an emphasis upon wild resources as a
logical explanation for the patterning in the data. How to
choose between alternatives?
The second and third choices seem less likely considering
the size and degree of permanency of the settlements, as re-
flected by the material culture and domestic architecture, and
the absence of any evidence for specialized production and
exchange between contemporary sites involving foodstuffs at
this time or even later in the Neolithic. Estimations of the
seasonality of death of various species based upon the age-
at-death of the various species at several sites also indicate
year-round occupation (Clason 1980; Arnold & Greenfield
2006). It is more difficult to distinguish between the first and
fourth choices. The choice varies depending upon the evid-
ence chosen to consider. In the Iron Gates, there is evidence
112
for human population, material culture, and subsistence con-
tinuity between Mesolithic and Early Neolithic levels, even
with the appearance of new subsistence strategies and mater-
ial culture. But comparable studies have not been undertaken
upon assemblages from Mesolithic sites outside of the Iron
Gates. Also, virtually all sites with a Mesolithic-type artefact
assemblage from the region outside of the Iron Gates are not
dated by radiocarbon techniques. They are assumed to date
earlier than Starčevo on the basis of the similarity of their
material cultural assemblages to other Mesolithic sites in the
Iron Gates and on the absence of Starčevo ceramic material
culture.
Conclusion
Previous studies of the earliest food producing communities
in the more temperate zones of Southeast Europe, such as the
central Balkans, have generally attempted to show that sub-
sistence economies were largely orientated towards domestic
animal economies (e.g. Murray 1970; Barker 1985). Faunal
data from Early Neolithic sites in the central Balkans are re-
considered here to demonstrate that animal exploitation
strategies during the Early Neolithic of the region was a rel-
atively complex matter and that taphonomic issues must be
considered if prehistoric economies are to be understood.
Even though such faunal assemblages can increase our un-
derstanding of Early Neolithic subsistence strategies, it is not
an easy leap from bones to human behaviour. A number of
studies have demonstrated the obvious and not so obvious
pitfalls of extrapolating from patterning in bone assemblages
to prehistoric human behaviour (e.g. Gifford 1981;
Greenfield 1986, 1988a, 1988b, 1991; Koster & Chang
1986). All of the analyses of Early Neolithic faunal as-
semblages and the ensuing comparative use of their results
have ignored the possible distorting effects of bone as-
semblage attrition (e.g. Champion et al. 1984; Lazić 1988).
There is a great deal of variability from site to site which af-
fects the representativeness of the sample and utility of the
final analysis, such as variability of the surrounding environ-
mental conditions (temperature, precipitation), local sedi-
mentary conditions (including soil type and pH), recovery
procedures, and methods of quantification. As a result of the
re-evaluation of the quality of the Early Neolithic faunal data
derived from such an approach, it is hoped that this will force
a reconsideration of the significance and value of conclusions
derived from previously analyzed faunal assemblages.
Notes
1. Manson (1990, this volume) established an age range of
6100–5100 cal BC, based on the archaeomagnetic intensity ana-
lysis of Starčevo and Körös ceramic sherds.
2. But note that Bökönyi’s (1988) faunal report does not deal with
this issue.
3. Skeletal series from Starčevo sites outside of the Iron Gates are
few and far between. However, they are now being re-analyzed
and show continuity with Mesolithic populations (see this
volume).
Acknowledgements
I should like to express my sincere gratitude to my numerous friends
 Haskel Greenfield: Early Neolithic faunal assemblages from the central Balkans
and colleagues throughout the central Balkans who made available
their assemblages for analysis or who helped in the excavation and
analysis. Unfortunatel, there are too many to list separately here.
Funding for this research came from a variety of sources, including
the National Science Foundation (#BNS8105358), the Wenner-
Gren Foundation for Anthropological Research (#4210), the
Fulbright-Hays Foreign Language and Area Studies Program
(Program no. 84.022; Project no. 022AH10048), the International
Research and Exchanges Board (1981–82, 1993–94), the University
of Manitoba, Winnipeg (1990–99), and the Social Science and
Humanities Research Council of Canada (1990–1996). I am grateful
to all of them for supporting the fieldwork, analysis and writing up
of this paper. In particular, I thank my wife, Tina Jongsma, who
shared much of the field and laboratory analysis that led to this pa-
per. Any errors, however, are my responsibility.
References
Ammerman, A. & Cavalli-Sforza, L.L. 1971: Measuring the rate of
spread of early farming in Europe, Man n.s. 6: 674–688.
Aranđelović-Garašanin, D. 1954: Starčevačka Kultura. Ljubljana:
Universitet v Ljubljana.
Arnold, E.R. & Greenfield, H.J. 2006: The Origins of Transhumant
Pastoralism in Temperate Southeastern Europe: A
Zooarchaeological Perspective from the Central Balkans.
British Archaeological Reports, International Series 1538.
Oxford: Archaeopress.
Barker, G. 1975: Early Neolithic land use in Yugoslavia.
Proceedings of the Prehistoric Society 41: 85–104.
— 1985: Prehistoric Farming in Europe. Cambridge: Cambridge
University Press.
Bogucki, P. 1988: Forest Farmers and Stockherders: Early
Agriculture and its Consequences in North-Central Europe.
Cambridge: Cambridge University Press.
Bökönyi, S. 1970: A new method for the determination of the num-
ber of individuals in animal bone material. American Journal of
Archaeology 74: 291–292.
— 1971: Animal remains from Lepenski Vir. Science 167:
1702–1704.
— 1974a: History of Domestic Mammals in Central and Eastern
Europe. Budapest: Akadémiai Kiadó.
— 1974b: The vertebrate fauna. In Gimbutas, M. (ed.) ‘Obre I and
II. Neolithic Sites in Bosnia’. Wissenschaftliche Mitteilungen des
Bosnisch-Herzegowinischen Landesmuseum, Arheologie 3, part
B: 55–154 (Sarajevo).
— 1976: The vertebrate fauna from Anza. In Gimbutas, M. (ed.)
Neolithic Macedonia as Reflected by Excavation at Anza,
Southeast Yugoslavia. Los Angeles: The University of
California, Los Angeles (Monumenta Archaeologica, vol. 1),
313–363.
— 1978: The vertebrate fauna of Vlasac. In Srejović, D. & Letica,
Z. (eds) Vlasac: A Mesolithic Settlement in the Iron Gates, vol. 2.
Belgrade: Serbian Academy of Science and Arts, 35–65.
— 1988: The Neolithic fauna of Divostin. In McPherron, A. &
Srejović, D. (eds) Divostin and the Neolithic of Central Serbia.
Pittsburgh: University of Pittsburgh, Department of
Anthropology (Ethnology Monographs no. 10), 419–446.
Bonfield, W. & Li, C.H. 1966: Deformation and fracture of bone.
Journal of Applied Physics 37: 869–875.
Casteel, R.W. 1973: Some biases in the recovery of archaeological
faunal remains. Proceedings of the Prehistoric Society 39:
382–388.
Champion, T., Gamble, C., Shennan, S. & Whittle, A. 1984:
Prehistoric Europe. London: Academic Press.
Childe, V.G. 1929: The Danube in Prehistory. Oxford: Clarendon
1
113
Press.
— 1958: The Prehistory of European Society. Harmondsworth:
Penguin.
Clason, A.T. 1980: Padina and Starčevo: game, fish and cattle.
Palaeohistoria 22: 142–173.
Clason, A.T. & Prummel, W. 1977: Collecting, sieving and archae-
ozoological research. Journal of Archaeological Science 4:
171–175.
Dennell, R. 1984: European Economic Prehistory: A New
Approach. London: Academic Press.
Djuričić, L. 1991: Surveying Paleolithic and Mesolithic Sites (in
Serbian). Glasnik Srpsko Arheolosko Društvo (Journal of the
Serbian Archaeological Society — Belgrade) 7: 43–51.
Garašanin, M. 1973: Praistorija na Tlu SR Srbija, 2nd edition.
Belgrade: Srpska Kniževna Zadruga.
— 1979: Centralbalkanska zona, In Basler, D., Benac, A.,
Gabrovec, S., Garašanin, M., Tasić, N., Cović, B. & Vinski-
Gasparini, K. (eds) Praistorija Jugoslavenskih Zemalja II:
Neolitsko Doba. Sarajevo: Akademija Nauka i Umjetnosti Bosne
i Hercegovine, 79–212.
— 1983: The Stone Age in the central Balkans. In Cambridge
Ancient History, volume 3, part 1. Cambridge: Cambridge
University Press, 75–135.
Gifford, D.P. 1981: Taphonomy and paleoecology: a critical review
of archaeology’s sister disciplines. In Schiffer, M.B. (ed.)
Advances in Archaeological Method and Theory, volume 4. New
York: Academic Press, 365–438.
Gilbert, A.S., Singer, B.H. & Perkins, D. 1982. Quantification ex-
periments on computer-simulated faunal collections. Ossa 8:
79–94.
Gimbutas, M. (ed.) 1976: Neolithic Macedonia as Reflected by
Excavation at Anza, Southeast Yugoslavia. Los Angeles: The
University of California, Los Angeles (Monumenta
Archaeologica, vol. 1).
Grayson, D.K. 1984: Quantitative Zooarchaeology. New York:
Academic Press.
Greenfield, H.J. 1986: The Paleoeconomy of the Central Balkans
(Serbia): A Zooarchaeological Perspective on the Late Neolithic
and Bronze Age (4500–1000 BC). BAR International Series 304.
Oxford: British Archaeological Reports.
— 1988a: Bone consumption by pigs in a contemporary Serbian
village: implications for the interpretation of prehistoric faunal
assemblages. Journal of Field Archaeology 15: 473–479.
— 1988b: The origins of milk and wool production in the Old
World: a zooarchaeological perspective from the central
Balkans. Current Anthropology 22: 573–593.
— 1991: Fauna from the Late Neolithic of the central Balkans: is-
sues in subsistence and land use. Journal of Field Archaeology
18: 161–186.
— 1993: Zooarchaeology, taphonomy, and the origins of food pro-
duction in the central Balkans. In Jamieson, R.W., Abonyi, S. &
Mirau, N.A. (eds) Culture and Environment: A Fragile Co-
existence. Proceedings of the 24th Chacmool Conference.
Calgary: University of Calgary Archaeological Association,
111–117.
— 1994: Faunal remains from the Early Neolithic Starčevo settle-
ment at Bukovačka Česma, Starinar n.s. 43/44: 103–114.
— 2000: Integrating surface and subsurface reconnaissance data in
the study of stratigraphically complex sites: Blagotin, Serbia.
Geoarchaeology 15: 167–201.
— n.d. a: ‘Vertebrate Fauna from Blagotin’. Unpublished
manuscript on file, University of Manitoba, Department of
Anthropology.
— n.d. b: ‘Vertebrate Fauna from Foeni-Salaş’. Unpublished
manuscript on file, University of Manitoba, Department of
Anthropology.
 The Iron Gates in Prehistory
Greenfield, H.J. & Jongsma, T.L. n.d.: Early Neolithic Blagotin: a
summary of recent research. In Bailey, D.W., Whittle, A. &
Cummins, V. (eds) Sedentism in the Central and East European
Neolithic. Oxford: Oxbow Books.
Greenfield, H.J., Jongsma,T.L. & Jezik, S. n.d.: Sedentary pastoral
gatherers in the Early Neolithic — architectural, botanical, and
zoological evidence for mobile economies from Foeni-Salaş,
SW Romania. In Bailey, D.W., Whittle, A. & Cummins, V. (eds)
Sedentism in the Central and East European Neolithic. Oxford:
Oxbow Books.
Halstead, P. 1989: Like rising damp? An ecological approach to the
spread of farming in south east and central Europe. In Milles, A.,
Williams, D. & Gardner, N. (eds) The Beginnings of Agriculture.
BAR International Series 406. Oxford: B.A.R., 23–53.
Hesse, B. & Wapnish, P. 1985: Animal Bone Archaeology: From
Objectives to Analysis. Washington D.C.: Taraxacum Press.
Jongsma, T. & Greenfield, H.J. 2001: Architectural technology and
the spread of early agricultural societies in temperate southeast-
ern Europe. In Tupakka, S., Gillespie, J. & de Mille, C. (eds)
Untrampled Ground — Untrammelled Views: Human
Exploitation of, and Settlement Patterns on, New Landscapes.
Proceedings of the 31st Annual Chacmool Conference, 1998.
Calgary: University of Calgary, Department of Archaeology,
181–200.
Klaić, B. 1982: Rjecnik Stranih Rijeci. Zagreb: Naklodni Zavod
Matice Hrvatske.
Koster, H. & Chang, C. 1986: Beyond bones: toward an archaeology
of pastoralism. In Schiffer, M.B. (ed.) Advances in
Archaeological Method and Theory, volume 9. Orlando:
Academic Press, 97–148.
Lazić, M. 1988: Fauna of mammals from the Neolithic settlements
in Serbia. In Srejović, D. (ed.) The Neolithic of Serbia. Belgrade:
University of Belgrade, 24–38.
Leković, V. 1985: The Starčevo mortuary practices — new per-
spectives. Godišnjak (Sarajevo: Centar za Balkanaloška
Istraživanja) 15: 157–172.
Lyman, R.L. 1994: Vertebrate Taphonomy. Cambridge: Cambridge
University Press.
Lyneis, M.M. 1988: Antler and bone artifacts from Divostin. In
McPherron, A. & Srejović, D. (eds) Divostin and the Neolithic of
Central Serbia. Pittsburgh: University of Pittsburgh, Department
of Anthropology (Ethnology Monographs no. 10), 301–319.
Manson, J.L. 1990: A Reanalysis of Starčevo Culture Ceramics:
Implications for Neolithic Development in the Balkans.
Unpublished Ph.D. dissertation. Southern Illinois University,
Carbondale, IL.
Meadow, R. 1978: Effects of context on the interpretation of faunal
remains. A case study. In Meadow, R. & Zeder, M.A. (eds)
Approaches to Faunal Analysis in the Middle East. Cambridge,
MA: Harvard University Peabody Museum Bulletin no. 2, 15–21.
114
Milojčić, V. 1949: Chronologie der Jungeren Steinzeit Mittel- und
Südosteuropas. Berlin: Gebruder Mann.
Murray, J. 1970: The First European Agriculture: A Study of the
Osteological and Botanical Evidence until 2000 B.C. Edinburgh:
Edinburgh University Press.
Naval Intelligence Division. 1944: Jugoslavia. Geographical
Handbook Series, vol. 1. London: Naval Intelligence Division.
Payne, S. 1972: Partial recovery and sample bias: the results of some
sieving experiments, In Higgs, E.S. (ed.) Papers in Economic
Prehistory. Cambridge: Cambridge University Press, 49–64.
Pounds, N. 1969: Eastern Europe. Chicago: Chicago University
Press.
Price, T.D. 1987: The Mesolithic of western Europe. Journal of
World Prehistory 1: 225–306.
Rimantiene, R. 1992: The Neolithic of the eastern Baltic. Journal of
World Prehistory 6: 97–143.
Schwartz, C. 1976: The vertebrate fauna from Rug Bair. In
Gimbutas, M. (ed.) Neolithic Macedonia as Reflected by
Excavation at Anza, Southeast Yugoslavia. Los Angeles: The
University of California, Los Angeles (Monumenta
Archaeologica, vol. 1), 364–374.
Srejović, D. 1979: Protoneolit kultura Lepenskog vira. In Basler, D.,
Benac, A., Gabrovec, S., Garašanin, M., Tasić, N., Cović, B. &
Vinski-Gasparini, K. (eds) Praistorija Jugoslavenskih Zemalja
II: Neolitsko Doba. Sarajevo: Akademija Nauka i Umjetnosti
Bosne i Hercegovine, 33–78.
Srejović, D. & Letica, Z. (eds) 1978: Vlasac: A Mesolithic
Settlement in the Iron Gates, 2 volumes. Belgrade: Serbian
Academy of Arts and Sciences.
Tringham, R. 1971: Hunters, Fishers and Farmers of Eastern
Europe, 6000–3000 B.C. London: Hutchinson University
Library.
Turnock, D. 1989: The Human Geography of Eastern Europe.
London: Routledge.
Whittle, A. 1985: Neolithic Europe: A Survey. Cambridge:
Cambridge University Press.
y’Edynak, G. 1978: Culture, diet and dental reduction in Mesolithic
forager-fishers of Yugoslavia. Current Anthropology 19:
616–618.
Živanović, S. 1975: A note on the anthropological characteristics of
the Padina population. Zeitschrift für Morphologie und
Anthropologie 66: 161–175.
— 1976: Cromagnon in the Iron Gate Gorge of the Danube. Nature
260: 518.
— 1979: Further evidence on Cro-Magnon in the Iron Gate Gorge
of the Danube. Current Anthropology 20: 805–806.
Zvelebil, M. & Dolukhanov, P.M. 1991: The transition to farming in
eastern and northern Europe. Journal of World Prehistory 5:
233–278.
 Section 2

SITE STUDIES
 Lepenski Vir animal bones: what was left in the houses?

Vesna Dimitrijević

Abstract: A collection of animal bones from the eponymous site of the Lepenski Vir culture in the Iron Gates is stored in the National
Museum in Belgrade. Formerly, four units from this collection were described, two of them related to the houses, another two to the areas
outside houses (Dimitrijević 2000). In this paper, faunal assemblages collected from the houses are analyzed. By analyzing contextual units
with animal bones that come from the houses an attempt is being made to understand activities of the house dwellers and significant aspects
of their relations to animals. A distinction is made between the remains related to the lifespan of a particular house, and those the inhabitants
left, intentionally, or unintentionally, behind them at the time of house abandonment. The animal species composition and skeletal parts
distribution are presented. The distribution of dog bones inside houses, the evidence of food preparation and consumption, skinning pro-
cedure, bone tool use and manufacturing, as well as the circumstances of the deposition and symbolic meaning of red deer antlers in the
houses of Lepenski Vir are discussed.
Key words: Lepenski Vir, animal bones, Mesolithic, Early Neolithic, house abandonment event, red deer antlers, dog

Introduction What was left in the houses?

The eponymous site of the Lepenski Vir culture in the Iron
Gates was excavated from 1965 to 1971 by Dragoslav
Srejović. Over the excavated area of 2500 m2, a 3.5 m-thick
cultural layer with a sequence of building horizons of
Mesolithic and Early Neolithic settlements was identified
(Srejović 1966, 1969; Radovanović 1996). Radiometric dates
associated with the houses indicate a time span from c. 6200
to 5500 cal BC (Quitta 1975; Whittle et al. 2002). The ex-
cavations were undertaken as a part of the vast Ðerdap (Iron
Gates) I rescue project in advance of the construction of the
dam near Kladovo. The time constraints imposed by the
construction project influenced the excavation strategy.
Collecting animal remains was not among the priorities.
Consequently, only a small, hand-collected, sample of animal
bones was saved.
Animal remains from the earlier excavation campaigns
were analyzed and published in a preliminary report
(Bökönyi 1969). These were a sample of the bones recovered
and comprised 2999 identified specimens. A selection of this
sample, as well as animal remains collected in subsequent
campaigns, are stored in the National Museum in Belgrade.
The animal bones curated in the National Museum in
Belgrade are packed in bags, with field data labels designat-
ing their origin. There are two main types of contextual unit
with osteological material — units related to structures, and
those related to various site deposits but unconnected with
the architectural structures. The latter are recorded by arbit-
rary excavation layers and their spatial position across the
site. Those units related to the architectural structures mainly
include the designations: ‘house’, ‘on the floor of the house’,
‘below the floor of the house’, ‘between the floors of the
houses’, ‘between the houses’, ‘hearth’, and ‘grave’ written
on the bag labels. This paper focuses on the contextual units
that were related exclusively to the houses.
By analyzing contextual units with animal bones from the
houses, we hope to approach the life of the house dwellers
and important aspects of their relations to animals. We shall
try to differentiate between those remains that are related to
the lifespan of a particular house, and those which dwellers of
the houses left, intentionally or unintentionally, behind them,
when abandoning a house.
It is necessary to remain aware of numerous obstacles
posed by excavation technique, stratigraphical problems, and
the nature of the archaeological record itself. The answer to
the question in the title of the paper is thus complex and
‘multi-layered’.
In the course of the excavations at Lepenski Vir not all the
bones were collected. This fact can best be affirmed by the
quantity of collected bones in relation to the large excavated
area. Bones were not collected from every house area. If
nothing else, some animal bone remains can be seen on the
photographs taken during the excavations that were neither
collected nor curated. Within single houses, not all the bones
found were gathered, also evident in unsaved material seen
on published photographs. A large part of the osteological
material analyzed by Bökönyi was discarded after his ana-
lyses, and only a small sample was curated in the collection
of the National Museum. Furthermore, we have to remain
aware that not a single contextual unit of collected animal
bones from Lepenski Vir represents a complete recovery of
the remains.
Also, field labels designating the contextual origin of an-
imal bones in relation to structures are not very precise. Many
labels indicate the number of a house only. It may be sup-
posed that these are bones that were found on the floor of a
house. However, it is difficult to be sure that these remains do
not relate to a layer above the floor that formed later, which
is certainly very relevant for their interpretation. In this re-

117
 The Iron Gates in Prehistory
Figure 1. Plan of the Lepenski Vir I houses (from Radovanović 1996).
spect, the units related to superimposed houses are particu-
larly problematic, with designations ‘between houses’ or
‘between house floors’ on the labels. It remains uncertain
whether the bones collected from these units relate to the
floor of the earlier house, or the foundation of the later house.
Even in those instances when field data confirm that bones
were found lying directly on the floor, in what capacity can
we relate that specific assemblage of bones to the life of the
house? In this respect, osteological evidence may often help
us to differentiate between what may be a heterogeneous as-
semblage, coming from a mixture of animal bones of the
layer deposited over an assemblage that represents a mean-
ingful or structured deposition within a house.
During the life of a house, animals, their carcasses or parts
of the carcasses, or even bones themselves, would be present
on particular occasions. Also, sometimes these remains
would have particular meaning and would merit a specific
explanation. Reasons for the presence of animal remains in-
side the houses may relate to a wide range of possibilities,
from the creatures which shared the space of a house with
humans, carcass parts as food that was consumed or left in-
side the house, accumulated food debris, and bone artefacts
or raw material discarded in the process of tool
manufacturing.
However, bones left by similar activities could be found,
and commonly are found, outside houses; while in the pro-
cess of burying a house the bones may be deposited on its
118
floors or in the layers above the floor. It may be supposed that
an abandoned house was either suddenly covered with a layer
of sediment when the base for the overlying house was being
constructed, or slowly buried over a period of time. In either
case, sediments containing animal remains would overlie the
house.
Being peculiar and extraordinary in many respects, the
houses of Lepenski Vir seem to contain in several instances
animal remains — offerings, that were intentionally left by
their occupants. The circumstances of these findings will be
emphasized in the following discussion. Other assemblages
of animal bones that can be reliably related to the houses will
also be described. The faunal assemblages are presented ac-
cording to the groups of superimposed houses and their loc-
ations as defined by Radovanović (1996) (Fig. 1), followed
by a description of assemblages from non-superimposed
houses. Any additional data found on field labels apart from
house numbers is mentioned. Only mammalian remains are
analyzed, while bird, fish and invertebrate remains are noted.
Remains of the following species were found in the houses
of Lepenski Vir: Erinaceus concolor Martin (hedgehog),
Castor fiber Linnæus (beaver), Canis familiaris Linnæus
(dog), Ursus arctos Linnæus (brown bear), Martes martes
(Linnæus) (pine marten), Meles meles (Linnæus) (badger),
Sus scrofa Linnæus (wild boar), Cervus elaphus Linnæus
(red deer), Capreolus capreolus (Linnæus) (roe deer), Bos
primigenius Bojanus (aurochs), Rupicapra rupicapra
 Vesna Dimitrijević: Lepenski Vir animal bones

Table 1. Animal remains found within the houses of Lepenski Vir: number of specimens, (minimum number of individuals), +
presence confirmed, * supplemented from the literature.

Rupicapra rupicapra

Capreolus capreolus
Erinaceus concolor

Bos primigenius
Canis familiaris

Cervus elaphus
Martes martes
Ursus arctos

Meles meles
Castor fiber

Sus scrofa
Location

molluscs
House

birds

fish
2 9 7(1) 17(1)

4 13 6(2) 8(1) +

30 1(1) + +
5
22 3(3)*

6 47 1(1) 1(1) 1(1)

36 1(1) +
7
35 1(1) 7(2) 1(1) 3(1) 5(2) 1(1) + + +

38 1(1)

8 37 1(1)

42 1(1)

34 2(2) 1(1)
10
27 1(1)

18 4(1)
11
23 1(1) 1(1) 12(2) 1(1) 2(2) 2(1) 1(1) + +

13 46 1(1)*

20 1(1) 1(1) 3(2) 3(2) + + +

14
32 1(1) 2(1) +

26' 2(1) 1(1) + +

26 5(1) 1(1) 1(1)

25 2(1) 1(1)

28 1(1) + +

40 10(2) 1(1) 2(1)

non-superimposed houses

44 2(1) 1(1) 1(1) + +

48 2(2)

49 1(1) 1(1) +

54 1(1) + +

61 1(1)*

62 2(1) + +

65 1(1) 2(1) + +

69 1(1) + +

70 1(1) 3(1) 3(1) +

71 + +

119
 The Iron Gates in Prehistory
Figure 2. Dog lumbar vertebrae from house 9: a. first or
second lumbar vertebra, arrow pointing to butchering
marks, b. the last or next to last lumbar vertebra, arrow
pointing to exostosis.
Figure 3. Right red deer antler from house 22.
120
(Linnæus) (chamois), birds, fish and molluscs (Table 1).
Superimposed houses
Location 2, houses 17, 8, 7 and 9
Bones originate from house 9. Some bones bear traces of fire
with varying degrees of carbonization/calcification. Remains
of dog and red deer are present, and numerous smaller frag-
ments of large mammal bones. The red deer remains consist
of a fragmented tine, many smaller antler fragments, and the
fragment of a second phalanx. All pieces are carbonized. The
dog remains comprise a carbonized fragment of lower jaw,
with the first premolar and preserved alveoli for the second
and the third premolar, an isolated third lower incisor, two rib
fragments, and a calcified metapodial fragment. In addition,
there are two dog lumbar vertebrae (Fig. 2), with no traces of
fire. The damage on their processes is recent. One of them,
probably the first or the second in the lumbar row (Fig. 2a),
bears butchery marks. Another, probably the last or next to
the last in the lumbar row (Fig. 2b), shows a pathological
feature — exostosis on the ventral side of the fusion between
the caput and the body of the vertebra, indicating an un-
healthy condition of the animal, possibly related to old age.
The assemblage contains red deer remains that could relate
to tool manufacturing and dog remains possibly originating
from one or two skeletons. Dogs were probably eaten.
Location 4, houses13, 14 and 15
Animal bones come from the earliest house at this location,
house 13. Red deer and wild boar remains were present,
while several identifiable fragments are of large mammals,
most likely these two identified species. The red deer remains
comprise two antler fragments, a fragment of the occipital
part of the skull, two fragments of thoracic vertebrae, a frag-
ment of a lumbar vertebra, a fragment of a metatarsal shaft,
and a carpal bone. Wild boar is represented by an upper third
molar, fragments of a proximal humerus and a humerus shaft,
a proximal ulna, a distal scapula, and the first phalanx. The
red deer antler fragments bear marks made by artefacts. The
wild boar remains could belong to at least two animals, on the
basis of the heavily worn third upper molar of an aged anim-
al, in contrast to the scapula which comes from a young an-
imal. Two humeral fragments bear butchery marks.
The assemblage is heterogeneous: the bones may originate
from carcass parts that were butchered for food consumption,
and skeleton parts used as raw materials.
Location 5, houses 30, 29, 22 and 21
At this location animal bones come from the floor of the
earliest house in the group, house 30, and also from house 22,
placed above house 29, and below house 21.
On the floor of house 30, a fragment of a red deer ilium
corpus fragment was found, with butchery marks on its
dorsal side. Large fish bones and teeth, and mollusc shells
were present too.
An interesting situation is attested on the floor of house 22.
Two pairs of red deer antlers were arranged here. First,
antlers of a young red deer attached to the skull were found
beside an extended human skeleton — burial 7 (Srejović &
Babović 1983: 18, upper right). This burial was cut through
 Vesna Dimitrijević: Lepenski Vir animal bones
Figure 4. Carbonized scapulae from house 47: a. wild boar right scapula; b. roe deer right scapula; c. red deer right
scapula.
the floor of house 21, which partly covered house 22 in its
rear part. Both the skeleton and red deer antlers were prac-
tically lying on the floor of house 22. Although the skeleton
and antlers are seemingly in connection, as shown on the
published photo, one of the antlers is just beneath and beside
the stone construction of house 21, while the other one is
stretching further in the direction below the floor of house 21.
It seems more probable that it was placed on the floor of
house 22, before house 21 was built and less probable that the
pit cut for burial 7 was of a width to permit the antlers to be
enclosed in the grave.
Another pair of red deer antlers, together with the skull can
be clearly seen on a photograph (Srejović & Babović 1983:
136, lower) beside the hearth of house 22, lying directly on
the floor. The right (Fig. 3) and left antlers are preserved in
the collection of the National Museum and correspond in size
and shape to those seen on the photograph, but the skull was
not saved. Most of their original length is preserved; the sur-
viving length of the left antler is 560 mm, and of the right
antler 630 mm. Brow tines, bez tines and trez tines are in
analogous positions and the antlers are of similar size. There
is a crown on the right antler with two tines that lack their
121
tips. The circumference of the right antler burr is 225 mm,
and its greater diameter 70 mm. The size of the antlers,
number of tines, and branching of the crown on the right
antler, point to an animal at least 5 years old; this is indicated
especially by the presence of the bez tine, which first appears
at that age, but could also appear when an individual is six
years old (Dragišić 1957).
Location 6, houses 58, 53, 47' and 47
The preserved animal bones come from the earliest house at
this location, house 47, more specifically, in its west (i.e.
narrow) side, as recorded on the field label. There are three
right scapulae of three different species: red deer, wild boar
and roe deer (Fig. 4). All three were carbonized and frag-
mented in a similar way: the glenoid part was preserved to-
gether with a smaller portion of the blade above the scapula
neck. Most likely these scapulae indicate remains of meat
consumption.
Location 7, houses 36 and 35
In the hearth of house 36 the third phalanx of a wild boar and
fish remains were found.
 The Iron Gates in Prehistory
Figure 5. Dog mandibles from house 34: a. left mandible of an adult animal; b. right mandible of a puppy.
In the later of the two houses, house 35, a large sample of
mammal bones was collected, consisting of 96 fragments of
various skeletal parts that can be attributed to several animal
species. Bird, fish and mollusc remains were collected, too.
Red deer remains consist of a tip fragment of an antler tine,
an upper fourth premolar, fragments of the left and right pel-
vis, and a fragment of an unfused radius distal epiphysis. The
aforementioned fragment of left pelvis and the fragment of a
radius distal epiphysis are gnawed. The antler fragment bears
many scars made by a blunt artifact. These bones belong to at
least one adult animal, and one young individual whose body
was still growing. The wild boar is represented by a fragment
of cranium, a fragment of scapula blade that bears gnawing
marks, and the third phalanx, all of them attributable to a ju-
venile animal. The dog is represented by two rib fragments,
a proximal ulna with gnawing pits at the olecranon, a prox-
imal metacarpal fragment, and the first phalanx. Two bones
of a neonate, a humerus and a radius, belong to a puppy. Also,
there are a carpal bone (intermedial) of a roe deer, a marten’s
lower jaw, and a beaver’s second phalanx. Two fragmented
artefacts were also found, one made on a fragment of a large
mammal long bone, the other on a medium size mammal’s
long bone.
The assemblage in house 35 is heterogeneous, i.e. it con-
sists of species whose body parts were utilized in different
ways by humans. Yet, some remains, such as the neonate dog
humerus and radius, and the bones of possible juvenile pig,
122
may originate from the same animal and would imply that
their disposal was contemporaneous. It should be noted that
a burial was discovered at this location (Radovanović 1996:
185; Stefanović & Borić, this volume: note 6). This is burial
70, described as the secondary disposal of human bones with
a dog mandible (Radovanović 1999).
Location 8, houses 41, 38, 37 and 42
In house 38 a fragment of a carbonized aurochs scapula was
found. In house 37, which cuts one corner of house 38, a dog
femur shaft fragment came to light. In house 42 a cattle lower
molar was found. The shaft fragment of the dog femur from
house 37 bears many short, irregular cuts, probably from the
procedure of disarticulating the skeleton, while both ends are
gnawed. The cattle lower molar from house 42, the first or the
second in the molar row, is of a size not distinctive for the
aurochs/domestic cattle separation.
Location 10, houses 34, 27, 51 and 43
Analyzed animal bones originate from the earliest house in
this group of houses, house 34, and the latest, house 27.
Two lower jaws of dogs were found in house 34, a red deer
antler fragment, and a metapodial fragment of a small ru-
minant. The lower jaws of dogs are of an adult dog and a
puppy (Fig. 5). The lower jaw of the adult animal (Fig. 5a)
has all teeth present in the alveoli except for the first incisor
and the last molar. Incisors and canine crowns are broken,
 Vesna Dimitrijević: Lepenski Vir animal bones
while the tips of the first three premolars are without enamel.
Longitudinal cracks are visible on the bone surface of the
corpus of the jaw. The jaw was not in direct contact with fire;
however, it suffered from substantial heating, which is evid-
ent from the reddish colouring of the basal part of the corpus
of the jaw, with cracking of the bone and ‘peeling’ of enamel
as a consequence.
The lower jaw of the puppy (Fig. 5b) has all three milk
molars present in the alveoli. There are gnawing marks on the
vertical ramus processes, which may have been made by a
small animal. In the incisive part, the first milk incisor alve-
olus is present; also fragments of the broken second and third
milk incisors as well as milk canines are in the alveoli. The
molars are completely unworn, while behind the last milk
molar, the crypt is opened and the permanent molar germ
may be seen in the process of developing. The milk dentition
in dogs is completed by the age of 5–6 weeks, while already
at the age of 3–4 months, the crowns of milk molar teeth
show considerable wear of prominent cusps and ridges; per-
manent teeth start appearing at this time, too (Habermehl
1975). Accordingly, the mandible from house 34 is of an an-
imal less than three months old.
Another two items found in house 34 are probably frag-
ments discarded in the process of tool manufacturing. A
fragment of a red deer antler bears pit-like depressions and
short grooves scattered over its surface, and traces made by
some artefact at the place where probably a tine was cut off
from the main branch. Another fragment originates from the
metapodial of a small ruminant, possibly roe deer. The sur-
face of a proximal joint is present on this fragment with the
shaft piece of 75 mm in length. There are traces made by an
artefact on the lower part of the shaft fragment: two half-
circular incisions, and two deep longitudinal cuts above the
upper incision. Obviously the bone was split longitudinally,
with possibly a tool manufactured out of the bone, while this
surviving fragment may have been a by-product.
A beaver lower incisor was found in house 27. Its aboral
part is broken. There is a natural lingual wear-mark, but in the
continuation behind it there is a polished surface, probably
intentionally modified.
Location 11, houses 19, 31, 18 and 23
Bones come from two later, superimposed houses in this
group, houses 18 and 23. House 23 partly covers house 18 in
its rear end. The bag label indicates a more precise contextual
position for the bones found in house 18 — it says ‘below
corner A’. Four bear vertebrae were found here: a thoracic
vertebra, probably the last one in the thoracic row (Th 14),
and three lumbar vertebrae, probably the first, the second and
the fourth. These vertebrae are from the same individual and
were probably buried together. This may be attested by the
same coloration, fossilization pattern, and similar age shown
by the unfused cranial and caudal vertebral body epiphyses.
There is a single filleting mark — longitudinal and sharp, cut
in the base of the neural spine of the first lumbar vertebra.
In house 23, the first lower molar of a dog was found. It is
a tooth crown, with roots incompletely developed, indicating
that the molar was in the process of erupting; the animal was
3.5–5 months old (Habermehl 1975).
The sample collected ‘between houses 18 and 23’, as noted
123
on the bag label, consists of mammal bones, large fish bones,
fish teeth, and molluscs. Among the mammals dog remains
are the most numerous. Dogs are represented by a lower jaw,
fragments of an upper jaw, an isolated canine, an isolated
molar, an epistropheus, two left ulnae fragments, a fragment
of radius, a fragment of pelvis and two metacarpals (the
second and the fourth). The olecrani of the ulnae are gnawed,
as well as one of the metacarpals. The metacarpals are of the
same individual, while two left ulnae confirm the presence of
at least two dogs. Remains of other species include the lower
jaw of a hedgehog, a lower incisor fragment of a beaver, a
proximal tibia of a marten, gnawed phalanxes of a wild boar
(an adult first phalanx and a juvenile second), a third phalanx
and a tooth fragment of an aurochs, a partly burnt thoracic
vertebra body of a red deer (with multiple longitudinal and
diagonal cuts made by a sharp artefact), and one fragmented
artefact made on a small ruminant metapodial.
The bear vertebrae ‘below corner A’ in house 18 are per-
haps related to the event of house construction. The as-
semblage ‘between houses 18 and 23’ consists of a large
number of species and different skeletal parts, obviously dis-
carded as a consequence of different activities (meat con-
sumption, tool manufacturing, possibly skinning). However,
some bones were probably deposited in a single event (most
likely the dog bones and perhaps the wild boar phalanxes),
and these, if not the whole assemblage, might have been re-
lated to house 18.
Location 13, houses 55, 56 and 46
Deer antlers attached to the fragmented skull were placed
over the floor of house 46. This is shown on the published
photograph of the hearth area of this house (Radovanović
1996: fig. 3.33).
Location 14, houses 20, 33 and 32
Bones were collected between houses 20 and 33, as noted on
the bag label, and from the floor of house 32.
‘Between houses 20 and 33’ a red deer lower molar was
found, two left astragali of red deer, two right astragali and a
metacarpal shaft fragment of roe deer, a fragmented proximal
ulna of a dog, the artificially modified thoracic vertebra of a
wild boar (Fig. 6), as well as bird bones, many fish bones and
mollusc shells.
On the floor of house 32, remains of a dog (an upper can-
ine), red deer (lower second incisor and a proximal metatars-
al), and fish were found.
Non-superimposed houses
House 25
Long bones of small carnivores, which can be characterized
as fur-bearing animals, were collected in this house. The fol-
lowing bones are present: a femur of a badger, and two long
bones of a marten (a femur and a humerus). The femurs of
both species bear cuts made by a sharp artefact. The beaver’s
femur shows two subparallel arched cuts on the cranial side
of the bone, below the proximal articulation. The marten’s
femur bears very narrow, longitudinal incisions on the shaft.
 The Iron Gates in Prehistory
Figure 6. Wild boar modified thoracic vertebra related to house 20: a. cranial view, arrow points to a groove made in the
vertebra body; b. lateral view — the position of an imaginable cord is shown, the arrow goes through the arch foramen.
House 26'
A tine of a red deer antler that was cut off at the base and re-
mains of a dog (the first upper molar and distal femur) were
found in house 26'. Fish remains and shells were also collec-
ted, as well as many small fragments of unidentifable mam-
mal bones.
House 26
A red deer antler on the pedicle and a fragment of a calvarium
were found in ‘house 26, beside the sanctuary’, as noted on
the bag label. Brow, bez and trez tines are cut off at their
bases, while the beam was broken above the trez tine, prob-
ably during the excavation. This cutting off of the tines was
patiently performed, by gradual thinning, as indicated by
flaking in the direction of the tine, and circumferential fur-
rows above them. Also, two fragmented tines are present,
with analogous traces of cutting off from the beam. On the
skull fragment, marks made by blows designed to separate
the antler from the rest of the skull are noticeable on the an-
terior basis of the pedicle. There are also multiple, deep,
elongated cuts in the area where the pedicles of the two
antlers were closest to one another, indicating skinning.
Dog remains were also found: an epistropheus, an upper
canine, an upper first molar, a distal femur and a distal tibia,
all possibly from the same individual.
House 28
Antlers attached to the skull of a young deer were found on
the floor of house 28 (Fig. 2). The skull was damaged when
it was lifted from the position where it was laying to such an
124
extent that it was not possible to reconstruct it. The frontal,
temporal and occipital parts of the skull were present, to-
gether with the left and right upper jaws with tooth rows. The
frontal, temporal and occipital parts of the skull were present,
together with the left and right upper jaws with tooth rows,
while the bones of the face, the nasals and praemaxillars,
were crushed (Dimitrijević 2000). Fish and molluscan re-
mains were deposited in the same place, evidenced by fish
bones and a gastropod shell fragment attached to the crushed
skull bones.
The antlers are preserved for a length of over 50 centi-
metres. They are asymmetrical: the left one has a brow tine
and simple crown with two tines, while the right antler con-
sists of a single branch, with just a slight protuberance at the
site of the brow tine. Although variability in red deer antlers
is well known (Dragišić 1957), antlers from house 28 should
be characterized as unusual in their length and asymmetry.
Due to their age they are extremely thin (right burr circum-
ference =126 mm, left burr circumference = 123 mm, right
beam circumference 100 mm from the burr = 80 mm; the
same measurement in the left beam = 76 mm).
The animal’s age is clearly indicated by its dentition, which
is in the last stage of milk/permanent tooth replacement and
with the last permanent teeth erupting. Both third milk mol-
ars still rise above the crowns of permanent fourth premolars,
although half of the crown of the one on the right side of the
jaw is worn; it would have been a matter of days before they
fell out. The second and third premolars are erupting, as well
as the third molar, preserved only in the left jaw. This stage of
dental development in the upper jaw should be analogous to
 Vesna Dimitrijević: Lepenski Vir animal bones
Figure 7 Young red deer antlers from house 28, frontal
view.
Figure 8 Wild boar left astragalus, calcaneus and distal
fibula in articulation from house 40: a. dorsal view;
b. plantar view, arrows point to the breakage of the fibula
and cut marks at the distal end of the calcaneus.
125
that in the lower jaw, and related to an age of 27 months
(Brown & Chapman 1991). Since deer give birth in
May/June (Bützler 1986), this means that the animal was
hunted in the early autumn, most probably in late September
or early October (Dimitrijević 2000).
House 40
Animal bones were found on the house floor, along its longer
right wall (corners B–C). There were wild boar remains
(fragment of an atlas, final lumbar vertebra, left scapula,
fragments of right and left pelvis, right proximal and left
distal radius, left ulna, astragalus and calcaneus (Fig. 8),
distal fibula and tibia, left distal femur with unclosed epi-
physeal lines (indicating a younger animal), together with a
fragment of a sacrum and gnawed distal radius of red deer,
and an aurochs’ scapula.
The wild boar bones are not just fragments, but complete or
almost complete bones of the cranial and sacral parts of the
backbone, parts of shoulder and pelvic girdles, and mostly
the upper parts of the fore and hind legs. None of the skeletal
elements is represented by more than a single specimen,
however, the assumption was made that the remains possibly
come from two animals. This is largely based on the distal
femur with the epiphyseal line of distal articulation, clearly
observable along its whole length, which would belong to an
animal that may be aged as almost adult, but not fully grown.
In contrast, all other wild boar bones are fully grown, and of
the size and proportions corresponding to a large adult male
individual.
The tarsal bones, astragalus and calcaneus, together with
the distal fibula, remained in articulation (Fig. 8a–b). In ad-
dition, the distal tibia and fibula’s shaft were found to fit per-
fectly the astragalus and the distal fragment of the fibula, in-
dicating that those bones were a continuation of the joint
described. There are no traces of synarthrosis, which would
indicate pathological fusing, so it is certain that they were
buried in situ as articulated, and not removed from the site
afterwards, when the organic tissue would have decomposed.
The quality of the deposit in which they were embedded,
probably enriched by carbonates, made their connection firm
even after their excavation. The final lumbar vertebra, as well
as the left and right pelvic girdle fragments, were attached to
the sacrum, while two long bones of the left front leg, the ulna
and radius, were also in articulation. Another important ob-
servation regarding this part of the skeleton is that the ulna is
complete, except for some damage to its central portion,
while the upper part of the radius is broken. The breakage of
both bones was caused by a strong blow, and probably
happened while the tendons, and probably also muscular tis-
sue, were still holding the bones together. The single frag-
ment of long bones in the right front leg, the proximal radius,
points to a similar breakage, a consequence of the patterned
manipulation of the carcass parts.
Butchery marks are clearly observable on the tarsal bones,
in the form of several deep transversal grooves on the cal-
caneus (Fig. 8b), and a few short grooves on the astragalus
made in the same direction, and by the same action indicating
disarticulation of the lower hind leg. There are no cut marks
on other bones, but there are indications that other butchery
techniques apart from cutting might have been applied, such
 The Iron Gates in Prehistory
as a blow to the middle of the radius and ulna shafts. The left
and right pelvis fragments attached to the sacrum were prob-
ably broken in the course of the disarticulation of the left and
right flanks.
While the primary butchery was performed, it seems that
the butchering process was not continued to the point of
completely stripping the meat from the bones. There are no
traces of filleting, which would be very difficult if not im-
possible to perform, by avoiding the contact of sharp arte-
facts with the bone, which is necessary for this operation.
This is especially true of the shoulder blade, on which fillet-
ing leaves characteristic long, longitudinal grooves (Binford
1981: fig. 4.06).
It is suggested that large carcass parts of a wild boar were
left in the house (Dimitrijević 2000).
House 44
A diaphysis shaft of a dog humerus and a fragment of a
gnawed proximal ulna of a dog, a third molar of a wild boar,
and a fragmented lower jaw of a roe deer were collected in
house 44.
House 48
Two dog lower jaws were found in house 48, one left, and one
right, but certainly from two different individuals, since wear
traces are more advanced on one of them.
House 49
An upper jaw of a roe deer and a red deer vertebra fragment
originate from house 48. Fish remains were also present.
House 54
A dog’s lower carnassial tooth and many fish remains (mostly
teeth) and molluscs were collected in house 54.
House 61
A deer’s skull with antlers is seen on the published photo-
graph of this house (Srejović & Babović 1983: 151, lower).
The skull is damaged, and antlers are partly behind a sculp-
ture and stones, so their morphology is not very clear, al-
though it is possible to say that they belonged to an adult
animal.
House 62
A radius and tibia shaft fragments of a dog, both with traces
of gnawing, come from house 62.
House 65
An antler fragment and a splinter of a metacarpal shaft of a
red deer, as well as a partly carbonized horn-core of a
chamois were found in house 65, together with fish teeth and
a sea snail.
House 69
According to the bag labels, a second metatarsal of a wild
boar was found behind the hearth, while fish remains and
mollusc shells come from the hearth of house 69.
House 70
In the hearth of house 70, mammal and fish bones were
126
found. The following mammal remains were found: an atlas
fragment of a dog, and several remains of red deer and
chamois. A fragmented epistropheus, a distal unfused epi-
physis of a radius, and a second phalanx are of red deer, and
chamois is represented by a tip fragment of a horn-core with
traces of manufacturing or use, and a complete second
phalanx. A fragment of a distal metapodial of a ruminant
could also belong to chamois.
House 71
Fish remains and a single bird bone were recovered from
house 71.
Interpretation of animal bone remains in the
houses of Lepenski Vir
Lepenski Vir houses by their size, solid floor construction,
and elaborately built hearth constructions might have been
multifunctional dwellings, where many activities were per-
formed regularly, being highly organized. Food preparation
and consumption, manufacturing and use of bone tools are
among those regular activities we would hope to find as
evidence among the studied animal remains collected in the
houses. Stone sculptures, altars, A-signs, partial burials and
other reflections of the highly developed symbolism of the
Lepenski Vir culture indicate that apart from practical activ-
ities, those related to complex social, symbolic and cult
practice might have taken place within the houses. Animal
remains may show special ‘status’ of a particular species, and
provide a material proof of an offering deposited in the house.
The question remains, whether the archaeological record
will reflect these activities, and how faithfully? For example,
if a highly organized life was carried out in the house, piles
of discarded animal bones left after the meal on the floor
would be a less plausible outcome then regular cleaning of
food debris. Of course, when a particular house was finally
left, the remains of the last meal could have been cleaned off,
or not, depending on the circumstances of departure or com-
mon habits of the dwellers.
It is important to stress that, when we are dealing with finds
that are certainly related to floors, we are looking upon the
end of the house’s life, i.e. at the time of its abandonment.
Intentional or forced, sudden or prepared departure from the
house may be reflected in the animal bones left on a house
floor.
The composition of the animal species found in the houses,
as well as the distribution of skeletal parts, may indicate some
of these activities, too, if different from those unrelated to
houses. In the following sections I shall discuss the pattern of
the appearance of dog bones and other evidence about the
presence of dogs in the houses, evidence of food preparation
and consumption, skinning procedure, use and manufactur-
ing of bone tools, and meaningful deposition of red deer
antlers within the houses of Lepenski Vir.
?Dogs in houses
Describing the faunal remains from Lepenski Vir, Bökönyi
(1969) points to a high frequency of dog remains (4.95% in
Lepenski Vir I, 11.6% in Lepenski Vir II, and 5.91% in
 Vesna Dimitrijević: Lepenski Vir animal bones
Lepenski Vir III) and their undoubtedly domesticated char-
acter. Furthermore, he notes that dog bones are mostly un-
broken and from adult animals. His observations that in the
houses of Lepenski Vir I and II ‘undisturbed skeletons of
body parts in anatomical arrangement are very often found’,
and that ‘dog bones are often found together with red deer
skull and scapula as well as wild boar bones’ are of special
importance here. He does not deny the possibility that dogs
were eaten, but supposes that their main role was as hunting
companions and watch-dogs, and that dog was also used as a
sacrificial animal (Bökönyi 1969).
A special human attitude toward dogs is also revealed by
human burials associated with dog remains. A grave with a
secondary disposal of human bones and a dog mandible is
described by Radovanović (1999). In her study of dog burials
and human burials associated with dog remains in the Iron
Gates Mesolithic, Radovanović supposes that a change oc-
curred at the second half of the VII millennium BC, in a
period coinciding with the first contact with Early Neolithic
groups, when a symbolic meaning was associated with dogs,
reflected in human burials with dog remains. At the same
time, throughout the Iron Gates Mesolithic ‘dog remains
scattered around the settlement show not only that they were
eaten, but also that they were not paid any special respect
(since their bones were left to be scavenged)’ (Radovanović
1999).
It is questionable how deeply we can understand the com-
plexity of human/dog relationship on the basis of the archae-
ological record in general, and especially in the light of ‘what
was left in the houses’ of Lepenski Vir. However, osteolo-
gical remains indicate that dogs were accorded special treat-
ment.
The abundance of dog remains among the bones collected
in the houses of Lepenski Vir is striking. Only red deer re-
mains are slightly more numerous in houses (Table 1). Often,
remains of more than a single dog were found (houses 35, 34,
23, 48, possibly house 9), and often different parts of a skel-
eton that are certainly (houses 35 and 23) or most probably
from the same individual (houses 9 and 26). This is in agree-
ment with Bökönyi’s note of ‘undisturbed skeletons of body
parts’.
In four houses, lower jaws of dogs were found, which
could be of a particular importance in light of their deposition
as grave goods in two burials at Vlasac and burial 70 at
Lepenski Vir (Radovanović 1999). There is a possibility that
this skeletal part was chosen to represent the presence or the
essence of the animal itself. Two mandibles were found in
house 34, and in house 48, originating from two different
animals respectively. One of those found in house 34 is of a
puppy, while the bones of a newborn were found in house 35.
That dogs were eaten is also highly probable. Bones with
butchering marks were collected in houses 9 (Fig. 2) and 37.
Cut marks on a dog vertebra from house 9 derive from the
removal of the tenderloin, which is regularly found on the
butchered vertebrae of meat-bearing animals (Binford 1981).
Many dog bones found in the houses are gnawed.
Although dog is not the sole animal that gnaws bones (some
rodents do, as well as pigs and humans), this habit is by far
most characteristic for canids. The size of the gnawing marks
on the bones from houses of Lepenski Vir — pitting, grooves
127
and compressions made by canines, as well as the pattern of
the epiphyses being gnawed in long bones strongly suggest
that they are made by dogs. Gnawed red deer and wild boar
bones are found in houses 35, 23, and 40. Dog bones gnawed
by dogs are found in houses 35, 37, 34, 23, 44, and 62. A dog
femur shaft fragment from house 37 with both ends gnawed,
bears many short, irregular cuts indicating disarticulating,
which means that only after butchering the animal did a dog
get his share.
There is something intriguing about the gnawed bones in
the houses of Lepenski Vir. When we find animal bones
within a house area, this does not mean that this animal was
ever in the space of the house alive. However, the gnawed
bones may suggest the opposite. Since there is really no
reason for humans to gather gnawed bone refuse and bring it
into a house, it would be reasonable to suppose that gnawed
bones were brought into the houses by dogs!
Parts of animal carcasses and remains of food consumption
Generally, bones that bear good portions of meat, and ori-
ginate from species that are of primary importance for meat
consumption, may be presumed to have been largely dis-
posed of as food remains. This presumption is further
strengthened in those instances when bones are grouped,
found in a meaningful context, and supported by the presence
of butchery and filleting marks on bones.
It is much more difficult to find proof of the disposal of
parts of animal carcasses. Yet, it seems that this is the case
with remains of wild boar in house 40, where evidence for
primary butchery is present on the bones of lower extremit-
ies, related to dismemberment, but further operations neces-
sary for separating individual bones and removing meat from
the bones were not carried out. On this evidence, it may be
concluded that carcass parts were left in the house, either as
raw meat, or dried, or prepared for drying (Dimitrijević
2000).
Another example of bones that bear large portions of meat
comes from house 18. However, this time, signs of stripping
meat from the bones are present. The carcass part comprised
a thoracic-lumbar part of the backbone of a brown bear. A
filleting cut mark was made on a single vertebra. Particularly
intriguing is the exact provenance of this skeletal part ‘below
corner A’ of the house, which signifies different reasons for its
deposition from the offerings on the floors and within houses.
This last instance possibly relates to the building of the house.
An indication of the ‘offerings’ left in the houses can be
found in Bökönyi’s note (1969) about seven scapulae of red
deer found in the houses and ‘often together with wild boar
bones and dog remains in anatomical position’. In the ana-
lyzed assemblages unbroken red deer scapulae were not
found, while a single red deer scapula was found in house 47,
together with the same skeletal part of a wild boar and a roe
deer (Fig. 4).
The scapula is among those parts of the skeleton that bear
the most desirable meat portions. When the bone is preserved
well enough, filleting marks, in the form of longitudinal cuts
made by a sharp artefact, may confirm the stripping off of
meat. Also, the fragmentation of a scapula discarded after
meat consumption is rather characteristic: the distal, solid
bone part may remain undamaged, while the rather thin blade
 The Iron Gates in Prehistory
is regularly broken. In this respect, it is safe to conclude that
the three scapulae from house 47 are remains of meat con-
sumption, which took place in the house and were probably
connected with a single event. The same degree of carboniz-
ation probably indicates that they were burnt at the same
time. On the other hand, the carbonization itself need not be
related to meat preparation; the bones might have been
thrown in the fire after the meat was consumed, or as a con-
sequence of house burning. The context where these were
found, in the west (i.e. narrow) side of the house, does not
relate them to the hearth.
A carbonized distal scapula from an aurochs was found in
house 38, and another was used as raw material in house 40.
Apart from scapulae found in house 47, bones that origin-
ate from carcass parts, either those bones that carry good
portions of meat or confirmed as butchered for consumption,
were found in house 13 (red deer and wild boar), house 30
(red deer) and house 35 (red deer and wild boar).
Skinning evidence
Skinning is recognizable only by cut marks on the bones. The
cut marks are not morphologically different from those made
for achieving other goals, such as butchering. Yet, their posi-
tion on certain bones is patterned (Binford 1981; Noe-
Nygaard 1995). The presence of particular species, tradition-
ally valued for their fur, is also indicative.
Clear evidence of skinning is found on the red deer skull
fragment with antlers in house 26. On the anterior basis of the
pedicle, impact marks of blows are found, directed to separ-
ate the antler from the rest of the skull, while multiple deep
elongated cuts stretch in the area where the pedicles of the
two antlers were closest to each other. A similar skinning
technique for deer has been recognized in the archaeological
and ethnoarchaeological records (Binford 1981; Becker
1986: Abb. 41–43; Noe-Nygaard 1995: 183, 188).
The presence of fur-bearing animals characterizes the as-
semblage of house 25. Although these species were probably
eaten, too, their prime economic importance ought to be in
connection with fur acquisition in particular. There is a single
bone from a badger, a femur. Two sub-parallel arched cuts
have been made on its cranial side below the proximal artic-
ulation. Of two bones of a marten, a femur and a humerus,
only the femur bears cut marks. They are in the form of very
narrow, longitudinal incisions on the shaft. Remains of
badger and marten are considered as being hunted mainly for
their fur at Vlasac and Padina, too, on the basis of the distri-
bution of skeletal parts and skinning marks (Bökönyi 1978;
Clason 1980).
Bones of marten were also found in houses 23 and 35, in
both cases associated with beaver remains, which is regarded
as another important fur-bearing animal.
Artefacts, by-products and bone and antler raw material
We may expect to find artefacts in houses, since their use
could be related to the life of a house, and houses are appro-
priate places where they could have been kept, or left and
forgotten. In the assemblages described it was mostly frag-
mented artefacts and parts of bones rejected as useless for
tool manufacturing that were found, since recognized arte-
facts were treated and packed separately during the
128
excavations.
Evidence of tool manufacturing comes from house 26 — a
red deer antler on the skull fragment, had all its tines cut off
at the base. The tines were obviously successfully separated
from the beam to be used probably as massive points. The
work had been done laboriously, possibly in order to be con-
tinued at the antler base and beam, which still contains parts
that can be used for tool manufacturing. Another example is
an aurochs scapula from house 40. It is cut off transversely
approximately on the distal third of its length. On the central
part of the blade, where the bone is the flattest, a deep half-
circular cut is made in order to get material, probably for a
very specific purpose, perhaps for something like an amulet.
Yet, the most interesting item relates to house 20. It is a
thoracic vertebra of a wild boar, with an artificial groove on
the body of the vertebra (Fig. 6a). The groove is inclined at an
angle so as to enable some sort of rod to be inserted and to
lean, with its upper side, on the neural spine (Fig. 6b), where
a similar, but shallower groove was made. There is a con-
spicuous polish on the neural spine at the place where an
imagined rod would have been leaned on. The polish is also
observable on the entire lateral surfaces of the spine, as well
as around arch foramina (openings on transverse processes).
These openings are probably the reason that the thoracic
vertebra of wild boar was chosen, since they are found only
in suids. They are positioned in the bases of transverse pro-
cesses and connect dorsal and ventral sides of the processes
with the vertebral channel. Via those openings a rod or a cord
might have been pulled vertically through the transverse
processes (Fig. 6b), while a cord (but not a rod, because of a
slightly different height in their position) might have been
pulled transversely through the vertebra. The bone around the
openings is also slightly polished, and the openings them-
selves somewhat widened. It is possible to imagine several
possible ways in which rods and cords were pulled through
those openings. It seems that this modified vertebra was a
part of a complex device with a special function. A wild boar
thoracic vertebra modified in almost exactly the same manner
has been recognized by the author in the assemblage from
Padina. Modified vertebrae were also found at Vlasac, where
they were interpreted as cult instruments (Srejović & Letica
1978, I: 106, pl. LXXI, CXXI no. 6, CXXIII no. 6, CXXIV
no. 1). The specimens from Vlasac are different from those of
Lepenski Vir and Padina in that their transversal processes
have been removed, and the neural spine ornamented.
Fragmented artefacts or fragments of bones with traces of
artificial modifications were also found in houses 23, 26', 27,
34, 35, and 70. In several houses (9, 13, 20, 65 and 70) bones
of skeleton parts traditionally used in tool manufacturing
were found, which could be regarded either as remains of
possible raw material, or as material rejected from such a
skeletal part when a desirable part was taken. These are
antlers and ruminant metapodials in the first case, and most
often carpals and tarsals, as well as phalanges in the second.
Antlers in houses
Not all the antlers and antler fragments found in the houses at
Lepenski Vir are taken into account here, but only those of
unquestionably symbolic significance.
Red deer antlers placed on the floors of houses are one of
 Vesna Dimitrijević: Lepenski Vir animal bones
the specific features of Lepenski Vir, whose significance must
be taken into consideration when trying to understand the
symbolism of the Lepenski Vir culture, together with its
unique architecture, stone sculptures, burial rites and other
material evidence. In the early phase of research on the
Lepenski Vir culture, Srejović (1969: 137) emphasized that
red deer antlers possess unique features of growing and being
rejected in a yearly cycle, as well as their amazing ability to
grow bigger, stronger and more beautiful every year and in
this way symbolising an existing connection between death
and renewed birth. Whether or not we relate their deposition
in the houses with the expression of the belief that the house
will live again, their presence in the houses of Lepenski Vir
does represent unique evidence of the house abandonment
event. It shows that this event was planned, prepared and
symbolically justified. In addition, animal bones and antlers
offer a possibility to find out about the season of the aban-
donment event.
Red deer skulls with antlers have been found in at least 13
houses at Lepenski Vir (Bökönyi 1969). Not all of them were
saved. In the collection at my disposal, pairs of antlers from
houses 28 and 22 are examined; on the published photo-
graphs two additional pairs of antlers on the skull can be de-
tected in house 22 (Srejović 1969: fig.69; Radovanović 1996:
fig. 4.3; Srejović & Babović 1983: 136, lower), and house 61
(Srejović & Babović 1983: 151, lower). Antlers related to
burials or below the floors are not taken into consideration
here (for example, those from house 61, Srejović 1969: fig.
71).
The antlers attached to the skull found on the floor of house
28 are of a young animal. They are very thin, with small brow
tine and crown on the right, and without tines on the left beam
(Fig. 7). The age is estimated from the dentition as 27
months, and accordingly, the time of the kill to the early au-
tumn (Dimitrijević 2000). We believe this is determining the
season of the house abandonment, too.
One of the pairs of antlers from house 22 (Srejović 1969:
fig. 69; Radovanović 1996: fig. 4.3) has a similar appearance,
and possibly a similar age and season of kill. Another antler
belonged to a prime adult (Srejović & Babović 1983: 136,
lower) (Fig. 3).
Shedding of red deer antlers is related to a restricted period
at the end of winter (February/March), while fully grown
antlers on the skull cover the season from the end of August
to the time of shedding in February/March. Still, the easiest
time to hunt is thought to be the rutting season
(September/October), when the deer population is gathered
within limited areas and the males are in excellent feeding
condition (Noe-Nygaard 1995). This is also the time of kill
already established for the deer skull with antlers from house
28, and possible time of kill for the skull with antlers in the
rear of the house 22.
Conclusion
It is necessary to begin these concluding remarks by repeat-
ing a sentence from the introductory chapter: “not a single
contextual unit of collected animal bones from Lepenski Vir
has complete recovery of the remains”. This paper thus con-
129
siders fragments of already fragmented archaeozoological
remains, while the overall degree of the fragmentation re-
mains unknown. Yet, the collection of animal bones related to
the houses of Lepenski Vir, offers some indications about
human/animal relationships and related activities. Moreover,
it offers some valuable reflections on particular events in the
life of these houses.
Probably the most important are those reflections to be
made in relation to the events of house abandonment. It is
attested that the abandonment of a particular house was a
prepared action pregnant with symbolism. Undoubtedly, red
deer skulls with antlers were placed in houses at the moment
of their inhabitants’ departure, most probably as symbols of a
belief in their return and renewal of the life of a particular
house. The pattern of this symbolic act may be similar in the
majority of the 13 recorded cases, but might have been dif-
ferent in the case of house 22, where more than one item was
deposited.
The abandonment of some houses was probably marked
symbolically in another way. Part of a wild boar carcass in
house 40, as well as dog mandibles, if not their whole bodies,
were possibly left in the houses as acts of abandonment, too.
It is certain that the repeated deposition of fish remains in
graves and houses of Lepenski Vir needs also to be taken into
account. However, this subject will receive more attention
when analyses of the fish remains have been completed. The
same is true for bird bones and invertebrate remains, which
could in addition offer supplementary data on the seasonality
of occupation.
Determination of seasonality on the basis of animal re-
mains is the only reliable source of detecting possible sea-
sonal abandonment of particular houses and the settlement of
Lepenski Vir as a whole. The early autumn date of the aban-
donment is confirmed in house 28, indicated in house 22, and
possible in houses 46 and 61. Specific methods of ageing an-
imal remains and identification of seasonality will show
whether this pattern is of general significance for the whole
site. Research in this direction is in progress (Borić 2002).
Some of the assemblages described offer evidence of vari-
ous activities that took place within a particular house, such
as food preparation and consumption, use and manufacturing
of bone tools, and skinning of animals. Does this evidence
imply that some of the houses were not abandoned after
elaborate preparations and symbolic acts, but suddenly, un-
willingly, or without according special respect to the house
left behind? This remains unclear, mainly as a consequence
of the incompleteness of the assemblages, and the strati-
graphic uncertainties already noted.
Finally, analysis of animal bones from the houses of
Lepenski Vir confirms that some animal species were accor-
ded a special status. In the case of dog, its close relationship
to humans and connection with the house area is reflected in
the large number of dog remains, and probable deposition of
its skeletal parts or even whole bodies in the houses. It may
also be supposed that they were present alive in the houses,
too. In fact, the extreme importance of the animal world for
the occupants of Lepenski Vir is beyond doubt, and most of
the animal species mentioned here certainly received some
sort of special treatment, special status and place in the sym-
bolic system of the Lepenski Vir culture. It is for us to try to
 The Iron Gates in Prehistory
find out the details of these relations in order to enhance our
understanding of the character of the Lepenski Vir culture.
Acknowledgements
I am very grateful to Ljubinka Babović, curator in the National
Museum in Belgrade, for allowing me to analyze osteological ma-
terial from Lepenski Vir. I also thank Ivan Stefanović for his skilful
restoration of antlers and conservation of bones, Predrag Filipović
for technical improvement of photographs, and Dušan Borić for
valuable comments on the paper.
References
Becker, C. 1986: Kastanas: Ausgrabungen in einem Siedlungshügel
der Bronze- und Eisenzeit Makedoniens 1975–1979, Die
Tierknochen-funde (Prähistorische Archäologie in Südosteuropa,
Band 5). Berlin: Volker Spiess.
Binford, L. 1981: Bones: Ancient Men and Modern Myths. New
York: Academic Press.
Bökönyi, S. 1969: Kičmenjaci (prethodni izveštaj). In Srejović, D.,
Lepenski Vir, Nova praistorijska kultura u Podunavlju. Belgrade:
Srpska književna zadruga, 224–228.
— 1978: The vertebrate fauna of Vlasac. In Srejović, D. & Letica,
Z. (eds) Vlasac: A Mesolithic Settlement in the Iron Gates, vol. 2.
Belgrade: Serbian Academy of Sciences and Arts, 35–65.
Borić, D. 2002: Apotropaism and the temporality of colours: col-
ourful Mesolithic–Neolithic seasons in the Danube Gorges. In
Jones, A. & MacGregor, G. (eds) Colouring the Past: The
Significance of Colour in Archaeological Research. Oxford:
Berg, 23–43.
Brown, W. & Chapman, N. 1991: Age assessment of red deer
(Cervus elaphus): from a scoring scheme based on radiographs
of developing permanent molariform teeth. Journal of the
Zoological Society of London 225: 85–97.
Bützler, W. 1986: Cervus elaphus Linnæus, 1758 — Rothirsch. In
130
Niethammer, J. & Krapp, F. (eds), Handbuch der Säugetiere
Europas, Band 2/II. Wiesbaden: Akademische Verlags-
gesellschaft, 107–139.
Clason, A.T. 1980: Padina and Starčevo: game, fish and cattle.
Palaeohistoria 22: 141–173.
Dimitrijević, V. 2000: The Lepenski Vir fauna: bones in houses and
between houses. Documenta Praehistorica 27: 101–117.
Dragišić, P. 1957: Jelen. Zagreb: Lovačka knjiga.
Habermehl, K-H. 1975: Die Altersbestimmung bei Haus- und
Labortieren: 2, vollständig neubearbeitete Auflage. Berlin: Paul
Parey.
Noe-Nygaard, N. 1995: Ecological, Sedimentary, and Geochemical
Evolution of the Late-glacial to Postglacial Åmose Lacustrine
Basin, Denmark (Fossils and Strata, no. 37). London: Taylor &
Francis.
Quitta, H. 1975: Die Radiocarbondaten und ihre historische
Interpretation. In D. Srejović, Lepenski Vir, Eine vorgeschicht-
liche Geburtsstätte europäischer Kultur. Bergisch Gladbach:
Gustav Lübbe, 272–285.
Radovanović, I. 1996: The Iron Gates Mesolithic. Ann Arbor:
International Monographs in Prehistory.
— 1999: ‘Neither person nor beast’ — dogs in the burial practice of
the Iron Gates Mesolithic. Documenta Praehistorica 26: 71–87.
Srejović, D. 1966: Lepenski Vir — a new prehistoric culture in the
Danubian region. Archaeologia Iugoslavica 7: 13–17.
— 1969: Lepenski Vir, Nova praistorijska kultura u Podunavlju.
Belgrade: Srpska književna zadruga.
Srejović, D. & Babović, Lj. 1983: Umetnost Lepenskog Vira.
Belgrade: Jugoslavija.
Srejović, D. & Letica, Z. 1978: Vlasac: A Mesolithic Settlement in
the Iron Gates. Vol. 1, Archaeology. Belgrade: Serbian Academy
of Sciences and Arts.
Whittle, A., Bartosiewicz, L., Borić, D., Pettitt, P. & Richards, M.
2002: In the beginning: new radiocarbon dates for the Early
Neolithic in northern Serbia and south-east Hungary. Antæus 25:
63–117.
 New-born infant burials underneath house floors at
Lepenski Vir: in pursuit of contextual meanings

Sofija Stefanović & Dušan Borić

Abstract: There are 41 infant (infans — an individual under one year of age) burials underneath houses at Lepenski Vir. Although there
have been several studies of these individuals in the course of physical anthropological analyses, no special attention has been paid to the
numerous burials of children at this site. Moreover, little attention has been focused on interpreting these interments in a wider social context.
Our intention is to contextualize fully this phenomenon, primarily in connection with the stratigraphic and architectural features and paying
particular attention to the osteometric properties of the buried bodies. The physical properties are: estimation of age at death, examination
of possible pre-mortem injuries on the skeletons, a detailed evaluation of represented body parts, and a discussion about taphonomic and
methodological problems in preservation and recovery of certain body parts in this context. Also, a general discussion of the mortality pro-
files of infants is contrasted with the picture of infant burials from Lepenski Vir concluding that these 41 individuals (with a predominant
age at death of 38–40 gestational weeks) cannot represent a natural mortality profile of a human population. We analyze these burials as
socially embodied and engendered individuals. In interpreting this phenomenon different spatio-temporal scales are considered: a building
with a burial, the site of Lepenski Vir with its respective phases, the region of the Danube Gorges with its Mesolithic and Neolithic sequences,
and the wider region of the central Balkans in the course of the Neolithic. Particular attention is paid to defining a chronological framework
for the phenomenon.
Key words: burials, neonates, houses, Lepenski Vir, Mesolithic, Neolithic

Introduction significance and embodiment (e.g. Meskell 1994; Moore &

What dimensions of life at Lepenski Vir do the new-born in-
fant burials open up? Buried mainly underneath the rear of
trapezoidal buildings at Lepenski Vir, their large number and
striking spatial patterning is unique to this site among the
other regional sites with chronologically (c. 9500–5500 cal
BC) overlapping sequences in the Danube Gorges (north-
central Balkans). Infant burials have remained under-
recognized for a long time unlike those other unique and
well-known features of Lepenski Vir — abundant artwork of
sculpted boulders and limestone building floors. Some ques-
tions spring immediately to the mind. Do the uncovered in-
fant burials reflect appalling religious/ritual practices of sac-
rifice made by family groups objectified through particular
‘houses’/‘shrines’, or is this determined inclusion into do-
mestic space the expression of a specific care and of what
kind? For what reasons did only certain buildings accom-
modate the deceased new-borns and how might this have re-
lated to the subsequent life history of these architectural
spaces? How do infant burials relate to the other burials at
this and the neighbouring sites? Where can we situate this
phenomenon historically and what diachronic changes oc-
curred? What sort of meanings might have surrounded the
practice and how fixed might these have been?
We shall aim at answering here a series of questions by
looking into the contextual richness of the site and the area.
However, this journey will take us away from Lepenski Vir,
also aspiring to its relevance for other places and times, ex-
ploring ways of making sense of the web of relations we have
started constructing below.
It is only recently that a more dedicated approach has
crystallized in the study of past infant lives, their social roles,
Scott 1997; Scott 1999). Burials are the most visible and most
effectively discussed aspect of the archaeological record re-
lating to children, although other material traces may have
some potency for revealing dimensions of infants’/children’s
actions (e.g. Finlay 1997; Politis 1999b) or actions specific-
ally related to children (e.g. Politis 1999b). Within our dis-
cussion of Lepenski Vir we shall highlight a possible ex-
ample of this kind, reaching to infants not only through the
perspective of mortuary data.
Infant burials, despite physical problems related to their
survival, do not represent a rare phenomenon among numer-
ous archaeological examples. Still it may be possible to
identify reasons that might have deterred archaeologists from
engaging fully in discussing and interpreting burials of this
age group (which should rather be viewed as consisting of
several age stages, cf. Sofaer Derevenski 1997). Possible
reasons would account for a frequent low elaboration of lack
of detail for infant burials when compared to adult burials,
difficulties in attributing sex to children and subadults (up to
the age of 15–20 years) on morphological grounds, which
blurs picturing infants’ social personae clearly, and a general
perception of infants’ lesser role in the (adult) life of a past
society that is studied by adult archaeologists.
In the case reported below, we have tried to overcome
some of these problems, such as that of the determination of
sex by using the newly acquired results of DNA analysis
(PCR method) on the studied material (Čuljković et al., this
volume). Other potential obstacles in the study of infants’
roles and meanings of their burials are solved by contextual-
izing their place and significance, i.e. by looking at relational
features of the studied site and the stratigraphic contexts, and
connecting existing evidence from neighbouring and con-

131
 The Iron Gates in Prehistory
Figure 1. Lepenski Vir – houses and burials (supplemented after: Srejović 1981: 20–21, and field documentation).
temporaneous sites in the wider region.
A more complex understanding is reached by examining
how different temporal and spatial contexts of human societ-
ies, described through numerous ethnographic records, re-
spond to, but also protect from, an infant’s death. This per-
spective, on the level of general analogies, explores
emotional (e.g. a period of bereavement and a need for the
physical proximity of the corpse), social (e.g. practices of
negotiation of meanings in delaying the act of burial) and
practical dimensions (e.g. the issue of reproduction in a des-
cent group) that constitute culturally inscribed attitudes spe-
cifically related to infants, which are context sensitive in re-
sponse to existential anxiety of this kind. A number of general
similarities transparently visible across various ethnographic
and archaeological contexts emphasize the need to bring and
discuss these examples together and to compare them. While
archaeological examples, such as the infant burials of
Lepenski Vir discussed here, with the addition of those from
the neighbouring and earlier site of Vlasac, can open up dia-
chronic depths in viewing localized transformations of par-
ticular practices, ethnographic records with thick observa-
tions of social life can in turn indicate new possibilities for
understanding the main themes of material culture rhetorics
of the archaeological examples studied.
The following discussion is in three main parts. Firstly, we
provide a discussion from the perspective of physical an-
132
thropology with related issues in the study of osteological
remains of infant burials. This includes taphonomic con-
cerns, details of identification, metric procedures aimed at
precise ageing of the burials and a discussion of subsequent
construction and interpretation of mortality profiles, as well
as an inquiry into the possible cause of death. Secondly, we
examine the archaeological context of these burials and their
relation to the architecture of hearths and trapezoidal build-
ing floors and cycles of their use and abandonment, other
burials of different age groups at the site and the possible
connection with the representational artworks of sculptured
boulders. By drawing attention to the stratigraphic and chro-
nological contexts of the infant burials at Lepenski Vir, we
touch upon the problem of a historically contested signific-
ance of the phenomenon. Lastly, possible meanings of the
phenomenon are sought through both a diachronic regional
perspective and across a wider anthropological background,
drawing upon local Balkan ethnographies.
Physical anthropology of infant burials
The meaning of the Latin word infantia is ‘unable to speak’,
and until recently infants remained quite silent in the field of
physical anthropology. It remains an open question at what
age infant becomes child, and definitions of these age stages
 Sofija Stefanović & Dušan Borić: New-born infant burials at Lepenski Vir

Table 1. Infant burials/houses at Lepenski Vir, their positions and orientations, and available long bone measurements
used for providing estimated age in gestational weeks.

House no. Burial no. Position Orientation Femur Humerus Age (gestational
length (mm) length (mm) weeks) & Sex
(PCR DNA)
1 3 112 ‡ ‡ – 65 ♂ 38–40
2 4′ 109 Back/flexed legs SE–NW 72 63 ♂ 38–40
3 109a ‡ ‡ 82 – ♀ 41–43
4 4 106 Back/legs splayed outwards c. SW–NE 81 74 ♂ 44–46
5 106(1) One bone only – – ? ? 41–43
6 107 Back (legs missing) N–S – 67 ♂ 38–40
7 108 Back/flexed legs N–S – 67 ? 38–40
8 110 ‡ ‡ – 70 ♀ 41–43
9 13 116 ‡ ‡ 62 (fibula) ♀ 38–40
10 19 98 Flexed/left side c. NW–SE – 61 ♀ 35–37
11 98a ‡ ‡ – 62 ♂ 35–37
12 103 Back/contracted legs S–N 75 65 ♂ 38–40
13 24 94 Flexed/left side NE–SW 79 66 ♂ 38–40
14 95 ‡ ‡ 79 – ♀38–40
15 101 Flexed/right side NW–SE 79 71 ? 38–40
16 102 Flexed/left side SE–NW 79 67 ♀ 38–40
17 26 63 Extended/slightly bent legs SE–NW 83 70 ♀41–43
18 27 128 ‡ ‡ 63 (tibia) ♀ 35–37
19 129 ‡ 78 68 ♂ 38–40
20 130 ‡ – these three burials were placed one over the other ‡ 74 66 ♂ 38–40
21 131 ‡ 75 65 ♂ 38–40
22 29 119 ‡ ‡ 79 – ♂ 38–40
23 120 ‡ – one placed on the top of the other SE–NW 76 65 ♂ 38–40
24 121 – 62 ♂ 35–37
25 36 114 Back/legs splayed outwards S–N 72 64 ♂ 38–40
26 115 Back/legs splayed outwards S–N 76 65 ? 38–40
27 37 132 ‡ ‡ – 65 ♂ 38–40
28 133 ‡ ‡ 75 67 ♂ 38–40
29 38 111 Back/legs splayed outwards S–N 76 – ♀ 38–40
30 111(1) One bone only – ≈70 – ? ≈35–37
31 43 96 ‡ c. SE–NW 78 69 ♂ 38–40
32 47 123 Contracted/left side SE–NW 77 – ♀ 38–40
33 124 Extended on its back S–N 74 64 ♂ 38–40
34 54 125 ‡ ‡ 74 65 ♂ 38–40
35 127 ‡ ‡ – 69 ♀ 38–40
36 62 118 ‡ ‡ ♀ 35–37
37 63′ 113 Back/legs splayed outwards NW–SE 77 66 ♀ 38–40
38 63 117 ‡ ‡ 80 – ♀ 38–40
39 48 134 ‡ ‡ Not preserved
40 See note 4 83b ‡ ‡ – 82 ? 48–50
41 See note 4 83b(1) ‡ ‡ – ≈ 65 ?≈38–40

‡ — position or orientation unclear

133
 The Iron Gates in Prehistory
vary across cultural contexts (Scott 1999: 2). Here, we use
one of the most common definitions, that an infant is an in-
dividual under the age of one year (Saunders 1992).
Most of the infant skeletons from Lepenski Vir discussed
here belong to the category neonata, new-born infants, i.e.
individuals who died during birth or soon after. We may ex-
pect some sort of special treatment in those instances when
the number of neonates is high in the archaeological record.
Although infant skeletons from Lepenski Vir were ex-
amined first by Zsuzsanna Zoffmann (1983) and sub-
sequently by Mirjana Roskandić (1999), few details are given
in those accounts. Zoffmann (1983) focused on the demo-
graphic facts, making the generalization that the “great num-
ber of newborn skeletons reflects the well-known high infant
mortality in earlier times” (Zoffmann 1983: 131). A more
detailed account was intended for the final publication on
Lepenski Vir that has never appeared (Zs. Zoffmann, pers.
comm.). Research on the Lepenski Vir human osteological
material by Mirjana Roksandic (1999) led to a revision of the
number of infant skeletons present in the assemblage whilst
concentrating on other aspects of the collection.
In this paper we offer a more detailed osteobiography of
infants at Lepenski Vir through taphonomic analysis, more
accurate estimation of age at death, new DNA findings on sex
attribution (cf. Čuljković et al., this volume), plotting, com-
paring and interpreting their mortality rate, and indicating
possible causes of death.
Taphonomy
To understand taphonomic processes and the extent of their
severance in relation to osteological remains may help in de-
termining sample bias (Eldredge 1991). For instance, the low
mineralization of bones of infant skeletons may lead to poor
preservation in the archaeological record. It is the relation-
ship between mineral and mechanical properties of bones
that explains the brittle nature of the bones of younger chil-
dren (Guy et al. 1997) — their tensile strength is fairly low
while their compressive strength and hardness are extremely
low (Currey & Butler 1975). Under some pressure these
skeletons would rarely escape being crushed in the ground. In
addition, even at marginal soil pH ranges it may be expected
that all or most of the infants would be eliminated systemat-
ically from the burial sample (Gordon & Buikstra 1981).
These generally pessimistic remarks in relation to the pre-
servation of infant skeletons do not apply to the infants dis-
covered at Lepenski Vir. Bones of almost all infants excav-
ated at Lepenski Vir are well preserved with a fairly high
percentage of bones present in individual skeletons and with
generally good preservation of single bones (even in the case
of fragmentary skeletons single bones are in a very good
state). Exceptions to these general rules are the bones of
Burials 98a and 119, which are poorly preserved (Figs 13 &
14). Cranial bones are frequently crushed, related to the fact
that these are not fused in the skull and are highly fragile.
Also, hand and foot bones are very rare most likely due to
hand-collecting and lack of sieving during excavation.
As Zoffmann’s (1983) and Roksandic’s (1999) estimates
of the number of infants present at Lepenski Vir differ, we
need to resolve their exact number on the basis of the pre-
served material. According to Zoffmann (1983), there are 80
134
individuals buried at Lepenski Vir and “36 individuals belong
to Inf.–Inf. I. age groups” (Zoffmann 1983: 131). Without
indicating a precise number of new-born infant skeletons,
Roksandic (1999: 78) notes the presence of “190 individuals
from 134 graves plus 42 individuals from unidentified con-
texts. Among them, there are 51 infant skeletons, 27% of all
buried individuals”. We re-examined only the skeletons of
neonates and came up with 41 individuals at Lepenski Vir,
accounting for over 20% of the entire burial assemblage (the
exact percentage cannot be given before all the skeletons
from the site are re-examined).
As Zoffmann (1983) does not list individual burials, the
discrepancy between our results and the two previous ac-
counts can be discussed only in relation to Roksandic’s
(1999: 78–80) findings. The difference in the number of
neonates present relates to those buried under Houses 4, 4′,
24, 35 and 47 at Lepenski Vir. The differences come from the
estimations made in the course of the osteological analyses
when in a few cases individual bones of several individuals
(often represented by only one ‘extra’ bone) are identified
within the material of a single archaeological burial and sub-
sequently marked as separate individuals (the identifications
were made by M. Roksandic). However, as burials of
neonates are found underneath building floors often close to
one another, excavated at the same time and re-packed and
analyzed several times we must account for all these potential
biases. Thus, we concentrated on discerning how these ‘ex-
tra’ bones/individuals might relate not only to the bones of
the burial within which they were found, but also to the bones
of other burials underneath the same house. These instances
are as follows:
Houses 4 and 4′ (Burials 106, 107, 109, 109a, 108, 110)
(see Fig. 7). Among the archaeologically identified burials of
this building, the following individuals were added: 106(1)
— one right humerus, 108(1) — one right femur, 110(1) —
one right scapula. The additionally marked Burial 108(1),
found among the bones of Burial 108, may belong to either
Burial 107 or 110 — both lacking the right femur. Similarly,
the right scapula marked as Burial 110(1) may belong to
either Burial 109 or 109a, both lacking the right scapula.
However, it is not clear how to explain one extra right hu-
merus in Burial 106, marked as Burial 106(1), as this burial
was found on the other side of House 4 (corner C). This may
be a sorting mistake; however, we have to account for this
individual (see Table 1).
House 24 (Burials 94, 95, 101, 102) (see Fig. 16). Two
more burials were separated: 101(2) — one right tibia and 4
ribs, and 102(1) — one pelvis. However, as there is no right
tibia among the bones of Burial 102, the ‘extra’ right tibia in
Burial 101 may well belong to Burial 102, buried close to one
another (although initial sorting and subsequent re-packing
and analyses must be accounted for again). The burial
marked as 101(1) is represented only by a rib of an adult in-
dividual. The pelvis fragment (ischium) marked as 102(1)
was rightly separated from Burial 102, but could belong to
Burial 101.
House 36 (Burials 114 and 115) (see Fig 9). Burial 114(1)
— one right ulna clearly belongs to Burial 115. These two
burials are buried one on top of the other.
House 38 (Burial 111) (see Fig. 5). The burial marked as
 Sofija Stefanović & Dušan Borić: New-born infant burials at Lepenski Vir
111(1) is a left femur and since Burial 111 has both femora
and is the only burial in House 38, this is clearly an ‘extra’
bone whose inclusion can be explained by some of the factors
already mentioned. However, we shall treat it as a separate
individual (see table 1).
House 47 (Burials 123 and 124) (see Fig. 4). Two more
individuals were separated: Burials 123(1) — a fragment of a
right femur and 124(1) — a fragment of a humerus and a
fragment of a left ulna. A fragment of a right femur marked
as Burial 123(1) most likely belongs to Burial 124 where
there is no right femur. Also, a fragment of humerus marked
as 124(1) refits with the humerus of Burial 124. A fragment of
a left ulna most likely belongs to Burial 123, which is missing
both ulnae. Although Burials 123 and 124 are not buried in
the immediate vicinity of one another, some of the factors
discussed may have caused the evident mixing.
Having in mind these corrections, the number of new-born
infants present in the assemblage is 40, to which may be ad-
ded infant burial 134 (House 48) which was noted and de-
scribed in the field documentation (see below) but was not
physically preserved after the excavation (noted in Field
Journal, 5/10/1970). Although there remains a possibility that
some of the ‘extra’ bones described do represent separate in-
dividuals, disturbed through activities upon the floors of the
Lepenski Vir’s buildings, the data currently available indicate
41 burials of neonates at this site. We see no reason to suggest
a different scenario to explain these instances (e.g. the pos-
sibility that the extra bones were incorporated into the burials
intentionally).
Age estimation
Bone size is a good indicator of age in perinatal skeletons.
Bone growth in the foetus may be influenced by extrinsic
factors, such as poor maternal nutrition, but these factors may
influence bone growth less after birth (Mays 1998). Maternal
malnutrition has to be quite severe to retard growth in the
developing foetus as in this situation the foetus is protected at
the expense of the mother (Bagchi & Bose 1962).
Scheuer et al. (1980) indicated the relationship between
length of long bones and age, using foetal material from
medical collections in England. Measuring the length of limb
bones, they came up with a relationship between age at death
and long bone length in foetal remains, thus giving an estim-
ated age at death within about two weeks (Mays 1998). Some
additional information on the growth of long bones in infants
and children has been gathered from the study of the Indian
Knoll skeletal sample by Johnston (1962) and a consideration
of the problem of estimating age at death only from a dia-
physis by Stewart (1968). It has been accepted that the gest-
ational ages of the foetal skeletons ranged from 27–46 weeks,
where an average gestational age at birth is 38–41 weeks
(Tanner 1989). This contrasts with Scott’s (1999: 67) scepti-
cism that it remains difficult “to determine the precise gesta-
tional age of infant skeletal material from antiquity”.
Estimates of the gestational age of infants from Lepenski
Vir were made according to the maximum length of
both/either femora and/or humeri (Bass 1987; Mays 1998).
Only in the cases of infant skeletons 116 (House 13) and 128
(House 27; Fig. 12), where femora and humeri are not
present, was the age estimated according to the maximum
135
length of either fibula (Burial 116) or tibia (Burial 128),
which were then compared to the length of these bones re-
spectively in other skeletons where estimations were based
upon femora and humeri. On the basis of these metrical
comparisons, the greatest number of those present indicate an
age of 38–40 gestational weeks (± two weeks), i.e. an average
gestational age of an infant at birth, indicating neonates.
Generally the mandibles of neonates are not well preserved in
the material studied and it was not possible to infer age on the
basis of dental development. There still remains the possibil-
ity of employing X-ray analyses on some of the well-
preserved mandibles of this age group.
In addition, we examined two other individuals (children)
buried within buildings — Burials 92 (House 28; Fig. 18) and
97 (House 31; Figs 19 & 20). As Burial 92 is well preserved,
it was possible to infer its age on the basis of dental develop-
ment (Ubelaker 1978: fig. 62) and the age is around 2 years
and 8 months. Burial 97 indicates age around 2–6 years on
the basis of the postcranial skeleton, as the mandible is not
well preserved.
Sex determination
The reliability of morphometric analyses in sex identification
of infants is low, and although pelvic sex differences are
present in infants at birth, they become obscured soon after
birth (Mays 1998). As a reliable substitute, DNA-based sex
identifications of 30 sampled infant burials from Lepenski
Vir were conducted. Of the 30 samples tested, 29 provided
results indicating 16 males and 13 females (see Table 1;
Čuljković et al., this volume). For further discussion on the
sex identifications and their relations to the spatial patterning
at the site see the discussion below.
Mortality
It is commonplace in discussions of infant mortality to em-
phasize high infant mortality in the past, and especially in
earlier prehistory. Similar generalizations indicate that “high
infant mortality in earlier and modern developing societies
contrasts with lower infant mortality in modern Westernized
societies” (Malhorta 1990: 315). Obviously, we can hardly
estimate the real number of infant deaths within the first year
of their lives in a population on the basis of the archaeolo-
gical sample. Also, a burial sample may be biased as in a
number of societies infants may be excluded from cemetery
burials (Mays 1998), with their bones usually being under-
represented in the studied assemblages (Roberts &
Manchester 1995). High chronological resolution of excav-
ated burials would be desirable before any attempt is made at
forming a synchronic picture of population demography.
Therefore, any attempt to calculate infant mortality rate for
an archaeological cemetery/burial site is a risky procedure, as
this hardly reflects a true picture of a past population, bring-
ing under scrutiny the relevance of sophisticated palaeo-
demographic statistics (contra Roksandic 1999).
However, we may try to offer a qualitative, comparative
and fairly general assessment of the high percentage of infant
burials at Lepenski Vir. For instance, in archaeological
cemeteries the proportion of infants often fluctuates around
5–6% of the whole assemblage (Guy et al. 1997: 221). We
have already indicated that the presence of infants at
 The Iron Gates in Prehistory
Lepenski Vir is around 20% out of the total number of burials
uncovered (bearing in mind the considerable time span prob-
ably represented by these burials, see below). It seems that
more new-born infants were chosen to be buried at Lepenski
Vir in comparison to all other age groups. But it can hardly
prove a higher mortality rate of neonates at Lepenski Vir over
a particular period of time. And as it is “hard to imagine that
the scarcity of infant remains in our cemeteries can be a true
reflection of a demographic fact” (Guy et al. 1997), we find
it hard to imagine that a large percentage of neonates at
Lepenski Vir is a true reflection of a demographic dictum.
Although generalizations about high infant mortality in
prehistory due to diseases and accidents (Scott 1999) may
hold some truth, with considerable fluctuations over historic
periods, we have to remind ourselves that the reasons for an
evident bias in the presence of any age group in the archae-
ological burial record are primarily functions of particular
and context-sensitive cultural practices.
Cause of death
Perinatal deaths indicate foetal deaths plus deaths of infants
who die within the first 28 days of life, while neonatal deaths
occur after 28 days of life but within the first year. Today,
some major factors associated with infant deaths include:
congenital malformations, perinatal conditions and infec-
tions, and SIDS (sudden infant death syndrome), with peaks
between the second and fourth month of life remaining
unexplained.
It may be assumed that in the past the percentage of infant
deaths attributable solely to congenital disease was probably
low and that many infants probably died from widespread
infections of gastroenteritis and pneumonia, deaths which
affected both normal and congenitally malformed infants
(Roberts & Manchester 1995). Since infants remain vulner-
able to acute gastrointestinal and respiratory infections today,
deaths in this age group may have been much higher in an-
tiquity without the availability of antibiotics (Roberts &
Manchester 1995). None of this would leave any visible trace
on the osteological material. Another natural cause of infant
deaths may be nutritional deficiency diseases (scurvy, rickets
and iron deficiency anaemia). Possible traces of classically
described nutritional deficiency are present. Burials 106 and
107 (House 4) had irregular growth of claviculae. Burial 120
(House 29) displays cribra orbitalia. Some other signs (ir-
N ational weeks). The cut mark is located above the distal end
Gestational Week Intervals
Figure 2. Absolute number of infant individuals (N=38)
studied at Lepenski Vir across stages of estimated age
in gestational weeks (Burials 83b and 83b[1] excluded).
136
regular growth of long bones and porosity) were noticed on
Burials 108 (House 4), 109, 109a (House 4′), 114 (House 36)
and 121 (House 29), and also on infant Burials 83b and
83b(1). However, the archaeologically visible palaeopatho-
logy in the case of anaemia is still questionable (Stuart-
Macadam 1989).
Another cause of infant deaths may be infanticide. It has
been suggested that some societies practise infanticide as a
method of population control owing to the traditional re-
quirement to control the number of females, or for general
population control (Roberts & Manchester 1995). Scott
(1999: 12), indicating appalling feelings held in relation to
the question of infanticide, notes: “It has struck me on a
number of occasions how discussion of infants in the past,
and especially infant burials, invariably turns to discussion of
infanticide”. As some sort of infanticide is for the most part
legal today through abortion, we can hardly reject the fact
that infanticide occurred in the past, in various forms and
with different reasons and connotations. There remains a
problem with its detection on the osteological material since
this might have included drowning, smothering, strangling,
exposure, or neglect (cf. Scott 1999). One of the indicators of
infanticide may be a strong peak of individuals aged about
38–40 weeks, following the logic that infanticide usually oc-
curs immediately after birth, and “the age distribution of vic-
tims of this practice would be expected to mimic the gesta-
tional age of all live births”, i.e. between 38–40 weeks (Mays
1998: 66).
Although the gestational age of infants from Lepenski Vir
has a strong peak between 38–40 weeks (see Fig. 2), we
cannot comment with any certainty about infanticide here.
Cultural and contextual reasons for this opinion are provided
below. Although it is to be expected that most births in the
past as well as in the present took place commonly between
38–40 gestational weeks, there may be a significant discrep-
ancy in the survival rate of new-born babies in the past and
today. It is almost certain that in the past the rate of infant
mortality in the course of delivery was much higher. In addi-
tion, as some authors have suggested, infanticide is evident
through the skewing pattern in relation to one sex. Data from
Lepenski Vir with 16 males and 13 females of successfully
DNA-analyzed infants (see Table 1; Čuljković et al., this
volume) do not fit this proposition.
Presently, the only ambiguous trace is a cut mark on the
left humerus of Burial 112 (in House 3 — male, 38–40 gest-
of the humerus, at its lateral side (length 10 mm, breadth
3 mm). It was made by a sharp tool, possibly flint. However,
rather than seeing this as proof of a violent death of this
neonate, it may be related to the practice of defleshing the
body. Equally, the skeleton may have been affected by post-
burial disturbance.
Stratigraphic and contextual relations
The following observations on the stratigraphic and contex-
tual position of the burials of new-borns at Lepenski Vir are
made on the basis of published information from the site
(Srejović 1969a, 1969b, 1970, 1972, 1981; Srejović &
 Sofija Stefanović & Dušan Borić: New-born infant burials at Lepenski Vir
Babović 1983; Radovanović 1996) and unpublished archive
field documentation from the excavation campaigns at
Lepenski Vir.1
Most of the infant skeletons from Lepenski Vir were un-
covered in the course of the protection-excavation work
throughout August, September and the start of October of
1970. During this period, a number of building floors and
hearths uncovered at Lepenski Vir were lifted from their ori-
ginal in situ location and relocated to a higher terrace where
the Museum of Lepenski Vir is situated today (Čanak-Medić
1970, 1971).2 During that campaign c. 40 new-born infant
burials were found at Lepenski Vir below the level of build-
ing floors in the rear parts of 19 buildings at the site. Although
these burials were mentioned briefly on several occasions by
the excavator (e.g. Srejović 1969b: 15, 1981: 40, 42 — at-
tributing this group of burials to phases LV Ib-e) and listed in
relation to buildings by I. Radovanović (1996: 175–185),
based on her insights into unpublished burial descriptions,
the whole phenomenon has escaped further attention. The
infant burials were not mentioned in the monographs pub-
lished about Lepenski Vir. One reason was the fact that their
discovery came after the first version of the Lepenski Vir
monograph — the Serbian edition published in 1969 — was
completed, with subsequent translations (into English, 1972;
and German, 1975) largely following the original text with
only minimal additional editing of paragraphs and with ap-
pendices to the previous text. However, the fact that infant
burials regardless of their large number and striking pattern-
ing at Lepenski Vir, have not been studied and described with
the same care and detail as some other features (such as art-
works), possibly reaffirms the already mentioned general
pattern of neglecting infants as a subject of lesser archaeolo-
gical importance and interest.
‘Houses’ vs ‘shrines’: the significance of domestic space
In the following discussion, referring to places where infant
burials were discovered we shall use the term ‘house’ to de-
scribe uncovered architectural features at Lepenski Vir with
trapezoidal bases and centrally-located rectangular hearth
slabs and front ‘thresholds’. This kind of terminology is not
without problems. We still lack in a published form detailed
knowledge of the contextual associations at these floor sur-
faces that would say more about both the character of occu-
pational activities that took place in connection with these
spaces and about scheduling of acts of abandonment of the
structures, witnessed through the intentional deposition of
animal bones and objects over the limestone floors or thrown
into the infill of these features (cf. Bökönyi 1969, 1972; Borić
2002a; 2002b; Borić & Dimitrijević 2005; Radovanović
1996).
One view is to see and interpret these features exclusively
as ‘shrines’ (also reflected in the terminology used in some
source publications, cf. Srejović & Babović 1983; see also
Babović 1997) where the supposedly sacred character of
activities and material culture associations completely ex-
clude ‘mundane’ aspects of life (Babović 1997: 103). On the
other hand, the excavator of the site, Dragoslav Srejović
(1969a: 67–69; 1972: 66–67), in a rather structuralist fashion,
interprets the space of a recurrent architectural form of the
structures found across the site, seeing the sacred part con-
137
fined to the area around the hearth (sometimes surrounded by
stone ‘tables’) of the trapezoidal buildings, with restriction of
movement in this area related to social boundaries, and the
rest of the building space serving profane functions — see
below, for similar suggestions regarding the sacred/profane
aspects of building space and social implications of indicated
oppositions (e.g. Eliade 1957; Bourdieu 1990[1970]; Hodder
1990). Interpretative attempts of this kind remain largely
synchronous and idealized, only a ‘frozen’ view of one mo-
ment — in the case of Lepenski Vir most likely the moment
of abandonment. Fixing oppositions in this way puts little
emphasis on the historical contingency of these features and
their individualized characters, the varied practices involved
in their furnishing, and the ways in which generations of hu-
mans who visited or inhabited these structures understood,
made sense of and altered their embodied and objectified so-
cial and cultural norms. Other suggestions have been made
along the lines of a more complex transformational character
of these spaces, seeing their life cycle trajectories as going
through a full circle — from domestic areas with a clearly
profane character to tombs and perhaps spaces for special
ritual purposes, all depending on their individual biographies,
i.e. events that took place in the course of individual histories
of these structures and their visitors/inhabitants (Borić 2003,
2005).
We will seek to understand the complex significance and
character of the features where infant burials occurred by
examining contextual associations reported within these
structures and by unfolding the web of possible meanings and
narratives created in the sequence of events that took place
during their histories — their life cycles. This is highly rel-
evant in interpreting a large number of infant burials at this
site and for tackling the issue of their restriction to only 19
out of at least 73 architectural structures uncovered at the
site.3
The term ‘house’ will be retained as it perhaps can best
capture the possible significance that these features may have
had for the complexity of kinship relations at Lepenski Vir.
Also, C. Lévi-Strauss’s (1983, 1987) original formulation of
‘house societies’ carries significant relevance, despite the de-
gree of difference expressed on the matter of its accuracy,
pointing to buildings as complex objects that frequently mir-
ror social relations and cultural identity (cf. Carsten &
Hugh-Jones 1995; Joyce & Gillespie 2000). We shall explore
this perspective here in relation to the significance of the ex-
isting association of infant burials and buildings found at
Lepenski Vir, especially regarding the fact that the buildings
at this site are the first examples of elaborated and localized
building forms that appeared in a specific historic period.
Burials found in the course of protection works in 1970,
when the floors were lifted, were assigned numbers between
94 and 134 (only neonate Burial 63 [Fig. 3] was found earlier
— in 1968). Apart from 19 buildings with neonate burials
underneath their floors, two more buildings are considered
closely (Houses 28 and 31), with burials of children of other
age groups.
Spatial patterning
The striking pattern in relation to the infant burials at
Lepenski Vir is that all of them are found exclusively in con-
 The Iron Gates in Prehistory
Figure 3. House 26 and neonate Burial 63.
nection with buildings. A number of burials of different age
groups (134 in all) found at the site were either placed in re-
lation to the trapezoidal buildings or were found in the space
around these structures (see Fig. 1). However, burials of
new-born infants were found below the rear ends of 18
buildings at Lepenski Vir only.4
There was no specific spatial clustering of neonate burials
to any particular area of the site (Fig. 1). Also, there was no
clear-cut patterning in relation to building size — they were
equally present in fairly small and less elaborate buildings,
such as House 43 (Fig. 11), as well as in much larger struc-
tures with elaborate architectural features, such as Houses 54
(Fig. 8) and 37 (Fig. 6), which are among the largest build-
ings at the site. One clear spatial pattern that is not only con-
fined to this age group but relates to all other burials at the
site, is their absence from the area with a cluster of individual
and overlapping buildings of the north-east part of the site
(superimposed Houses 10, 11 and 12, Houses 5 and 6, Houses
7, 8, 9 and 17, House 16 and House 64 — see Fig. 1; cf.
138
Radovanović 1996; Borić 2002a: fig. 7).
On the other hand, there was again no clear-cut spatial
patterning in the distribution of infants in relation to their sex
(enabled by recent DNA analyses; see Table 1, cf. Čuljković
et al., this volume); neither in relation to a possible clustering
across the site nor within those buildings that contained more
than one infant interment. Only in two buildings, with more
than one infant burial, were individuals of exclusively one
sex found — House 29 (3 males) and House 37 (2 males).
Seven buildings contained infant burials of both sexes —
House 4′ (one male and one female), House 4 (two males and
one female), House 19 (two males and one female), House 24
(two females and one male), House 27 (three males and one
female) House 47 (one male and one female) and House 54
(one male and one female); in addition, in House 36, with two
burials, there is one male and the other of unknown sex, and
another nine buildings had only one infant burial respect-
ively. The number of male infants slightly outnumbers the
female ones (16 males and 13 females). There are no indica-
 Sofija Stefanović & Dušan Borić: New-born infant burials at Lepenski Vir
tions of a burial preference in relation to one sex among the
studied individuals. This fact may find its relevance within an
overall interpretation of the phenomenon.
Another striking spatial pattern is that all infant burials,
except one (Burial 63), were found below the ‘rear’ of the
buildings (the area behind the rectangular hearth). The
neonates and children were found a) underneath the lime-
stone floorings at the rear of buildings, b) in the area next to
the limestone floorings at the rear of buildings, sometimes
below the construction stones, which at certain buildings
surrounded the limestone floors, and c) in one case (Burial
63) below the south-west corner (corner A), i.e. in the front
part of the building (House 26; Fig. 3).
a) Experience of building space at Lepenski Vir
In order to add a contextual dimension to these patterns, it
would be necessary to describe briefly the original topo-
graphy of the site and some of the constructional elements of
the buildings at Lepenski Vir.
The site is placed on a sloping alluvial terrace covered by
deposits of loessic sand (for the further discussion on the
geological formation of this deposit see Marković-
Marjanović 1969, 1978; Brünnacker 1971). As the excavator
described (Srejović 1969a: 62, 1972: 62) and as emphasized
recently (Borić 1999, 2002a, 2002b), the structures with the
trapezoidal footprints were actually objects cut into the slope.
In this way, their builders made a semi-subterranean levelled
space that had a trapezoidal shape, making an elaborate ref-
erence to the trapezoidal Treskavac Mountain that is situated
across the Danube in front of Lepenski Vir, being an im-
pressive landmark (for discussions on the significance of this
landmark, see Srejović 1969a, 1972; Chapman 1993, 2000;
Borić 1999, 2003). The base of the space dug out in this way
was further furnished with a limestone flooring of red to
white colour c. 5–25cm thick (Srejović 1969a: 57, 1972: 54;
for a discussion on the possible technology used for this
flooring, see Ney 1971). In a number of instances the lime-
stone flooring did not cover the entire area levelled for the
floor. Instead, rows of irregular stone blocks surrounded the
floors in a number of buildings assigned by Srejović to the
Lepenski Vir I phase. On the other hand, the Lepenski Vir II
phase, as defined by the excavator, represents a later phase
(Srejović 1969a: 92 calls it ‘maniristic’), overlying the set-
tlement of phase I. This phase was defined as consisting of
buildings with a trapezoidal shape that is defined by rows of
irregular stone blocks only, but without specific limestone
flooring and mostly lacking rectangular hearths. However, it
has been suggested that this stratigraphic interpretation is
perhaps misleading (Borić 2002a). This can be shown clearly
by superimposing outlines of stone walls of supposedly
Lepenski Vir II phase buildings (marked in Roman numbers)
over the outlines of limestone building floors of Lepenski Vir
I phase. As the outlines of phase II buildings strikingly match
the outlines of limestone building floors of phase I, the fea-
tures identified as belonging to the Lepenski Vir II phase can
rather be seen as walls made in a dry stone technique to pro-
tect the dug loessic sides of the cuts of Lepenski Vir I build-
ings, forming part of the lower wall portion of a building.
Also, probably, these served to support the upper part of the
construction and the roof, made of more perishable materials
139
such as wood and possibly reeds and/or hides (Borić 2002a:
figs 7–9).
This description of the three-dimensional outline of the
main architectural features of the buildings at Lepenski Vir,
especially in relation to the offered re-interpretation of some
of their constructional elements, is intended to better contex-
tualize the possible significance of neonate and child burials
in the rears of the buildings.
Although the rear of the building might have been in-
scribed with meanings in relation to the structural normative
division of space at Lepenski Vir (prescribed for certain em-
bodied actions, although with possibly fluid and transient
boundaries in space division), it might have gained some
significance relating to its ‘cave atmosphere’ (Srejović 1981:
22; Srejović & Babović 1983: 43), placed deep into the slope,
i.e. cut into the virgin, sterile deposits of the original sloping
terrace. The fact that infant (and other) burials were not found
within the buildings of the north-east part of the site may re-
late to the relatively flat terrain and shallow deposits (quickly
reaching bedrock) in this area, where buildings would have
lacked the darkness and protection of steeply cut sides.
The importance of this spatial dimension becomes even
more evident through the example of the only deviation from
the rule of placing infants in the rear of the building. This is
the case of Burial 63 interred at the front corner (corner A) of
House 26 (Fig. 3) — the only building at Lepenski Vir (ex-
cept miniature House 49, which will be discussed below) that
was orientated parallel to the Danube, in contrast to the other
buildings at the site the wider front parts of which mainly
faced the Danube. This unusual orientation may have been a
consequence of the topography of the north-central area of
the site where House 26 was situated, which slopes slightly
toward the south-east. Also, several very large rocks origin-
ally surrounded the building on its north-east side (see Fig.
1), i.e. the side its front part would have been had it followed
the way other buildings were oriented. Thus, we have some
reasons to speculate that the way this building was oriented is
related to the unusual and exceptional interment of Burial 63
in the south-west, front corner of the building. A part of the
floor in this corner was cut in order to place the burial into the
virgin sandy loessic soil of the slope (Borić 1999: fig. 20).
Having in mind a topographic spatial dimension, placing
Burial 63 in this area of the building is not different from
other infant interments, i.e. away from the Danube and cut
into the slope of the terrace. This instance indicates the im-
portance of the ‘topographic rule’ behind the way of reason-
ing of where to inter an infant’s body at Lepenski Vir.
b) Engendered space and its interpretations
The last point does not undermine the fact that all other in-
fants were restricted exclusively to the rear end of the build-
ings. This may indicate the presence of embodied social, age
and gender boundaries in the use of building space (both in
life and death), restricting activities and social rights in rela-
tion to differences that were mapped onto the building space.
Ethnographic records abound with examples indicating spa-
tial relations of this kind (e.g. Bourdieu 1990[1970]; Lane
1994; Carsten & Hugh-Jones 1995; Joyce & Gillespie 2000).
Also, for instance, on the basis of detailed micromorpholo-
gical analyses of floor use at the Early Neolithic site of
 The Iron Gates in Prehistory
Figure 4. House 47 and neonate Burials 123 and 124.
Çatalhöyük (south-central Anatolia), similarly prescribed en-
gendered restrictions in the use of a building are suggested
(Hodder & Cessford 2003). Different building parts may
have had very different meanings and complex connotations,
as will be discussed in more detail below in the context of the
scale of building associations.
Despite the suggested general patterning, where the rear of
the building at Lepenski Vir frequently served for infant in-
terments, in the case of each building the decision of where to
place a burial was made according to the topography of the
surrounding terrain and constructional elements of each
house. Moreover, specific meanings assigned to particular
areas of ‘houses’, their constantly shifting and individualized
re-interpretation and the events that took place in the course
of the house history, all influenced decisions about where to
inter a deceased infant. Although people at Lepenski Vir must
have obeyed culturally determined spatial norms, burial
practices following an infant’s death probably reflected a
140
particular constellation of events within the house and the
community. More general or collective reasons for the spe-
cial and patterned burial rite in the case of this age group at
Lepenski Vir, shared among the houses and over generations,
are constituted by the noise of individual actors. The provi-
sion of burials may release to some extent the noise spectrum
of individual/‘house’ (re-)interpretations of these collective
and shared norms, including dimensions of diachronic
changes.
Body treatment and elaboration of burials
The neonate and child burials at Lepenski Vir are relatively
poorly furnished. This is not surprising as generally this age
group is rarely accompanied by rich grave goods (e.g. Scott
1999: 90ff.). None of the infant burials at Lepenski Vir was
accompanied by any sort of non-perishable grave offering.
However, we can focus on other elements of the burial, such
as the orientation and position of a skeleton and elements of
 Sofija Stefanović & Dušan Borić: New-born infant burials at Lepenski Vir
Figure 5. House 38 and neonate Burial 111.
burial construction.
There were problems in determining the exact orientation
and especially the position of the easily disturbed infant
burials (see Table 1). This is due to the taphonomic processes
that may have affected these burials regardless of their good
preservation, as indicated in the taphonomic description
above. Also, in several cases it seems that the placing of a
new interment in the same area disturbed some of the earlier
skeletons, affecting their preservation, as in the cases of
Burial 98a disturbed by Burial 98 (House 19; Fig. 13), Burial
110 disturbed by Burial 107 (House 4; Fig. 7) and Burial
109a disturbed by Burial 109 (House 4′; Fig. 7); this is also
evident from the nomenclature used in marking these burials
with an additional letter ‘a’ to the number of burial that dis-
turbed them. Moreover, even in those houses where there was
only one infant burial, we may assume that their occasional
poor preservation is a consequence of later occupational and
141
building activities that disturbed them. These later disturb-
ances might have caused the reportedly poor preservation of
Burials 95 (House 24; Fig. 16), 112 (House 3; Fig. 10), 117
(House 63), 118 (House 62), 119 (House 29; Fig. 15), 132
and 133 (House 37; Fig. 6), 125 and 127 (House 54; Fig. 8)
and 128 (House 27; Fig. 12). In the case of Burials 129, 130
and 131 (House 27; Fig. 12) and Burials 120 and 121 (House
29; Fig. 15), it remains an interesting possibility that their
interments took place at the same time, although the very
complicated and to some extent messy position in which they
were found may still be a consequence of sequential inter-
ments, with a time lapse, at the same spot within the building.
a) Re-enacting the body position
In those cases where it was possible to detect and reconstruct
the anatomical position of infant skeletons, the excavators
noted the following positions: a) extended, b) contracted/
 The Iron Gates in Prehistory

Figure 6. House 37 and neonate Burials 132 and 133.

flexed and c) a position with legs splayed outwards (see Table
1). Only two burials were found in the extended position —
63 (House 26; Fig. 3) and 124 (House 47; Fig. 4; cf.
Radovanović 1996: fig. 4.4). In addition, around 2 years and
8 months old child Burial 92 (House 28; Fig. 18) was found
in the extended position on its back. A number of skeletons
were found in flexed/contracted positions — Burials 94
(flexed/left), 101 (flexed/right) and 102 (flexed/left) in House
24 (Fig. 16), Burials 98 (flexed/left) and 103 (on the back,
contracted legs) in House 19 (Fig. 13), Burial 108 (on the
backs, contracted legs) in House 4 (Fig. 7), Burial 109 (on its
back, flexed legs) in House 4′ (Fig. 7) and Burial 123
(contracted/left) in House 47 (Fig. 4). From the description of
these burials it is difficult to discern any pattern that would
relate to the difference between those flexed burials lying on
the side and those lying on their back with only the legs con-
tracted. Also, there is no clear pattern in the choice of the side
for placement of the flexed/contracted burials. The third
group of burials are infant skeletons found on their backs
with legs splayed outwards and heals below the pelvis (in the
1970 Field Journal noted as sitting) — new-born Burials 114
and 115 (House 36; Fig. 9), 113 (House 63′; Fig. 17), 111
(House 38; Fig. 5) and 106 (House 4; Fig. 7). Child Burial 97
(2–6 years old) in House 31 was found in this last position too
(Figs 19 & 20).
To interpret this variety of positions we find the procedure
of re-enacting and re-fleshing the body a very useful one (cf.
Hawkes & Molleson 2000). The manipulation of the body
prior to its burial (e.g. excarnation, dismemberment,
binding/wrapping or burials in bags and baskets), processes
of decay as well as later disturbances may all be relevant for
the position in which we find the body (Hawkes & Molleson

142
 Sofija Stefanović & Dušan Borić: New-born infant burials at Lepenski Vir

Figure 7. House 4 and neonate Burials 106, 107, 108 and 110 and House 4′ and neonate Burials 109, 109a.

2000). In addition, it is reasonable to assume that in some placed change through time. The position of infants with
instances the position of the deceased can be explained by a their legs splayed outwards at Lepenski Vir indicates that
chronological difference between interments, where customs possibly we need to account for some sort of treatment of the
about the way the deceased infants, as well as adults, are body prior to the burial. We suggest that at least some infants

143
 The Iron Gates in Prehistory
might have been placed in a bag (which would hardly leave
any visible trace) prior to their interment in a small burial pit.
In the process of decay when a movement of the body parts
reduces a corpse to bones that slump in a certain way, de-
pending on the way the bag with the neonate’s body was
placed in the ground, we may expect to find the deceased in-
fants in varied positions. This would solve the unusual posi-
tion of the legs splayed outwards, as in these burials an in-
fant’s body was put contracted in a bag and might have been
placed on its back; in the process of the decay of the bag and
the body the bones of the legs would slump either outwards
to both sides or both legs to one side. The latter instance, also,
explains burials with the torso on the back while only the legs
were found in contracted positions, as well as the reported
unclear body positions. Furthermore, the bag might have re-
duced the immediate disturbance of the body, thereby ac-
counting for the relatively good preservation of infant skel-
etons at Lepenski Vir.
However, there remains a possibility that not all neonates
and older infants were placed in bags prior to burial and
treated in the same way, with or without chronological signi-
ficance. Extended burials may have been placed directly in
the ground. Also, is it a coincidence that three infant burials
in House 24, all three found in the flexed/contracted position,
are in the same building where two adult burials were placed
on the floor (or possibly one in the infill) in flexed/contracted
positions (Burials 8 and 9, cf. Srejović 1969a: fig. 72)? In
chronological terms, the contracted/flexed position largely
connects to the burial practices that characterize the Early
Neolithic of the wider region (cf. Stanković 1992; Borić
1999; Minichreiter 1999). On the other hand, in House 47,
Burial 123 was placed in the contracted position (perhaps
placed in a bag) and Burial 124 in the extended position (cf.
Radovanović 1996: fig. 4.4). Can this mean a change of
burial custom/norm during the life cycles and rebuilding of
House 47? Also, burials found with their legs laying splayed
outwards were found mainly in possibly older buildings,
completely or partly overlapped by later buildings (Burials
114 and 115 in House 36, Burial 113 in House 63′, Burial 111
in House 38 and Burial 97 in House 31 — see Fig. 1, cf.
Radovanović 1996). In this case does the treatment of the in-
fant’s body by putting it in a bag prior to burial represent a
relatively older burial practice? All these questions related to
the chronological significance of any patterning can best be
answered through the application of AMS dating to infants’
bones. However, later we shall discuss the significance of the
practice possibly identified here of putting the dead infant’s
body into a bag in relation to an instance from recent Balkan
ethnography.
b) Orientation
In those cases where it was possible to establish their orient-
ation, it seems that a number of burials were oriented with the
head toward the south east or the south, i.e. approximately
parallel with the Danube with the head pointing downstream
(see Table 1). This is important in connecting infant burials to
other burials at this and other sites in the Danube Gorges.
This particular orientation may have a chronological signi-
ficance and may relate to an older phase of burial practices
(Radovanović 1996), or even may indicate a hybrid re-
144
invention that kept its potency through narratives about past
practices, preserving certain beliefs and conveying particular
meanings. However, in several cases infant burials were
placed with their heads pointing to the north west, north or
north east. These variations were also noticed among the
skeletons within the same building. These differences pos-
sibly reflect the ‘noise’ of individualized actions, realizations,
and re-interpretations of memories about past practices, their
effectiveness and potency (cf. Borić 2003).
c) Burial pits
Infant burials at Lepenski Vir were placed in pits that were
dug a) mostly immediately behind the rear part of the lime-
stone flooring (a number of instances), b) through the dam-
aged parts of flooring in the rear end of a building, such as
Burials 128–131 (House 27; Fig. 12) and 96 (House 43; Fig.
11), c) through the limestone floors, for example in the cases
of Burials 111 (House 38; Fig. 5), 113 (House 63′; Fig. 17),
103 (House 19; Fig. 13), or d) within the stones of the rear of
the building, as Burials 94, 95, 101 and 102 (House 24; Fig.
16). A number of burial pits were clearly visible in the virgin
soil and had different shapes — the pits for Burials 113
(House 63′) and 94 (House 24) were rectangular, Burial 103
(House 19) was placed in an oval burial pit, while pits for
Burials 106 and 107 (House 4), 128–131 (House 27) and 132
(House 37) were reported by the excavators as having irreg-
ular shapes. Burial pits were most clearly visible in those
cases where a skeleton was placed in the virgin soil. It was
much more difficult to spot the outline of a burial pit in the
case of Burials 123 and 124 (House 47) that were dug
through the floor of House 47 and placed in deposits that
were covering an older building (House 47′) (see photo in
Radovanović 1996: fig. 4.4). The same goes for those burials
interred into deposits of occupational activities, possibly
considerably older from those of the building floors (cf. Borić
1999: 53–54; Borić & Dimitrijević 2005; Whittle et al.
2002), such as neonate Burials 116 (House 13; Fig. 10), 119
(House 29), 133 (House 37) and child Burial 97 (House 31;
Fig. 19). The burials were placed at differing depths, meas-
uring from the level of the floor — from as little as 10 cm in
the case of Burial 63 (House 26), 20–30 cm in the case of
Burials 128–131 (House 27), 118 (House 62), 132 (House
37) and 134 (House 48), up to 60 cm in the case of Burial 116
(House 13).
d) Use of stone in burials
In several instances stones were used in different ways in the
elaboration of a burial. In the case of Burial 63 (House 26)
stones were placed vertically, two on the sides of the head and
one on the side of the legs. In two more instances special
concern for the infant’s head was expressed — the head of
Burial 106 (House 4) was fixed between two stones and two
small stones were placed next to the head of Burial 125
(House 54). There was a stone construction in the case of the
rear of House 24 (Fig. 16) where Burials 94, 95, 101 and 102
were placed. A similar stone construction was found within
the rear of House 43 where Burial 96 was interred. In the case
of Burial 113 (House 63′), where the rectangular burial pit
was cut through the floor of the building, the burial was
covered by a stone slab after placing the body (Fig. 17). Also,
 Sofija Stefanović & Dušan Borić: New-born infant burials at Lepenski Vir

Figure 8. House 54 and neonate Burials 125 and 127 and mandible Burial 126.

child Burial 92 (House 28; Fig. 18) was partly lying on a
small stone slab of irregular shape while above it, on the level
of the building floor, there was a large stone block. The ex-
cavators note that a piece of floor, with a depression on its
upper side, was used to cover Burial 103 (House 19, Fig. 13;
1970 field diary).
Discussion
An important conclusion to be drawn from the described in-
stances of elaboration of infant burials at Lepenski Vir sug-
gests that probably all of these burials were dug from already
existing buildings with furnished limestone floors. On the
basis of data presented, it is difficult to sustain the opposing

145
 The Iron Gates in Prehistory

Figure 9. Houses 35 and 36 and neonate Burials114 and 115.

view that the burials were placed before the buildings were
furnished, i.e. before the floors were laid (contra
Radovanović 2000: 340, note 7), as some kind of foundation
deposits (similar to the examples from Neolithic settlements
in the Carpathian Basin, cf. Makkay 1983), or even as some
sort of ritual sacrifice, indicating a practice of infanticide —
for instance, Benac (1973) offered an interpretation along
these lines for infant burials at the Early Neolithic site of
Obre in central Bosnia. Although burial pits for infant inter-
ments were not recognized immediately from the floor level
during the excavation of Lepenski Vir’s buildings, this can be
attributed to the fact that the burial pits were small, perhaps
wide and deep enough just to place an infant’s body (possibly
in a bag). Also, as indicated above, in many instances infants
were buried just off the floor area in the rear of a building,
frequently under the stone wall, which would tend to obscure
an unexpected and small burial pit. To these reasons, one
should add the occasionally crude excavation method. Here,
we have to disagree with Srejović’s opinion that infant Burial
63 (House 26) (see Borić 1999: fig. 20) and an older child
Burial 61 (House 40) were buried prior to the construction of
these buildings (Srejović 1969a: 136; 1972: 119). Both buri-
als were clearly recognized and excavated in the course of the
excavation of these buildings. This would suggest that the

146
 Sofija Stefanović & Dušan Borić: New-born infant burials at Lepenski Vir
Figure 10. House 13 and neonate Burial 116.
burials were noticed from the level of the building floors, and
indicates that they were dug from this level. Furthermore, a
representational boulder was placed above Burial 61 (see be-
low), which is not a coincidence (contra Srejović 1969a:
136), and the limestone floor in the rear of House 40, where
the burial took place, was significantly damaged (see Fig. 1;
Srejović 1969a: fig. 65). It seems that in some instances in-
fants were placed in already existing damaged areas of the
building floor (e.g. House 27 — Fig. 12, House 43 — Fig.
11). These damaged areas may indicate that buildings were
used over a long period of time prior to the burial of an infant.
On the other hand, it is difficult to ascertain if perhaps these
damaged parts of the floor expanded exactly as a con-
sequence of digging a burial pit. Also, at present we have no
way of telling if the buildings continued to be used after the
burial of an infant took place. Combined AMS dating of the
abandoned animal bones on building floors and infants’ skel-
etons may be a solution to the problem.
147
A final point in this part concerns two small fragments of
Early Neolithic pottery found in the fill of Burial 113 (House
63′). As already indicated, this pit was covered with a stone
slab and by the floor of later House 63 (Fig. 17). In the ori-
ginal field notes it was indicated that the pottery fragments
within the burial infill were found in the part of the burial pit
that was not covered with the stone slab and the possibility
was advanced that these were deposited by the Danube. In
our opinion, there is a slim chance for this scenario, and the
fragments may represent either an accidental content of the
pit infill or some sort of (fragmented) grave goods (cf.
Chapman 2000). This instance, also, opens up the very con-
tentious issue of the chronological position of the infant
burials and buildings of Lepenski Vir I phase (cf. Borić 1999,
2002a, 2002b; Borić & Dimitrijević 2005; Whittle et al.
2002; Radovanović 2000). We shall return to this question.
The contextualization of infant burials at Lepenski Vir
takes us further to consider other features found within the
 The Iron Gates in Prehistory
Figure 11 House 43 and neonate Burial 96.
buildings. Here we focus on two types of finds in particular
which may relate to the infant burials on the level of building
associations. These are articulated and disarticulated burials
of other, non-infant, age groups found placed on or below
limestone floors and the artworks of ornamented boulders
and ‘altars’ (most likely serving as pestles) found in connec-
tion with building floors.
a) Adult burials/disarticulated bones in buildings
Disarticulated bones of adult individuals were found fre-
quently in relation to the rectangular hearths in buildings. For
instance, a human mandible (Burial 22) was found next to
Proto-Lepenski Vir Hearth a (see Fig. 1; cf. Srejović 1969:
132). A striking case is Burial 21, a human mandible placed
next to the hearth slab of House 40, the teeth facing down
with a flat stone plaque blocking the open area between the
caudal rami of the mandible (Srejović 1969a: fig. 70, 1972:
119, fig. 64; Radovanović 1996: fig. 3.48). This made a re-
cognizable ‘A-support’ which, in the case of all other build-
ings, is made out of stone plaques only. These supports were
placed in varying numbers around a number of the building
hearths at Lepenski Vir.5
A human mandible (Burial 126) was found lying on a large
rock below the rear of the hearth of House 54, with neonate
Burials 125 and 127 buried under its rear (Fig. 8). Below the
floor of House 31, where child Burial 97 was found, another
isolated human mandible (Burial 105) was found approxim-
ately below the area of the hearth (Fig. 19). Also, a human
skull with no mandible (Burial 122) was found in the rear part
148
of the hearth of House 47′ (Fig. 4), underneath the floor of
House 47 that had neonate Burials 123 and 124 (cf.
Radovanović 1996: fig. 4.4). It is important to emphasize the
connection of human mandibles/skulls with building hearths.
In other buildings with infant burials, disarticulated or
semi-articulated and partly preserved extremital bones of
human adults were found, again closely related to building
hearths, and only rarely found in the other parts of buildings.
In House 43, where neonate Burial 96 was buried, partly
preserved and probably articulated bones of a human torso
(Burial 104 consisting of clavicle, sternum and ribs) were
found below the floor on the rear part of the hearth (Fig. 11).
Also, in House 19 with neonate Burials 98, 98a and 103, on
the sides of a shallow pit dug for the sculpted boulder and
placed behind the hearth, disarticulated human bones (Burial
99 consisting of scapula and clavicle on one side and broken
humerus on the other side of the pit) were found (Fig. 13). In
House 24, with neonate Burials 94, 95, 101 and 102, besides
three adult burials placed on the building floor or alternat-
ively dug into the already existing infill of the building (these
are articulated contracted Burials 8 and 9 and human skull
Burial 10 — Srejović 1969a: fig. 72, 1972: fig. 63), partly
preserved and semi-articulated human adult bones (Burial
100) were found below the rear of the building (Fig. 16).
Another disarticulated human humerus (no burial number)
was found in the occupation layer of dark soil (with animal
bones) below the floor of House 29 with neonate Burials
119–121. Bones of a human foot were found below the floor
of House 62′ in the north corner (corner B). This instance
 Sofija Stefanović & Dušan Borić: New-born infant burials at Lepenski Vir

Figure 12. House 27 and neonate Burials 128, 129, 130 and 131.

may relate to neonate Burial 118 found below the rear of
House 62 that overlapped older House 62′. Also, Srejović
(1969a: 134–135, 1972: 118) mentions that in both Houses 3
(neonate Burial 112) and 54 (neonate Burials 125 and 127)
one human femur was embedded in the floor behind the
hearth construction. In some instances, disarticulated human
bones were placed over building hearths. A disarticulated
humerus (Burial 23) was placed over the hearth of House 48
with neonate Burial 134. On the floor of House 36 (neonate
Burials 114 and 115), at the rear side of the hearth, disarticu-
lated human adult bones were found (Burial 70) together with
a dog mandible (cf. Radovanović 1999: 74).6
Certainly, some of these examples of disarticulated and
semi-articulated burials found below the building floors may
come from burials considerably older than the occupation
inside these spaces (Borić 1999: 59), disturbed by subsequent
building activities. For instance, Burial 100, below the rear of
House 24, or the bones of a human foot found underneath
House 62′ may perhaps represent older, disturbed burials.
However, the intentional placement of disarticulated human
bones is indicated by their frequent placement in the area at
the rear, narrow end of the hearth (below or over it). These
bones possibly both circulated during the life of a building
and/or are a consequence of an intentional deposition indic-
ating the event of building abandonment (Borić 2003; 2005).
Considering the choice of manipulated body parts, as pre-

149
 The Iron Gates in Prehistory

Figure 13. House 19 and neonate Burials 98, 98a and 103.

viously noted (Srejović 1969a: 143–144), one can underline
a special concern in the manipulation of heads and mandibles
of the deceased. Mandibles and skulls of the deceased in
many cases were found separated, as mentioned in some of
the described examples of disarticulated burials. However,
some kind of dismemberment of skulls from mandibles was
noticed in articulated burials too. For instance, a skull of
child Burial 92 (House 28) was detached from the mandible
and adult Burial 26 (House 34) was found with the skull de-
tached from the body and turned to face the east, i.e. the

150
 Sofija Stefanović & Dušan Borić: New-born infant burials at Lepenski Vir
Figure 14. Neonate Burials 98, 98a and 103 after lifting the floor of House 19.
Danube,7 while the mandible slumped as if left in the ana-
tomical position (Srejović 1969a: 135; cf. Radovanović
1996: fig. 4.5). Also, articulated Burial 28 (‘House’ XXXIII)
was found with the mandible, but no skull (Srejović 1969a:
fig. 68). At Lepenski Vir, several burials were found in their
anatomical position but with complete skulls detached from
their bodies, such as extended Burial 54e (House 65/XXXV,
Srejović 1969a: 137, fig. 67), extended Burials 15 and 16
(marked as in ‘House’ XXVII of Lepenski Vir II phase,
Srejović 1969a: 137, fig. 18) and contracted Burial 19 (House
54/XLIV) where the skull of the deceased was placed on the
stone slab that covered the burial (Srejović 1969a: 165, 1981:
8).
All these instances, together with the practices mentioned
in relation to infant burials, point to the possible metonymic
importance of the head and the mandible, standing for the
deceased individuals. The importance of human (but also
animal) skulls and mandibles that stand for the whole body
and the person, and their manipulations though cultural
practices were widespread in the whole eastern
Mediterranean during the Mesolithic and Neolithic periods
(e.g., Mellaart 1967; Hodder 1990; Cauvin 1994).
We may find some relevance in Srejović’s (1969a: 140,
1972: 117) inclination to see some disarticulated burials
connected to building hearths as ‘hearth guardians’. Indeed,
these might have been individuals whose death occurred long
before the time of their deposition in buildings of Lepenski
Vir. Their actual antiquity and possible place(s) of origin may
be deduced in the future by applying AMS dating and trace
151
element analyses. Although skulls and mandibles had a spe-
cial significance among the body parts, other skeletal parts
were also manipulated and similar practices have been ob-
served in other contexts of the European Neolithic (cf.
Whittle 1996). At Lepenski Vir, we may claim with some
certainty that these might have been observed as ancestors,
whose identity and recognition varied extensively — from
those very recognizable to members of a building through
direct memory and/or generational narratives to almost an-
onymous individuals from a distant past. Furthermore, we
may speculate that placement of parts of ancestors’ bodies
next to the hearths might have meant that the area of the
hearth was particularly inhabited with ancestral potency.
These bones might have acted as second-class agencies (cf.
Gell 1998) with powerful apotropaic potencies (Borić 2003).
These, possibly protective, ancestors might have had special
significance and importance for buildings and their
members/visitors, perhaps especially for those building lin-
eages where infants and children were dying frequently or
were faced with other misfortunes. We shall elaborate further
on this argument, but first we need to pay attention to the
sculpted boulders that might have anchored ancestral poten-
cies in a similar way, focusing social and collective attention.
b) Artworks and their potencies
Sculpted and aniconic boulders and pestles (‘altars’) were
found mostly in the apparently fixed contexts of buildings at
Lepenski Vir and only occasionally in a loose context (cf.
Srejović & Babović 1983). They were found concentrated
 The Iron Gates in Prehistory

Figure 15. Houses 21, 22 and 29 and neonate Burials 119, 120 and 121.

152
 Sofija Stefanović & Dušan Borić: New-born infant burials at Lepenski Vir

Figure 16. House 24 and neonate Burials 94, 95, 101 and 102.

mainly at the rear of the hearths (similar to the disarticulated
burials). In several cases, boulders were found also at the rear
of the building and only rarely in its front part. Also, some
buildings (including a few of those with infant burials) lacked
both/either boulders and/or pestles. Their placement in
buildings at Lepenski Vir, as well as their spatial patterning,
must have depended on specific events related to the prac-
tices that took place immediately before a building was
abandoned.
The representational, purely ornamented and aniconic
boulders and so-called ‘altars’ (i.e. pestles, sometimes orna-
mented too) might have been conceived as important heir-
looms of individual buildings, closely fixed to the building
hearths. We may underline the monumentality of some of the

153
 The Iron Gates in Prehistory
Figure 17. House 63′ and neonate Burial 113.
boulders, being architectural parts of buildings, as in the case
of the hearth of House 43 (Srejović & Babović 1983: 133)
where the boulder acted as a side of the stone hearth. Srejović
(1969a, 1972) drew attention to the perfectly spherical and
‘closed’ boulder as pregnant with a mysterious sacredness.
We may also view them as bodies with animate characterist-
ics, objectifying narratives and myths about past events, re-
lated to the whole community as well as individual buildings
(Borić 2005). In particular, they might have been connected
to mythical dimensions of anadromous fish (cf. Radovanović
1997), but also to heroes and petrified ancestors from the
distant past. Being artworks with a special investment of in-
tention and desire (cf. Gell 1998), these objects might have
embodied special powers important for building members,
both deceased and living.
Here, we shall focus on those instances where sculpted
boulders were placed on the level of the building floor im-
mediately above the head of the deceased buried within
buildings. In the case of Burial 7/I, a decorated boulder
(Srejović & Babović 1983: 136, cat. no. 30; Srejović 1967)
was placed on the forehead of the deceased male. The already
indicated connection of sculpted boulders and disarticulated
154
burials as pregnant with ancestral (protective) potencies may
be further affirmed, as a detached human (ancestral?) skull
(Burial 7/II), with strong eyebrow ridges and heavily worn
(possibly from handling), was placed next to the head of the
deceased, facing him. Also, an aurochs skull was placed over
the other shoulder of the deceased (Srejović 1969a: 136–137,
fig. 69, 1972: fig. 61, 1981: 43a; Srejović & Babović 1983:
136; Radovanović 1996: fig. 4.3). Both the boulder and the
human and animal skulls may have accompanied the de-
ceased for apotropaic reasons. In addition, he was placed next
to the hearth, behind its rear. Also, the burial pit for 7/I was
clearly cut through the floor of House 21, while the skeleton
was placed in the extended position with the head pointing
downstream. In another instance already discussed, child
Burial 61, around 7 years old (see above and note 7), was
placed in the extended position with the head pointing
downstream in a similar way, within a burial pit cut through
the floor in the rear of House 40. Above the head of the de-
ceased, a small sculpted boulder was placed, representing a
human (child?) face with schematic mouth, nose and en-
graved eye circles (Srejović 1969a: fig. 65; 1972: fig. 59;
Srejović & Babović 1983: 108). In both examples (with the
 Sofija Stefanović & Dušan Borić: New-born infant burials at Lepenski Vir

Figure 18. House 28 and child Burial 92.

addition of Burial 26 cut through the floor of House 34 and
placed in the extended position with its head pointing down-
stream, see above and note 7) the building floor was cut to
place the deceased. This may be relevant for the way infants
were buried by cutting through the floor or though the virgin
soil in the area next to the floor, in a similar fashion and in
some cases with the same orientation of the body, i.e. with the
heads pointing downstream. We may underline that Burials
7/I, 61 and 26 were treated in a particular way and were spe-
cifically furnished/manipulated.
Only child Burial 92 (House 28; Fig. 18) clearly shares
this orientation, position and association with a sculpted
boulder. Approximately above Burial 92 two sculpted
boulders were placed on both sides of a large stone block in
the rear of House 28 (Srejović 1969a: fig. 24, 1972: fig. 14;
Srejović & Babović 1983: 107). This skeleton was found
without its skull (only the mandible was present), placed in
the extended position with the head pointing downstream.
The representational boulder ‘Adam’ (Srejović & Babović
1983: 107, cat. no. 1) placed in relation to this burial is of
particular interest. It shows a human (child’s?) face with a
schematic representation of its mouth and nose similar to
those represented on the sculpture above the head of Burial
61. What appears different is the way the eyes are represen-
ted. Instead of carved circles, only two horizontal lines were
carved, possibly indicating shut eyes. In this context, one
must emphasize the fact that all other representational
boulders found at Lepenski Vir showing a human face bear

155
 The Iron Gates in Prehistory
Figure 19. House 31 with child Burial 97 and isolated mandible Burial 105.
eyes as carved circles (cf. Srejović & Babović 1983).
The differences mentioned may have some importance in
indicating cultural attitudes toward infants’ lives and deaths
at Lepenski Vir through the medium of their representation,
breaking down age stages of childhood at this site. There may
be an important connection between the way the eyes are
represented in the case of boulders above Burials 61 and 92
and the respective age stages of these individuals. Seen from
an anthropological perspective in the treatment of artworks,
eyes are ‘orifices’ that first open up the aniconic form of an
idol and do not have representational significance only
(cf. Gell 1998: 132, 147). First and foremost they serve to
penetrate the mind, into the invisible ‘inwardness’ of an idol
(Gell 1998: 132). From all the representational features of the
sculpted face, eyes most strongly express the personhood
represented. Hence it is no coincidence that an older child
Burial 61 had eyes carved as circles, possibly indicating that
on death it had reached a socially recognizable personhood,
as a consequence of the rites of passage he experienced on the
way to adulthood (cf. van Gennep 1960; Turner 1974).
Although other representational boulders found at Lepenski
Vir were not associated with any burial in particular, it may
be possible to assume that they represent metamorphosed
adults/elders/ancestors and this state of being might have
been emphasized with two semi-circled carved lines that ap-
pear immediately below the engraved eyeholes on these
sculptures (Srejović & Babović 1983: 113, 116, 118).
Younger child Burial 92 in House 28 can be meaningfully
connected to the sculpted boulder with ‘shut’ eyes, represen-
ted as two carved horizontal lines. At the time of the child’s
156
death, it might have been lacking socially achievable per-
sonhood, although recognizable in the social world of the
community. Metaphorically and socially at the time of death,
its worldly persona was not fully formed and hence the eyes
could not have been represented opened up, facing the world.
We may possibly indicate that it might have been still partly
conceived as ‘boneless’ (cf. Astuti 1998; see below), prone
and vulnerable to the malevolent forces abounding in the
world.
c) Identification of ‘ancestral’ potencies
The examples of disarticulated burials and boulders de-
scribed above indicate a possible connection of these features
in the buildings of Lepenski Vir with the function of anchor-
ing ancestral powers and elaborating the arena of domestic
space with strong apotropaic associations. The placement of
infant burials in a patterned way may have accounted for
these potencies in the expression of lineage and communal
care toward the deceased. Moreover, this care for the descent
group is obvious in examples of the burials of other age
groups, possibly depending on the constellation of events that
surrounded their deaths. Considerable age differences of
these deceased individuals and the infants were played out
through spatial patterning of their respective burials within
the building space. Some of the child and adult burials were
commemorated with sculpted boulders, the iconographic
representational features of which were context sensitive.
Infants of neonate age were never accompanied with a
boulder and the memory of their short existence was almost
neutralized by placing them back into the earth (possibly in a
 Sofija Stefanović & Dušan Borić: New-born infant burials at Lepenski Vir
Figure 20. Child Burial 97 (House 31).
bag). However, significantly this happened within the space
of the building.
d) House 49 — visible infants?
One example that possibly opens a perspective on the visib-
ility of infants in the society of Lepenski Vir and, at the same
time, makes reference to the elaborate building space is
House 49. This is a miniature structure (1.4 m2) that is a pro-
portionally reduced image of a normal-size building
(Srejović 1969a: fig. 37, 1972: fig. 27). House 49 was placed
below a large rock on the north-west side of House 47 and
was conspicuously and exceptionally oriented with its front
part facing the north-west, i.e. almost completely opposite to
the way all other buildings at the site were oriented (see Fig.
1). The function of this building, along with another example
of a miniature building, though less well preserved and
slightly bigger, House 55 (1.9 m2), remained an enigma for
the excavator of Lepenski Vir (Srejović 1969a: 71, 1972: 69).
Srejović notes that nothing was found on the building floors
although the small plaques, that formed the hearth, were
burnt by fire. He suggests that these examples are architec-
tural prototypes for other buildings at the site, representing
blueprints (Srejović 1969a: 71, 1972: 69). However, a dif-
157
ferent interpretative account is offered here. Their construc-
tion may relate to children at Lepenski Vir with various reas-
ons in mind. Some indications for the scope of reasons can be
gathered from an ethnographic case describing the Nukak
hunter-gatherers from the Amazonian rainforest (cf. Politis
1999a, 1999b). An instance was recorded among the Nukak
where they constructed a miniature hut at a residential camp
which in every specific constructional detail followed the
outline of a normal size hut. Also, a photograph was taken of
a child sitting within this miniature hut. The anthropologist
notes a special reason for the construction of this hut and
placement of the child under the hut as it may have been sick
or faced with some other sort of danger, indicated by the
child’s red painted face (Politis, pers. comm.; see the picture
on the back cover of Cárdenas & Politis 2000). In the case of
House 49 we may speculate about a possible window into
practices at Lepenski Vir that might have related to children
during their lives, involving issues of protection with a strong
reference to the building space.
Before continuing to draw upon wider meanings of similar
cultural practices through other ethnographies, we turn to the
diachronic perspective of the phenomenon of infant burials at
Lepenski Vir, their regional and historical context.
 The Iron Gates in Prehistory
Chronological and regional perspectives
We arrive at the question of the chronological position of in-
fant burials at Lepenski Vir. Clearly, this is inseparable from
the question of the chronological position of buildings with
trapezoidal limestone floors of Srejović’s Lepenski Vir I
phase. It has already been noted that the previously suggested
schemes for stratigraphic phasing of buildings and other
contexts at this site by both Srejović (1969a, 1972) and
Radovanović (1996) are problematic in a number of points
(see above, notes 3 and 6; Borić 1999, 2002a, 2002b; Borić &
Dimitrijević 2005). Also, their schemes partly ignore the ex-
isting radiometric 14C dates on charcoal (Quitta 1975; Borić
1999: fig. 7, 2002a: fig. 5) which indicate a time span of c.
6300–5500 cal BC, i.e. the period that coincides with the ap-
pearance and the duration of the Early Neolithic in the wider
region of the central Balkans. This chronological position of
occupation on the building floors was further reaffirmed by
new AMS dates (Whittle et al. 2002). Among the new dates,
one date (OxA-9055: 8445±60 BP), also, gives the first clear
indication of much older deposits concealed by limestone
floors and represented by a number of contexts across the site.
There have been a few published accounts that discussed
the presence of Early Neolithic pottery at Lepenski Vir in re-
lation to the trapezoidal buildings, particularly by analogy
with the neighbouring and contemporaneous site of Padina
(Jovanović 1969; Borić 1999; 2002a), but also by revealing
photos of in situ pottery on building floors of Lepenski Vir
(Garašanin & Radovanović 2001). This contextual associ-
ation that has a chronological significance may further be
supported by the instance of the Early Neolithic pottery
fragments found in neonate Burial 113 in House 63′ which, as
already mentioned, is difficult to understand as intrusions
from a separate and overlying Early Neolithic layer.
By analogy with the buildings at Lepenski Vir and con-
sidering the fact that the deceased infants were buried by
cutting through their floors, we can largely relate these inter-
ments to the same period that is attributed to the dated build-
ing, i.e. 6300–5500 cal BC. A recent extensive AMS dating
programme even suggests that the period can be confined to
6300–5900 cal BC (Borić & Dimitirjević 2007).
Localized diachronic changes
However, changes in cultural practices of infant burial rites
considering much deeper time depths may be observed in the
micro-regional context of Lepenski Vir, within the Upper
Gorges of the Danube. This is related to the site of Vlasac,
situated c. 3 km downstream from Lepenski Vir. The deposits
of this site are of considerable antiquity, especially in view of
some of the AMS dated burials (Bonsall et al. 1997; Borić
2006) and possibly cover the period from at least 9500 to
5900 cal BC, indicating a Mesolithic as well as
transitional/Early Neolithic development in the region.
Around 87 burials with more then 119 individuals of different
age groups were found at Vlasac (Srejović & Letica 1978:
53–82), buried over the indicated time span. There are 22
burials of infants (re-examined by S.S.). However, their
mortality profile is different from and less uniform than the
infant interments from Lepenski Vir, but also the treatment of
their bodies prior to the burial and their burial contexts are
158
considerably different (Borić & Stefanović 2004).
At Vlasac, the mortality profile is not dominated by any
particular category of infants expressed in gestational weeks
on the basis of the morphometric analyses (see above) and we
shall discuss elsewhere the results of these findings. It is suf-
ficient to say that the age in gestational weeks of the infants
from Vlasac varies between 26–28 and 35–37 weeks for
those that would correspond to the foetal category, and 38–40
to more than 47 weeks for neonates and older infant categor-
ies. Infants of foetal age most likely indicate miscarriages
and in some instances this is reflected through the burial rite.
The most dramatic example is Burial 67, a female who prob-
ably died during pregnancy or in giving birth, found with the
bones of a foetus in the pelvic area. The skeleton was covered
with ochre, also buried in the red earth (Srejović & Letica
1978: 57). In other cases, infant bones were frequently found
within the burials of adults (Burials 4b, 6a — on the shoulder
of Burial 6 — ochre, 50/1 — ochre, 50a(1) — ochre and
Cyprinidae teeth, 49(2) — completely covered with ochre,
55(2), 58b —adjacent to the legs of Burial 58, 60(1) — pelvic
area covered with Cyprinidae [fish] teeth). These infant buri-
als were sprinkled with red ochre over the pelvic area or over
other body parts and/or placed in red soil, with addition of
Cyprinidae teeth in several cases. On the basis of previous
morphological analyses some of these burials were sexed as
males and it remains necessary to re-examine their sex de-
termination in this light. For instance, in Burial 60, determ-
ined as male and with the pelvic area covered with
Cyprinidae teeth, a foetus — Burial 60(1) — was found.
Furthermore, this individual was lying on the right side, fa-
cing three small pits in the immediate proximity with infant
burials completely covered with red ochre — Burials 59, 61
and 62, containing more individuals (Srejović & Letica 1978:
60–61). Burial 62–62a with an adult individual and an infant
was covered with red ochre and Cyprinidae teeth, with the
addition of a flint nucleus (Srejović & Letica 1978: 61). In
other instances there was no obvious connection between the
burials of infants and those of older individuals (infant
Burials 21, 42, 5, 10, 12b, 35a–b, 19 and 66a). Some of these
were furnished in a particular way. For instance, particles of
graphite were found within Burial 5 (Srejović & Letica 1978:
68). Infant Burial 42 was found without its head and accom-
panied with ochre and Cyprinidae teeth over its pelvis
(Srejović & Letica 1978: 58, fig. XXXV/2). Also, infant
Burial 21 was accompanied by some Cyprinidae teeth around
its abdomen and with a necklace of 50 perforated snails
(Cyclope neriteia) on its chest (Srejović & Letica 1978: 58).
Infant burials from Vlasac indicate cultural practices that
are different from those at Lepenski Vir, and among other
things this is the function of diachronic changes. We may ac-
count for a considerable importance of expressing a
mother–child tie in a number of burials at Vlasac, an em-
phasis on the danger of pregnancy and possibly a communal
focus on a misfortune in respect to miscarriages. Some of the
accents of practices, thus, differ from the underlying concep-
tual schemes at work in Lepenski Vir. Yet we may recognize a
much wider underlying theme in these practices that might
have been shared over a considerable time span with power-
ful elements of signification. Red ochre in particular, as well
as some other elements present, such as Cyprinidae teeth and
 Sofija Stefanović & Dušan Borić: New-born infant burials at Lepenski Vir
possibly graphite, might have been used as powerful means
of acting upon the body of the deceased infant and the ac-
companied adult (in some cases possibly mothers) with the
intention of guarding their liminal experience of death, guid-
ing them toward the underworld. The use of red ochre at
Vlasac almost exclusively relates to infants and adult burials
accompanied by an infant. Signifying qualities of red colour
imminently relate to the body (cf. Turner 1967). The concepts
of care, protection, but also danger in connection with birth,
infants and pregnant women, are clearly present at Lepenski
Vir too. We can advance the hypothesis that there remained a
residual significance of red, used at Vlasac through the use of
red ochre and reproduced at Lepenski Vir through the quality
of red limestone floors, acting as the elaborated and powerful
domestic arena. Still, this connection might have rather come
from an unconscious production of signification, than as a
conscious and meaning-laden realization (cf. Deleuze &
Guattari 1984).
Wider regional context
Presently, no other infant burials are known from contem-
poraneous sites in the Danube Gorges. Infant burials do ap-
pear among a number of other sites in the central Balkans and
across Southeast Europe, dated to the Early Neolithic (e.g.
Benac 1973; Stanković 1992; Minichreiter 1999; Borić
1999). These sites are largely contemporaneous with the
elaboration and occupation of trapezoidal floors at Lepenski
Vir, also sharing similar pottery styles (cf. Whittle et al.
2002).
For instance, a new-born infant burial was placed in a
pithos that was found at the Early Neolithic site (c.
6100–5500 cal BC) of Anzabegovo in FYR Macedonia
(Gimbutas 1976: plate 47; Garašanin 1982: 89; Nemeskéri &
Lengyel 1976: 396, fig. 242). The pithos itself, with inten-
tionally broken bottom and handles, was placed beneath the
two adult burials arranged symmetrically in contracted posi-
tions. Also, infant bones were found in a central hole of one
of two smaller quadrangular buildings of tamped clay excav-
ated at this site.
Another striking example is the Early Neolithic site of
Obre in central Bosnia (Benac 1973). At this site 8 burials
were found, and all belonged to infants and children
(Nemeskéri 1974). Burials 1–4 were placed in contracted or
flexed positions, Burial 5 was represented by a skull, while
Burials 6–8 were largely scattered or it was difficult to de-
termine their exact positions and orientations (Benac 1973:
347–363). It was suggested that Burial 7 was placed in the
seated position (Benac 1973: 351). Grave offerings were
found possibly accompanying Burial 1 (Starčevo type clay
‘altar’) and Burials 5 and 6 (most likely residual remains —
potsherds, flint tools and animal bones), and clearly in Burial
7 (‘sun discs’ — one from fired clay and one from stone and
potsherds) and Burial 8 (broken amber ornament, semicircu-
lar Spondylus ornament, two polished stone axes and two
pots — one of the Adriatic type and another ornamented with
‘barbotine’ decoration with stone pebbles inside). Also,
Burial 3 was placed over the area of burned clay and, as well
as Burial 4, covered with a thin layer of clay. In the case of
Burial 2, stones were placed around the skeleton, and Burial
5 was also surrounded with a circle of burned stone slabs.
159
Burial 7, even more elaborately, was placed on a stone base
and surrounded by stone slabs (representing a stone con-
struction of some kind). Scattered bones of Burial 8 were
found on the clay floor, surrounded with stone slabs and the
whole area covered by small stone pebbles, almost forming a
small tumulus. Burial 8 with finds of two stone axes, largely
influenced the excavator to interpret all the burials as sacri-
fices, possibly related to a sun cult (Benac 1973: 357–363).
This interpretation may be regarded as rather naïve and
without a serious grounding in the described instances. In the
context of the present discussion we may assume some sig-
nificance for the elaboration of clay floorings and, perhaps,
the practice of covering some of the buried infants with clay.
Single infant burials were found at several other Early
Neolithic sites in the Balkans. At Obrež in the region of
Vojvodina, a contracted burial of an infant with no grave of-
ferings was found, being the only burial found at the site
(Brukner 1960). Also, an infant burial was discovered at the
site of Divostin in central Serbia (Zoffmann 1988). In the
same region, at the site of Blagotin, an infant burial was
found in the infill of a large pit-dwelling, covered with a thick
layer of ash. This burial is AMS dated (OxA-8609: 7270±50
BP; Whittle et al. 2002), being contemporaneous with the
earliest dates for occupation remains on building floors of
Lepenski Vir (see above). In the layers of this feature some
other finds were exceptional — two large figurines of fired
clay, a grain model with incisions, a clay ‘altar’ above the
layer of ash and amulets (Stanković & Leković 1993; Whittle
et al. 2002).
In the described examples across the central and northern
Balkans, roughly in the same historical context, an occasion-
al focus on infants and children is evident. At some of the
sites infant burials, as at Lepenski Vir, are related to building
floorings or less elaborated areas, but frequently to ‘special’
features of some sort. The focus of these practices may be of
wider regional as well as cross-cultural importance.
At other sites of the eastern Mediterranean infant burials
are similarly found in buildings and below their floors in the
pre-Neolithic–Early Neolithic phases. At the Early Neolithic
site of Nea Nikomedeia in northern Greece, a woman and two
children were found underneath the floor of one building
(Rodden 1965; Hourmouziadis 1973; Gallis 1996). At
Franchthi Cave in southern Greece, burials were found in
both Mesolithic and Early Neolithic levels. Among nine ar-
ticulated Mesolithic burials two were infants (Cullen 1995).
However, in the Early Neolithic levels of the cave, eight
burials belonged to infants or children. Also, one of these
burials, a several weeks old infant, was accompanied by a
fine marble vessel and a broken-in-half clay pot (Jacobsen
1969: 373–374, 380–381, 1976: 140, 142).
In south-central Anatolia at the site of Çatalhöyük (c.
7200–6300 cal BC) infant burials were found during excava-
tions in 1960s (Mellaart 1967). Describing burials with red
ochre painting over the skull or the body (mostly females),
Mellaart (1967: 207–208) singles out a burial of a prema-
turely born infant (VI.A.14) and a female burial with a child
on top of her (VIII.1) which, among other goods in the burial,
had “fresh water mussels filled with red ochre”. Also, red
ochre covered the whole body of a young girl who had
suffered a broken femur, with addition of cinnabar paint over
 The Iron Gates in Prehistory
her skull (IX.1). Another child’s skull and the upper part of
the body were again painted with red ochre (E.IV.8). In the
course of new excavations at this site, more burials were
found underneath building floors. For instance, in Building 1
in the North Area of the site a very large number of infants,
young children and juveniles (more than half of all buried in
this building) were found (especially confined to the early
phases of the building’s use) among around 60 buried indi-
viduals (Hamilton 1996, 1997; Molleson and Andrews 1997:
fig. 32–34; Hodder 1999; Archive Reports 1996–2000, http:
//www.catalhoyuk.com/). In Building 1 under the north-west
platform, an adult burial (1924) with an infant on top of it was
uncovered (Hamilton 1996) while, similarly, in the founda-
tion layers of Building 1, a neonate (2532) was placed on the
head of an adult (2527), with a fragment of red painted plaster
found next to the head of this individual (Hamilton 1997;
Hawkes & Molleson 2000: fig. 13.6). Infant burial 2105,
found under the north-west platform, was possibly wrapped
with several hundred white and pink limestone beads
(Hamilton 1997). It was suggested that the posture of neonate
burial 2197, found buried on a grinding stone in the founda-
tion layer of Building 1 (Hamilton 1997), with legs and arms
splayed outwards, indicates that it was placed in a bag
(Hawkes & Molleson 2000: 160). Another practice related to
infants in this building is the placement of three neonates
(2199, 2197 and 2515) in a row at the threshold of the crawl
hole (connecting two parts of the room). A specific physical
intimacy is also seen in placing an infant (1450) facing an old
female skeleton (1450) (Molleson & Andrews 1997). Also,
continued work within the South Area of Çatalhöyük (where
Mellaart’s excavation took place) confirmed the same pattern
of a relatively high proportion of neonates/infants found in
buildings (Hamilton 1999). In Building 6 of the South Area,
six infants, two adults and one adolescent were excavated.
Two infants had red pigment (ochre) applied over them,
while three of them were placed in baskets. One of these
(4406), covered with ochre too, had strings of beads around
each wrist and an ankle. Another infant placed in a basket
(4927) was covered by a yellow substance (ochre?). In
Building 17, two other babies were covered one with a yel-
low substance and the other with red pigment; the latter was
placed in a basket. Also, it was indicated that in several in-
stances infants were placed behind ovens and fire installa-
tions (such as the only neonate burial in Building 18, i.e.
Mellaart’s shrine X.8). In the South Area, in Space 112
(Mellaart’s Shrine VII.9) a partly preserved infant 2017 and,
also, infant 2779 were found associated with a fire installa-
tion. Within the same space, again, the posture of infant
burial 2728 indicated that it was possibly placed in a bag
prior to its burial. Within Space 109, a neonate (2772) was
placed between the platform and the fire installation. Its legs
were splayed and it was overlaid with a horncore (Molleson
et al. 1998). It is interesting to note that some of the burials
with traces of chronic bone pathologies had red ochre applied
to their skulls (Molleson et al. 1999). A very similar pattern
of a relatively large number of buried infants, children and
juveniles has been confirmed also in the course of the ongo-
ing excavations of Building 3 (BACH Area) at the site
(Hamilton 1999; Molleson et al. 2000).
At sites of the Natufian culture, such as Mallaha (Eynan)
160
(Boyd 1995), infant burials were found associated with ar-
chitectural features too.
There is an important connection in the accentuated ap-
pearance of infant burials within elaborated
structures/buildings, spaces with both domestic and sacred
elements, across the eastern Mediterranean in the period prior
to and in the course of the development of the Neolithic phe-
nomenon. All the differences and specific developments in-
volved in these shared practices and meanings would merit a
much more detailed discussion elsewhere. Now, we need to
go beyond the temporal and spatial context of our case study
and beyond archaeological vestiges that limit accessing im-
material aspects of social life in their contextual richness and
diversity.
Possible meanings: Balkan ethnographies and
beyond
It does not require moving far from the hinterlands of the
Danube Gorges and the central Balkans to start an ethno-
graphic pursuit for the meanings of infant burials at Lepenski
Vir. The ethnography of the Saracatsans and the Vlachs, pas-
toralist groups of the various mountainous regions of the
Balkans (eastern Serbia, Thrace, Macedonia, Epirus,
Thessaly), who culturally (and possibly genetically) repres-
ent one of the oldest surviving populations of the Balkans, is
extremely rich (Høeg 1925; Kavadias 1965; Campbell 1964;
Antonijević 1982; for a general history of the Vlachs see
Winnifrith 1987). In our opinion, this ethnography carries
specific significance for understanding numerous practices
and patterns seen throughout the prehistory of the Balkans.
Although it would be difficult or even impossible to pin down
the ways of transmission of distant cultural practices and
memories through narrative and material forms or to follow
their genealogies in fine detail, it seems that in many in-
stances the former practices stay rooted in the local land-
scapes long after the generations of past inhabitants are gone
and their dwelling places abandoned (cf. Connerton 1991;
Fentress & Wickham 1992). We are left with two legitimate
possibilities in this context, that is either to create imaginary
genealogical narratives (cf. Tringham 2000: 126)8 or to study
rich details of local ethnographies against archaeological ex-
amples in a more conventional way. We explore the latter
possibility here.
The religious beliefs of the Saracatsans and Vlachs are
overwhelmingly occupied with issues of death and protection
from evil forces. Their mythology, although under strong in-
fluence from Christianity, took a specific eclectic form with
numerous hybrid features. Firstly, there are intrinsic elements
that connect domestic hearth and fire to mother and new-born
child in the ethnography of these groups. Thus, among the
customs of the Saracatsans a woman would give birth on the
ground next to the hearth and fire, which is believed to protect
the mother and the child. Another noted custom is to draw a
hearth on the body of a new-born child with a stone, previ-
ously put in fire. These customs, together with many others,
may fall under the rites of protection and it has been noted for
some time now that among the Bulgarians and the Slavs in
general, but also other European peoples, the emphasis was
 Sofija Stefanović & Dušan Borić: New-born infant burials at Lepenski Vir
frequently placed on practices that serve to shelter pregnant
women, foeti and children from the dangers of malevolent
forces (van Gennep 1960: 45). For these reasons, a number of
rites of separation and incorporation that frequently relate to
buildings, i.e. to the time of their occupation, show incredible
richness across ethnographic contexts. The main aim of these
practices is, first, to introduce a woman and the society into
the ‘abnormal’ state caused by a woman’s pregnancy and,
after the birth, to slowly incorporate the mother back into the
normal flow of life, as a “social return from childbirth” (van
Gennep 1960: 46).
However, in the event of a new-born’s death, more dra-
matic practices may take place. According to one recorded
custom that was widespread among the Saracatsans up to
1940 (Hatzimichali 1957: 68, cited by Antonijević 1982:
134), in the event of the death of a new-born or a miscarriage
the child would be put in a rawhide bag full of salt and hung
above the parents’ marital bed for approximately forty days,
until the child’s corpse dried out. A rite of separation of the
mother (she could not leave the hut during this period) from
the community is involved in the custom. Also, every morn-
ing, the mother would draw a cross (a sign believed to pos-
sess special protective powers) on the earthen floor of the hut.
After this period the corpse of the deceased child in the bag
is buried inside the hut, in a burial pit dug in one corner.
This custom strikingly resembles the pattern of the previ-
ously described infant burials from Lepenski Vir. But apart
from a possible formal similarity, we may discern elements
that can be of wider importance. Firstly, the burial takes place
inside the hut, where life continues, and may relate to a belief
that by this practice the fertility of the conjugal couple would
be increased and facilitate a future delivery of a healthy child.
The practice of keeping the deceased child in the domestic
hut and spatially close to the parents for forty days (but also
its subsequent burial inside this space) indicates a strong tie
of the deceased child with the parents, and especially the
mother. It is possible to interpret this as an act of specific care
and solidarity, among the Saracatsans and the Vlachs very
pronounced in relation to all the dead (Antonijević 1982:
137) and specifically in relation to children. Secondly, al-
though the salt used in this custom may have the practical
function of embalming the child’s body to some extent
(Antonijević 1982: 135), in relation to local beliefs the salt
also has a strong demonic power and through its use positive
contact would be established with the demons (on the similar
significance of salt in other parts of the eastern
Mediterranean, see Welters 1999c: 64). Hence, the issue of
protection is involved again. The third element in this partic-
ular ethnographic example that may be of some relevance for
our archaeological case relates to the meaning of the earthen
floor of the hut under which the corpse of the deceased infant
is buried. As the earth may be conceived of as the place of the
dead, by burying the deceased through/under the earthen
floor of the hut the deceased infant is returned to the place of
its origin.
Van Gennep (1960: 52) points out that similar practices of
protection in relation to mother and infant may relate to the
fact that the infant who dies before its introduction to and in-
corporation into both social and celestial worlds lacks spir-
itual powers, i.e. a ‘layered’ social persona gained through
161
social experience. It relates primarily to the age of an infant,
and across ethnographic examples world wide (including the
Saracatsans and the Vlachs) special care is expressed in rela-
tion to the liminal (transitional) period after the birth. Thus,
by customs such as preserving the umbilical cord or placenta
after the birth and burying them in a distant place (where
nobody can touch it) or under the threshold, hearth, etc., it
was intended to preserve the child’s personality, its soul (van
Gennep 1960: 52, see below). In various ethnographic ex-
amples, the crucial emphasis is on the time that elapses be-
fore an infant acquires full individuality, which can take days
or weeks after the birth, and sometimes also can be relevant
much later throughout the period of the child’s growth toward
adulthood. During this whole period, the child remains
highly vulnerable and different rites of protection are needed.
This may equally apply to the soul of the deceased infant that
requires special care in the liminal/transitional phase on the
journey to the underworld. Another significance of the cus-
tom whereby the deceased new-born stays above the marital
bed for forty days may relate to the belief held across the
Balkans that forty days represent the first phase on the way to
the underworld, a voyage that ends only after one year (e.g.
Bandić 1997). For instance, a crucial event after birth that
determines the treatment of an infant among the Slavic in-
habitants of the Balkans and Eastern Europe relates to
whether the infant is baptized or not at the moment of its
death (Čajkanović 1924: 60-61, 128).
Beliefs among the Saracatsans and the Vlachs focus on the
supernatural and on practices that involve black magic, with
a specific belief about the ‘evil eye’ as a source of danger
(Antonijević 1982). The belief in a harmful ‘evil eye’ (espe-
cially threatening those in liminal stages of their lives, such
as new-borns, newly-married, pregnant women, etc.) is
widespread across the Mediterranean (cf. Welters 1999c:
64ff.; Mladenović 1999: 101ff.). As already emphasized
above in the discussion on the significance of the representa-
tion of eyes on the artworks of Lepenski Vir, these orifices of
the body can both expose the hidden interior of the
being/mind (Gell 1998: 147) but also penetrate the body and
possibly harm it.
The world of the Balkan pastoralists and their whole
mythology can be described as inhabited by demons, ghosts,
monsters and fairies, with both malevolent and benevolent
intentions, although the Saracatsans believe that malevolent
forces are dominant. Such forces inhabit the landscape and
are present in the air, mountains, rocks, streams, wells, trees,
leaves — almost everywhere – and are called different names
and possess powers that can be in varying degrees harmful to
human health, good fortune or life (Antonijević 1982:
162–170).
It has been recorded that in the region of Homolj (eastern
Serbia) among the Christian Serbs, during the periods when
frequent deaths of new-borns occur in a buildinghold, imme-
diately after a birth, a midwife takes the new-born and passes
it though the circle frame of a wheel, sways it three times to-
ward the east and asks the mother while standing on the
building’s threshold (where the ancestors of the building
reside): ‘Do you like the sun or the moon?’. And the mother
each time answers from the building: ‘the sun’ (Čajkanović
1924: 130), i.e. this world. In the mythology of the Balkan
 The Iron Gates in Prehistory
pastoralists, the moon, although sometimes signifying fertil-
ity, equally relates to the underworld since it goes through
periodical changes, indicating the termination of life cycles
and rebirth (Antonijević 1982: 60–61). The practice of
passing the child through, in the previous example, indicates
the performance of an early, ‘prematurely’ evoked rite of
passage to achieve the ‘transference of evil’ (cf. van Gennep
1960: 52, 59). It ensures the child’s smooth incorporation into
the society as the first step in preparation of later rites of
passage on the way to gaining a personhood.
However, a different connotation of the rites of protection
among Balkan pastoralist groups, but also the Slavic inhabit-
ants of these areas, in relation to new-borns and pregnant
mothers is the notion that these individuals can be dangerous
for other members of the society (Bandić 1980, 1997). This
comes from the ambiguous associations of pregnant women
and infants who, although vulnerable and in need of protec-
tion, can be of considerable danger for the community, caus-
ing serious illness and even death. In order to protect the
collectives, a number of customs relate to the isolation of a
mother and child during the period of pregnancy and for
some period after the birth (Bandić 1980, 1997).
The custom of presenting children to ancestors of the
building among some groups (van Gennep 1960: 61) is
grounded in the notion of apotropaic powers held by the an-
cestors, that can protect the child. In Balkan ethnographies
the threshold and the hearth are the main places in a building
where the spirits of the ancestors reside (e.g. Čajkanović
1924: 127ff.; for a cross-cultural example of the significance
of the threshold, see Eliade 1957: 25–27, 181).
The cultural attitudes toward pregnant mothers, new-born
babies and deceased infants in the aforementioned examples
from Balkan ethnographies indicate a very strong emphasis
on rites of protection. Similarly, a recent comparative study
of folk dress in numerous examples across the Balkans and
Anatolia, also explicitly emphasizes the obsession with the
intertwined issues of protection and fertility, especially re-
lated to mothers and children (Welters 1999a, 1999b).
Going beyond the Balkan ethnographies, there are two
relevant issues that can enrich the present discussion with a
wider cross-cultural and comparative perspective. The first
concerns the powerful place of architecture, domestic spaces
and buildings in transcending issues about protection and
apotropaic potency anchored at these places. As already
noted, if we accept that buildings objectify social units, serve
the role of outer shells to descent groups that belong to these
places, both dead and alive, reminding of complex biograph-
ies of places and past inhabitants, their multifaceted nature
reaffirms the significance of the notion of ‘building societies’
formulated by C. Lévi-Strauss (cf. Carsten & Hugh-Jones
1995; Gillespie 2000a). Houses encapsulate tensions of
complex social webs of blood and affinal kin relations and
transcend long memories, in a way fixing the ‘essence’ of a
descent group. Still their gendered and androgynous images
remain unfixed as they move through transformations that
alter their ‘essence’, inseparable from the flux of the lives of
people that inhabit or visit these places.
The building of a new building is the birth and growth of
an objectified descent group and this notion is transparent in
the rituals involving the placenta of new-born babies. For in-
162
stance, in Tana Toraja, South Sulawesi, Indonesia, through
successive generations, fathers would bury placentas (seen as
a ‘twin’ to the baby) of new-born babies always at the same
spot at the east side of a building, which is associated with
life and the rising sun (Waterson 1990: 198, 2000: 180, 182).
In other Indonesian societies, such as the Tetum, the umbil-
icus and placenta would be placed in a bag and hung on the
central (ancestral) pillar of the main room in a building.
Similarly, in Tanimbar the placenta would be buried in the
building floor (Waterson 2000: 180). In another example
from the same region, in the village of Ara, among the
Makassarese of South Sulawesi, Indonesia, the birth is sur-
rounded by extensive magic rituals intended to facilitate the
delivery and protect the baby (but also the mother) from evil
forces and spirits that can harm the baby while still “in the
extremely vulnerable condition of having an unhealed navel
and soft fontanelle” (Gibson 1995: 137). The following birth
customs make a reference to the building:
“The bloodied banana leaf on which the child was born is folded up
and placed in a bamboo pole with a basket at one end called a tom-
pong. ... The tompong is planted in the ground beneath the building
with its ‘mouth’ open to the sky. It is thus able to catch the dalle,
‘good fortune’, that descends from heaven. ... The tompong is left
under the building for seven days, by which time the child’s navel is
healed and the next ritual can be performed. During this time, the
child must remain in the building, and the mother is not supposed to
leave it either, for fear her ‘open’ state may attract evil spirits to enter
her which she would then bring back with her into the building.”
(Gibson 1995: 136–137)
In Ara, a number of apotropaic rituals that are performed
in the construction, i.e. ‘closure’ of the building are closely
analogous to those rituals that are performed with the new-
borns in attempts to ‘close’ their navel and fontanelle (Gibson
1995: 144–145).
Away from Indonesia, among the Mam Maya of
Chimaltenango in highland Guatemala, to bury the afterbirth
in the father’s sweat building (the focal ritual structure of a
family) emphasized the patrilateral extended family in which
future ritual activities would take place (Gillespie 2000b:
219, note 8).
Differently in terms of kinship associations, among the
Zafimaniry of Madagascar a woman, who is pregnant for the
first time, leaves the husband’s building and goes to her natal
building. Here, another protective layer is made by creating a
‘building’ on the bed after the woman gives birth. The child
is connected to the mother’s building as the ritual of putting
the soot from the hearth of the building on the child’s fore-
head and burying of the placenta (here, as in Indonesia, be-
lieved to represent a twin) are done in the mother’s parent’s
building. Only the third child would be born in the parents’
own building and through the same ritual materially associ-
ated with that building, which at the same time becomes
complete (Bloch 1995: 77). In the same society, Bloch (1995:
83) records an instance where the ill new-born was cured
through a ritual medicine soup cooked on the place of the
hearth of the holy building (where previously the head of the
village was buried) by feeding the baby with the soup and
pouring it in places where the posts once stood.
Frequently, reasons to abandon a building are related to
events that are interpreted as the result of misfortune. Thus,
among the Ye’cuana of Venezuela, the building is abandoned
if either illness or the deaths of infants or a village leader is
 Sofija Stefanović & Dušan Borić: New-born infant burials at Lepenski Vir
associated with the building. Also, the deceased can be buried
under the floor of the building (Rivière 1995: 197).
The second relevant cross-cultural issue relates to the so-
cial construction of age and gender in human societies. As the
examples from Lepenski Vir indicate, there was a significant
difference in the way that neonates of the same age are buried
in comparison to the burial rite that accompanied other age
groups. As emphasized, the construction of these age (and
possibly gender) differences (Borić 2005) might have been
inscribed in the boulder artworks found at this site.
It is of relevance here to describe the notion of a ‘boneless’
baby that comes from the Vezo of Madagascar (Astuti 1998).
The Vezo think that people are not born as humans, but they
become humans, with a rather non-essentialist view on iden-
tity. They see new-born babies as strongly tied with mothers,
vulnerable and not fully human. Moreover, new-born babies
are malleable and plastic, especially in regard to their facial
features, and their bodies are soft and weak, i.e. ‘boneless’,
and only have the potential to become ‘fully-boned’ adults
(Astuti 1998: 36). Both mother and baby are vulnerable after
the event of birth as their bodies can be penetrated by the
deadly ‘air’ and the babies must be protected by wrapping up
in layers of clothes (Astuti 1998: 35). The babies must not be
left alone as they are unprotected from the harmful influence
of passing spirits known as angatse, and the angatse are the
reason that eyelids of babies tremble and roll sideways every
time the spirits are around. Another kind of danger comes
from vengeful ancestors who can be bad-tempered and can
easily harm a ‘boneless’ baby. If the baby dies before it is one
year old, it cannot be buried in a family ancestral tomb as it
has not become a human, i.e. its social person has not been
‘created’ yet (Astuti 1998: 36–37). It is interesting to note
that in the case of the Zafimaniry of Madagascar, buildings,
like people, also acquire ‘bones’ through their slow build up
with more permanent material (from woven bamboo to
massive wooden planks) and elaboration of specific decorat-
ive wood carvings, where the process of ‘beautifying’ the
timbers forms a part in the process of the growth and suc-
cessful marriage of the founding couple (Bloch 1995: 78ff.).
The views held by the Vezo in regard to infants are in-
structive and this example triggers a more complex discus-
sion on the way the social person is constructed in different
cultural contexts, relations of sex and gender, individuality
and dividuality of personhood, etc., which we do not pursue
directly here. However, this example shows some aspects of
a non-essentialist view of identity that the Vezo themselves
maintain, related to both collective and individual self-
realization, with the body as the arena of this expression. It
points to a constant flux of becoming in the construction of a
social person, in this respect similar to the way the
Melanesian personhood is constructed (cf. Strathern 1988;
LiPuma 1998).
Conclusions
The study of burials of neonates and children from Lepenski
Vir, together with apparent diachronic changes at the micro-
regional level in burial rite and its elaboration in relation to
these age groups at the neighbouring and earlier site of
Vlasac, brings an invaluable corpus of data that triggers
163
questions of much wider significance. We have tried to trace
an ‘ethnography’ of neonates’ and children’s burials in the
Upper Gorge of the Danube that probably involves at least
part of the timerange from 9500 to 5500 cal BC.
We can point to several important findings that the data
and their contextual analyses imply:
1. On the basis of osteometric analyses of 38 infant burials
found underneath the floors of buildings at Lepenski Vir, it
is possible to confirm a high peak of individuals aged at
38–40 gestational weeks (± 2 weeks) (see Fig. 2), indicat-
ing that most if not all of the infants buried underneath the
rear of building floors at this site were neonates, i.e. new-
born babies who died soon after birth.
2. The neonates were confined to 19 (out of c. 73) buildings
at Lepenski Vir, with the addition of two younger children
— Burials 97 (House 31) and 92 (House 28) — buried in
two other buildings in a similar way. As with other burials,
neonates and infants are absent from groups of buildings
in the north east of the site. Houses accommodated from
one to six neonate burials and there was no clear pattern-
ing related to the deceased’s sex, as mostly individuals of
both sexes were found in buildings with more than one
interment.
3. Almost all neonates were buried in the rear of buildings,
either through their floors or immediately next to the
limestone floors into the virgin deposits of the loessic
slope into which the trapezoidal buildings were dug. The
only exception, Burial 63 (House 28), relates to the un-
usual orientation of this building, indicating that the topo-
graphic perspective was of considerable significance con-
cerning the burial rite of this age group. On the basis of
present findings, we reject the possibility that the infant
burials were interred before the building floors were fur-
nished (contra Srejović 1969a: 136; Srejović 1972: 119;
contra Radovanović 2000: 340, note 7).
4. There were no grave offerings in neonates’ burials. On the
basis of the varied positions in which they were found, and
with the help of the perspective of re-enacting the buried
body, we suggest that some, if not all neonates, and pos-
sibly younger children as well, might have been buried in
bags. These were buried in small burial pits, sometimes
covered with stones and stone slabs, or in one case with
part of the floor. Both their positions and orientations var-
ied even within the same building.
5. In a number of buildings with neonate burials, both artic-
ulated burials on the floor level and, especially, disarticu-
lated bones of adult individuals were found in the area
around the hearth of the building. In the latter case, these
were detached skulls and mandibles, but also other disar-
ticulated bones of the postcranial skeleton. We suggest
that these bones might have encapsulated powerful, apo-
tropaic forces and agencies that might have served varied
purposes in the lives of the inhabitants of, and/or visitors
to, these spaces, and possibly specifically relating to the
buried remains of the neonates. However, significantly, no
disarticulated adult (ancestral?) bones were found in the
neonate burials.
6. Apart from the neonates and children, only three other
burials at Lepenski Vir (7/I, 26 and 63) were dug through
the building floor. Strikingly, all three were placed in ex-
 The Iron Gates in Prehistory
tended positions on their backs oriented with the heads
pointing downstream, and behind/next to the rear of the
hearths of the buildings, respectively. We suggest that this
patterning in their placement significantly relates to their
age (probably socially fully embodied persons), hence the
difference compared to the neonates’ and younger chil-
dren’s burials. However, rites involved in their burial
might have carried the same prophylactic concerns toward
the deceased’s corporeal selves, similar to the neonates’
burials. Incorporation of ‘ancestral’ bones, such as the
skull (Burial 7/II) in Burial 7/I points in this direction.
7. There were no representational/decorated boulders
(mainly concentrated around hearths) commemorating
any of the buried neonates. In contrast, above two- to
three-year-old child Burial 92 two sculpted boulders were
found, one with a unique iconic expression (visible but
‘closed’ eyes — carved horizontal lines), possibly related
to his unfinished social embodiment. Significantly, Burial
63 (House 40) of an approximately seven-year-old child
was commemorated with a small boulder, representing a
(child?) face with eyes ‘opened up’ as engraved circles, as
on all other representational boulders (cf. Srejović &
Babović 1983), possibly indicating stages of socially in-
scribed personal embodiment.
8. At the neighbouring site of Vlasac, there are proportion-
ately fewer infants and children among the discovered
burials (Borić & Stefanović 2004). On the basis of osteo-
metric parameters there is no dominant infant category
expressed in gestational weeks. A number of foetuses and
neonates were found, in several cases spatially closely tied
to adult individuals, possibly mothers (also within the
mother’s pelvic area), indicating miscarriages that may
have caused the death of the mother too. A number of
whole infant burials and frequently groin areas of females
were sprinkled with red ochre and/or in several cases fish
(Cyprinidae) teeth. The red ochre might have been of
apotropaic significance. We may assume that there is in-
trinsic, if only not fully consciously-made, connection
between the ochre of red colour used in Mesolithic burials
at Vlasac and the red colour of later limestone floors at
Lepenski Vir. The Vlasac examples indicate significant
diachronic changes within the same cultural milieu.
9. The neonates and children discovered at Lepenski Vir
were buried through already existing building floors.
Considering a number of radiometric dates for the occu-
pation and wooden posts of upper structures of these
buildings, we may assume an Early Neolithic date for the
discussed burials, i.e. 6300–5500 cal BC. This is also
confirmed by the reported presence of fragments of Early
Neolithic Starčevo-type pottery in the hardly disturbed
context of neonate Burial 113 (House 63′). This conclu-
sion can be of significance in the attempt to situate histor-
ically (and possibly culturally) the particular phenomenon
of infant/child burials and buildings, i.e. their placement
within the elaborated domestic space. This is related to a
number of examples of infant/child burials found under-
neath building floors across Pre-Neolithic and Early
Neolithic societies of the eastern Mediterranean.
10. Cross-culturally, C. Lévi-Strauss’s notion of ‘building
societies’ brings fruitful perspectives in considering the
164
connection of the paradoxical and illusory struggle to
achieve ‘fixed’ descent groups, objectified through build-
ings (more precisely through the acts of their
creation/becoming), and the ways in which social persons
are constructed in a society, involving negotiation of per-
sonal and group identities and their engendered embodi-
ments. These issues are central to the practices that sur-
rounded the burials of infants and other age groups at
Lepenski Vir.
In addition to the ‘exclusivity’ of the conclusions listed
above, we offer several less sharply defined concluding
comments on the basis of the forgoing discussion.
In his discussion of Lepenski Vir, Ian Hodder (1990: 25)
indicates that death dominates the buildings of this site. In his
opinion, at Lepenski Vir, like Çatalhöyük, ‘the domestication
of death’ takes place by bringing the dead body into the
building (Hodder 1990: 29). The power discourse is involved
in this interpretation and the dead body is controlled and ac-
ted upon. Hodder (1990: 40) sees a significant relationship
between the dying and decaying individual body and more
permanent social units. As Meskell (1996: 7) argued in her
critique of the social constructionists’ view of the body, es-
pecially in relation to the works of Michel Foucault, this is
“the body as the scene of display”. Meskell (1996: 9) sug-
gests that archaeological inquiries into lived and experienced
bodies should escape the discourse of domination and con-
trol, something that was so frequently applied in interpreting
evidence from the Danube Gorges and Lepenski Vir (e.g.
Handsman 1991; Chapman 1993, 2000). The emotional force
that might have surrounded the death of a new-born baby in
a building of Lepenski Vir, causing feelings such as grief,
bereavement and rage (cf. Rosaldo 1989[1993]: 1–20), al-
though not directly penetrable for an archaeological study,
must be emphasized, indicating a possible porosity of all
previously established ‘fixed’ cultural patterns.
We offer an alternative to the discourse of power and con-
trol over both the dead and the living body. We emphasize
that the proliferation of neonate burials in buildings of
Lepenski Vir expresses particular care in the face of a dis-
rupting existential anxiety of facing the death of a new-born
baby. On the other hand, it is believed that the deceased in-
fant, already partly embodied, needs the protection of the
building, ancestral powers and apotropaic potencies
anchored in these structures, their hard limestone floors,
hearths and sacred heirlooms, such as sculpted boulders.
Personal embodiment probably played a significant part in
determining the rites performed at Lepenski Vir. For instance,
several older age burials placed within cuts made through the
building floor were close to the hearths and were literally ac-
companied by ‘ancestral’ bones (Burial 7/I), while neonate
and child burials were always placed away from the hearths,
avoiding their direct contact with ‘ancestral’ bones. This may
indicate a realization of a harmful influence of a direct phys-
ical contact between the deceased neonates and younger
children with possibly easily angered ancestors (cf. Astuti
1998).
We may assume that neonates and very young infants at
Lepenski Vir were not born with an already formed social
persona. It is possible that individuals who died immediately
after birth were seen by society as not fully human and, had
 Sofija Stefanović & Dušan Borić: New-born infant burials at Lepenski Vir
they lived longer, only with the passage of time would they
have layered their respective social personae, creating the
‘essence’ of their selves and socially formed personhood.
Still, there must have been some ‘essence’ of their embodied
selves that was socially ‘invested in’ by burying them within
buildings. Future work on the DNA-based formulation of re-
latedness among the burials of Lepenski Vir will provide
further clues to the direction of what influenced the structur-
ing of infant and other burials at the site in particular
buildings.
It may also be that although perceived as humans,
neonates, infants and younger children were not considered
capable of being protected on a dangerous journey to the un-
derworld. This brings us to the question, what elements might
have constituted a person at Lepenski Vir? Some of the eth-
nographic examples (e.g. Astuti 1998; LiPuma 1998) as well
as those archaeological examples aided by textual informa-
tion (e.g. Meskell 1996) indicate that persons might have
been “multiply constituted throughout life and death and that
the spiritual and corporeal selves were inextricably bound,
rather than the Classical notion of the soul as entity im-
prisoned within flesh” (Meskell 1996: 13). This perspective
gives significant weight to the materiality of the dead body as
simultaneously it belongs both to the world of the living and
to the ‘other’ world. As the body is invested through varied
forms of social actions (e.g. initiation rites, tattoos) during its
life, and rites that transform and guide the individual through
the stages of social becoming (cf. Gell 1998: 86ff.), it re-
mains similarly important to help the deceased body, inex-
tricably connected with immaterial components that consti-
tute it, to pass the thresholds of its metamorphosis in death.
Although Meskell (1996: 11) critiques the preoccupation
of current archaeological discourse with the exteriority, sur-
faces, treatment, elaboration and decoration of the body,
which in her opinion neglects the embodied individuality,
exactly these practices are central in the creation/becoming
of the socially embodied person. All these practices of layer-
ing and wrapping of the human body (also by the building)
bring to the forefront the processes of becoming and indicate
the apotropaic salience of cultural practices.
Particular examples of burials at Lepenski Vir may indeed
indicate “drama and performance of some plot” (Hodder
1990: 29). This drama may be obvious in adult Burial 7/I as
it was accompanied by animal and human skulls that might
have been intended to aid his journey to the underworld. We
may only assume that this individual might have died in a
violent or unusual way, not ready for the perils of the journey
to the ‘other’ world, desperately needing the guiding of this
possibly ‘ancestral’ skull in particular. It seems that similarly
the burials of infants and children at Lepenski Vir needed the
sheltering proximity of the building and anchored potencies
that were spatially played out in a specific way for this age
group.
As Balkan and other ethnographies inform us, a safe jour-
ney to the underworld was not of importance for the deceased
only but also for the living, as the spirits of those deceased
that are trapped in the liminal plane of existence, between the
world of the dead and the living, can be of a considerable
danger to the living.
A
The notion of apotropaism underlying a number of cultur-
165
al practices is emphasized in the offered interpretation of
neonate and infant burials at Lepenski Vir. Prophylactic con-
cerns related to the human body are also significantly con-
nected to the construction and elaboration of building and
domestic area. This may be related to the pronounced em-
phasis on the descent-group and concerns on its ‘fixation’,
historically constituted (Borić 2003). However, we must not
stay detached, only looking at the transcended cross-cultural
issues of political and social strategies at large. It is also ne-
cessary to approach empathy, admitting that practices such as
events of infant burials at Lepenski Vir, were constituted by
acts that are individual, emotional and experiential (cf.
Meskell 1994). Thus, the ontology of care is highlighted in
this interpretation, connecting concerns of larger social
structures, buildings as objectified lineages on one side, with
very individual and ‘noised’ actions of experiencing and
emotionally driven individuals on the other. People at
Lepenski Vir intentionally engaged in significant repetition
of cultural practices, subscribing to and believing in their
potencies.
Notes
1. The archive field documentation used is stored at the Centre for
Archaeological Research of the Department of Archaeology,
Faculty of Philosophy, University of Belgrade. The late Prof. D.
Srejović, principal excavator of Lepenski Vir, was affiliated to
this Department. Permission to use the archive documentation
was kindly granted by the Head of the Department of
Archaeology, Prof. Živko Mikić.
2. The floors of almost all excavated buildings were relocated in the
course of protection work in 1970 (Čanak-Medić 1971). It is
important to note that the floors of Houses 13, 18, 27, 32, 37, 48
and 51 were moved as complete blocks (whereas some of the
buildings owing to their large size were cut into several blocks),
together with c. 0.75 m thickness of the deposits underneath the
floor (Čanak-Medić 1971: 14). Although infant burials were also
found underneath some of the buildings moved in this way (13,
27, 37 and 48), one could speculate that those buildings like
Houses 18 and 51, where the whole building was moved in a
single block, still conceal unexcavated infant burials together
with other finds.
3. It is not possible to indicate the exact number of buildings for
several reasons. First, some of the open-air hearths (e.g. Hearth
a and d, see Fig. 1) have no floor around them and perhaps are
much older than the later limestone floorings (Srejović’s
Lepenski Vir I phase) and although these do not correspond to the
notion of an elaborate ‘house’, they may indicate a domestic
space. Second, in the cases when an older feature is overlaid with
a later flooring (e.g. Houses 62/62′, 63/63′, 37/37′, 47/47′ and
26/26′; see Fig. 1) it is difficult to specify if the earlier features
should be regarded as separate buildings. And third, as we largely
reject Srejović’s Lepenski Vir II phase (Borić 2002a) the separate
Roman numbers that he assigned to the phase II buildings do not
indicate separate buildings but dry stone walls around the cuts of
Lepenski Vir I buildings (see discussion in the text).
4. The only exception to this is Burial 83a–b. This burial was found
in the occupation layer (quad. d/II, next to the corner of point c,
excavation level XI, in the virgin soil; Field Diary 30/06/1970).
Burial 83a is an older child (around 5 years old) placed in a con-
tracted position, NW–SE orientation. Several potsherds, two
pieces of flint, and snail shells were found in the burial. Together
with this individual, bones of two infants (83b and 83b(1); see
Table 1) were also found, partially preserved. After a taphonomic
 The Iron Gates in Prehistory
examination of the bones from this burial by one of us (S.S.), it
was noted that their fossilization (colour and bone consistency) is
markedly different from the rest of the infants at this site.
5. There is again no clear-cut pattern in the relationship between
these supports and buildings with infant burials. For example, in
some buildings, such as Houses 24, 47, 27, 43, 19 and 4, there
were a number of these supports surrounding the hearths and, at
the same time, there were from one to four or more new-borns
buried in each building. On the other hand, infants were also
buried in buildings with no supports found around the hearths
(Houses 54, 37, 38, 29, 62, 26, 36 and 48). We may simply ob-
serve that relatively more infants were buried in buildings with
more hearth supports (Houses 19, 4, 24 and 27). Srejović (1969:
140–141, 1972: 121–122) suggested that these supports stand for
the deceased members of a building regardless of whether or not
they were buried in the buildings. In his opinion, the community
cared for the deceased who were represented by the supports (as
schematic mandibles) and by placing the supports around the
hearth the deceased would be able “to enjoy the warmth of the
hearth and the food prepared on it” (Srejović 1969: 141, 1972:
122). Radovanović (1996) used the supports and the form of
building thresholds as the main architectural features of the
buildings for her re-phasing of Lepenski Vir’s stratigraphy, as-
suming uniform change through time reflected especially
through the quantity of these supports around hearths. In her di-
vision, buildings with a large number of hearth supports would
represent chronologically later features, while their smaller
number or absence would suggest older buildings. This can only
be partly relevant as it is obvious from the published photographs
that a number of buildings that once had stone plaques were left
with triangular holes as the stone plaques were removed (Borić
2000b). Although some profane function for these supports can-
not be excluded (cf. Jovanović 1969), it may be that they were
not placed once and forever and their presence or absence to
some extent could indicate the changing status of a building in its
life history, while indeed they can perhaps be connected to the
burials and/or the deceased members of a building (Borić 2003,
2005).
6. Ivana Radovanović (1999: 74) discusses Burial 70 in connection
with dog ‘burials’, i.e deposition of dog bones at Lepenski Vir
and Vlasac. Radovanović incorrectly attributed, discussed and
phased this burial in relation to House 32 at Lepenski Vir, al-
though it was actually found in House 36 (see Radovanović
1996: 185).
7. A similar practice of turning the head to the east, i.e. to the
Danube, is expressed differently in relation to Burial 61 in House
40. Although the skull of this burial is in the anatomical position,
the sculpted boulder in the form of a human head that was placed
directly above the skull of the deceased child was turned to face
the Danube (cf. Srejović 1969a: fig. 65–66; Srejović & Babović
1983: 108) in the same way as the skull of Burial 26.
8. Since the local Vlachs represented the excavation team in the
course of excavation work at Lepenski Vir, this fact could have
been used to view their interaction with and interest in the ex-
cavated residues of the past excavated in their local surround-
ings. This possibility is now largely lost.
Acknowledgements
We are grateful to Prof. Dr Živko Mikić for his kind permission to
work on the infant burials from Lepenski Vir. We also thank the or-
ganizers and participants of the conference, The Iron Gates in
Prehistory (Edinburgh, 31 March–2 April, 2000) for their reactions
and comments on the presentation of this paper. For valuable com-
ments and information we would like to thank Katarina Novaković.
A
166
References
Antonijević, D. 1982: Obredi i običaji balkanskih stočara.
Belgrade: Balkanološki institut.
Astuti, R. 1998: ‘It’s a boy’, ‘it’s a girl!’: reflections on sex and
gender in Madagascar and beyond. In Lambek, M. & Strathern,
A. (eds) Bodies and Persons. Comparative Perspectives from
Africa and Melanesia. Cambridge: Cambridge University Press,
29–52.
Babović, Lj. 1997: Položaj i funkcija svetilišta na Lepenskom Viru.
In Antidoron Dragoslavo Srejović completis LXV annis ab
amicis collegis discipulis oblatum. Belgrade: Centre for
Archaeological Research, University of Belgrade, 95–108.
Bagchi, K. & Bose, A.K. 1962: Effect of low nutrient intake during
pregnancy on obstetrical performance and offspring. American
Journal of Clinical Nutrition 11: 586–592.
Bandić, D. 1980: Tabu u tradicionalnoj kulturi Srba. Belgrade:
Beogradski izdavački zavod.
— 1997: Carstvo zemaljsko i carstvo nebesko. Ogledi o narodnoj
religiji. Belgrade: Biblioteka XX vek.
Bass, W. 1987: Human Osteology: a Laboratory and Field Manual
(3rd edition). Columbia, Mo: Missouri Archaeological Society.
Benac, A. 1973: Obre I. A Neolithic settlement of the
Starčevo–Impresso and Kakanj culture at Raskršće.
Wissenschaftliche Mitteilungen des Bosnisch–Hercegowinisch
Landesmuseums IIIA: 327–430.
Bloch, M. 1995: The resurrection of the building amongst the
Zafimaniry of Madagascar. In Carsten, J. & Hugh-Jones, S. (eds)
About the House, Lévi-Strauss and Beyond. Cambridge:
Cambridge University Press, 69–83.
Bökönyi, S. 1969: Kičmenjaci (prethodni izveštaj). In Srejović, D.
Lepenski Vir: Nova praistorijska kultura u Podunavlju.
Belgrade: Srpska književna zadruga, 224–228.
— 1972: The vertebrate fauna. In Srejović, D. Europe’s First
Monumental Sculpture: New Discoveries at Lepenski Vir.
London: Thames and Hudson, 186–189.
Bonsall, C., Lennon, R., McSweeney K., Stewart C., Harkness, D.,
Boroneanţ, V., Bartosiewicz, L., Payton, R. & Chapman, J. 1997:
Mesolithic and Early Neolithic in the Iron Gates: a palaeodietary
perspective. Journal of European Archaeology 5: 50–92.
Bourdieu, P. 1990 [1970]: The Kabyle House, or the world reversed.
In Bordieu, P. (ed.) The Logic of Practice. Cambridge: Polity
Press, appendix
Borić, D. 1999: Places that created time in the Danube Gorges and
beyond, c. 9000–5500 BC. Documenta Praehistorica 26: 41–70.
— 2002a: The Lepenski Vir conundrum: reinterpretation of the
Mesolithic and Neolithic sequences in the Danube Gorges.
Antiquity 76: 1026–1039.
— 2002b: Seasons, Life Cycles and Memory in the Danube Gorges,
c. 10,000–5500 BC. Unpublished Ph.D. Dissertation, University
of Cambridge.
— 2003: ‘Deep time’ metaphor: mnemonic and apotropaic practices
at Lepenski Vir. Journal of Social Archaeology 3: 41–75.
— 2005: Body metamorphosis and animality: volatile bodies and
boulder artworks from Lepenski Vir. Cambridge Archaeological
Journal 15: 35–69.
— 2006: New discoveries at the Mesolithic–Early Neolithic site of
Vlasac: preliminary notes. Mesolithic Miscellany 18(1): 7–14.
Borić, D. & Dimitrijević, V. 2005: Continuity of foraging strategies
in Mesolithic–Neolithic transformations: dating faunal patterns
at Lepenski Vir (Serbia). Atti della Società per la Preistoria e
Protostoria della regione Friuli-Venezia Giulia 15 (2004–05):
33–80.
Borić, D. & Dimitrijević, V. 2007. When did the ‘Neolithic package’
arrive at Lepenski Vir? Radiometric and faunal evidence.
Documenta Praehistorica 35:.53–72.
 Sofija Stefanović & Dušan Borić: New-born infant burials at Lepenski Vir
Borić, D. & Stefanović, S. 2004: Birth and death: infant burials from
Vlasac and Lepenski Vir. Antiquity 78: 526–546.
Boyd, B. 1995: Houses and hearths, pits and burials: Natufian mor-
tuary practices at Mallaha (Eynan), Upper Jordan Valley. In
Campbell, S. & Green, A. (eds) The Archaeology of Death in the
Ancient Near East. Oxford: Oxbow Books, 17–23.
Brukner, B. 1960: Rezultati zaštitnog iskopavanja lokaliteta
‘Baštine’ kod sela Obreža. Rad vojvodjanskih muzeja 9: 81–111.
Brünnacker, K. 1971: Geologisch–pedologische Untersuchungen in
Lepenski Vir am Eiserner Tor. In Schwabedissen, H. (ed.) Die
Anfänge des Neolithikums vom Orient bis Nordeuropa, Teil 2:
Östliches Mitteleuropa. Köln: Böhlau, 20–32.
Campbell, J.K. 1964: Honour, Family and Patronage. Oxford.
Cárdenas, D. & Politis, G. 2000: Territorio, movilidad, etnobotán-
ica y Manejo del Bosque de los Nukak orientales. Bogotá:
Universidad de los Andes.
Carsten, J. & Hugh-Jones, S. (eds) 1995: About the House, Lévi-
Strauss and Beyond. Cambridge: Cambridge University Press.
Cauvin, J. 1994: Naissance des divinités, Naissance de l’agricul-
ture: La Révolution des symboles au Néolithique. Paris: Éditions
CNRS.
Chapman, J.C. 1993: Social power in the Iron Gates Mesolithic. In
Chapman, J. & Dolukhanov, P. (eds) Cultural Transformations
and Interactions in Eastern Europe. Aldershot: Avebury, 71–121.
— 2000: Fragmentation in Archaeology. People, Places and
Broken Objects in the Prehistory of South-eastern Europe.
London: Routledge.
Connerton, P. 1991: How Societies Remember. Cambridge:
Cambridge University Press.
Cullen, T. 1995: Mesolithic mortuary ritual at Franchthi Cave,
Greece. Antiquity 69: 270–289.
Currey, J.D. & Butler, G. 1975: The mechanical properties of bone
tissue in children. Journal of Bone and Joint Surgery 57A:
810–814.
Čajkanović, V. 1924: Život i običaji narodni. Studije iz religije i
folklora. Belgrade: Srpska Kraljevska Akademija.
Čanak-Medić, M. 1970: Projekat za spasavanje Lepenskog Vira.
Saopštenja 7: 7–12.
— 1971: Radovi na spasavanju Lepenskog Vira u 1970. godini.
Dopunska istraživanja. Saopštenja 9: 5–17.
Deleuze, G. & Guattari, F. 1984: Anti-Oedipus. London: Athlone.
Eldredge, N. 1991: Fossils: the Evolution and Extinction of Species.
New York: Harry N. Abrams.
Eliade, M. 1957: The Sacred and the Profane. The Nature of
Religion. New York: Harvest.
Fentress, J. & Wickham, C. 1992: Social Memory. Oxford:
Blackwell.
Finlay, N. 1997: Kid knapping: the missing children in lithic ana-
lysis. In Moore, J. & Scott, E. (eds) Invisible People and
Processes. Writing Gender and Childhood into European
Archaeology. London: Leicester University Press, 201–212.
Gallis, K. 1996: Burial customs. In Papathanassopoulos, G.A. (ed.)
Neolithic Culture in Greece. Athens: Nicholas P. Goulandris
Foundation, 171–174.
Garašanin, M. 1982: The Stone Age in the central Balkan area. In
Boardman, J., Edwards, I.E.S., Hammond, N.G.L. & Sollberger,
E. (eds) The Prehistory of the Balkans; the Middle East and the
Aegean world, Tenth to Eighth Centuries BC. (The Cambridge
Ancient History, Volume III, Part I. 2nd edition). Cambridge:
Cambridge University Press, 75–135.
Garašanin, M. & Radovanović, I. 2001: A pot in building 54 at
Lepenski Vir I. Antiquity 75: 118–125.
Gell, A. 1998: Art and Agency. An Anthropological Theory. Oxford:
Clarendon Press.
Gibson, T. 1995: Having your building and eating it: buildings and
siblings in Ara, south Sulawesi. In Carsten, J. & Hugh-Jones, S.
A
167
(eds) About the House, Lévi-Strauss and Beyond. Cambridge:
Cambridge University Press, 129–148.
Gillespie, S.D. 2000a: Beyond kinship: an introduction. In Joyce,
R.A. & Gillespie, S.D. (eds) Beyond Kinship. Social and
Material Reproduction in House Societies. Philadelphia:
University of Pennsylvania Press, 1–21.
— 2000b: Maya ‘nested buildings’: the ritual construction of place.
In Joyce, R.A. & Gillespie, S.D. (eds) Beyond Kinship. Social
and Material Reproduction in House Societies. Philadelphia:
University of Pennsylvania Press, 133–160.
Gimbutas, M. (ed.) 1976: Neolithic Macedonia as Reflected by
Excavation at Anza, Southeast Yugoslavia. Los Angeles: Institute
of Archaeology, University of California.
Gordon, C. & Buikstra, J. 1981: Soil pH, bone preservation and
sampling bias at mortuary sites. American Antiquity 46: 566–571.
Guy, H., Masset, C. & Baud C-A. 1997: Infant taphonomy.
International Journal of Osteoarchaeology 7: 221–229.
Hamilton, N. 1996: The figurines and other small finds. Çatalhöyük
1996 Archive Report. http://www.catalhoyuk.com/archive_reports/
— 1997: Figurines. Çatalhöyük 1997 Archive Report.
http://www.catalhoyuk.com/archive_reports/
— 1999: Figurines, burials/grave goods and beads. Çatalhöyük
1999 Archive Report. http://www.catalhoyuk.com/
archive_reports/
Handsman, R.G. 1991: Whose art was found at Lepenski Vir?
Gender relations and power. In Gero, J. & Conkey, M.W. (eds)
Engendering Archaeology: Women in Prehistory. Oxford:
Blackwell, 329–365.
Hatzimichali, A. 1957: Sarakatsanoi [Sarakatsans], 2 volumes.
Athens.
Hawkes, L. with Molleson, T. 2000: Refleshing the past. In Hodder,
I. (ed.) Towards Reflexive Method in Archaeology: The Example
at Çatalhöyük. Cambridge: McDonald Institute for
Archaeological Research, 153–166.
Hodder, I. 1990: The Domestication of Europe. Structure and
Contingency in Neolithic Societies. Oxford: Blackwell.
— 1999: Symbolism at Çatalhöyük. In Coles, J., Benley, R. &
Mellars, P. (eds) World Prehistory. Studies in Memory of
Grahame Clark (Proceedings of the British Academy 99).
Oxford: Oxford University Press, 177–191.
Hodder, I. & Cessford, C. 2004: Daily practice and social memory
at Çatalhöyük. American Antiquity 69: 17-40.
Høeg, C. 1925: Les Saracatsans. Une Tribu nomade Grècque.
Paris–Copenhague.
Hourmouziadis, G. 1973: Burial customs. In Theocharis, D.T. (ed.)
Neolithic Greece. Athens: National Bank of Greece, 201–212.
Jacobsen, T.W. 1969: Excavations at Porto Cheli and vicinity, pre-
liminary report, II: the Franchti Cave. Hesperia 38: 343–381.
— 1976: 17,000 years of Greek prehistory. In Lamberg-Karlovsky,
C.C. (ed.) Hunters, Farmers and Civilizations: Old World
Archaeology. Readings from Scientific American. San Francisco:
W.H. Freeman, 133–151.
Johnston, F.E. 1962: Growth of the long bones of infants and young
children at Indian Knoll. Human Biology 23: 66–81.
Jovanović, B. 1969: Chronological frames of the Iron Gate group of
the Early Neolithic period. Archaeologica Iugoslavica 10: 23–38.
Joyce, R.A. & Gillespie, S.D. (eds) 2000: Beyond Kinship. Social
and Material Reproduction in House Societies. Philadelphia:
University of Pennsylvania Press.
Kavadias, G.B. 1965: Pasteurs-nomades Méditerranéens, Les
Saracatsans de Grèce. Paris: Gauthier–Villars.
Lane, P.J. 1994: The temporal structuring of settlement space among
the Dogon of Mali: an ethnoarchaeological case study. In Parker
Pearson, M. & Richards, C. (eds) Architecture and Order.
Approaches to Social Space. London: Routledge, 196–216.
Lévi-Strauss, C. 1983: The Way of the Masks. London: Jonathan Cape.
 The Iron Gates in Prehistory
— 1987: Anthropology and Myth: Lectures 1951–1982. Oxford:
Blackwell.
Lipuma, E. 1998: Modernity and forms of personhood in Melanesia.
In Carsten, J. & Hugh-Jones, S. (eds) About the House, Lévi-
Strauss and Beyond. Cambridge: Cambridge University Press,
53–79.
Makkay, J. 1983: Foundation sacrifices in Neolithic buildings of the
Carpathian Basin. Valcamonica Symposium 3 (1979): 157–167.
Malhorta, K.C. 1990: Changing patterns of disease in India with
special reference to childhood mortality. In Swedlund, A. &
Armelagos, G. (eds) Disease in Populations in Transition.
Anthropological and Epidemiological Persectives. London:
Bergin & Garvey, 313–332.
Marković-Marjanović, J. 1969: Geologija i stratigrafija. In Srejović,
D. Lepenski Vir – Nova praistorijska kultura u Podunavlju.
Belgrade: Srpska književna zadruga, 185–192.
— 1978: Geologija i stratigrafija. In Srejović, D. & Letica, Z. (eds)
Vlasac: A Mesolithic Settlement in the Iron Gates, vol. 2
Belgrade: Serbian Academy of Science and Arts, 11–27.
Mays, S. 1997: ‘New directions in the scientific study of infant
skeletons from archaeological sites’. Paper presented to the TAG
Conference, University of Bornemouth 1997.
— 1998: The Archaeology of Human Bones. London: Routledge.
Mellaart, J. 1967: Çatal Hüyük: a Neolithic Town in Anatolia.
London: Thames & Hudson.
Meskell, L. 1994: Dying young: the experience of death at Deir el
Medina. Archaeological Review from Cambridge 13: 35–45.
— 1996: The somatization of archaeology: institutions, discourses,
corporeality. Norwegian Archaeological Review 29: 1–16.
Minichreiter, K. 1999: Ranoneolitički ukopi i pogrebni običaji u
naseljima starčevačkog kulturnog kompleksa. Prilozi instituta za
arheologiju u Zagrebu 15/16: 5–17.
Mladenović, V. 1999: Threads of life: red fringes in Macedonian
dress. In Welters, L. (ed.) Folk Dress in Europe and Anatolia:
Beliefs about Protection and Fertility. Oxford: Berg, 97–110.
Molleson, T. & Andrews, P. 1997: The human remains. Çatalhöyük
1997 Archive Report. http://www.catalhoyuk.com/archive_reports/
Molleson, T., Andrews, P., Boz, B., Sofaer Derevenski, J. &
Pearson, J. 1998: Human remains up to 1998. Çatalhöyük 1998
Archive Report. http://www.catalhoyuk.com/archive_reports/
Molleson, T., Andrews, P. & Boz, B. 1999: Report on human
remains recovered from the South Area, together with a
summary of material from the BACH area and the
KOPAL trench. Çatalhöyük 1999 Archive Report.
http://www.catalhoyuk.com/archive_reports/
Molleson, T., Andrews, P., Boz, B. & Hager, L. 2000: Human re-
mains. Çatalhöyük 2000 Archive Report.
http://www.catalhoyuk.com/ archive_reports/
Moore, J. & Scott, E., eds. 1997: Invisible People and Processes.
Writing Gender and Childhood into European archaeology.
London: Leicester University Press.
Nemeskéri, J. 1974: Outline on the anthropological finds of a
Neolithic site. Wissenschaftliche Mitteilungen des
Bosnisch–Herzegowinischen Landesmuseums, Archäologie IVA:
37–46.
Nemeskéri, J. & Lengyel, L. 1976: Neolithic skeletal finds. In
Gimbutas, M. (ed.) Neolithic Macedonia as Reflected by
Excavation at Anza, Southeast Yugoslavia. Los Angeles: Institute
of Archaeology, University of California, 375–410.
Ney, P. 1971: Mineralogische Analyse einer Fussbodenprobe von
Lepenski Vir I. In Schwabedissen, H. (ed.) Die Anfänge des
Neolithikums vom Orient bis Nordeuropa, Teil 2: Östliches
Mitteleuropa. Köln: Böhlau, 33–34.
Politis, G.G. 1999a. Plant exploitation among the Nukak hunter-
gatherers of Amazonia: between ecology and ideology. In
Gosden, C. & Hather, J. (eds) The Prehistory of Food. Appetites
A
168
for Change. London: Routledge, 99–125.
— 1999b. La actividad infantil en la producción del registro ar-
queológico de Cazadores-Recolectores. Revista do Museu de
Arqueologia e Etnologia Suplemento 3: 263–283.
Quitta, H. 1975: Die Radiocarbondaten und ihre historische
Interpretation. In Srejović, D. Lepenski Vir. Eine vorgeschicht-
liche Geburtsstätte europäischer Kultur. Bergisch Gladbach:
Gustav Lübbe, 272–285.
Radovanović, I. 1996: The Iron Gates Mesolithic. Ann Arbor:
International Monographs in Prehistory.
— 1997: The Lepenski Vir Culture: a contribution to interpretation
of its ideological aspects. In Antidoron Dragoslavo Srejović
completis LXV annis ab amicis collegis discipulis oblatum.
Belgrade: Centre for Archaeological Research, University of
Belgrade, 85–93.
— 1999: ‘Neither person nor beast’ — dogs in the burial practice of
the Iron Gates Mesolithic. Documenta Praehistorica 26: 71–87.
— 2000: Houses and burials at Lepenski Vir. European Journal of
Archaeology 3: 330–349.
Rivière, P. 1995: Houses, places and people: community and con-
tinuity in Guiana. In Carsten, J. & Hugh-Jones, S. (eds) About the
House, Lévi-Strauss and Beyond. Cambridge: Cambridge
University Press, 189–205.
Roberts, C. & Manchester, K. 1995: The Archaeology of Disease.
Ithaca (N.Y.): Cornell University Press.
Rodden, R.J. 1965: An Early Neolithic village in Greece. In Avenues
to Antiquity. Readings from Scientific American. San Francisco:
W.H. Freeman, 151–159.
Roksandic, M. 1999: Transition from Mesolithic to Neolithic in the
Iron Gates Gorge: Physical Anthropology Perspective.
Unpublished Ph.D. thesis, Department of Anthropology, Simon
Fraser University, Burnaby, B.C.
Rosaldo, R. 1989 [1993]: Culture and Truth. The Remaking of
Social Analysis. Boston: Beacon Press.
Saunders, S.R. 1992: Subadult skeletons and growth related studies.
In Saunders, S.R. & Katzenberg, M.A. (eds) Skeletal Biology of
Past Peoples: Research Methods. New York: Wiley-Liss, 1–20.
Scheuer, J.L., Musgrave, J. & Evans, S.P. 1980: Estimation of late
foetal and perinatal age from limb bone lengths by linear and
logarithmic regression. Annals of Human Biology 7: 257–265.
Scott, E. 1999: The Archaeology of Infancy and Infant Death (BAR
International Series 819). Oxford: Archaeopress.
Sofaer Derevenski, J. 1997: Engendering children, engendering ar-
chaeology. In Moore, J. & Scott, E. (eds) Invisible People and
Processes. Writing Gender and Childhood into European
Archaeology. London: Leicester University Press, 192–202.
Srejović, D. 1967: Lepenski Vir. Katalog izložbe. Belgrade: Narodni
muzej.
— 1969a: Lepenski Vir: Nova praistorijska kultura u Podunavlju.
Belgrade: Srpska književna zadruga.
— 1969b: The roots of the Lepenski Vir culture. Archaeologia
Iugoslavica 10: 13–21.
— 1970: Lepenski Vir, Boljetin — predneolitska i neolitska naselja
i nekropole. Arheološki pregled 12: 72–73.
— 1972: Europe’s First Monumental Sculpture: New Discoveries at
Lepenski Vir. London: Thames & Hudson.
— 1981: Lepenski Vir: Menschenbilder einer frühen europäischen
Kultur. Mainz: von Zabern.
Srejović, D. & Babović, Lj. 1983: Umetnost Lepenskog Vira.
Belgrade: Jugoslavija.
Srejović, D. & Letica Z. 1978: Vlasac: A Mesolithic Settlement in
the Iron Gates. Vol. 1, Archaeology. Belgrade: Serbian Academy
of Sciences and Arts.
Stanković, S. 1992: Kultna mesta centralnobalkanskog područja.
Unpublished Ph.D. dissertation. Department of Archaeology,
University of Belgrade.
 Sofija Stefanović & Dušan Borić: New-born infant burials at Lepenski Vir
Stanković, S. & Leković, V. 1993: Neolithic settlement at Blagotin.
Glasnik Srpskog arheološkog društva 9: 177–179.
Stewart, T.D. 1968: Identification by the skeletal structures. In
Camps, F.E. (ed.) Gradwohl’s Legal Medicine, 2nd Edition.
Bristol: Wright, 123–154.
Strathern, M. 1988: The Gender of the Gift. Problems with Women
and Problems with Society in Melanesia. Berkeley: University of
California Press.
Stuart-MacAdam, P.L. 1989: Nutritional deficiency diseases: a sur-
vey of scurvy, rickets and iron-deficiency anaemia. In Işcan,
M.Y. & Kennedy, K.A.R. (eds) Reconstruction of Life from the
Skeleton. New York: Alan Liss, 201–221.
Tanner, J.M. 1989: Foetus into Man (2nd edition). Ware:
Castlemead.
Tringham, R. 2000: The continuous building: a view from the deep
past. In Joyce, R.A. & Gillespie, S.D. (eds) Beyond Kinship.
Social and Material Reproduction in House Societies.
Philadelphia: University of Pennsylvania Press, 115–134.
Turner, V. 1967: The Forest of Symbols. Aspects of Ndembu Ritual.
Ithaca: Cornell University Press.
— 1974: Dramas, Fields and Metaphors. Symbolic Action in
Human Societies. Ithaca: Cornell University Press.
Ubelaker, D.H. 1978: Human Skeletal Remains: Excavation,
Analysis, Interpretation. Chicago: Aldine.
Van Gennep, A. 1960: The Rites of Passage. Chicago: University of
Chicago Press.
Waterson, R. 1990: The Living House: an Anthropology of
169
Architecture in South-East Asia. Kuala Lumpur: Oxford
University Press.
Waterson, R. 2000: House, place and memory in Tana Toraja
(Indonesia). In Joyce, R.A. & Gillespie, S.D. (eds) Beyond
Kinship. Social and Material Reproduction in House Societies.
Philadelphia: University of Pennsylvania Press, 177–188.
Welters, L. (ed.) 1999a: Folk Dress in Europe and Anatolia: Beliefs
about Protection and Fertility. Oxford: Berg.
— 1999b: Introduction: folk dress, supernatural beliefs, and the
body. In Welters, L. (ed.) Folk Dress in Europe and Anatolia:
Beliefs about Protection and Fertility. Oxford: Berg, 1–12.
— 1999c: The Peloponnesian ‘Zonari’: a twentieth-century string
skirt. In Welters, L. (ed.) Folk Dress in Europe and Anatolia:
Beliefs about Protection and Fertility. Oxford: Berg, 53–70.
Whittle, A. 1996: Europe in the Neolithic. The Creation of New
Worlds. Cambridge: Cambridge University Press.
Whittle, A., Bartosiewicz, L., Borić, D., Pettitt, P. & Richards, M.
2002: In the beginning: new radiocarbon dates for the Early
Neolithic in northern Serbia and south-east Hungary. Antaeus 25:
63–117.
Winnifrith, T.J. 1987: The Vlachs: The History of a Balkan People.
London: Duckworth.
Zoffmann, ZS. 1983: Prehistorical skeletal remains from Lepenski
Vir. Homo 34(3/4): 129–148.
Zoffmann, ZS. 1988: Human skeletal remains from Divostin. In
McPherron, A. & Srejović, D. (eds) Divostin and the Neolithic of
Central Serbia. Pittsburgh: University of Pittsburgh, 447–455.
 DNA-based sex identification of the infant remains from
Lepenski Vir
Biljana Čuljković, Sofija Stefanović & Stanka Romac
Abstract: At Lepenski Vir 41 infant skeleton were excavated from under the houses. Infant mortality could be affected by a broad range
of economic, social and health conditions, and could be regarded as an indicator of overall quality of life in past communities. Sex determ-
ination presents a difficult problem in anthropology in cases of fragmentary or child remains. We performed DNA-based sex identification
of 30 infant remains. Of the 30 samples tested, 29 specimens provided results: 16 were found to be males and 13 females.
Key words: DNA, sex determination, infant skeletons, Lepenski Vir
Introduction
Human remains from Lepenski Vir consist of 190 individuals
from 134 graves plus 42 individuals from unidentified con-
texts (Roksandic 1999: 78). Among them there are 51 infant
skeletons, most of them found underneath the house floors
when these were lifted in 1970.
Infants are defined as individuals under 1 year of age and
the study of their skeletons in palaeodemography is import-
ant because infant mortality could be affected by a broad
range of economic, social and medical conditions, and could
be regarded as an indicator of overall quality of life of past
communities. Although this material represents the most
demographically variable and sensitive portion of the human
life cycle (Roth 1992: 177), there are many problems associ-
ated with the assessment of infant mortality from skeletal
material. The reasons why these skeletons might not be rep-
resented in osteoarchaeological material are: the effect of soil
acidity (as soil acidity increases so the preservation of bone
decreases, especially infant bones), burial practices (infants
may be excluded from cemetery burial), and excavation
strategy. In this context we can state that the number of in-
fants at Lepenski Vir is high (27% of all buried individuals).
We estimated the gestational age of infants from Lepenski
Vir, according to maximum length of femora and humeri.
Most of them belong to 38–40 gestational weeks.
The anthropological criteria for sex determination are
based on qualitative assessment of the morphological fea-
tures of the skull, long bones and pelvis. The problem of sex
determination persists when dealing with fragmentary and/or
infant burials. The reliability of morphometric analyses for
gender identification in infants is low, although pelvic sex
differences are present at birth, they become obscured soon
after birth.
New developments in molecular biology, and especially in
analysing DNA recovered from ancient bones, have provided
reliable methods for gender determination based on ampli-
fication of DNA sequences specific to the X and/or Y chro-
mosomes (Mannucci et al. 1994; Lassen et al. 1996). The
development of the polymerase chain reaction (PCR) (Saiki
et al. 1985) has allowed extremely small amounts of highly
degraded DNA to be analysed.
171
In this study we performed DNA-based sex identification
of 30 infant remains in order to obtain information on the sex
distribution of infants at Lepenski Vir.
Materials and methods
The extractions were performed from 0.2–0.3 g powdered
bone (ribs). The bone was ground either in a coffee grinder or
in a mortar. In order to avoid contamination with contempor-
ary DNA, all instruments and working areas were exposed to
UV light. Powdered bone was mixed with lysing buffer in the
ratio: 1 g of powdered bone and 2 ml of lysing buffer (Hughes
& Galau 1988). The sample was then incubated at 550 °C
overnight followed by phenol/chloroform extraction and iso-
propanol precipitation.
The same experimenter conducted all experiments. All
buffers and water were autoclaved. A large amount of PCR
premixes containing all the necessary components, except
DNA and Taq polymerase, were prepared and frozen in small
sealed aliquots at –200 °C. A ‘quality control’ PCR without
added DNA but only Taq polymerase was performed on ran-
dom aliquots.
The preparation of bone samples and the DNA extractions
were performed in one room and stored in a freezer. On ar-
rival in the laboratory, the specimens were cleaned and UV
irradiated (254 nm) for 60 minutes. Specimens were sub-
sequently handled under sterile conditions. Small samples of
bone were extensively cleaned by cutting off approximately
20 mm of the entire bone surface with scalpel blades. Bones
ware then placed in disposable sterile tubes further used for
DNA extraction. Preparations of buffers and PCR set-up
were performed in a dedicated sterile hood under constant
UV illumination (254 nm). Dedicated pipettes with aerosol
resistant plugged tips were used throughout. PCR experi-
ments as well as the analysis of PCR products were per-
formed in a separate room. For each set of PCR experiments
reaction blanks and mock extraction controls were conducted.
The PCR was performed in a total volume of 25μl reaction
mixture containing 10mM Tris-HCl pH 8.3, 50mM KCl,
1.5mM MgCl 2 , 1% Triton X-100, 0.2mM dNTPs each,
0.2mM each primer and 1.25U Taq DNA Polymerase
 The Iron Gates in Prehistory

A 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17

150bp

100bp

B 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19

150bp

100bp

C 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17

150bp

100bp

Figure 1. Acrylamide gel electrophoresis of PCR products of amelogenin loci for sex determination. A: lanes 1 and 2:
sample 63, lanes 3 and 4: sample 71, lanes 5 and 6: sample 95, lane 7: blank reaction control, lanes 8, 9 and 10: sample
110, lane 11: 50bp DNA marker (Pharmacia LKB), lanes 12, 13 and 14: sample 111, lane 15: mock extraction control, lane
16: female’s contemporary DNA, lane 17: male’s contemporary DNA; B: lane 1: 50bp DNA marker, lanes 2 and 3: sample
102, lane 4: sample 103, lane 5: sample 109A, lane 6: sample 107, lane 7: mock extraction control, lane 8: female’s con-
temporary DNA, lane 9: blank reaction control, lane 10: 50bp DNA marker, lane 11: sample 113, lanes 12, 13 and 14: sample
112, lanes 15 and 16: sample 110, lanes 17 and 18: sample 111, lane 19: mock extraction control. C: lane 1: 50bp DNA
marker, lanes 2 and 3: sample 113, lane 4: mock extraction control, lanes 5 and 6: sample 118, lanes 7 and 8: sample 119,
lane 9: 50bp DNA marker, lane 10: sample 114, lane 11: sample 123, lanes 12, 13 and 14: sample 120, lanes 15 and 16:
sample 133, lane 17: blank reaction control.

(Boehringer). Samples were amplified through 40 cycles in a
Techne Progene Thermal Cycler. The PCR temperature pro-
files and sequences of primers for the X–Y homologous am-
elogenin genes (AMELX and AMELY) and CD4 gene were
previously described (Mannucci et al. 1994; Edwards et al.
1991).1 PCR products were visualized on 12% polyacrylam-
ide gels by silver staining.
Results and discussion
For each DNA extract we performed at least three PCR reac-
tions for the amelogenin gene. Out of 30 samples tested, 29
specimens provided results. Sixteen specimens were found to
be males and 13 females. The results for 18 specimens are
shown in Figure 1. The success rates of PCRs are given in
Table 1. In total, data were obtained for 68 of 118 PCRs of 30
specimens. There were no inconsistencies or conflicting data
for any of the specimens. The same type of analysis was
successfully performed on aDNA samples from adult skelet-
ons of known sex.
DNA extracted from these samples was also analysed by
agarose gel electrophoresis (data not shown), and all of them
showing smear pattern, indicating DNA degradation. A large
portion of the isolated aDNA samples came from bacteria, so
quantification does not guarantee success for PCR amplifica-
tion, which probably is the reason why we observed low
success rates of PCR amplification for many samples.
The authenticity of DNA samples was verified by ampli-
fication of CD4 loci (commonly used for human identifica-
tion — Edwards et al. 1991) and comparing these DNA pro-
files to the profile of the analyst who conducted all the
experiments. Results of CD4 profiling for eight specimens
from Lepenski Vir and the experimentator’s DNA are shown
in Table 2. These experiments prove that the amplified frag-
ments when aDNA is used as a template are not the result of
contamination with present day DNA. If the amplified frag-
ments were the result of contamination with contemporary
DNA, we would obtain similar or identical results for the
CD4 gene. Although isolation of DNA was performed under
stringent conditions, which included scraping the surface of
bone samples (see materials and methods), and DNA profil-
ing showed differences among analysed samples and experi-
mentator DNA, there is still a chance of contamination with
contemporary DNA coming from excavators and all other
people who handled the samples prior to aDNA analysis.

172
 Biljana Čuljković, Sofija Stefanović & Stanka Romac: DNA-based sex identification at Lepenski Vir

Table 1. Data on DNA-based sex identification of the infants from Lepenski Vir.

Specimen Successful Results

PCRs*

63 2 Female
71 3** Male
94 0 –
95 2 Female
96 2** Male
98 1 Female
98A 2 Male
102 2** Female
103 2 Male
106 2** Male
107 4** Male
109 2** Male
109A 2 Female
110 3 Female
111 3** Female
112 3** Male
113 4** Female
114 1 Male
116 2 Male
117 4** Female
118 2 Female
119 2 Male
120 3** Male
123 2 Female
124 2 Male
125 2** Male
127 2 Female
128 2 Female
132 1** Male
133 3** Male

*3 PCRs were performed **5 PCRs were performed

Table 2. Results of DNA profiles for the CD4 gene.

Sample CD4 profile

71 5/4
95 9/9
109 6/6
114 9/9
118 10/10
120 5/10
124 9/5
133 7/7
Experimentator DNA 4/4

Reasons for infant mortality can be various, from congen-
ital diseases to infanticide. Whatever the reason for the deaths
of infants at Lepenski Vir, we did not find significant differ-
ences in the sex distribution of those tested.
Note
1. Twenty-two out of our 23 pairs of chromosomes are called auto-
somes and do not differ at the gross level between the sexes. The
23rd pair — the sex chromosomes (XX in females and XY in
A
males) — determine the sex of an individual. Overall, the X
chromosome is much larger and has many more genes than the Y
chromosome, but one region, the pseudoautosomal region is
shared between these two chromosomes. One gene located there,
amelogenin (AMEL), encodes a tooth enamel protein. An ancient
6 base deletion mutation in the X chromosome version of the
gene, AMELX, compared to the Y chromosome version, AMELY,
has been exploited by researchers as a means to identify the
chromosome complement, and hence the gender of any DNA
sample. The polymerase chain reaction (PCR) is a technique that
amplifies large amounts of a selected sequence even if it is ini-
tially present at vanishingly small quantities (such as ancient

173
 The Iron Gates in Prehistory
DNA samples). Sex determination works by using a PCR assay
over the AMELX/AMELY region to amplify the DNA sample and
then examining the length of the PCR products (short — deletion
carrying — products indicate female DNA (XX), whereas a
mixture of long and short products indicate male DNA (XY)).
The advantage of this system is that rather than using separate
X-specific and Y-specific tests and combining the results to de-
termine the sex, the choice of amelogenin gene PCR necessitates
just a single amplification process — more efficient and tech-
nically robust. [Ed.]
References
Edwards, M.C., Clemens, P.R., Tristan, M., Pizzuti, A. & Gibbs,
R.A. 1991: Pentanucleotide repeat length polymorphism at the
human CD4 locus. Nucleic Acids Research 19: 4791.
Hughes, D. & Galau, G. 1988: Preparation of RNA from cotton
leaves and pollen. Plant Molecular Biology Reporter 6: 253–257.
Lassen, C., Hummel, S., Herrmann, B. 1996: PCR based sex iden-
tification of ancient human bones by amplification of X- and Y-
174
chromosomal sequences: a comparison. Ancient Biomolecules 1:
25–34.
Mannucci, A., Sullivan, K., Ivanov, P.L. & Gill, P. 1994: Forensic
application of a rapid and quantitative DNA sex test by ampli-
fication of the X/Y homologous gene amelogenin. International
Journal of Legal Medicine 106: 190–193.
Roth, E. 1992: Applications of demographic models to paleodemo-
graphy, In Saunders, S.R. & Katzenberg, M.A. (eds) Skeletal
Biology of Past Peoples: Research Methods. New York: Wiley-
Liss, 175–188.
Roksandic, M. 1999: Transition from Mesolithic to Neolithic in the
Iron Gates Gorge: Physical Anthropology Perspective.
Unpublished PhD thesis, Simon Fraser University, Burnaby, B.C.
Saiki, R.K., Scharf, S., Faloona, F., Mullis, K.B., Horn, G.T., Erlich
H.A. & Arnheim, N. 1985: Enzymatic amplification of β-globin
genomic sequences and restriction site analyses for diagnosis of
sickle cell anemia. Science 230: 1350–1354.
Saunders, S.R. 1992: Subadult skeletons and growth related studies.
In Saunders, S.R. & Katzenberg, M.A. (eds) Skeletal Biology of
Past Peoples: Research Methods. New York: Wiley-Liss, 1–20.
 Dating burial practices and architecture at Lepenski Vir

Clive Bonsall, Ivana Radovanović, Mirjana Roksandic, Gordon Cook,

Thomas Higham & Catriona Pickard

Abstract: Previous attempts to establish a chronology for Lepenski Vir using three different methods (stratigraphy, radiometric 14C dating of
bulk charcoal samples, and AMS 14C dating of human bone collagen) produced inconsistent results. Discrepancies between the human bone
and charcoal ages were found to result from a reservoir effect in the bones of people who ate significant quantities of Danube fish. When a
reservoir ‘correction’ is applied, the human bone 14C dates are consistent with the charcoal dates, and this raises questions about the excav-
ator’s relative and absolute chronology based on stratigraphy and inter-site comparisons. Single-entity dating of surviving archaeological
materials offers the best hope of constructing a reliable chronological framework for Lepenski Vir. This paper presents the results of a further
programme of AMS 14C dating of human remains. Direct dating of 24 burials confirms that different burial practices characterized the Final
Mesolithic and Early Neolithic. Previous attempts to assign burials to Mesolithic or Neolithic phases, based on stratigraphic observations,
are shown to be broadly correct but not always accurate in detail. The evidence from radiocarbon dating and stratigraphy is used to calculate
‘minimum’ and/or ‘maximum’ ages for certain of the trapezoidal buildings, which suggest that this architectural form was in use during the
Final Mesolithic and Early Neolithic. The implications of the human bone 14C dates and associated stable isotope measurements for the
timing of the Mesolithic–Neolithic transition in the Iron Gates are also discussed.

Key words: burials, architecture, Mesolithic, Neolithic, AMS 14C dating, stable isotopes

Introduction structures ‘stratigraphically’ were among the latest chronolo-

Lepenski Vir stands out among the Stone Age sites of the Iron
Gates (Fig. 1) by virtue of its architecture characterized by
distinctive trapezoidal structures with lime plaster floors
containing numerous large stone sculptures and other sym-
bolic artefacts (among other finds), together with a record of
complex burial practices associated with the structures and
the areas between them. Its unusual features have led many
archaeologists, including the excavator Dragoslav Srejović
gically, according to their radiocarbon ages.
Radiometric 14C dating of charcoals from the Lepenski Vir
I–II structures gave 14C ages between c. 7360 and 6560 BP
(Quitta 1972, 1975). These ages were similar to those for
Early Neolithic (Starčevo–Körös–Criş sites in the surround-
ing regions and were rejected by Srejović (1971, 1972a,
1972b) as being too young. He suggested an age range for
Lepenski Vir I–II of c. 500 years earlier, based on radiometric
14 C ages of c. 7930–7440 BP for charcoal samples from

to view Lepenski Vir as not just another settlement on the
edge of the Danube, but as a site that came into being early in
the Mesolithic and later assumed an important role in the
lives of the local population as an aggregation, sacred and/or
ceremonial site.
The conventional interpretation of the occupation se-
quence at Lepenski Vir was proposed by Srejović (1969,
1972a), who recognized a succession of Mesolithic settle-
ments: Proto-Lepenski Vir, Lepenski Vir I, and Lepenski Vir
II, followed by the Neolithic settlements of Lepenski Vir IIIa
and IIIb. The main Mesolithic occupation phase of Lepenski
Vir I was divided into 5 sub-phases (Ia–e) based on strati-
graphic observations and material culture content, while the
Early Neolithic Lepenski Vir IIIa phase was divided into two
sub-phases (IIIa1–2). Srejović (1969, 1972a) argued for a
significant time gap between Lepenski Vir II and III.
This periodization of Lepenski Vir became controversial
even before the excavations were completed in 1970, due
primarily (though not exclusively) to the discrepancies
between Srejović’s stratigraphic observations and the first
published radiometric 14C dates on charcoal samples collec-
ted from the Lepenski Vir I structures (Quitta 1969). The 14C
results suggested that some of the earliest Lepenski Vir I
Mesolithic contexts at the neighbouring site of Vlasac.
Other researchers have accepted the radiocarbon meas-
urements for Lepenski Vir as valid and have interpreted
phases I–II as either Early Neolithic (e.g. Jovanović 1969,
1972) or a Late Mesolithic survival in an area unsuitable for
farming but rich in natural resources capable of supporting a
hunter-gatherer population (e.g. Voytek and Tringham 1989).
Many of the burials at Lepenski Vir were also assigned to
the five main phases (Zoffmann 1983), but the basis on which
this was done was never adequately explained in the literat-
ure. It is assumed to be partly on the basis of stratigraphy,
including the stratigraphic relationship with buildings, and
partly on the basis of burial type. The human remains from
Lepenski Vir were re-analyzed by Roksandic (1999, 2000;
Roksandic et al. 2006, this volume), but a reconsideration of
the phasing of the burials was beyond the scope of her study.
Subsequent AMS dating of human remains assigned to
phases IIIa and IIIb, produced ages between c. 7770 and 6910
BP (Bonsall et al. 1997), which are earlier than expected on
the basis of Srejović’s stratigraphic interpretation and older
than the charcoal ages for preceding phases I and II. Bonsall
et al. (1997) suggested, and subsequent research (Cook et al.
2001) has demonstrated, that the radiocarbon ages on the

175
 The Iron Gates in Prehistory
Figure 1. Lepenski Vir and the Iron Gates.
bones of humans whose diets included regular consumption
of aquatic food sources (mainly freshwater fish) are subject to
a reservoir effect of up to approximately 500 years. Cook et
al. (2001, 2002) devised a method for correcting the 14C ages
made on human bone samples for this reservoir effect using:
(1) the δ15N values for human bone collagen samples to es-
timate the percentage freshwater aquatic diet, based on a
knowledge of the approximate δ15N end members for 100%
freshwater aquatic and terrestrial diets, assuming a linear re-
lationship between δ15N and percentage freshwater diet and
(2) the age offsets between human bone ages and closely as-
sociated ungulate bone ages for particular δ15N values, based
on excavated material from another Iron Gates site, Schela
Cladovei, on the Romanian bank of the Danube. When this
reservoir correction is applied to the human bone ages from
Lepenski Vir obtained by Bonsall et al. (1997) the resulting
age range (7310 to 6720 BP) is very similar to that of the bulk
charcoals (7360 to 6560 BP). This age range and the fact that
some of the Lepenski Vir III burials had ‘Mesolithic’ dietary
signals raises questions about the phasing of the burials. It
also raises doubts about the stratigraphic integrity of
Lepenski Vir III, because the ages are no different from
phases I and II.
In parallel with the AMS 14C dating of human remains,
Radovanović (1996a) attempted a revision of the Lepenski
Vir stratigraphy. Using information in Srejović’s published
accounts together with previously unpublished field docu-
mentation, she proposed a revised sequence of occupation
phases: Proto-Lepenski Vir, Lepenski Vir I (comprising sub-
176
phases 1–3, with 11 building levels), Lepenski Vir II, and
Lepenski Vir III. Subsequently, several authors have ques-
tioned the stratigraphic integrity or even the existence of
some of Srejović’s phases (e.g. Borić 1999, 2002b;
Radovanović 2000; Garašanin & Radovanović 2001; Bonsall
et al. 2002a; Perić & Nikolić 2004).
The greater part of the Lepenski Vir site was excavated in
1965–70 .With the impounding of the Danube by the Iron
Gates I dam the site was flooded and is no longer accessible,
apart from a number of structures from Lepenski Vir I that
were rescued and preserved for display in 1970. Realistically,
therefore, the disagreements over the relative and absolute
chronology of Lepenski Vir can now only be addressed by a
detailed programme of scientific analysis, including single-
entity dating, of the surviving archaeological materials. To
the authors’ knowledge, no charcoal or other carbonized
plant material is available for analysis. However, large col-
lections of animal and human bones are still available for
scientific study, as are low-fired, organic-tempered Starčevo
pottery sherds that, potentially, are datable by AMS 14C, ar-
chaeomagnetic intensity, and luminescence techniques
(Bonsall et al. 2002b).
The dating programme
Aims and objectives
The main aims of the AMS 14C dating programme reported
here were to:
 Clive Bonsall et al.: Dating burial practices and architecture at Lepenski Vir
1. Investigate changes in diet and subsistence during the oc-
cupation of Lepenski Vir;
2. Establish the chronological contexts of particular forms of
burial represented at the site;
3. Test the phasing of the Lepenski Vir burials proposed by
Srejović (1969) and Zoffmann (1983) based on strati-
graphy and burial custom;
4. Test previous hypotheses of the age and phasing of the
trapezoidal buildings by dating human burials that were
either cut through or sealed by their plaster floors; and
5. From the above, provide new information bearing on the
timing of the Mesolithic–Neolithic transition in the Iron
Gates gorge.
Why date the burials?
Two approaches have been used in an effort to establish a
more reliable chronology for the trapezoidal structures at
Lepenski Vir using single-entity AMS 14C dating of bone re-
mains. One approach has involved the dating of terrestrial
animal bones (e.g. Borić & Miracle 2004; Borić &
Dimitrijević 2005, 2007), the other the dating of human
bones from articulated skeletons (Bonsall et al. 2004). Both
approaches rely on the existence of a clear stratigraphic rela-
tionship between the dated bone and the building.
There are uncertainties associated with both approaches,
but we suggest that the uncertainties are far greater with the
first approach. The bases of the Lepenski Vir buildings were
made by cutting more or less horizontally into the sloping
bank of the Danube, effectively creating pit features, and
these pits became infilled by one means or another after the
buildings were abandoned.
A disarticulated bone found on the floor of a pit could have
reached that position in any of a number of ways. It may be
in a primary context, deposited when the building was in use,
or shortly after its abandonment. On the other hand, the bone
could be older than the pit (i.e. in a secondary context) de-
posited there after its abandonment as a result of slumping at
the sides of the pit, movement of material from upslope due
to gravity or hillwash (probably a common event on steep
valley side slopes in the Iron Gates gorge), or deliberate in-
filling of the pit with soil material containing earlier archae-
ological objects. It is possible for younger bones to be intro-
duced into ancient pit features as a result of bioturbation,
including animal burrowing, earthworm activity and root
penetration, or of post-depositional disturbance by humans.
Moreover, unless animal bones bear manufacturing traces or
butchery marks, there will always be an element of doubt
about whether their presence in an archaeological site is the
result of primary human activity or natural processes. These
are just some of the initial steps in the taphonomic process
that can affect bones prior to deposition; further loss of in-
formation (e.g. water transport or the destruction of datable
collagen) may be expected in the deposit itself during fossil
diagenesis. In addition, secondary human effects in the form
of errors of excavation or curation (see below) may result in
older or younger material being wrongly attributed to an ar-
chaeological feature. A possible example of this from the Iron
Gates was discussed by Bonsall et al. (2002b).
It has been argued that the taphonomic biases associated
with the dating of animal bones are lessened by dating bones
177
that are still in articulation (skeletons or partial skeletons) on
the assumption that they are in a primary context and were
deposited soon after the death of the animal (e.g. Borić &
Dimitrijević 2007: 55). However, these are rare occurrences
on most archaeological sites, including Lepenski Vir.
This argument applies equally to human skeletons, which
were much more frequent than animal skeletons/partial skel-
etons at Lepenski Vir. While 14C age measurements on hu-
man bones from this site are less precise because of the
reservoir age associated with them (see above), often the ar-
ticulated bones (more-or-less complete skeletons, rather than
typical food remains from animals) are from clearly-defined,
archaeological contexts and are subject to better stratigraphic
control. Moreover, in contrast to the limited information
available for animal remains, the stratigraphic relations
between burials and buildings at Lepenski Vir are reasonably
well documented by photographs, plans and written records.
Thus, given that we have already developed a reservoir cor-
rection for 14C measurements on human bones from the Iron
Gates (Cook et al. 2001, 2002, in press) this becomes our
preferred material for a detailed dating programme.
Curatorial issues
Recovery methods and post-excavation practices can have as
much taphonomic effect on bones as the diagenetic processes
that led to their fossilization, and must be taken into account
when assessing the 14C dates and stable isotope results from
Lepenski Vir. The following observations are based on the
study undertaken by Mirjana Roksandic in 1996–8 for her
PhD thesis (Roksandic 2000).
In many respects the Lepenski Vir excavation was a re-
markable achievement. Between 1965 and 1969 in five cam-
paigns lasting a total of 12 months, approximately 2500m2 of
the site, with deposits averaging 3.5 m deep, were excavated
to reveal architecture, monumental sculpture and graves of
the ‘Lepenski Vir culture’. But all this was done at great
speed, for the most part without the use of fine sieving,1 and
much detailed information on burial practices and relation-
ships between burials and architecture undoubtedly was
overlooked.
Judging from photographic evidence and testimonies of
archaeologists who worked at Lepenski Vir as students, when
burials were identified their excavation and recording was
usually quite meticulous. But not all burials were recognized,
or were only recognized after they had been disturbed by the
excavators. In some cases (e.g. burials 29, 30, 49 and 60) only
skulls and long bones were collected, even though the burials
(according to the available photographs) were intact and the
bones well preserved. In other cases, contextual data are im-
precise. Some groups of bones were recorded only as coming
from a particular ‘cultural layer’ and were not registered as
‘burials’; yet these are sometimes represented by more
skeletal elements than some of the identified burials.
Furthermore, the material was washed on site by unskilled
workers, and it is possible that some smaller bones and bone
fragments were lost in the process. However, the frequency
of small-sized human carpal bones and phalanges in the col-
lection would seem to exclude this as a major factor affecting
bone presence and preservation.
Since the excavations, the human bones from Lepenski Vir
 The Iron Gates in Prehistory

have been housed in the Faculty of Philosophy, University of
Belgrade, where they have been studied by generations of
researchers. Frequent handling took its toll on the collections,
but most damage was a consequence of the economic and
social problems that afflicted the former Yugoslavia in the
1980s and 1990s. During this period cardboard boxes and
packaging materials were in short supply and storage space
was at a premium. The collections could not be housed to-
gether in one area. There were not enough boxes to store the
skeletons individually, and bones from two or more skeletons
were occasionally placed together in the same box.
Moreover, shelf space was limited with the result that boxes
were piled on top of one another, which caused some to col-
lapse. For a time a large part of the collection was housed in
a damp basement, causing boxes and wrappings to disinteg-
rate and mould to develop on some of the bones. Labels were
also affected to some extent, but in the majority of cases were
nevertheless still legible.
In 1997 with the support of the Wenner-Gren Foundation
and logistical help from Prof. Ž. Mikić and students at the
University of Belgrade, Roksandic was able to bring the col-
lection together in one place, with proper shelving, and to
embark on the task of restoring the collection to its original
curatorial condition — work that took seven months. By
1998 she had reconstructed as much of the collection as she
could, removed the mould from the bones, and recorded bone
presence and preservation for each of the skeletons. A num-
ber of burials were found to be missing; some skeletons that
appeared more or less intact in excavation photographs had
few bones remaining in 1996–8; and discrepancies with the
documentation and probable mixing of bones from different
skeletons were consistently recorded.
This knowledge guided selection of samples for 14C dating
and stable isotope analysis in 2000. Even so, not all problems
were resolved it seems, as illustrated by burials 4 and 7/I (see
below).
Methods and results
Bone samples were prepared for AMS 14 C dating at the
Oxford Radiocarbon Accelerator Unit (ORAU) using routine
collagen extraction procedures (Law & Hedges 1989; Bronk
Ramsey et al. 2000). An additional ultra-filtration pretreat-
ment step was used to further purify the bone gelatin and re-
tain only the >30-kD molecular weight fraction for 14C assay
(Brown et al. 1988; Bronk Ramsey et al. 2000). The <30-kD
fraction may include degraded collagen fragments, salts, and
contaminants that may be of a different 14C age to the gelatin.
We used the ratio of carbon to nitrogen (C:N ratio) to de-
termine the chemical integrity of the extracted gelatin. All
ratios were within the 2.9–3.6 range of acceptability for bone
collagen used at ORAU. Yields of ultra-filtered gelatin that
are below 10 mg g–1 are not dated routinely because they in-

Table 1. AMS radiocarbon ages and stable isotope values for human burials from Lepenski Vir. Body position:
Es – extended supine; Dd – dorsal decubitus; C – crouched; D - disarticulated; ? – uncertain

Burial Body Skeletal element Lab ID 14C age BP δ13C ‰ δ15N ‰ Reservoir- Calibrated age Median
No. position corrected range (2σ) probability
14C age BP BC/AD BC/AD

60 Es R or L femur OxA-11715 9470±55 -18.9 15.5 9020±80 8445–7953 8226

61 Es R femur OxA-11698 7860±50 -18.8 16.1 7374±80 6406–6071 6247
14 Es R femur OxA-11704 7830±45 -18.6 15.7 7368±75 6396–6072 6239
54c Es L femur OxA-11696 7610±45 -19.6 12.4 7346±57 6365–6074 6205
26 Es R tibia OxA-11693 7380±45 -20.1 9.6 7284±47 6233–6056 6148
54e Es L femur OxA-11697 7550±45 -19.1 13.0 7250±59 6227–6017 6126
7/I Es R femur1 OxA-12979 7697±38 -17.5 16.6 7183±77 6225–5907 6056
7/I Es R femur2 OxA-11692 7710±50 -18.1 16.2 7218±81 6243–5917 6094
62 Es R femur or R tibia OxA-11718 445±63 -18.5 9.3 — AD 1327–1635 AD 1465
29 Es L femur OxA-11706 445±31 -18.4 9.4 — AD 1415–1606 AD 1445
30 Es R humerus OxA-11717 477±34 -18.4 10.3 — AD 1404–1463 AD 1432
69 Dd L femur OxA-11703 9180±50 -19.0 14.6 8784±72 8202–7609 7869
32a C L femur OxA-5828 7270±90 -19.6 11.9 7036±95 6066–5728 5911
9 C L femur OxA-11695 7150±45 -19.4 10.8 6982±50 5983–5747 5866
88 C Femur or L tibia OxA-5831 7130±90 -20.2 10.5 6980±92 6023–5677 5862
8 C L femur OxA-11694 7050±45 -19.7 9.8 6942±47 5972–5729 5821
2 C R femur OxA-11719 5425±50 -19.5 10.6 5269±54 4236–3974 4108
4 C L fibula OxA-11699 485±31 -18.2 9.2 — AD 1405–1451 AD 1430
54d D L femur OxA-11700 7785±45 -17.7 15.2 7353±72 6385–6067 6219
45b D L femur OxA-11701 7805±50 -18.5 15.8 7337±79 6388–6053 6198
79a D Scapula OxA-11705 7780±50 -18.6 15.8 7312±79 6366–6029 6172
31a D L humerus OxA-5827 7770±90 -18.7 15.7 7308±108 6401–5997 6177
44 D R humerus OxA-5830 7590±90 -18.9 15.3 7152±106 6233–5797 6027
89a D Tibia OxA-11702 7595±45 -18.1 15.7 7133±75 6208–5845 6008
18 ? L ulna OxA-11716 1874±40 -18.4 10.5 1724±44 AD 219–420 AD 316
35 ? Long bone OxA-5829 6910±90 -19.7 11.2 6730±93 5792–5483 5643

1 Sample collected in 2000

2 Sample collected in 1989

178
 Clive Bonsall et al.: Dating burial practices and architecture at Lepenski Vir
Figure 2 Plan of Lepenski Vir showing the locations of the AMS 14C dated burials (based on a drawing by Dušan Borić).
dicate poor levels of collagen preservation. All of the ultra-
filtered samples from Lepenski Vir were above this threshold.
The >30-kD fraction was lyophilized and analyzed using a
Europa Scientific ANCA-MS system consisting of a 20–20
IR mass spectrometer interfaced to a Roboprep CHN sample
converter unit operating in continuous flow mode. CO2 from
the combustion was trapped cryogenically and graphite was
prepared by reduction of CO2 over iron within an excess H2
atmosphere. Graphite targets were then measured by AMS
(Bronk Ramsey & Hedges 1997). Small samples of CO2
(<1.6 mg C) were dated directly using the ORAU gas ion
source (Bronk Ramsey & Hedges 1997). δ13C values in this
paper are reported in per mil (‰) with reference to VPDB
and δ15N results are reported with reference to AIR (Coplen
1994).
The radiocarbon results, both uncorrected and corrected
for reservoir age, and the corresponding stable isotope values
are presented in Table 2 together with results from five
samples previously dated (Bonsall et al. 1997), and these 24
samples form the basis of the following discussion. The loc-
ations of the burials are shown in Figure 2.
The problem of burial 7
This burial included in the dating programme is currently the
source of some debate. The burial, which according to
Srejović (1972a: 120, 156) was inserted through the plaster
floor in the rear of building 21, contained bones from at least
two individuals. There was an almost complete, articulated
179
skeleton (7/I) placed in the extended supine position.
Originally published as female by Nemeskéri (1969), the
skeleton was reassessed as that of an adult (middle-aged)
male by Zoffmann (1983) and as ‘male?’ by Roksandic
(1999). In addition, the isolated cranium (7/II) of an adult,
classified as ‘female’ by both Zoffmann (1983) and
Roksandic (1999), was found above the left shoulder of the
skeleton. Animal bones were also recovered from the grave
and assumed to have been deliberately placed there at the
time of burial. These included the skull and horns of an
aurochs, which lay on the right shoulder and upper rib cage of
the skeleton (Fig. 6).
Samples of bone were collected in 1989, and again in
2000, from the right femur of the skeleton. Part of the sample
collected in 1989 was analyzed for stable isotopes at the
Scottish Universities Research Reactor Centre, East Kilbride
(Bonsall et al. 2000), while the remainder of the sample col-
lected in 1989 and the sample collected in 2000 were ana-
lyzed for AMS 14 C and stable isotopes at the Oxford
Radiocarbon Accelerator Unit. The results of these analyses
are presented in Table 2. The 14C age (see Table 1) and the
stable isotope data are consistent with the body position in
indicating a Late Mesolithic context for the burial.
A sample of bone from burial 7/I was also analyzed for
stable isotopes at Gisela Grupe’s laboratory in Munich
(Grupe et al. 2003; Borić et al. 2004). The results (δ13 C,
-19.7‰; δ15N, +11.5‰) differ markedly from those obtained
at East Kilbride and Oxford. The difference between the two
 The Iron Gates in Prehistory
Table 2 Stable isotope results for Burial 7/I (right femur)
measured at the East Kilbride and Oxford Radiocarbon
Laboratories.
Date collected Laboratory δ13C ‰ δ15N ‰
1989 East Kilbride -18.4 +15.8
1989 Oxford -18.1 +16.2
2000 Oxford -17.5 +16.6
sets of stable isotope measurements (East Kilbride/Oxford vs
Munich) is substantial, and not easily explained in terms of
inter-laboratory differences in analytical procedures. Grupe
et al. (2003) did not report which skeletal element was
sampled for their study. If the sample was also taken from the
right femur, then the discrepancy in the stable isotope results
must be due either to analytical error or to a curatorial error
between the time of sample collection and processing. The
consistency of the results from East Kilbride and Oxford im-
plies that the samples collected in 1989 and 2000 would not
have been the source of the error. On the other hand, if Grupe
et al. (2003) sampled a different skeletal element, then a third
possibility exists — that the right femur sampled in
1989/2000 and the (unidentified) bone analyzed by Grupe did
not belong to the same individual, perhaps reflecting curat-
orial problems prior to sample collection (see above).
Borić et al. (2004) attempted to explain the discrepancy in
the stable isotope results in terms of misreporting of the
samples that were analyzed. The burials in grave 7 were re-
ferred to by Nemeskéri (1969, 1972) and Zoffmann (1983) as
burial 7/I (skeleton) and burial 7/II (isolated cranium), re-
spectively. Other authors have referred to them as burials 7a
and 7b. Borić et al. (2004) equate burial 7a with 7/I.
However, when the skeleton (7/I) was sampled in the
Department of Archaeology in Belgrade in 1989 and 2000 it
was clearly labelled ‘burial 7b’; the equivalence of ‘7/I’ and
‘7b’ was also reported by Roksandic (1999). Regardless of
whether burial 7/I equates to ‘7a’ or ‘7b’, it can be stated
categorically that the bone samples taken in 1989 and 2000
did not come from the cranium (7/II) as suggested by Borić
et al. (2004).
Discussion
Burial practices
A number of distinct forms of burial are represented at
Lepenski Vir. They include: (i) extended (supine) inhuma-
tion; (ii) supine inhumation in the ‘butterfly position’;
(iii) crouched inhumation; and (iv) burial of disarticulated
bones.
Extended supine inhumations
This burial mode is where the individual is laid on their back
with the legs extended straight out (Figs 3–9). It appears to
have been a characteristic burial form of the Mesolithic of the
Iron Gates, directly dated examples from other sites in the
region ranging in age between c. 9800 and 7500 BP (Bonsall
et al. 1997; Boroneanţ et al. 1999; Borić & Miracle 2004).
Ten examples of this burial type from Lepenski Vir (buri-
180
als 7/I, 14, 26, 29, 30, 54c, 54e, 60, 61, and 62) were included
in the current study. The results fall into three distinct periods:
1. By far the earliest in the series is burial 60 with a
reservoir-corrected 14C age of 9020±80 BP, which places it
in the Early Mesolithic. The skeleton is probably that of an
adult male, aged between 25 and 40 years at the time of
death (Roksandic 1999). Located in the south-east part of
the site, just beyond the zone with later trapezoidal build-
ings (which may explain why the grave remained more or
less undisturbed prior to excavation) this burial is unusual
at Lepenski Vir in being oriented more or less perpendic-
ularly to the Danube (SW–NE) with the head away from
the river. Radovanović (1996a: 186) reported only one
other adult extended inhumation of presumed Mesolithic
age from the site (burial 28) with the same orientation.
Extended inhumations laid out perpendicular to the
Danube are better represented at other sites in the Iron
Gates gorge, such as Hajdučka Vodenica, Padina and
Vlasac but, according to the 14C evidence, were not con-
fined to the Early Mesolithic. Burial 60 is the earliest dated
burial from Lepenski Vir, and one of oldest known from
the Iron Gates region — burials dating before 9000 BP
were also found at Padina (Borić & Miracle 2004) and
Vlasac (Bonsall et al. 1997). The 14C date for burial 60,
together with stable isotope values reflecting a diet heavily
dependent on aquatic resources (cf. Bonsall et al. 1997),
suggests that Lepenski Vir was already being used as a
residential site or seasonal fishing camp in the very early
Holocene.
2. Six burials (7/I, 14, 26, 54c, 54e, and 61) have reservoir-
corrected 14 C ages ranging between 7374±80 and
7200±562 BP, and the 1-sigma calibrated age ranges over-
lap indicating that the results are statistically indistin-
guishable. The dates suggest that the burials coincided
with the well-documented cold phase between
c. 6400–6000 cal BC, known as the ‘8200 cal BP event’
(Alley et al. 1997; Clark et al. 2001; Magny 2003) — a
period that is not well represented by radiocarbon dates
from other Iron Gates sites (Bonsall et al. 2002a).
Archaeological opinion is divided on whether this phase in
the use of Lepenski Vir should be interpreted as ‘Final
Mesolithic’ (e.g. Bonsall et al. 2004), ‘Neolithic’ (e.g.
Borić 2002b), or ‘transitional’ (e.g. Roksandic 2000; Borić
et al. 2004). Of the six dated burials, five were adults (in-
cluding both males and females) and one (burial 61) was a
child aged 2–6 years (Roksandic 1999). With the excep-
tion of burial 14, they occurred within trapezoidal build-
ings of ‘LVI–II’. In contrast to the much earlier burial 60,
all were oriented roughly parallel to the river with the
heads downstream, although there was some variation in
body position, notably in the placement of the arms
(Appendix 1). Burial 54c is distinctive in that the skull was
missing; in contrast isolated crania were found in several
contexts at Lepenski Vir, including grave 7 (see above).
Since there are clear parallels in earlier contexts elsewhere
in the region, including Schela Cladovei and Vlasac, buri-
al 54c may represent the continuation of a Mesolithic tra-
dition of skull removal and skull caching in the Iron Gates.
Burials in the extended, supine position and oriented par-
allel to the Danube are very well-represented at Lepenski
 Clive Bonsall et al.: Dating burial practices and architecture at Lepenski Vir

Figure 3. Burial 26 inserted through the plaster floor of building 34. Photo: Institute of Archaeology, Belgrade.

Figure 4. Building 34 and the outline of grave 26 prior to excavation. Photo: Institute of Archaeology, Belgrade.

181
 The Iron Gates in Prehistory

Figure 5. Burial 7 (an extended supine

inhumation) inserted through the plaster
floor of building 21. The photograph shows
the burial in process of excavation and what
appears to be the outline of the ‘grave pit’.
Photo: Institute of Archaeology, Belgrade.

Figure 6. Burial 7 after excavation. The

‘grave pit’ has been artificially enlarged, and
the positions of various items interpreted as
burial goods can be seen — a human
cranium (7/II) above the left shoulder of the
skeleton, an aurochs skull with horns above
the right shoulder, and below the hearth a red
deer skull with antlers attached. Photo:
Institute of Archaeology, Belgrade.

182
 Clive Bonsall et al.: Dating burial practices and architecture at Lepenski Vir

Figure 7. Burial 54c (extended supine

inhumation) in building 65, lacking the skull.
Disarticulated bones from other individuals
can be seen at a higher level, to the right of
the articulated skeleton. Photo: Institute of
Archaeology, Belgrade.

54e

54c
Figure 8. Burials 54e and 54c (extended
supine inhumations) in building 65. 54e lies
directly on the floor of the building; a disar-
ticulated skull and bones belonging to one or
more other individuals occur around and
resting on the upper body of the skeleton.
Photo: Institute of Archaeology, Belgrade.

183
 The Iron Gates in Prehistory
Figure 9. Burial 61 (extended supine
inhumation of a juvenile) in building 40.
Photo: Institute of Archaeology, Belgrade.
Vir. Radovanović (1996a: 174–187) lists a number of oth-
er examples (both adults and children, including
neonates), and some of these, though not all, were found
within or under trapezoidal buildings. It remains to be es-
tablished by direct 14C dating how many of them also be-
long to the period between 6400 and 6000 cal BC.
3. The remaining three burials are much later in date. The 14C
ages for burials 29, 30 and 62 are very similar at 445±31,
477±34 and 445±63 BP, respectively. These and a fourth
burial (49) were identified in the excavation records from
Lepenski Vir as probably belonging to the ‘Slavic’ period,
i.e. after the 7th century AD. All four burials are described
by Roksandic et al. (2007). The three dated individuals
were found close together in the north-west corner of the
site (Fig. 2). The skeletons were oriented more-or-less
E–W, with the heads to the west — similar to Early
Mesolithic burial 60, and distinct from the ‘period 2’
burials. The excavation records suggest that they occurred
at about the same depth with no associated grave goods.
Although the burials were described as well-preserved,
only a few bones appear to have been kept by the excav-
ators. Osteological analysis suggests that all four indi-
viduals were young or middle-aged adult males. The skull
of one individual shows signs of an injury caused by a
sharp instrument, probably a heavy metal-bladed weapon,
although signs of healing of the wound indicate that this
was not the cause of death. Roksandic et al. (2007) have
suggested that the four ‘Slavic’ period burials were those
of soldiers involved in border warfare along the Danube,
which was prevalent in the late 14th and early 15th cen-
turies, except for a brief period of peace between 1403 and
1425. Soldiers would not necessarily have been born or
lived locally and if no correction is applied for the River
Danube reservoir effect, then the pooled mean of the 14C
ages of the three dated individuals, which is 458±22 BP
(1415–1455 cal AD), is consistent with the interpretation
proposed by Roksandic et al. (2007).
1
184
Dorsal decubitus inhumation in the ‘butterfly’ position
Burial 69 (Fig. 10) is the only example of a burial from
Lepenski Vir in this position where the individual is lying on
the back (dorsal decubitus) and the legs are flexed and
splayed with the soles of the feet together, reminiscent of the
butterfly pose in yoga. Like burial 60, burial 69 was found in
the south-east corner of the site outside the zone with
trapezoidal buildings, and was oriented perpendicularly to
the river with the head upslope. The burial occurred in ‘vir-
gin’ soil, beneath a pile of stones. According to Srejović
(1972a: 117–118) the body was placed in a trapezoidal grave
pit c. 1.10m long and the position of the head, bent forward
on the chest, was taken to indicate that it had been supported
by the rear wall of the grave.
The similarity in ‘shape’ and orientation between burial 69
and the majority of the trapezoidal plan buildings at Lepenski
Vir led Srejović to suggest that the burial belonged to a phase
immediately preceding the construction of the first plaster-
floored buildings (Srejović 1972a: 117). However, the
reservoir-corrected 14C age for burial 69 (8784±72 BP) is
substantially older than the ages of the charcoal samples from
the buildings, implying that it derives from a much earlier
period in the use of the site. Thus burial 69 can be regarded
as Early Mesolithic, but on the radiocarbon evidence prob-
ably somewhat later than burial 60.
Borić & Miracle (2004) identified parallels for burial 69 at
a number of other Iron Gates sites, specifically Padina (buri-
als 15 and 16), Vlasac (burial 17), Kula (burial 5), Velesnica
(burial 2G), and Ostrovul Corbului (burial 25). In most of
these cases, however, the resemblance is not exact. For ex-
ample, burials 15 and 16 at Padina are in a sitting position
with the legs crossed (Radovanović 1996a: fig. 4.1; Borić &
Miracle 2004: fig. 8), and burial 5 at Kula (Sladić 1986) also
has the legs crossed. Only Velesnica burial 2G (Vasić 1986,
this volume; Roksandic, this volume) is laid on the back
(dorsal decubitus position) with the soles of the feet together.
Currently, the only directly dated burial from this group is
 Clive Bonsall et al.: Dating burial practices and architecture at Lepenski Vir
Figure 10. Burial 69 (dorsal decubitus inhumation in the ‘butterfly’ position). Left: the burial with its original covering of
stones. Right: after removal of the stones. Photos: Institute of Archaeology, Belgrade.
burial 15 from Padina, with a reservoir-corrected 14C age of
9138±71 BP; another ‘seated’ burial from Padina (burial 21)
was dated at 9729±73 BP (Borić & Miracle 2004: table 3).
On this evidence, the seated burials from Padina are Early
Mesolithic, but somewhat older than burial 69 at Lepenski
Vir.
Crouched inhumations
There are a number of examples of burials at Lepenski Vir
where the body was laid out in a crouched or tightly flexed
position (as defined by McKinley & Roberts 1995: 4). In the
wider region of the central and northern Balkans crouched
inhumation was characteristic of the Early Neolithic Starčevo
culture (Tringham 1971), but was also common in the later
Neolithic to Early Bronze Age (e.g. Ciugudean 1996; Whittle
1996).
We dated six crouched inhumations from Lepenski Vir
(Table 1; Fig. 11). One of these, burial 2, was associated with
pottery of the Salcuţa culture. The reservoir-corrected 14C age
of 5269±54 BP confirms the Chalcolithic (Eneolithic) dating.
Burials 8, 9, 32a and 88, which were assigned by Srejović
(1969) to the Starčevo culture, gave reservoir corrected ages
between 7036±95 and 6942±47 BP. These ages are statistic-
ally indistinguishable and confirm the Early Neolithic attri-
bution. The stable isotope (δ13C and δ15N) values for these
four burials and burial 2 (Table 1) are indicative of diets
based largely on terrestrial food sources, further supporting
the post-Mesolithic dating.
Borić and Dimitrijević have also reported reservoir-
corrected dates for crouched inhumations, though only in
graphical form (Borić & Dimitrijević 2007: fig. 3). Burial 19
in building XLIV/57 has a calibrated age of c. 5950 BC
185
(which equates to c. 7080 BP, uncalibrated) consistent with
the Early Neolithic. Burial 5, found in grid square A/VI (Fig.
2), may also be of Early Neolithic date. A bone tool reported
to have been found with this skeleton has a calibrated 14C age
of c. 6015 BC (which equates to c. 7135 BP, uncalibrated)
(Borić & Dimitrijević 2007: fig. 3). However, there is some
doubt about the association of the skeleton and the bone
artefact; in the Field Burial Record for Lepenski Vir it is re-
corded that no grave goods were found with burial 5, while an
entry in the Field Journal for 19 July 1967 includes a note
that a “ bone awl was found in the pelvic area”.
Other, undated, crouched inhumations from Lepenski Vir
were assigned by Srejović to his IIIa and IIIb (Starčevo cul-
ture) phases. However, in the absence of direct 14 C age
measurements, it would be premature to conclude that they
are all of Early Neolithic date. Although crouched burials are
rare in Southeast Europe after the Early Bronze Age, they do
occur occasionally even as late as the Medieval and post-
Medieval periods where the rite seems to be associated with
social outcasts such as ‘witches’ or ‘mad’ people, the bodies
being bound up to prevent the individuals from ‘causing
trouble’ (A. Choyke, pers. comm.; Taylor 2002).
It is interesting, therefore, that another crouched inhuma-
tion included in our dating programme, burial 4 (Fig. 11,
right), produced the unexpectedly late 14C date of 485±31 BP
(OxA-11699). This is very similar to the dates obtained on
bones from burials 29, 30 and 62 (see above), which were
buried in the Late Medieval (Christian) tradition, lying on the
back, fully extended, with the head to the east. The associated
stable isotope values are also in agreement with a Late
Medieval date; the relatively heavy δ13 C (combined with
light δ15N) is characteristic of post-Bronze Age populations
 The Iron Gates in Prehistory
Figure 11. Crouched inhumations. Left: burial 8 (bottom) and burial 9 (top), above building 24. Right: burial 4, above
building 25. Photos: Institute of Archaeology, Belgrade.
in parts of Southeast and Central Europe, and probably re-
flects a farming economy where millet (a C 4 plant) was a
staple crop grown for human consumption and/or animal
fodder (Murray & Schoeninger 1988; Bonsall et al. 2000,
2004, 2007). The chances of the paired 14C and stable isotope
results being erroneous due to analytical error, or that one of
the Late Medieval samples submitted for dating was inad-
vertently measured twice, are extremely small. A more likely
explanation is that either burial 4 is Late Medieval in date, or
the bone that was sampled came originally from another
burial and was wrongly assigned to burial 4. In view of this
uncertainty, OxA-11699 should be regarded as suspect and
new 14 C measurements undertaken to confirm the date of
burial 4.
Disarticulated bone burials
At Lepenski Vir there were many finds of disarticulated hu-
man bones, individually or in groups. It is likely that some
(perhaps the majority) of these occurrences result from unin-
tentional post-depositional disturbance of articulated burials
and dispersal of the bones, either by later human actions (e.g.
the digging of new graves or building foundations) or by
natural processes. However, disarticulated bones were some-
times found in contexts suggesting deliberate burial of the
bones or, in some cases, of body parts still held together by
soft tissue. These were either buried separately or added to
graves containing an intact body. They are generally inter-
preted as instances of ‘delayed’ or secondary burial linked to
the practice of excarnation (Srejović 1972a: 117), which
raises the possibility of storage and even ‘use’ of the (ances-
tral) bones for a time prior to burial. This treatment was af-
186
forded to the bones of adults (males and females) and
children.
Samples from six disarticulated bone burials were in-
cluded in our dating programme. These were from burial 54d
in building 65 (Fig. 8), burial 45b in building 61 (Fig. 12),
and burials 31a, 44, 79a, and 89a. With the exception of
burial 79a where a scapula was sampled, all the samples for
dating were taken from the shafts of long bones.
Burial 54d has been the source of some debate. It consists
of a cranium and several long bones which were found in
various positions around or resting on the upper body of a
more or less complete, articulated skeleton, burial 54e (Fig.
8). According to the field notes made at the time of excava-
tion (see also Srejović 1972a: 118), burial 54d was con-
sidered to be the earliest burial in building 65 and originally
articulated, but subsequently disturbed by the insertion of
burial 54e. However, the fact that the bones of burial 54d are
so few in number and may belong to more than one indi-
vidual (Roksandic 1999) casts doubt on this interpretation. It
is also worth noting that the detached skull of 54d rests on the
left shoulder of skeleton 54e, similar to the position of the
detached skull (7/II) with respect to the articulated skeleton
(7/I) in grave 7 (compare Figs 6 & 8). Thus, it seems much
more likely that burial 54d represents an instance of deliber-
ate secondary disposal of disarticulated bones or body parts,
that were emplaced at the same time as burial 54e (see also
Radovanović 2000: 336).
In cases of secondary disposal, the time interval between
the death of the individual and final burial of the bones is
unknown. Thus, radiocarbon dating will only provide an ap-
proximation of the time of death but not of final interment,
Clive Bonsall et al.: Dating burial practices and architecture at Lepenski Vir

Figure 12. Disarticulated bone burial (45b)

inserted through the plaster floor of building
61. Photo: Institute of Archaeology, Belgrade.

Figure 13. Disarticulated human cranium

(burial 122) above the hearth of building 47′.
Photo: Institute of Archaeology, Belgrade.

187
 The Iron Gates in Prehistory

Table 3. Srejović–Zoffmann phasing of the burials. Body burial of a detached skull (Borić & Stefanović 2004: fig. 7),
position: Es – extended supine; Dd – dorsal decubitus; However, the field notes pertaining to this burial record that
C – crouched; D – disarticulated; ? – uncertain. it was only partially excavated; it occurred at the very edge of
a trench, several bones including a skull were removed, but
Burial Reservoir-corrected Body position Srejović–Zoffmann other bones were left “in the section” unexcavated (Appendix
14C age BP phasing
1). This description implies that burial 35 was not an instance
60 9020±80 Es Ic of skull caching, but possibly an articulated (crouched?)
69 8784±72 Dd Proto-LV burial. Another possible, post-Mesolithic example of a sec-
61 7374±80 Es Ic
ondary burial that was included in our dating programme is
14 7368±75 Es I–II
54d 7353±72 D Ib
burial 18. Dated to the Roman period (Table 1), it is described
54c 7346±57 Es Ib in the field notes as ‘disarticulated’. However, judging from
45b 7337±79 D I the field description, the bones could be from a disturbed
79a 7312±79 D — context rather than a case of deliberate interment of disartic-
31a 7308±108 D IIIb ulated remains.
26 7284±47 Es I
54e 7250±59 Es Ib Phasing of the burials
7/I 7218±81 Es I
The validity of the phasing of the Lepenski Vir burials pro-
44 7152±106 D IIIb
89a 7133±75 D II posed by Srejović and Zoffmann (Srejović 1969; Zoffmann
32a 7036±95 C IIIb 1983; Radovanović 1996a; Srejović, pers. comm. 1989) was
9 6982±50 C IIIb first questioned by Bonsall et al. (1997) on the basis of bone
88 6980±92 C IIIa collagen stable isotope analyses. They noted that several
8 6942±47 C IIIb burials assigned to LVIIIa or IIIb yielded heavy δ15N values,
35 6718±93 ? IIIb similar to those obtained from the majority of skeletons as-
2 5269±54 C Salcuţa
signed to the ‘Mesolithic’ phases LVI and LVII (Bonsall et al.
18 1724±44 ? IIIb
62 445±63* Es Medieval
1997: table 5). Moreover, two of the ‘LVIII’ burials produced
14C ages that were more consistent with the Mesolithic, al-
29 445±31* Es Medieval
30 477±34* Es Medieval though at that time a reservoir correction was not being ap-
plied to the 14C ages.
* No reservoir correction applied.
The extension of this dating programme, detailed here,
provided a further opportunity to test the Srejović–Zoffmann
which may have occurred weeks to centuries later. The phasing of the burials (Table 3). The results show that when
reservoir-corrected ages of the six disarticulated burials in- the human remains had a clearly identifiable burial rite and/or
cluded in this study range from 7353±83 to 7133±75 BP, diagnostic burial goods, the phasing proposed by
while the δ13C and δ15N ranges are -17.7 to -18.6‰ and +15.2 Srejović/Zoffmann was broadly correct. For example, all the
to +15.8‰ respectively, indicating diets that were heavily extended supine burials are assigned to phases I or II and all
dependent on aquatic food sources. This evidence indicates the crouched burials are assigned to phase III, except for
that the practice of ‘delayed’/secondary burial was certainly burial 2, which was associated with typical Eneolithic
applied to the bones of Mesolithic people. Even if the bones (Salcuţa) pottery and assigned to that period.
are from ‘Mesolithic’ people, it is conceivable (though un- However, when the disarticulated burials are considered,
likely) that final burial of the remains took place after the there is a poor correlation between the Srejović–Zoffmann
Mesolithic. Similar evidence of secondary burial is well phasing and the radiocarbon ages. Similarly, within the group
documented from older Mesolithic contexts elsewhere in the of extended supine burials, there is a poor correlation
Iron Gates, notably at Schela Cladovei on the Romanian between the sub-phase assignments (LVIa–e) and the ra-
(left) bank of the Danube (Boroneanţ et al. 1999; Bonsall diocarbon ages (to a lesser extent, the same is perhaps true of
2003). the crouched burials with respect to the LVIIIa–b sub-
It is an interesting question whether the practice of sec- phasing). The most likely explanation for this is that phasing
ondary burial continued into later periods at Lepenski Vir. of these ‘undiagnostic’ burials relied on stratigraphic obser-
Zoffmann (1983: 130) reported that disarticulated human vations or the level at which a burial occurred — further
bones were found in “Starčevo pits and Starčevo layers” at evidence, perhaps, of the stratigraphic complexity of
Lepenski Vir. The Early Neolithic crouched inhumation Lepenski Vir.
(burial 19) from building XLIV/57 dated by Borić and A special case is represented by burial 69, which was as-
Dimitrijević (2007: fig. 3) lacked the skull. According to the signed by Srejović (1969, 1972a) to his ‘Proto-Lepenski Vir’
Field Burial Record, a separate cranium occurred on a stone phase. Srejović’s phasing of the Lepenski Vir site (as opposed
plaque above the skeleton, and a maxilla was found to the to the burials) was based primarily on architectural evidence.
south at the same level as the skeleton. The skeleton, cranium The ‘Proto-Lepenski Vir’ phase was represented by remnants
and maxilla were assumed by the excavators to belong to the of stone-bordered hearths, which occurred in a narrow zone
same individual, but there are no 14C dates or stable isotope along the riverbank between the normal high water mark and
measurements for the cranium or maxilla to support this in- the trapezoidal buildings of ‘Lepenski Vir I–II’. There are no
terpretation. Burial 35, dated to the Starčevo period (Table 1; radiocarbon measurements on organic materials associated
Bonsall et al. 1997) has been interpreted as the secondary with the ‘Proto-Lepenski Vir’ hearths and no direct strati-

188
 Clive Bonsall et al.: Dating burial practices and architecture at Lepenski Vir
graphic relationship was observed between these hearths and
the buildings of Lepenski Vir I. Equally, there is no docu-
mented stratigraphic relationship between the hearths and
burial 69. Srejović assigned burial 69 to ‘Proto-Lepenski Vir’
partly on the basis of the level at which it was found (said to
be equivalent to the level of the ‘Proto-Lepenski Vir’ hearths)
but primarily on the basis of the body position and the shape
and orientation of the grave. As noted above, this led Srejović
to suggest that the burial could be seen as the precursor of the
architectural tradition represented by the trapezoidal plaster-
floored buildings, and that it belonged to a phase immediately
preceding the earliest phase of their construction (Srejović
1972a: 117–118). Radiocarbon dating has shown that, while
burial 69 does indeed predate the earliest known trapezoidal
buildings, it does not immediately precede them in time, nor
is it necessarily the earliest burial on the site.
Architecture
The dating of the trapezoidal, plaster-floored buildings at
Lepenski Vir rests on several lines of evidence: (i) the ra-
diometric 14C ages of charcoal samples from the buildings
that could be interpreted as representing structural timbers or
fuel; (ii) AMS 14C ages of human burials with a direct strati-
graphic relationship to a building; (iii) AMS 14C ages of ter-
restrial animal bone material that was potentially related to a
building; and (iv) the stratigraphic relationships of the build-
ings themselves, i.e. where one building was superimposed
(partially or wholly) on another. In principle, the charcoal
ages provide absolute dates for the construction or use of the
buildings, while the other lines of evidence provide informa-
tion on the relative ages of the buildings.
Charcoal ages
The first attempt to establish an absolute chronology for the
major architectural features (the trapezoidal buildings) at
Lepenski Vir was based on radiometric 14C dating of charcoal
recovered from a number of the buildings. Charcoal samples
from 14 of the trapezoidal-plan, plaster-floored structures
that are unique to this site gave 14 C ages ranging from
7430±160 to 6560±100 BP. This provides a possible time-
range for the buildings of between 7750 and 6360 BP
(c. 6590–5320 cal BC). According to Quitta (1972), the
charcoal samples were recovered from:
“... house floors or from the occupation layers immediately
above them. Only in a few cases were they recognisable as
elements of a house: parts of a burnt beam, for instance, from
houses 36 and 37” (Quitta 1972: 205).
However, Borić (2002a: appendix 1) has provided informa-
tion based on the original excavation records, which suggests
that for at least 16 of the samples dated the charcoal came
from contexts contemporaneous with the construction or use
of the buildings. On the basis of the data presented by Borić
(2002a: appendix 1), the buildings were built, or at least in
use over a period from at least 7335±70 BP (weighted mean
of two determinations for building 36) to 6620±100 BP
(single date for building 51), equivalent to an overall time-
range from 7475 to 6420 BP (6374–5380 cal BC).
This interpretation is complicated by two potential sources
of error, (i) the charcoals from the ‘houses’ were from long-
lived species — oak (Quercus sp.) and elm (Ulmus sp.) —
189
which could produce an ‘old wood effect’, and (ii) some
samples representing structural timbers could have been re-
cycled from earlier buildings. Both of these would result in
14C ages that are older than the true construction ages of the
buildings. It should also be borne in mind that only a relat-
ively small number of contexts were dated and, therefore, the
range of ages may not be representative of the entire
timespan over which construction of the trapezoidal build-
ings took place.
Dating buildings in relation to burials
A significant number of the burials at Lepenski Vir occurred
in a direct stratigraphic relationship to trapezoidal structures
and thus potentially provide additional information on the
ages of the buildings. Nine such burials were included in our
dating programme; they comprise seven examples of articu-
lated burials, and two disarticulated bone burials. The fol-
lowing general principles are adopted here. A known-age
burial that was sealed by a plaster floor establishes a terminus
post quem (TPQ or maximum age) for the overlying building.
A known-age articulated burial that cuts through a plaster
floor can be presumed to be no older than the floor and
therefore establishes a terminus ante quem (TAQ or minim-
um age) for the building.3 However, a disarticulated bone
burial that was inserted through a plaster floor does not es-
tablish a TAQ for the building, since the bones may have
been stored for a significant number of years before burial.
Similarly, a known-age articulated burial that overlies a
plaster floor provides a TAQ for the building, but a disarticu-
lated burial overlying a plaster floor does not.
a) Burials inserted through plaster floors
Two articulated burials fall into this category: burial 7/I in
building 21 (Figs 5 & 6), and burial 26 in building 34 (Figs
3 & 4). The reservoir-corrected 14C ages of these burials are
very similar: 7218±81 BP and 7284±47 BP, respectively. In
the case of burial 26 there is clear photographic and docu-
mentary evidence that the grave was dug through the plaster
floor of building 34 (see Fig. 4). We are not aware of any
photograph or field drawing of building 21 prior to the ex-
cavation of burial 7/I that shows a similar grave-sized dis-
turbance in its plaster floor. However, if the excavator’s in-
terpretation (Srejović 1972: 120) is correct, then burials 7/I
and 26 must be younger than the buildings to which they re-
late. Thus building 34 was probably built no later than 6056
cal BC and building 21 no later than 5917 cal BC (the
respective younger limits of the 2-sigma calibrated age
ranges). Since building 21 is superimposed upon buildings
22, 29 and 30 (Fig. 2), the 14 C age of burial 7/I also
establishes a TAQ for these structures.4
b) Burials overlying building floors
Articulated burials overlying the plaster floors of two
trapezoidal buildings at Lepenski Vir were included in our
dating programme.
Burials 8 & 9 in building 24 have reservoir corrected 14C
ages of 6942±47 and 6982±50 BP, respectively, and exhibit
the crouched body position characteristic of Early Neolithic
(Starčevo culture) burials from the central and northern
Balkans. Judging from photographic evidence (Fig. 11, left),
 The Iron Gates in Prehistory
the skeletons lay above the plaster floor but were not in con-
tact with it, suggesting that both individuals were buried in
the infilling of the ‘house pit’ — it is not known if the corpses
were placed in grave pits, or laid directly on the ground and
covered with soil material or stones. Thus, the interments
must have occurred sometime after building 24 was aban-
doned, although their apparently deliberate placement with
respect to the structure suggests the ‘house pit’ was still vis-
ible (i.e. not completely filled in) at the time of burial. On this
evidence, building 24 is probably no younger than 5729 cal
BC (the younger limit of the 2-sigma calibrated age range of
burial 8), but it could be earlier than that date.
Other burials were found within or adjacent to building 24.
They include the (articulated?) skeletons of four neonates
(burials 94, 95, 101 and 102). These were not discovered un-
til 1970, when the floor of building 24 was lifted as part of
conservation work at the site — that is, several years after
building 24 was first exposed and burials 8 and 9 excavated.
Judging from the positions of burials 94 and 101 with respect
to burial 8,5 they may have been emplaced before burial 8.
However, an AMS 14C date for burial 94 reported by Borić &
Dimitrijević (2007: fig. 3) is not significantly different from
that for burial 8, i.e. any age difference is too small to be re-
solved given the precision on the AMS measurements. Borić
& Stefanović (this volume; see also Stefanović & Borić
2004: fig. 10) have suggested that the neonates were buried in
pits among stones at the rear of building 24. However, it is
not clear from the original excavation records whether the
stones represent a stone construction built to accommodate
the burials, as suggested by Borić & Stefanović (this volume,
page 144), or were part of the foundations of building 24, or
elements of an earlier trapezoidal structure (building 24a —
cf. Srejović 1969: 89) on which building 24 was superim-
posed. Whatever the explanation, the stratigraphic relation-
ship of the neonate burials to building 24 is unclear, and so
they provide no reliable additional information about the
dating of building 24.
Building 65, in the south-west corner of the site, contained
two articulated burials (54c and 54e) and several disarticu-
lated burials (54a, 54b, 54d, and 47). The articulated burials
are extended inhumations, with the same orientation, parallel
to the Danube (see above). Judging from photographic evid-
ence (Figs 7 & 8), skeleton 54e lay directly on the floor of
building 65, while skeleton 54c occurred within the infill of
the ‘house pit’. As noted above, skeleton 54c lacked the skull
— one of several examples of ‘headless’ skeletons from
Lepenski Vir. Since there is no evidence in any of these cases
of actual decapitation (in the form of cut marks), it is likely
that the skulls were removed after decomposition of the soft
tissue. This raises the question of how defleshing and skull
removal were achieved without otherwise disturbing the
skeleton. Deep burial followed by exhumation seems un-
likely. Shallow burial or surface exposure to allow the flesh to
rot away carries the risk of disturbance by scavengers;
moreover, removal of the flesh by mammalian and some avi-
an scavengers would be expected to leave marks on the
bones, for which there is no evidence. Another method would
have been to lay the body out on the ground and cover it with
stones. Figure 7 shows building 65 at an early stage of ex-
cavation. Visible on the photograph is an elongated heap of
190
stones in a position that appears to correspond with that of
extended burial 54e at a lower level (cf. Fig. 8). This suggests
that burial 54e was laid directly on the floor of building 65
and covered with stones. The same may have been done with
burial 54c and the cairn subsequently dismantled or partially
dismantled to allow the skull to be removed. Cairn burial was
not uncommon in the Iron Gates Mesolithic; examples are
known from several sites, including Lepenski Vir and Padina,
dating back to the Early Mesolithic.
If articulated burials 54e and 54c were cairn burials, their
respective positions in relation to the floor of building 65
would imply that 54e is the older burial. The 14C ages of
7250±59 BP (54e) and 7346±57 BP (54c) are not signific-
antly different and therefore neither support nor contradict
this interpretation, but they do suggest that the two burials
were not widely separated in time. Since the burials overlie
the plaster floor, the 14C data suggest that building 65 was
constructed some time prior to 6017cal BC (the younger limit
of the 2-sigma calibrated age range of burial 54e).
c) Burials sealed by plaster floors
A significant number of the burials from Lepenski Vir were
reported as having been found below the plaster floors of
trapezoidal buildings. Many of these were neonates (see
Stefanović & Borić 2004; Borić & Stefanović, this volume),
but they include the burials of older individuals. Often,
however, it is not clear whether a burial was sealed by a
plaster floor, or was inserted through the floor. Also, there is
always the possibility of a plaster floor being repaired fol-
lowing interment, and the repair not recognized during
excavation.
Burial 61, the extended supine inhumation of a child
(Fig. 9), has sometimes been cited as an example of an artic-
ulated burial found below the plaster floor of a trapezoidal
building, and for that reason was included in our dating pro-
gramme. Srejović (1969, 1972a) described burial 61 as
“...buried beneath the rear wall of building no. 40” He went
on to observe that “the skeleton ... takes up the whole of the
‘rear’ wall, and a representational sculpture was placed on the
floor above its head” Accordingly, he suggested that, “the
burial took place immediately prior to the construction of the
house” (Srejović 1972a: 119). If this were the case, then the
reservoir corrected age of 7374±80 BP would indeed estab-
lish a TPQ for building 40. However, some aspects of this
interpretation seem open to question. Building 40 was ex-
posed during the 1967 excavation season. Burial 61 appears
to have been found in 1968 when digging adjacent to the
south-west ‘corner’ of the structure. This can be understood
in terms of the method of excavation adopted at Lepenski Vir,
which usually involved exposing the floor of a building and
then continuing the excavation vertically beyond the peri-
meter of the structure, leaving the remains of the building on
a soil pedestal — as can be seen on numerous photographs of
the site (e.g. Srejović 1972a: plate III). Since the burial ex-
tended ‘under the floor’ of building 40 (Field Burial Record,
August 1968) and no trace of a grave pit was observed when
the floor was first exposed, it was assumed that the burial
must be older than building 40 (Srejović 1972a: 119).
However, judging from the available field drawings and
photographs, building 40 was not well preserved. The precise
 Clive Bonsall et al.: Dating burial practices and architecture at Lepenski Vir
limits of structure were unclear, and there were gaps in the
plaster floor around the hearth and at the margins of the
structure, including at the rear where burial 61 was later
found. From this, it is not clear if the floor at the rear was
damaged by the insertion of burial 61, or whether the floor
was laid after the burial was emplaced and was damaged at a
later date. Thus, it is not possible to say whether burial 61 is
older or younger than building 40.
Borić and Dimitrijević (2007: fig. 3) report two AMS 14C
dates on a human cranium (burial 122) found in the south-
west corner of the hearth of building 47′ (Fig. 13), underneath
the floor of building 47. Provided that the burial was sealed
by the plaster floor of building 47, and not inserted through
the floor (no information is available to us that would confirm
or refute this) then the 14C dates for the cranium would es-
tablish a TPQ for building 47. In other words, building 47
would be no older than the older limit of the 2-sigma calib-
rated age range of the skull, c. 6200 cal BC (based on graph-
ical information presented by Borić & Dimitrijević 2007: fig.
3). However, if the cranium had been curated for some con-
siderable time prior to burial, the TPQ could substantially
overestimate the age of building 47.
Dates on animal bones
Borić & Dimitrijević (2007) presented 32 AMS 14C dates on
mammalian bones from Lepenski Vir, of which two had been
published previously (Whittle et al. 2002: 113). Details of the
bones and their 14 C ages were provided for only seven
samples. In the other 25 cases the radiocarbon dates were
presented only in graphical form (Borić & Dimitrijević 2007:
fig. 3), although from this it is possible to determine the ap-
proximate 2-sigma calibrated age ranges and median prob-
ability ages (to within 10–35 yr). Borić & Dimitrijević used
the animal bone ages to establish: 1) the main periods of
Mesolithic and Neolithic occupation at Lepenski Vir, 2) the
main period of construction/use of the trapezoidal buildings,
and 3) the timing of the introduction of agriculture (domestic
livestock) at the site.
Table 4. Summary of the ages of trapezoidal buildings at
Lepenski Vir based on radiocarbon dating of associated or
stratigraphically related finds. Numbers in italics and un-
derlined signify buildings with pottery.
Buildings older than → Date (cal BC) ← Buildings younger than
5500
5600
5700
24, 24a 5800 51
5900 32, 37
21, 22, 29, 30, 30b 6000 1, 9, 16, 47
34, 65 6100
6200
6300 54
6400
191
Of the 32 dates on animal bones, 25 were on bone samples
that were found above, below, or on the floors of buildings.
Unfortunately, information relating to the species, state of
articulation, or exact position of the bones was not provided,
as this will be discussed in a forthcoming paper (Borić &
Dimitrijević, in press). For present purposes, we assume that
all their samples were from species with largely terrestrial
diets (i.e. minimal aquatic input). In addition, we have to base
our conclusions on the very limited contextual information
that is provided for most of the samples.
We focus here on the utility of these samples for dating the
trapezoidal buildings, and apply the following principles:
1. Disarticulated bones found on or above a building floor
cannot be used to establish the age of the building, because
their taphonomic histories are unknown. Such samples
could be, a) closely contemporaneous with the building, if
the bones were deposited on the floor during use or on
abandonment of the structure, b) older than the building,
in the case of re-deposited or curated objects, or
c) younger than the building, in the case of bones intro-
duced by later bioturbation.
2. A bone found underneath a building in a context that was
clearly sealed by the plaster floor would provide a TPQ
(maximum age) for the building.
As many as 16 of the animal bones dated by Borić &
Dimitrijević (2007) potentially fall into the category of dis-
articulated bones found on or above a building floor. These
are: OxA-8618, OxA-15999, OxA-16000, OxA-16007,
OxA-16009, OxA-16071, OxA-16073, OxA-16075, OxA-
16076, OxA-16077, OxA-16081, OxA-16082, OxA-16083,
OxA-16084, OxA-X-2176-18, and OxA-X-2176-19.
Eight animal bone samples dated by Borić & Dimitrijević
(2007) are listed as having been found “between [two]
houses” or “underneath a house” — OxA-8610 (under build-
ing 23), OxA-15998 (between building 20 and building 33),
OxA-16001, OxA-16002 (between building 26 and building
26′), OxA-16003 (between building 35 and building 36),
OxA-16004, OxA-16005, OxA-16072 (underneath building
47′). If these samples were sealed by the plaster floors, then
the 14C ages would establish TPQs (maximum ages) for the
overlying buildings. However, the plaster floors of the
Lepenski Vir buildings were rarely intact; there were often
gaps in the plaster which could have been caused by post-
depositional disturbances, with the potential to introduce
younger material into ‘sub-floor’ contexts. Unless the exact
position of a bone in relation to an overlying plaster floor is
known (e.g. from photographic evidence, detailed plans, or
3-D coordinates), then it may not be possible to say for cer-
tain when the bone was emplaced relative to the construction
of the floor. Without access to this kind of information, we
feel unable to comment further on the chronological signific-
ance of the animal bone samples found below building floors
reported by Borić and Dimitrijević (2007).
Arguably, articulated bones (partial skeletons) are less
susceptible to post-depositional movement in certain cir-
cumstances (but see Coard & Dennell 1995), and so may
have greater potential for dating the trapezoidal buildings at
Lepenski Vir. Even so, problems can still arise. For example,
in the series reported by Borić and Dimitrijević (2007),
OxA-16078 dates a red deer skull with antlers found on the
 The Iron Gates in Prehistory
floor of building 28. Dimitrijević (this volume) has inter-
preted this as the result of a symbolic act on abandonment of
the building. Whilst this interpretation is reasonable, the ab-
sence of the mandible suggests that the deer skull was already
de-fleshed at the time of deposition. If so, then it may rep-
resent a curated item (e.g. a trophy) from an animal that died
long before its skull was deposited in the building, in which
case the 14C age reflects the death of the animal and not the
act of deposition in building 28.
Reviewing the evidence
Appendix 2 summarizes the data bearing upon the dating of
the trapezoidal buildings, including the 14C ages of the char-
coal samples, the stratigraphic information that relates the
charcoal to the buildings, and the stratigraphic relationships
between the buildings themselves and between buildings and
14C dated burials. Using this information it is possible to es-
tablish a terminus ante quem (TAQ) or terminus post quem
(TPQ) for certain of the buildings (summarized in Table 4).
The TAQ and TPQ values are based on the extremities of the
2-sigma calibrated age ranges. For example, building 1 has a
charcoal 14 C age of 6860±100 BP. This calibrates to
5982–5572 cal BC (95.4% probability range). The charcoal
is assumed to be derived from timber used in the construction
of the building or burnt in the hearth, but the 14C age of the
charcoal itself could have an ‘old wood effect’ associated
with it (or derive from re-used timber), and thus may overes-
timate the age of the building. Therefore, the maximum age
(TPQ) of building 1 would equate to the older limit of the
2-sigma calibrated age range of the charcoal (5982 cal BC),
while the minimum age (TAQ) would be equal to or less than
the younger end of the 2-sigma age range (≤5572 cal BC),
given the possibility of an old wood effect.
A number of problems arise when attempting to interpret
these data. Firstly, the dating of the trapezoidal structures still
relies heavily on the original charcoal 14C ages, and it could
be argued that many of the taphonomic issues that surround
the use of disarticulated bones as a source of chronological
information apply equally to charcoal. For example, charcoal
is just as susceptible to post-depositional movement, al-
though if a sample came from a beam (i.e. thought to have
been used in the construction of a building) or a hearth, we
can perhaps have greater confidence that it was in a primary
context. Secondly, there are a number of apparent conflicts
among the data which need to be considered:
1. The 14C age (6820±100 BP) of the charcoal sample from
building 34 is significantly younger than the reservoir-
corrected 14C age (7284±47 BP) of burial 26 that was in-
serted through the plaster floor. In other words, the
2-sigma calibrated age ranges of the charcoal (5972–5554
BC) and the burial (6233–6056 BC) do not overlap. The
charcoal sample was taken from a beam at the rear of
building 34 (Borić 2002a: appendix 1). Judging from
published plans (e.g. Srejović 1969: fig. 19; Borić 2002b:
fig. 7) the rear ‘wall’ of building 34 was more or less con-
tiguous with the front of building 43, which was inter-
preted as the later of the two structures (Srejović 1969,
1972a). If the charcoal sample (beam) was associated with
building 43 rather than building 34, then this could explain
the difference in the 14C ages of the charcoal and burial 26.
192
2. Charcoal from the hearth of building 27 has a 14C age of
7210±200 BP. Building 27 is reported as having been su-
perimposed on building 34, which has a charcoal age of
6820±100 BP (timber beam from rear — cf. Borić 2002a:
appendix 1). It is not clear from published plans/photos
whether this was in fact the case. Regardless, the 2-sigma
calibrated age ranges of buildings 27 and 34 (6445–5720
BC and 5972–5554 BC, respectively) overlap, and so are
not inconsistent with the stratigraphic interpretation.
3. Building 21 is older than burial 7/I (7186±56 BP) that cuts
through its plaster floor. Borić and Dimitrijević (2007) re-
ported an AMS 14C date on a red deer skull that they in-
terpret as a grave offering with Burial 7/I. The 2-sigma
calibrated age ranges of the burial (6212–5931 BC) and
deer skull (c. 5890–5730 BC, estimated from Borić &
Dimitrijević 2007: fig 3) do not overlap, which suggests
the burial and the deer skull were not associated.6
If we accept the validity of the charcoal 14C dates in relation
to the plaster floored buildings (except in the case of building
34) and accept burial 7/I as establishing a TAQ for building
21, then the following observations may be made:
a. The data indicate that the trapezoidal buildings were con-
structed over a maximum time-range from c. 6400–5550
cal BC.
b. The implied time-ranges of individual buildings are quite
broad because of the low precision on the charcoal ages
and the reservoir-corrected human bone ages.
c. Of the directly or indirectly dated trapezoidal buildings,
only two (34 and 65) can be shown to have been con-
structed before 6000 cal BC, while seven buildings (1, 9,
16, 32, 37, 47, and 51) were clearly constructed after 6000
cal BC, including three (32, 37 and 51) after 5900 cal BC.
d. Although the age ranges for the charcoal dates are typic-
ally post-6000 BC, this does not exclude the possibility
that a substantial number of the trapezoidal buildings were
constructed before 6000 BC, since those that can be dated
in relation to charcoal samples or burials represent only
about 10% of the total number of such structures that were
identified at Lepenski Vir.
Are the trapezoidal buildings Mesolithic or Neolithic?
Pottery and domesticates
Srejović (1969, 1972a) regarded the trapezoidal structures of
Lepenski Vir as a purely Mesolithic phenomenon. However,
difficulties with this interpretation were apparent from the
outset. Pottery, as well as ground-edge artefacts and imported
Balkan flint, with clear parallels in the Early Neolithic
Starčevo culture, occurred in some of the buildings. Different
interpretations have been placed on this evidence:
1. Srejović argued that the pottery was derived from a later,
Starčevo cultural layer, which he divided into two phases,
LVIIIa (‘Proto-Starčevo’) and LVIIIb (‘Classical
Starčevo’). To accommodate this view, he rejected the
charcoal 14 C dates that carry the implication that the
trapezoidal buildings were at least partly contemporan-
eous with the earliest Neolithic settlements in the sur-
rounding regions.
 Clive Bonsall et al.: Dating burial practices and architecture at Lepenski Vir
2. Other researchers have accepted the association between
the trapezoidal buildings and Early Neolithic artefacts, al-
though opinion is divided on whether the people involved
were farmers (Jovanović 1969), or hunter-gatherers who
had adopted some elements of Neolithic technology
(Borić 1999, 2002b; Budja 1999, 2004) or simply acquired
them through exchange with neighbours (Voytek &
Tringham 1989; Radovanović 1996b).
Srejović (1969: 153, 1972a: 134) listed 15 trapezoidal
buildings with pottery, although the (unpublished) excava-
tion records suggest that pottery was also found in several
more buildings. In some cases, it seems clear that the pottery
was derived from later contexts. Irrespective of whether a
Starčevo ‘cultural layer’ actually existed at Lepenski Vir,
pottery formed the largest category of finds from the site;
much of it was recovered from the many pit features that
were recorded across the site, and which occasionally were
observed to cut into earlier trapezoidal structures. Thus the
potential for ‘mixing’ of material between contexts is clear.
Perić and Nikolić (2004) have argued persuasively that the
pottery found in building 5 was in fact intrusive, associated
with a large Starčevo pit (‘hut A’) that was dug down to the
floor of the trapezoidal structure and had a ‘U-shaped stove’
at its base.
Whole pots were found on the floor of building 4 and in
the ‘ashplace’ of building 54, and these have been assumed to
be in situ (Radovanović 2000; Garašanin & Radovanović
2001; Borić 2002). This interpretation is reasonable, al-
though the possibility that they were later intrusions cannot
be excluded entirely — complete pots also occurred at the
bases of Starčevo pits that reached the floors of earlier
trapezoidal buildings, as in building 5 (Perić & Nikolić 2004:
170–173).
While pottery and other Neolithic artefacts were reported
from some of the trapezoidal structures, evidence of food
production in the form of domesticated animal and plant re-
mains was not. There was not even accidental ‘mixing’ of
animal bones between archaeological contexts it seems,
which is surprising in view of the evidence for disturbance of
the trapezoidal structures by Starčevo pit features, and the
very rapid and imprecise nature of the excavation of the site.
The first direct AMS 14C dates on bones of domesticates
from Lepenski Vir have been provided by Borić and
Dimitrijević (2007). They dated 5 bones belonging domestic
cattle, pig and goat from Starčevo contexts; the dates range
between 7043±37 and 7008±38 BP (6002–5798 cal BC).
From this evidence, and the apparent absence of bones of
livestock from any of the trapezoidal buildings, Borić and
Dimitrijević concluded that this architectural form had
ceased to be constructed at Lepenski Vir by 5900 BC. They
further argued that the use of pottery must have preceded the
introduction of livestock at Lepenski Vir. Thus, Borić and
Dimitrijević identified a pre-agricultural, ‘Early Neolithic’
phase at Lepenski Vir between 6300–5900 cal BC, with
trapezoidal buildings, pottery, ground-edge tools and
‘Balkan’ flint (corresponding to Srejović s LVI and LVII),
which was succeeded by a ‘Middle Neolithic’ phase (equi-
valent to LVIII) distinguished by the appearance of animal
domesticates, crouched burials, and new styles of pottery
decoration (Borić & Dimitrijević 2007: table 2). There are
193
several problems with this interpretation:
1. In the first place, the charcoal 14C dates (discussed above)
imply that the trapezoidal buildings continued to be con-
structed until after 5700 BC. To accept the hypothesis that
the trapezoidal buildings all predate 5900 cal BC would
therefore require us to reject the 14C ages younger than
5900 BC as in error, or the charcoal as intrusive. This
would be special pleading and, logically, there seems no
more reason to dismiss a charcoal sample as intrusive,
than the pottery or other small finds from the same context.
2. Secondly, there is no evidence that pottery was in use
throughout the period represented by the trapezoidal
buildings (as implied by Borić and Dimitrijević). In fact,
the dated buildings with Starčevo pottery mainly have 14C
ages younger than c. 6000 cal BC. Building 54, where a
complete pot was found in the ‘ashplace’ in front of the
hearth (dated typologically to a very early phase of the
Starčevo culture on the basis of the spiral decorative mo-
tif: Garašanin & Radovanović 2001; Whittle et al. 2002),
has a 14C age of 7132±64 BP (c. 6007 cal BC), but this
should be regarded as a maximum age because of the pos-
sibility of an ‘old wood’ effect and/or reuse of the timber.7
3. Also at variance with the suggestion of an early appear-
ance of pottery at Lepenski Vir is the general lack of
evidence for Neolithic settlements near to the Danube be-
fore c. 6000 BC. 14C dates for the earliest ceramic sites in
Romania (‘Pre-Criş’ or ‘Criş I’) and south-east Hungary
(early Körös) fall between 6000 and 5900 cal BC (Whittle
et al. 2002; Biagi et al. 2005). South of the Danube, there
have been claims for a very early Neolithic in eastern
Croatia (Krajcar Bronić et al. 2004), central Serbia
(‘Proto-Starčevo’ sites — e.g. Srejović 1988) and northern
Bulgaria (‘Early Monochrome’ phase — e.g. Bojadžiev
1995; Vajsov 1998). However, the majority of the 14C ages
from sites belonging to the so-called ‘Proto-Starčevo’ or
‘Early Monochrome’ phases are later than 7150 BP/6000
cal BC; the few ‘older’ 14C ages from these sites are mostly
on charcoal and may show an ‘old wood’ effect. An ex-
ception is the site of Poljanica-platoto in north-east
Bulgaria which has four radiometric 14C dates on pottery
sherds ranging between 7535±80 and 7140±80 BP (Quitta
1978; Todorova 1989; Görsdorf & Bojadžiev 1996).
However, these dates were done in the 1970s before the
development of stepped-combustion techniques for separ-
ating the different organic fractions in pottery; therefore, it
is possible that the Poljanica dates have been affected by
contamination from ‘old’ carbon present in the clays that
were used to make the pots (cf. Bonsall et al. 2002b).
There are other potentially early sites in northern Bulgaria,
notably Dzhuljunitsa and Koprivets (Gurova 2008), but so
far no 14 C dates have been published. Currently, the
nearest Neolithic site geographically to Lepenski Vir that
appears securely dated to before 6000 cal BC is Blagotin
in central Serbia, some 120km from the Danube (Whittle
et al. 2002).
Returning to the question of the dating of the introduction
of domesticates vis-à-vis the trapezoidal buildings, another
potential problem for Borić and Dimitrijević’s (2007) hypo-
thesis is sample size. The bones of domestic livestock were
generally scarce at Lepenski Vir, being far outnumbered by
 The Iron Gates in Prehistory
those of wild mammals, dogs and fish, as well as by the
quantity of Starčevo pottery sherds. In Bökönyi’s (1969,
1972) analysis of the faunal material collected during the first
few seasons of excavation (up to 1968?), the bones are not
listed according to individual archaeological features and it is
not made clear on what basis material was assigned to the
‘Mesolithic’ (LVI–II) or ‘Neolithic’ (LVIII), respectively. A
smaller faunal sample that was recovered mainly in later field
seasons was analyzed by Dimitrijević (Borić & Dimitrijević
2005). Of the 120 ungulate bones among the material studied
from trapezoidal structures (Borić & Dimitrijević 2005: table
4), only 15 came from contexts that according to Srejović
(1969, 1972a) also contained pottery. Thus, sampling bias
cannot be ruled out as a possible explanation for the absence
of domesticates from those contexts.
Stable isotope data
The bone collagen stable isotope values of the burials dis-
cussed in this paper (Table 1) have some bearing on the
question of the timing of the Mesolithic–Neolithic transition
at Lepenski Vir. On the basis of these data, it may be sugges-
ted that two distinct dietary patterns are represented among
the individuals buried at Lepenski Vir between 6300 and
6000 cal BC (Bonsall et al. 2004). The majority of individu-
als show very heavy δ15N (+15.2 to +16.2‰) and relatively
heavy δ13C (-17.7 to -18.8‰) values suggesting diets that
were dependent on riverine food sources, especially fish —
similar to the dietary pattern found throughout the Mesolithic
in the Iron Gates (Bonsall et al. 1997, 2000). Three directly
dated skeletons (26, 54c, and 54e) have much lighter δ13C
and δ15 N values, similar to values recorded for crouched
burials belonging to the period after 5900 cal BC at Lepenski
Vir (Table 1). Such lighter values are consistent with the in-
clusion of much higher proportions of terrestrial foods in the
diets compared to the traditional Mesolithic pattern, although
the diets of at least two of the three individuals probably in-
cluded significant amounts of protein of aquatic origin.
Higher-level interpretation of these data is complicated by
several factors. One is the possibility that between 6300 and
6000 cal BC Lepenski Vir had the status of ‘sacred site’, that
is, one with special religious significance for the people who
lived within its ‘catchment’ or zone of influence (Srejović
1969, 1972a; see also Gimbutas 1991; Bonsall et al. 2002a;
Radovanović 2006); thus some or all of the people buried at
Lepenski Vir may not have lived there, but at outlying settle-
ments along the river or in the hinterland. Another issue is the
dating of burials 26, 54c and 54e relative to the group with
strongly ‘aquatic’ diets — were they contemporaneous, or
did the three individuals with ‘more terrestrial’ diets belong
to a later phase within the 6300–6000 cal BC age range? The
relative ages of the two dietary groups cannot be determined
precisely, because of the 14C date uncertainties of the indi-
vidual skeletons and the shape of the calibration curve around
that time (Bonsall 2007). However, there is some limited
evidence to support the idea that the ‘terrestrial’ group is
later, although the evidence is not conclusive. Remains of
individuals belonging to both groups occurred in building 65,
and were included in the dating programme (Table 1: skelet-
ons 54c, 54d, and 54e). The chronological sequence of the
burials has already been discussed (see above). More im-
194
portant in the present context, however, is the sequence in
which the individuals died. Disarticulated skeleton 54d (with
δ15N = 15.2‰, δ13C = -17.7‰) is a ‘secondary’ burial, and
the individual almost certainly died before individuals 54e
(δ15N = 13.0‰, δ13C = -19.1‰) and 54c (δ15N = 12.4‰, δ13C
= -19.6‰). Thus, in one context at Lepenski Vir (building 65)
an individual with a typical ‘Mesolithic’ dietary signature can
be shown to be very probably older than two skeletons that
have C- and N-isotope values closer to the ‘Neolithic’ pattern
(i.e. intermediate between ‘aquatic’ and ‘terrestrial’ diets).
There are a number of possible explanations for the pres-
ence of individuals with ‘intermediate’ and ‘terrestrial’ diets
at Lepenski Vir between 6300–6000 cal BC (cf. Bonsall et al.
2004; Radovanović 2006; Bonsall 2007). They could (be):
1. Members of a local population, or even a single co-
resident group, with highly variable dietary preferences;
2. Belong to a phase when farming started to make a signi-
ficant contribution to the local economy — presumably in
the latter part of the 6300–6000 BC time-range;
3. Belong to a time(s) when there was a shift in the
Mesolithic subsistence base, toward greater reliance on
wild terrestrial resources;
4. Local Mesolithic foragers who moved away to live with
hunter-gatherers (with more terrestrial-based diets) or
farmers, but were returned to Lepenski Vir for burial;
5. Incomers who originated among either, a) hunter-
gatherers or b) farmers in the hinterland, and subsequently
moved into the Lepenski Vir locality (e.g. on marriage, or
as slaves/war captives);
6. Members of outlying a) hunter-gathering or b) farming
communities who had some connection with Lepenski Vir
and were brought there for burial.
Further research, including higher-resolution dating of the
skeletons, comparison of δ13C and δ15N values for teeth and
bone to help identify dietary change between childhood and
adulthood, as well as Sr- and O-isotope analysis of teeth to try
to assess geographic origin, may help to narrow the range of
possibilities. In the meantime, we offer the following general
observations: Hypotheses 1 and 3 seem unlikely, since there
is no evidence of equivalent dietary variability among earlier
or later populations in the Iron Gates, and we are unaware of
appropriate ethnographic or archaeological parallels else-
where. Hypothesis 2 would be consistent with the evidence
from building 65 of a time trend, with later individuals
showing lower C- and N-isotope values suggestive of an in-
crease in the consumption of terrestrial protein. It would also
imply that farming had begun in the Lepenski Vir locality
before 6000 BC and, significantly, before the change from
traditional Mesolithic extended supine inhumation burial to
the characteristic Starčevo burial rite of crouched inhuma-
tion. Hypotheses 5b and 6b would carry the implication that,
as the agricultural frontier advanced towards the Danube,
there came a point before 6000 BC when the Lepenski Vir
‘catchment’ included both Mesolithic (hunting-gathering)
and Neolithic (farming) settlements. In this scenario, the
Neolithic settlements would probably not have been very far
from Lepenski Vir — the distance over which a corpse could
be transported intact presumably was limited by a number of
factors, including the rate of decomposition and ease/mode of
transport (overland or by water).
 Clive Bonsall et al.: Dating burial practices and architecture at Lepenski Vir
Conclusions
This paper has presented new AMS radiocarbon and stable
isotope results for human remains from Lepenski Vir, and
discussed the implications of these data for our understand-
ing of temporal changes in burial practices, the chronology of
the distinctive trapezoidal buildings, and the timing of the
Mesolithic–Neolithic transition at the site.
Attention has been drawn to a number of factors that
complicate interpretations of the data. These include: a) the
lack of precision in the human bone 14C dates because of the
need to apply a reservoir correction; b) doubts about the reli-
ability of certain archaeological observations made at the
time of the excavations; c) the possibility of curatorial errors
leading to mixing of bones from different burials; and d) the
effects of post-depositional taphonomic processes. The prin-
cipal findings of the study are as follows:
1. A number of changes in burial customs are documented
during the period of site use. Extended inhumation, skull
removal/caching, and disarticulated bone burial (‘partial
burial’) were all characteristic of the Mesolithic. Two ar-
ticulated burials dated to before 7500 cal BC (‘Early
Mesolithic’) were both oriented perpendicularly to the
Danube with the head away from the river, while articu-
lated burials dated to the period 6400–6000 cal BC (‘Final
Mesolithic’) were all laid out more or less parallel to the
river with the head downstream. Crouched burial replaced
extended inhumation after c. 6000 cal BC, and was still in
evidence during the Chalcolithic, c. 4100 cal BC. The
burial customs of the later prehistoric period are unknown,
but extended inhumation is attested again in the Middle
Ages.
2. The 14C results show that previous interpretations of the
phasing of the burials from Lepenski Vir were flawed. This
is most obvious in the case of the disarticulated burials
where the original dating relied mainly on ‘stratigraphy’.
3. At least six burials included in the dating programme were
found within trapezoidal buildings, either inserted through
the plaster floor or deposited on or above the floor. The 14C
ages of these burials establish minimum ages for four of
the buildings, indicating that at least three of these four
buildings are older than 6000 cal BC. This reinforces the
impression gained from other lines of evidence that many
of the plaster-floored stuctures recorded at Lepenski Vir
are older than that date. However, if a previous series of
radiometric 14C dates on wood charcoal samples are valid,
then trapezoidal buildings with lime plaster floors contin-
ued to be erected at Lepenski Vir until 5700 cal BC or later.
4. The paired 14C and stable isotope measurements on human
bone indicate that a significant change in diet occurred at
some point between 6300 and 6000 cal BC, and that it took
place at a time when the Mesolithic tradition of extended
inhumation was still practised at Lepenski Vir. If this di-
etary shift reflects the beginning of farming in the
Lepenski Vir catchment area, then other Neolithic markers
might be expected to occur around the same time. On
present evidence it is difficult to argue for the appearance
of pottery, ground-edge artifacts, domestic livestock, and
crouched inhumation before 6000/5900 cal BC. However,
if the dietary shift occurred close to 6000 cal BC, then the
195
archaeological and isotopic indicators of change are not
necessarily irreconcilable. Resolving the issues surround-
ing the Meso–Neolithic transition at Lepenski Vir is an
important goal for future research, but one that may re-
quire the application of higher resolution dating, including
direct dating of pottery, as well as advances in radiocarbon
and stable isotope analyses of human bone, if not new field
research to locate and investigate evidence of
Mesolithic–Neolithic settlement in the hinterland.
Notes
1. Some sieving was undertaken at Lepenski Vir but only, it seems,
of the infills of certain ‘house pits’ excavated after 1967; the
mesh size used is not reported in the field documentation.
2. Weighted mean of OxA-12979 and OxA-11692 after reservoir
correction (see Table 1).
3. Strictly speaking, it is the act of burial that postdates the plaster
floor. Death may have preceded burial by several days and it is
conceivable, though unlikely, that the plaster floor was laid in the
interim. Moreover, the 14C age of the skeleton is that of the bone
collagen — in adults the turnover rate for collagen is in the order
of 1.5–4% per year, for adolescents the rate is typically higher
(up to 15% per year) (Hedges et al. 2007). Therefore, the 14C age
may overestimate the time of death by years to decades, de-
pending on the age-at-death of the individual.
4. If burial 7 was not inserted through the floor of building 21, then
an entirely different set of chronological interpretations would
follow.
5. Comparison of excavation plans and photographs suggests that
burial 8 occurred more or less directly above infant burials 94
and 101.
6. While there is the possibility of curatorial error in the dating of
burial 7/I (see discussion on pp. 177–178), there is also uncer-
tainty over which deer remains were actually dated by Borić and
Dimitrijević and the relationship of those remains to burial 7/I.
Srejović (1972: 120) states that a deer skull was found “by [the]
right hand” of skeleton 7/I; however, this is not visible in any
photograph or field drawing we have seen. According to the
Field Journal a single antler was found next to the legs of skel-
eton 7/I, but there is no mention of a deer skull at that location.
A red deer skull with antlers attached was found below the floor
of building 21 to the east of skeleton 7/I (Fig. 6). Some distance
separates this find from the skeleton, and it seems it was only
discovered when the ‘grave pit’ of burial 7/I was artificially en-
larged during excavation (cf. Figs 5 & 6). This raises doubts that
the red deer skull/antlers was really deposited with the burial.
Furthermore, careful scrutiny of field drawings and photographs
suggests the plaster floor above the deer skull/antlers was ori-
ginally intact; in fact one of the antlers attached to the deer skull
appears to have extended below the hearth of building 21 (see
also Stefanović & Borić this volume: fig. 15). Therefore, it seems
unlikely that the red deer skull/antlers was inserted through the
floor of building 21, and more likely perhaps that it relates to
underlying building 22. It is worth noting that another red deer
skull with antlers attached was found on or above the floor of
building 22 — this can be seen in the photograph published by
Srejović & Babović (1983: 136), and the plan produced by
Stefanović & Borić (this volume: fig. 15). We are unsure if the
red deer skull/antlers near the hearth of building 21 is that dated
by Borić and Dimitrijević (2007). If it is, then the 14C measure-
ment would have important implications for the dating of build-
ing 21 and could conflict with our dating based on the 14C result
for burial 7/I (see also Note 4).
7. Charcoal from building 54 was dated at three different radiocar-
1
bon laboratories, Berlin, Köln, and Zagreb. Two dates, KN-407
 The Iron Gates in Prehistory
and Z-143, have very large errors (>100yr) and should, perhaps,
be disregarded. The weighted mean of the three dates with errors
of 100 yr or less is 7079±57 BP (c. 5950 cal BC).
Acknowledgements
We would like to thank the Director of the Institute of Archaeology
in Belgrade for permission to reproduce photographs from the
Lepenski Vir archive. The senior author would also like to thank
Dušan Borić for permission to use his plan of Lepenski Vir for
Figure 2.
References
Alley, R.B., Mayewski, P.A., Sowers, T., Stuiver, M., Taylor, K.C. &
Clark, P.U. 1997: Holocene climate instability: a prominent
widespread event 8200 yr ago. Geology 25: 483–486.
Biagi, P., Shennan, S. & Spataro, M., 2005: Rapid rivers and slow
seas? New data for the radiocarbon chronology of the Balkan
Peninsula. In Nikolova, L. & Higgins, J. (eds) Prehistoric
Archaeology & Anthropological Theory and Education. Salt
Lake City: International Institute of Anthropology, 43–51.
Bojadžiev, J. 1995: Chronology of prehistoric cultures in Bulgaria.
In Bailey, D.W. & Panajatov, I. (eds) Prehistoric Bulgaria.
Madison: Prehistory Press, 149–192.
Bökönyi, S. 1969: Kičmenjaci (prethodni izveštaj). In Srejović, D.,
Lepenski Vir, Nova praistorijska kultura u Podunavlju. Beograd:
Srpska književna zadruga, 224–228.
— 1972: The vertebrate fauna. In Srejović, D., Europe’s First
Monumental Sculpture. New Discoveries at Lepenski Vir.
London: Thames and Hudson, 186–189.
Bonsall, C. 2003: The Iron Gates Mesolithic. In Bogucki, P. &
Crabtree, P. (eds) Ancient Europe 8000 B.C. to A.D. 1000:
Encyclopedia of the Barbarian World. New York: Scribner,
175–178.
— 2007: When was the Neolithic transition in the Iron Gates? In
Spataro, M. & Biagi, P. (eds) A Short Walk through the Balkans:
the First Farmers of the Carpathian Basin and Adjacent
Regions. Trieste: Società per la Preistoria e Protostoria della
Regione Friuli-Venezia Giulia, 53–65.
Bonsall, C., Lennon, R., McSweeney, K., Stewart, C., Harkness, D.,
Boroneanţ, V., Bartosiewicz, L., Payton, R. & Chapman, J. 1997:
Mesolithic and Early Neolithic in the Iron Gates: a palaeodietary
perspective. Journal of European Archaeology 5: 50–92.
Bonsall C., Cook G., Lennon R., Harkness D., Scott M.,
Bartosiewicz, L. & McSweeney, K. 2000: Stable isotopes, ra-
diocarbon and the Mesolithic–Neolithic transition in the Iron
Gates. Documenta Praehistorica 27: 119–132.
Bonsall, C., Macklin M.G., Payton, R.W. & Boroneanţ, A. 2002a:
Climate, floods and river gods: environmental change and the
Meso–Neolithic transition in southeast Europe. Before Farming:
the Archaeology of Old World hunter-gatherers 3_4(2): 1–15.
Bonsall, C., Cook, G., Manson, J. & Sanderson, D. 2002b: Direct
dating of Neolithic pottery: progress and prospects. Documenta
Praehistorica 29: 47–59.
Bonsall, C., Cook, G.T., Hedges, R.E.M., Higham, T.F.G., Pickard,
C. & Radovanović, I. 2004: Radiocarbon and stable isotope
evidence of dietary change from the Mesolithic to the Middle
Ages in the Iron gates: new results from Lepenski Vir.
Radiocarbon 46: 293–300.
Bonsall, C., Horvat, M., McSweeney, K., Masson, M., Higham,
T.F.G., Pickard, C. & Cook, G.T. 2007: Chronological and diet-
ary aspects of the human burials from Ajdovska Cave, Slovenia.
Radiocarbon 49: 727–740.
Borić, D. 1999: Places that created time in the Danube Gorges and
beyond, c. 9000–5500 BC. Documenta Praehistorica 26: 41–70.
1
196
— 2002a: Seasons, Life Cycles and Memory in the Danube Gorges,
c. 10000–5500 BC. Unpublished Ph.D. Dissertation, University
of Cambridge, Cambridge.
— 2002b: The Lepenski Vir conundrum: reinterpretation of the
Mesolithic and Neolithic sequences in the Danube Gorges.
Antiquity 76: 1026–1039.
Borić, D. & Dimitrijević, V. 2005: Continuity of foraging strategies
in Mesolithic–Neolithic transformations: dating faunal patterns
at Lepenski Vir (Serbia). Atti della Società per la preistoria e
protoistoria della regione Friuli-Venezia Giulia, Trieste 15
(2004–2005): 33–107.
Borić, D. & Dimitrijević, V. 2007: When did the ‘Neolithic package’
reach Lepenski Vir? Radiometric and faunal evidence.
Documenta Praehistorica 34: 53–72.
Borić, D. & Dimitrijević, V. In press. Apsolutna hronologija i
stratigrafija Lepenskog Vira (Absolute chronology and strati-
graphy of Lepenski Vir). Starinar.
Borić, D. & Miracle, P. 2004: Mesolithic and Neolithic (dis)con-
tinuities in the Danube Gorges: new AMS dates from Padina and
Hajdučka Vodenica (Serbia). Oxford Journal of Archaeology 23:
341–371.
Borić, D. & Stefanović, S. 2004: Birth and death: infant burials from
Vlasac and Lepenski Vir. Antiquity 78: 526–546.
Borić, D., Grupe, G., Peters, J. & Mikić, Ž. 2004: Is the
Mesolithic–Neolithic subsistence dichotomy real? New stable
isotope evidence from the Danube gorges. European Journal of
Archaeology 7: 221–248.
Boroneanţ, V., Bonsall, C., McSweeney, K., Payton, R. & Macklin,
M. 1999: A Mesolithic burial area at Schela Cladovei, Romania.
In Thévenin, A. (ed.) L’Europe des Derniers Chasseurs: Épipa-
léolithique et Mésolithique. Paris: Éditions du Comité des
Travaux Historiques et Scientifiques, 385–390.
Bronk Ramsey, C. & Hedges, R.E.M. 1997: Hybrid ion sources: ra-
diocarbon measurements from microgram to milligram. Nuclear
Instruments and Methods in Physics Research B 123: 539–545.
Bronk Ramsey, C., Pettitt, P.B., Hedges, R.E.M., Hodgins, G.W.L.
& Owen, D.C. 2000: Radiocarbon dates from the Oxford AMS
system: Archaeometry Datelist 30. Archaeometry 42: 459–479.
Brown, T.A., Nelson, D.E., Vogel, J.S. & Southon, J.R. 1988:
Improved collagen extraction by modified Longin method.
Radiocarbon 30: 171–177.
Budja, M. 1999: The transition to farming in Mediterranean Europe
— an indigenous response. Documenta Praehistorica 26:
119–141.
— 2004: The neolithisation of South-eastern Europe. In Knutsson,
H. (ed.) Coast to Coast — Arrival: Results and Reflections.
Uppsala: Uppsala University, 377–417.
Ciugudean, H. 1996: Epoca Timpurie a Bronzului în Centrul şi
Sud-vestul Transilvaniei. Bucureşti: Institutul Român de
Tracologie.
Clark, P.U., Marshall, S.J., Clarke, G.K.C., Hostetler, S.W.,
Licciardi, J.M. & Teller, J.T. et al. 2001: Freshwater forcing of
abrupt climate change during the Last Glaciation. Science 293:
283–287.
Coard, R. & Dennell, R.W. 1995: Taphonomy of some articulated
skeletal remains: transport potential in an artificial environment.
Journal of Archaeological Science 22: 441–448.
Cook, G.T., Bonsall, C., Hedges, R.E.M., McSweeney, K.,
Boroneanţ, V. & Pettitt, P.B. 2001: A freshwater diet-derived 14C
reservoir effect at the Stone Age sites in the Iron Gates gorge.
Radiocarbon 43: 453–460.
Cook, G.T., Bonsall, C., Hedges, R.E.M., McSweeney, K.,
Boroneanţ, V., Bartosiewicz, L. & Pettitt, P.B. 2002: Problems of
dating human bones from the Iron Gates. Antiquity 76: 77–85.
Cook, G.T., Bonsall, C., Pickard, C., McSweeney, K., Bartosiewicz,
L. & Boroneanţ, A. In press: The Mesolithic–Neolithic transition
 Clive Bonsall et al.: Dating burial practices and architecture at Lepenski Vir
in the Iron Gates, southeast Europe: calibration and dietary is-
sues. In Crombé, P. (ed.) Chronology and Evolution in the
Mesolithic of Northwest Europe.
Coplen, T.B. 1994: Reporting of stable hydrogen, carbon and oxy-
gen isotopic abundances. Pure and Applied Chemistry 66:
273–276.
Garašanin, M. & Radovanović, I. 2001: A pot in house 54 at
Lepenski Vir I. Antiquity 75: 118–125.
Gimbutas, M. 1991: The Civilization of the Goddess. San Francisco:
Harper.
Görsdorf, J. & Bojadžiev, J. 1996: Zur absoluten Chronologie der
bulgarischen Urgeschichte. Eurasia Antiqua 2: 105–173.
Grupe, G., Mikić, Ž, Peters, J. & Manhart, H. 2003: Vertebrate food
webs and subsistence strategies of Meso- and Neolithic popula-
tions of central Europe. In Grupe, G. & Peters, J. (eds)
Decyphering Ancient Bones: The Research Potential of
Bioarchaeological Collections. Rahden (Westf): Marie Leidorf,
193–214.
Gurova, M. 2008: Towards an understanding of Early Neolithic
populations: a flint perspective from Bulgaria. Documenta
Praehistorica 35: 111–129.
Hedges, R.E.M., Clement, J.G., Thomas, C.D.L. & O’Connell, T.C.
2007: Collagen turnover in the adult femoral mid-shaft: modeled
from anthropogenic radiocarbon tracer measurements. American
Journal of Physical Anthropology 133: 808–816.
Jovanović, B. 1969: Chronological frames of the Iron Gate group of
the Early Neolithic period. Archaeologica Iugoslavica 10: 23–38.
— 1972: The autochthonous and migrational components of the
Early Neolithic in the Iron Gates. Balcanica 3: 49–58.
Krajcar Bronić, I., Minichreiter, K., Obelić, B. & Horvatinčić, N.
2004: The oldest Early Neolithic (Starčevo culture) settlements
in Croatia: Zadubravlje-Dužne and Slavonski Brod-Galovo. In
Higham, T., Bronk Ramsey, C. & Owen, C. (eds) Radiocarbon
and Archaeology: Fourth International Symposium, St
Catherine’s College, Oxford, 9–14 April 2002. Oxford: Oxford
University School of Archaeology, 229–245.
Law, I.A. & Hedges, R.E.M. 1989: A semi-automated bone pre-
treatment system and the pretreatment of older and contaminated
samples. Radiocarbon 31: 247–253.
Magny, M., Bégeota, C., Guiot, J. & Peyrona, O. 2003: Contrasting
patterns of hydrological changes in Europe in response to
Holocene climate cooling phases. Quaternary Science Reviews
22: 1589–1596.
McKinley, J.I. & Roberts, C. 1995: Excavation and Post-excavation
Treatment of Cremated and Inhumed Human Remains (IFA
Technical Paper 13). Birmingham: Institute of Field
Archaeologists.
Murray, M. & Schoeninger, M. 1988: Diet, status and complex so-
cial structure in Iron Age Central Europe: some contributions of
bone chemistry. In Gibson, D.B. & Geselowitz, M.N. (eds) Tribe
and Polity in Late Prehistoric Europe: Demography, Production
and Exchange in the Evolution of Complex Social Systems. New
York: Plenum, 155–176.
Nemeskéri, J. 1969: Stanovništvo Lepenskog Vira. In Srejović, D.,
Lepenski Vir: Nova Praistorijska Kultura u Podunavlju.
Belgrade: Srpska Knjizevna Zadruga, 239–257.
— 1972: The inhabitants of Lepenski Vir. In Srejović, D., Europe’s
First Monumental Sculpture. New Discoveries at Lepenski Vir.
London: Thames and Hudson, 190–204.
Perić, S. & Nikolić, D. 2004: Stratigraphic, cultural and chronolo-
gical characteristics of the pottery from Lepenski Vir 1965 ex-
cavations. In Vasić, M., Vucković, P. & Cvetković, B. (eds) The
Central Pomoravlje in Neolithization of South East Europe.
Belgrade: Archaeological Institute in Belgrade, Regional
Museum in Jagodina and Regional Museum in Paraćin, 157–217.
Quitta, H. 1969: Datiranje radiokarbonskih proba. In Srejović, D.,
1
197
Lepenski Vir: Nova Praistorijska Kultura u Podunavlju.
Belgrade: Srpska Knjizevna Zadruga, 229–238.
— 1972: The dating of radiocarbon samples. In Srejović, D.,
Europe’s First Monumental Sculpture. New Discoveries at
Lepenski Vir. London: Thames and Hudson, 205–210.
— 1975: Die Radiocarbondaten und ihre historische Interpretation.
In Srejović, D., Lepenski Vir, Eine vorgeschichtliche
Geburtsstätte europäischer Kultur. Bergisch Gladbach: Gustav
Lübbe, 272–285.
— 1978: Radiocarbon dates and three chronological systems.
Interdisciplinary Research 1: 12–24. [in Bulgarian]
Radovanović, I. 1996a: The Iron Gates Mesolithic. Ann Arbor:
International Monographs in Prehistory.
— 1996b: Mesolithic/Neolithic contacts: a case of the Iron Gates
region. Documenta Praehistorica 23: 39–48.
— 2000: Houses and burials at Lepenski Vir. European Journal of
Archaeology 3: 331–350.
— 2006: Further notes on Mesolithic–Neolithic contacts in the Iron
Gates region and the central Balkans. Documenta Praehistorica
33: 107–124.
Roksandic, M. 1999: Transition from Mesolithic to Neolithic in the
Iron Gates Gorge: Physical Anthropology Perspective.
Unpublished PhD thesis, Department of Anthropology, Simon
Fraser University, Burnaby, B.C.
— 2000: Transition from Mesolithic to Neolithic in the Iron Gates
gorge: physical anthropology perspective. Documenta
Praehistorica 27: 1–100.
Roksandic, M., Djurić, M., Rakočević, Z. & Seguin, K. 2006:
Interpersonal violence at Lepenski Vir Mesolithic/Neolithic
complex of the Iron Gates Gorge (Serbia–Romania). American
Journal of Physical Anthropology 129: 339–348.
Roksandic, M., Wood, C. & Vlak, D. 2007: Death in the line of duty:
a late Medieval burial at the site of Lepenski Vir (Serbia).
International Journal of Osteoarchaeology 17: 635–642.
Sladić, M. 1986: Kula près Mihajlovac — un site préhistorique.
Đerdapske sveske 3: 432–442.
Srejović, D. (ed.) 1969: Lepenski Vir, Nova praistorijska kultura u
Podunavlju. Beograd: Srpska Knjizevna Zadruga.
— 1971: The roots of the Lepenski Vir culture. Archaeologia
Iugoslavica 10: 13–21.
— 1972a: Europe's First Monumental Sculpture: Lepenski Vir.
London: Thames and Hudson.
— 1972b: Kulturen des frühen Postglazials in südlichen
Donauraum. Balcanica 3: 11–44.
— 1988: The Neolithic of Serbia: a review of research. In Srejović,
D. (ed.) The Neolithic of Serbia. Archaeological Research
1948–1988. Belgrade: Centre for Archaeological Research,
University of Belgrade, 5–20.
Srejović, D. & Babović, Lj. 1983: Umetnost Lepenskog Vira.
Beograd: Jugoslavija.
Taylor, T. 2002: The Buried Soul: How Humans Invented Death.
London: Fourth Estate.
Todorova, H. 1989: Das Frühneolithikum Nordostbulgariens im
Kontext des ostbalkanischen Neolithikums. In Heger, N. &
Hiller, S. (eds) Tell Karanovo und das Balkan Neolithikum.
Gesammelte Beiträge zum Internationalen Colloquium in
Salzburg, 20.–22. Oktober 1988. Salzburg: Institut für
Klassische Archäologie der Universität Salzburg, 9–25.
Tringham, R. 1971. Hunters, Fishers and Farmers of Eastern
Europe 6000–3000 BC. London: Hutchinson.
Vajsov, I. 1998: The typology of the anthropomorphic figurines
from northeastern Bulgaria. In Stefanovich, M., Todorova, H. &
Hauptmann, H. (eds) In the Steps of James Harvey Gaul, vol. 1:
James Harvey Gaul – In Memoriam. Sofia: The James Harvey
Gaul Foundation, 107–141.
Vasić, R. 1986: Compte-rendu des fouilles du site préhistorique à
 The Iron Gates in Prehistory

Velesnica 1981–1982. Đerdapske sveske 3: 264–285.
Voytek, B. & Tringham, R. 1989: Rethinking the Mesolithic: the
case of South-east Europe. In Bonsall, C. (ed.) The Mesolithic in
Europe. Edinburgh: John Donald, 492–500.
Whittle, A. 1996: Europe in the Neolithic: the Creation of New
Worlds. Cambridge: Cambridge University Press.
Whittle, A., Bartosiewicz, L., Borić, D., Pettitt, P. & Richards, M.
2002: In the beginning: new radiocarbon dates for the Early
Neolithic in northern Serbia and south-east Hungary. Antæus 25:
63–117.
Zoffmann, Zs. 1983: Prehistorical skeletal remains from Lepenski
Vir (Iron Gate, Yugoslavia). Homo 34: 129–148.

198
 Appendix 1. Descriptions of the burials included in the AMS 14C dating programme.

Body position Burial Location Relationship to trapezoidal structure Summary descriptions taken from the Field Burial Record (BR) and Field Journal (FJ)

Extended 7/I Building 21 Burial cuts through floor, to rear of hearth BR: Burial no. 7, extended, S–N orientation, arms along the body, hands on pelvis; to the west of the
skull of the skeleton was another human skull; to the east of the skull of the skeleton was a bovid skull
and antlers.
FJ: Oct 24 1967: skull belonging to skeleton no. 7 is labeled ‘Skull I’ ; the separate skull next to it (to
the west) is labeled ‘Skull II’ (described as having prominent supra-orbital ridges).
14 Grid square A/VIII–IX Upstream of the zone with trapezoidal buildings BR: Burial 14 occurred in an oval pit with bones of other individuals (it was originally labeled burial
no. II from the ‘ossuary’); it is described as extended S–N, upper arms along the body, lower arms on
the body.

26 Building 34 Cuts through floor of building 34, to rear of BR: Body extended S–N, hands on pelvis.
hearth
29 Grid square b/VI, VII n/a (medieval) BR: Body extended W–E, left arm along the body, right arm bent across the torso, hand touching the
chin.
30 Grid square b/VI, VII n/a (medieval) BR: Body extended W–E, lower right arm across the torso, left arm bent, hand on the pelvis.
54c Building 65/XXXV Above floor of building 65 (field interpretation: BR: Body extended S–N, right arm bent with hand on left shoulder, left arm along the body, no skull.
cut through burials 54a–b)
54e Building 65 On the floor BR: Body extended S–N, lower arms across the pelvis (field interpretation: burial 54e disturbed
skeleton 54d).

199
60 Near building 40 and Proto-LV Next to corner A of building 40, ‘in virgin soil’ BR: Extended SW–NE, right arm bent touches the chin, lower left arm across the pelvis.
hearth. Grid square B-C/11
61 Building 40 Burial cuts through floor, to rear of hearth BR: Body extended S–N, right arm along the body, hand on pelvis, lower left arm missing.
62 Grid square b/VI, VII n/a (medieval) BR: Extended W–E, lower right arm on pelvis, left arm bent with hand on the right shoulder.

Dorsal decubitus 69 Grid square a/13 In ‘virgin soil’, downstream of the zone with BR: Covered with stones, W–E orientation, back almost vertically placed, legs bent, left foot over the
trapezoidal buildings right foot, arms along the body over the upper legs, left arm slightly bent.

Crouched 2 Grid square A/II–III, 2nd BR: S–N orientation, crouched on left side; damaged – only mandible, maxilla and femur recovered.
Clive Bonsall et al.: Dating burial practices and architecture at Lepenski Vir

excavation level.
4 Grid square A/V, 8th & 9th BR: In rectangular stone construction, covered by stones. Orientation W–E, head to W, crouched on
excavation levels, relative the right side, well preserved. Around the head – lithics, river shells and animal bones; bone awl in area
depth 1.57–2.06m of legs.
8 Building 24 Above the floor, in rear part of building BR: S–N orientation, crouched on right side, knees at level of upper body, arms bent, hands in front of
face.
9 Building 24 0.30 m above floor in front part of building BR: N–S orientation, crouched on the right side.
32a Grid sq. c–d/1–I On flat rock behind (upslope of) buildings in BR: Found with burials 32b–c. Body in crouched position, in burial pit bordered with several stones.
the central part of the site [orientation and ‘side’ not recorded]
 Appendix 1 (cont.). Descriptions of the burials included in the AMS 14C dating programme.

Body position Burial Location Relationship to trapezoidal structure Summary descriptions taken from the Field Burial Record (BR) and Field Journal (FJ)

Crouched 88 Grid sqare e/I, in a pit bordered To the rear of the central part of the settlement, BR: Burial pit bordered with stones. Skeleton crouched with E–W orientation, left arm bent below the
with stones upslope of the zone with trapezoidal buildings backbone, right arm extended along the body touching a fragmentary bowl. [‘side’ not recorded]

Disarticulated 1 31a Grid square a/VI BR: Above pit 2. Burial 31 consisted of left mandible with 5 teeth, 3 separate teeth, and head of a
femur.
44 Grid square d/I BR: In pit. Comprises ulna, radius, part of pelvis, a few vertebrae, femur head
45b Building 61 BR: In destroyed hearth construction with burials 45a and 45c (45a is recorded as a skull, 45b is
recorded as a skull found within the hearth but below the level of the floor, and 45c is recorded as
consisting of a pelvis, femur, and vertebrae).
54d Building 65 On the floor; comprising disarticulated bones BR: Field interpretation: extended inhumation with S–N orientation dislocated by the interment of
adjacent to articulated skeleton 54e (burials individual 54e; lower extremities preserved.
54a–c occurred at higher levels in infilling of
‘house’ pit )
79a Grid square e/4 At rear of central part of the site, upslope from BR: In burial pit bordered with stones, with burials 79b–c. Burial 79a described as comprising
trapezoidal structures fragments of a skull and the left side of the mandible; 79c also described as parts of a skull and the left
side of a mandible.
89a Grid square f/I–II, next to Behind (upslope) buildings in the central part BR: Burial 89 consists of bones from 2 individuals, an adult (89a) and an infant 3–4 years old (89b);

200
profile f of the site adult comprises fragments of skeleton and skull. [field interpretation: extended burial, S–N orientation,
arms along the body, damaged, legs missing, Bos primigenius horns around the head]
Uncertain 18 Grid square A/I BR: Comprised fragments of a skull and a tibia.
The Iron Gates in Prehistory

35 Grid square d/III, section d/II BR: a fragment of a skull, shoulder bone, and ribs from the right side of the body collected from
section d/II in sq. d/III on 18th July 1967. The rest of the body remained within the section – not
excavated.
1 The bones sampled for dating from burials in the ‘disarticulated’ and ‘uncertain’ categories (cf. Table 1) do not always correspond with the descriptions of the burials in
the Field Burial Record or Field Journal. It is not clear if this reflects post-excavation curatorial errors, or errors in the field documentation.
 Appendix 2. Chronological and stratigraphic relationships of the trapezoidal buildings at Lepenski Vir, and presence/absence of pottery and ground-edge tools.

Calibrated TAQ TPQ Ground

Building Sample type Context 14C age BP age range (minimum (maximum Younger than Older than Pottery edge Comments
BC (2σ) age) cal BC age) cal BC tools

1 Charcoal From floor 6860±100 5982–5572 ≤5572 5982 Building 2 x x

2 ≤5572 >5982 Building 1
4 x
5 Buildings 5a, 6 ? ?
and 17
5a Building 6 Building 5 The existence of building 5a was
disputed by Perić & Nikolić (2007)
6 Buildings 5
and 5a
7 ≤5565 >5978 Buildings 8 and 17 Building 9 x
8 ≤5565 >5978 Buildings 9
and 7
9 Charcoal From structural beam and 6845±100 5978–5565 ≤5565 5978 Buildings 7, 8 x
floor and 17

201
13 Buildings 14
and 15
14 Building 13 Building 15
15 Buildings 14 x
and 13
16 Charcoal From floor 6820±100 5972–5554 ≤5554 5972 x
17 ≤5565 >5978 Buildings 9, 7
and 5
18 Buildings 31 Building 23
and 19?
Clive Bonsall et al.: Dating burial practices and architecture at Lepenski Vir

19 Building 31? x
20 ≤5571 >5867 Building 66 Buildings 32 x
and 33
21 Bone From burial inserted 5931 >6212 Buildings 22, 29, Burial 7/I ? Borić & Dimitrijević (2007)
through plaster floor and 30 reported an AMS date of c. 5800 cal
BC on a red deer skull interpreted as
a grave offering with burial 7/I
22 >5931 >6212 Buildings 29 and Building 21,
30 and burial 7/I
 Appendix 2 (cont.). Chronological and stratigraphic relationships of the trapezoidal buildings at Lepenski Vir, and presence/absence of pottery and ground-edge tools.

Calibrated TAQ TPQ Ground

Building Sample type Context 14C age BP age range (minimum (maximum Younger than Older than Pottery edge Comments
BC (2σ) age) cal BC age) cal BC tools

23 Buildings 18, 31
and 19?
24 Bone From crouched burials in 5729 >5983 Building 24a Burials 8 and 9 x x TAQ/TPQ based on 2-sigma
the infill of ‘house’ pit calibrated age range of burial 8
24a >5729 >5983 Building 24
26 Building 26′ x x
26′ Building 26
27 Charcoal From the hearth? 7210±200 6445–5720 ≤5720 6445 Building 27b, 34 x TAQ/TPQ based on charcoal age
and 52 (but context of charcoal uncertain).
27 <5554 <5972 x TAQ/TPQ based on charcoal age of
stratigraphically older building 34.
27b >6445? Buildings 34 and Building 27 TPQ based on charcoal age of
52? building 27 (but context of charcoal
uncertain).
28 x

202
29 Building 30 Buildings 21
and 22, and
burial 7/I
The Iron Gates in Prehistory

30 >5931 >6212 Building 30b Buildings 21, x

22 and29, and
burial 7/I
30b >5931 >6212 Building 30
31 Building 19? Buildings 18 x
and 23
32 Charcoal From the hearth 6814±69 5867–5571 ≤5571 5867 Buildings 33, 20 x
and 66
33 ≤5571 >5867 Buildings 20 and Building 32
66
34 Charcoal From a beam 6820±100 5972–5554 ≤5554 5972? Buildings 27, 27b x Conflict between charcoal age and
and 43? age relative to burial 26. Association
between charcoal sample and
building is suspect.
34 Bone From burial inserted >6056 >6233 Burial 26 ? TAQ/TPQ are based on stratigraphic
through plaster floor relationship to burial 26.
 Appendix 2 (cont.). Chronological and stratigraphic relationships of the trapezoidal buildings at Lepenski Vir, and presence/absence of pottery and ground-edge tools.

Calibrated TAQ TPQ Ground

Building Sample type Context 14C age BP age range (minimum (maximum Younger than Older than Pottery edge Comments
BC (2σ) age) cal BC age) cal BC tools

35 ≤6061 <6374 Building 36 x x

36 Charcoal From a beam 7335±70 6374–6061 ≤6061 6374 Building 35
37 Charcoal From a beam and floor 6870±65 5890–5640 ≤5640 5890 Buildings 38 and Building 42 x
41
38 ≤5640 >5890 Building 41 Buildings 37
and 42
40 ? Burial 61 claimed to be sealed by
building 40, but no corroborating
evidence.
41 ≤5640 >5890 Buildings 42,
37 and 38
42 ≤5640 >5890 Buildings 37, 38
and 41
43 ? ? ? Building 34? Not clear if any overlap existed
between building 43 and building 34.

203
46 Building 55 Building 56 x
47 Charcoal From a beam 6970±60 5983–5735 ≤5735 5983 Buildings 47′, 58 x Burial 122 has 2-sigma cal BC age
& 53, and burial ranges of 6208–5982 and
122 6020–5838. These are not in conflict
with the charcoal age.
47′ ≤5735 >5983 Buildings 58 and Building 47
53
48 ?
51 Charcoal From a beam and floor 6620±100 5718–5376 ≤5376 5718 Building 52 x
52 ≤5376 >5718 Buildings 51,
Clive Bonsall et al.: Dating burial practices and architecture at Lepenski Vir

27, 27b and

43?
53 ≤5735 >5983 Buildings 47
and 47′
54 Charcoal From a beam and hearh 7132±64* 6205–5849 ≤5849 6205 x x * Weighted mean of 5 14C
determinations (one sample with
uncertain context excluded)
55 Buildings 46
and 56
 Appendix 2 (cont.). Chronological and stratigraphic relationships of the trapezoidal buildings at Lepenski Vir, and presence/absence of pottery and ground-edge tools.

Calibrated TAQ TPQ Ground

Building Sample type Context 14C age BP age range (minimum (maximum Younger than Older than Pottery edge Comments
BC (2σ) age) cal BC age) cal BC tools

56 Buildings 46 and
55
57 Borić & Dimitrijević (2007: fig. 3)
reported a calibrated 14C age of c.
5950 cal BC for burial 19 (crouched,
headless) “found at the floor level”
of building XLIV/57. The number of
buildings at this location, and the
stratigraphic relationship(s) between
the building(s) and the burial, are
unclear.
58 ≤5735 >5983 Buildings 47
and 47′
62 Charcoal From a beam 7430±160 6592–6006 Context of charcoal sample
(7430±160 BP) is uncertain.
63 Building 63′
63′ Building 63

204
65 >6017 >6365 Burials 54e
and 54c
66 ≤5571 >5867 Buildings 32,
The Iron Gates in Prehistory

33 and 20
 The vertebrate fauna from Hajdučka Vodenica in the
Danubian Iron Gates: subsistence and taphonomy from
the Early Neolithic and Mesolithic

Haskel J. Greenfield

Abstract: This paper presents the results of the analysis of the vertebrate fauna from Hajdučka Vodenica, an Early Neolithic site in the
Danubian Iron Gates along the border between Serbia and Romania. The site was initially misreported as an Early Iron Age settlement, but
recent studies have redated its lower levels to both the terminal Mesolithic and Early Neolithic. Vertebrate faunal remains from the site are
divided between these two periods and analyzed from the perspective of subsistence and taphonomy. None of the other assemblages from
the gorge belonging to these periods were analyzed from the perspective of taphonomy. Wild fauna dominate the assemblage in both periods,
but domestic species (pigs and cattle) are found in both as well. The data point to the borrowing of domestic fauna by later Mesolithic oc-
cupants of the region, rather than indigenous domestication. The types and frequencies of both domestic and wild fauna from the lower levels
are compared to other Early Neolithic and Mesolithic settlements in the gorge.
Key words: Iron Gates, Hajdučka Vodenica, vertebrate fauna, subsistence, taphonomy, Mesolithic, Neolithic

Introduction the gorge were carried out preceding the construction of a

Most considerations of the transition from the Mesolithic to
the Early Neolithic among the indigenous societies of tem-
perate Southeast Europe focus upon a series of sites in the
Iron Gates of the Danube, where the river cuts through the
mountains on its way to the Black Sea. The Iron Gates is the
only area in temperate Southeast Europe where there is
abundant archaeological and physical anthropological evid-
ence for the continuous presence of the same population of
humans from the end of the Palaeolithic into the Early
Neolithic (Živanović 1976a, 1976b; y’Edynak 1978;
y’Edynak & Fleisch 1983; Prinz 1987). During the
Mesolithic, there is very little evidence for hunter-gatherer
occupation in temperate Southeast Europe beyond the con-
fines of the Iron Gates (e.g. Tringham 1973; Srejović 1975;
Voytek & Tringham 1989; Radovanović 1996). With the ad-
vent of the Neolithic, the meagre evidence for Mesolithic
cultures disappears except in the Iron Gates. According to
radiocarbon dates, the terminal Mesolithic sites of the Iron
Gates (Fig. 1) are contemporary with the earliest phases of
the Early Neolithic (Starčevo-Körös-Criş) food producing
cultures in the surrounding region (Radovanović 1996;
Bonsall et al. 2000). It would appear that the Mesolithic so-
cieties of the Iron Gates eventually adopt the material culture
of food production into their subsistence systems, while re-
taining much of their hunting-gathering-fishing Mesolithic-
style subsistence system. As such, they provide the only sig-
nificant Southeast European example of an indigenous
population of Mesolithic hunter-gatherers assimilating to an
Early Neolithic lifestyle (Voytek & Tringham 1989;
Greenfield 1993).
In the restricted confines of the Iron Gates gorge, several
sites dated to the later Mesolithic and Early Neolithic were
found during the 1960s when large-scale salvage surveys of
large hydroelectric dam at the eastern end of the Iron Gates.
The later Mesolithic occupation of the Iron Gates appears to
be contemporary with the Early Neolithic Starčevo-Körös-
Criş cultures in the surrounding regions (Manson 1990;
Radovanović 1996). Faunal remains were collected and ana-
lyzed from only a few of them — four from the settlements
located on the Serbian side of the gorge (Lepenski Vir I–III,
Padina IA and IB, Vlasac, and Hajdučka Vodenica) and at
least one on the Romanian side (Icoana II) (Fig. 1). The fauna
from all of the sites, with the exception of Hajdučka
Vodenica, have been reported on in greater or lesser detail
(Bökönyi 1971, 1978; Bolomey 1973; Clason 1980).
This paper presents the results of the analysis of the ver-
tebrate fauna from Hajdučka Vodenica. Even though
Hajdučka Vodenica was one of the first of the Early Neolithic
sites in the gorge to be located and excavated, its fauna only
recently became available for analysis and the results are
presented here.1 The site was initially misreported as an Early
Iron Age settlement, but recent studies have redated its lower
levels to the Terminal Mesolithic and the Early Neolithic. In
the original faunal analysis, unfortunately, the faunal remains
were not temporally subdivided since the stratigraphic in-
formation was not available (Greenfield 1984). The faunal
remains were thought to belong to the Early Neolithic hori-
zons because they were found in artefact bags that also con-
tained Starčevo-type ceramics (which was considered to be a
hallmark of Early Neolithic cultures). Many of the same bags
also contained ornamented pebbles like those described from
some other sites in the Iron Gates gorge (cf. Jovanovic 1969).
Recent reanalysis of the Iron Gates assemblages has demon-
strated that the appearance of pottery takes place during the
second half of the 7th millennium BC throughout the region
in otherwise Mesolithic cultural contexts (Radovanović
1996). As a result, the faunal assemblage is reconsidered in

205
 The Iron Gates in Prehistory

Figure 1. Map of sites mentioned in the text: A–Anza; B–Blagotin; BC–Bukovačka Česma; D–Divostin; G–Gomolava;
H–Hajdučka Vodenica; I–Icoana; L–Lepenski Vir; O–Obre; P–Padina; S–Starčevo; V–Vlasac.

the light of this new information. In this analysis the data are
divided between the Mesolithic and Early Neolithic levels,
and analyzed separately.
The data from Hajdučka Vodenica form an important
comparative database that can increase our understanding of
the other Mesolithic and Early Neolithic assemblages from
the Iron Gates gorge. The problems in how the data were
collected are common to all samples from the Iron Gates.
While the problems compromise its utility for traditional re-
constructions of subsistence economy, the sample can be
used to elucidate issues common to all of the excavations in
the gorge, such as assemblage recovery and taphonomy. It
can also be used to corroborate the evidence from the other
sites, such as distinguishing between Mesolithic and Early
Neolithic patterns of faunal exploitation. The types and fre-
quencies of both domestic and wild fauna from the lower
levels are compared to those from other settlements in the
gorge.
The site
Site location and surrounding environment
Hajdučka Vodenica (‘brigands water mill’) is located on the
right bank of the River Danube as it passes through the Iron
Gates on its way to the Black Sea. The Iron Gates is a 100-km
long gorge cut by the Danube through the southern arm of the
Carpathian Mountains where it crosses from Romania to
Serbia.
The gorge is bordered by high and rugged mountains.
Steep cliffs often reach over the water’s edge. The mountains
bordering the gorge effectively isolate the area from the out-
side world. Most traffic is forced into a narrow corridor par-
allel to the river.
The Danube narrows in width from 1–2 km at the mouth of
the gorge, to only a couple of hundred metres at places inside
the gorge. The river becomes extremely rapid and turbulent
inside the gorge. Its flow is punctuated by whirlpools that
have long been favourite fishing spots. The rugged landscape
is interspersed by small and large coves. It is in these coves
that several prehistoric settlements were found during the
surveys of the 1960s.
Hajdučka Vodenica is located in the eastern half of the
gorge, on the right (Serbian) bank of the Danube, in a broad
cove between the Mali Štrbac hill and the Danube. The pre-
historic section of the site is located on the riverbank and the
first terrace above the river. The terrace slopes up gradually at
first, but it ascends rapidly after a short distance into the
mountains bordering the gorge. It extends both east and west
of the Roman/Medieval castellum (Jovanović 1966a: 102).

206
 Haskel Greenfield: Reanalysis of the vertebrate fauna from Hajdučka Vodenica

The local environmental zones range within a kilometre
from riverine to shoreline, to small meadows in the moun-
tains, to high altitude deciduous and evergreen forests. Soils
in the region include alluvium along the river’s edge and
skeletal podzols at higher altitudes. Most of the prehistoric
settlements along the river’s edge are cut back into loess de-
posits on the river’s terraces (Mišić et al. 1970; Gigov 1972;
Cârciumaru 1978).
At the present time the area is used for the grazing of do-
mestic stock (primarily sheep and goat, with some cattle).
Wild game still abounds throughout the region. Hunting,
gathering and fishing continue to be important subsistence
supplements in the region. Shellfish collecting was important
until the local ecology was devastated by massive flooding of
the gorge when the hydroelectric dam was completed.
Site description and excavation history
Excavations at Hajdučka Vodenica took place between 1966
and 1969 coincident with the excavation of many other sites
in the gorge. It is a multi-period site with deposits dating to
the Roman, Early Iron Age (EIA), Early Neolithic, and pos-
sibly Terminal Mesolithic (Živanović 1976b; Srejović &
Letica 1978; Jovanović 1984; Vasić 1983; Prinz 1987;
Radovanović 1996). The chronological position and cultural
affinities of the prehistoric layers at Hajdučka Vodenica have
long been a source of controversy. Originally, all the prehis-
toric levels were reported to be from an Early Iron Age set-
tlement and cemetery (Jovanović 1966a, 1966b, 1969a,
1969b). It is now recognized that the prehistoric part of the
site can be divided between Early Iron Age, Early Neolithic
and Mesolithic occupations (Srejović & Letica 1978;
Jovanović 1984; Radovanović 1996). The excavations ex-
tended between 2.0 and 2.5 metres in depth (Jovanović
1966a: 102).
The upper level (nivo 1) included an EIA settlement and
necropolis, and associated architectural features and arte-
facts. The EIA level is stratigraphically superimposed above
the Early Neolithic levels, but EIA burials occasionally in-
trude into them. The EIA horizon is stratigraphically and
culturally distinct and separated by a sterile sandy layer from
the Early Neolithic and Mesolithic cultural layers. This ulti-
mately allowed the material remains from the lower levels
(Early Neolithic and Mesolithic) to be separated from the
upper levels at the site (Iron Age and later). The soil matrix
from the cultural levels was also mixed with sand.
Recently, the lower horizons at the site have been tempor-
ally reassigned. Nivo 2 has been reassigned to the Early
Neolithic Starčevo-Criş culture (Hajdučka Vodenica II). The
finds from levels (nivo) 3–4 and the level with stone con-
structions have been reassigned to the Terminal Mesolithic
(Jovanović 1984; Radovanović 1996) (see Appendix 3). The
Mesolithic occupation has been subdivided into an earlier
occupation (Hajdučka Vodenica Ia — end of the 8th and first
half of the 7th millennium BC), and a later (Hajdučka
Vodenica Ib — first half of the 6th millennium BC)
(Radovanović 1996). The latter phase is contemporary with
the earliest phases of the Early Neolithic cultures (Starčevo-
Criş) beyond the Iron Gates (c. 7450–6450 BP — Garašanin
1983; 6100–5100 cal BC — Manson 1990, this volume).
The architecture and sculptures found in levels 2–4 include
free standing stone sculptures, trapezoidal houses, and
stone-lined rectangular hearths often used for inhumations
(Jovanović 1966a, 1966b, 1969a, 1969b, 1984). These are
typical of Mesolithic culture in the region and are found at
Lepenski Vir I–II, Padina IA, and Vlasac I–III (Radovanović
1996). Early Neolithic Starčevo-type ceramics and lithic and
ground stone tools are present, however. These are similar to
those from Early Neolithic Starčevo culture levels at sites in
the region which have been associated with pit houses (at
Lepenski Vir III and Vlasac IV) and trapezoidal houses and
stone-lined rectangular hearths (at Padina IB — Srejović
1970, 1972; Jovanović 1974; Srejović & Letica 1978). Early
Neolithic levels at these sites are usually distinguished by the
presence of domestic animals (Bökönyi 1970; Clason 1980:
143). Because of continuity between Late Mesolithic and
Early Neolithic sites in the chipped stone tool assemblages
and architecture (Voytek & Tringham 1989), the Early
Neolithic in the Iron Gates is considered one of the few ex-
amples of the indigenous adoption by the Mesolithic inhab-
itants of Southeast Europe of an Early Neolithic lifestyle
(Greenfield 1993). The appearance of pottery has often been
used to distinguish Mesolithic from Neolithic occupations.
Yet, pottery similar to that of the Early Neolithic Starčevo
culture in the surrounding region appears in the first half of

Table 1. Percentage of tools in the assemblage compared with NISP.

Period unknown Early Neolithic Mesolithic Grand Total

# of % of # of % of # of % of # of % of
Tools NISP NISP Tools NISP NISP Tools NISP NISP Tools NISP NISP

Taxon
Bos taurus 3 2 5 40.00 2 8 25.00
Cervus elaphus 7 22 31.82 5 23 21.74 43 130 33.08 55 175 31.43
Ursus arctos 1 1 1 100.00 1 2 50.00
Large-sized mammal 1 1 100.00 1 1 100.00
Bos/Cervus 1 4 25.00 4 22 1 30 3.33
Unknown 0 9 0.00 2 2 100.00 1 2 12 16.67
Grand Total 8 35 22.86 8 34 23.53 46 159 28.93 62 228 27.19

207
 The Iron Gates in Prehistory
the 7th millennium BC (calibrated radiocarbon dates) in the
Mesolithic sites of the Iron Gates (Radovanović 1996). Soon
afterwards, architecture similar to that used by the Starčevo
culture (pit houses) makes its appearance in the gorge
(Lepenski Vir III).
Analytical methodology
The following analysis will provide a description of the bone
finds from Hajdučka Vodenica, and the foundation for some
cautiously drawn conclusions based on the sample. All of the
bones described here derive from the pre-Early Iron Age
levels.
Excavation and recovery methodology
Excavation was conducted in two ways. First, after clearing
of the overlying humus and Roman layers, excavation was
conducted by following the architectural features within
trenches. Most of the artefact bags are labelled by trench and
feature. The most pronounced features were trapezoidal
houses, with stone-lined rectangular hearths, and horizontal
clay floors cut back into the terraces in a manner similar to the
houses at Lepenski Vir, Padina, and Vlasac. Second, where
architecturally defined units were not visible, strata were ex-
cavated by a system of horizontal excavation units of arbit-
rary depth (usually 10 cm) that potentially cut across the
sloping natural strata. This may have resulted in the mixture
of materials from different periods. Where possible, artefacts
from different coloured soil deposits within the same level
were collected and analyzed separately. As a result, the Early
Iron Age material was separated from the Early Neolithic (by
the excavator — B. Jovanović, pers. comm. 1982).
Hajdučka Vodenica was one of the first early prehistoric
sites to be excavated in the gorge. As such, its importance
was not recognized. Artefacts were haphazardly collected
without the aid of any mechanical devices (e.g. sieves).
Consequently, most bones were probably not recovered.
Bone tools and other bones with evidence of modification
were preferentially collected (B. Jovanović, pers. comm.
1982). The bias in recovery is obvious from the high per-
centage of tools in the assemblage (Table 1). No well-
collected collection from the region has such a high percent-
age of bone tools (Greenfield 1986, 1991).
As with most other Mesolithic and Early Neolithic sites
from the Iron Gates (Vlasac, Padina and Lepenski Vir — as
cited in Bökönyi 1978; Clason 1980: 142), the sample from
Hajdučka Vodenica is small. The exception is Vlasac, which
yielded a much more substantial faunal assemblage. Vlasac
was the last of the Mesolithic and Early Neolithic sites to be
excavated in the gorge. At the time of its excavation, there
was an increased awareness on the part of excavators from
the region of the importance of faunal assemblages.
Excavations at Vlasac were therefore conducted with a more
systematic rigour that had been practised during earlier ex-
cavations in the region (S. Bökönyi, pers. comm. 1982).
Because of its relative size and the improved recovery tech-
niques employed, Vlasac provides a comparative sample for
the scale and quality of the collections that were lost or dis-
carded at earlier excavations in the Iron Gates area. Recovery
208
of faunal remains, nevertheless, was not fully systematic in
any of the samples (Greenfield 1993).
Bone preservation
A variety of factors, other than recovery methodology, affect
the quality of the bone assemblage. Almost all bones show
signs of surface erosion, many with deep pitting revealing the
trabecular structure. Soil pH levels were not measured since
the excavation took place prior to an interest in taphonomy in
the region. Location of the site on a loess terrace, the high
sand component of the soil matrix, and the poor preservation
of the surfaces of recovered bones indicate that extensive at-
trition of the bone assemblage occurred prior to excavation.
Other factors affecting the nature of the assemblage include
disturbance and reworking of deposits through down slope
movement, river flooding and erosion, and post-occupational
interments.
The presence or absence of the various bone elements from
each species does not reflect the differential butchery or dis-
posal practices of the ancient inhabitants of the site. Instead,
they are a direct reflection of the recovery procedures used
during excavation. Given the problems recognized above, it
is necessary to ask why any attention should be devoted to
such a poor data set? Several reasons exist. First, the gorge
has been completely flooded since the completion of the hy-
droelectric dam and comparable additional data sets are un-
likely ever to appear. Second, it is the last of the collections
to go unreported and can add to our discussion of early cul-
tural adaptations in the region. Third, the vast majority of the
faunal remains from Lepenski Vir and Vlasac were discarded
after they were identified by Bökönyi (1971, 1978 — per-
sonal communications from Sandor Bökönyi, Svetozar
Stanković, and Dragoslav Srejović, 1982). Only some selec-
ted samples were curated for posterity. Only a preliminary
report was published on the Icoana sample before the pre-
mature death of the analyst (Bolomey 1973). Aside from the
limited published data available, little new information can
be squeezed out of the reports. Only the faunal remains from
Padina and Hajdučka Vodenica, of the Serbian sites, have
been entirely curated. Therefore, Hajdučka Vodenica is only
one of two faunal data sets from the region currently avail-
able for reanalysis. Hence, the utility of this report lies in de-
scribing potentially useful material still available for analys-
is. Finally, the faunal collection can, as will be shown below,
contribute to the discussion of the chronological position of
Hajdučka Vodenica.
Quantification procedures
Quantification of bone assemblages has been the subject of
numerous debates (Grayson 1984). Several techniques have
been proposed, but only two are commonly used for Early
Neolithic assemblages in the central Balkans: number of
identified specimens (NISP) and minimum number of indi-
viduals (MNI) (Greenfield 1986, 1991). Both quantification
procedures were used for the Hajdučka Vodenica fauna.
During the NISP calculations, each bone or bone fragment
was separately identified and analyzed using procedures out-
lined in Greenfield (1986). In brief, when articulations
between bones or fragments were recognized, they were
analyzed as belonging to the same individual and quantified
 Haskel Greenfield: Reanalysis of the vertebrate fauna from Hajdučka Vodenica

as equal to one specimen regardless of the number of bones
or fragments that were present. Thus, each separate specimen
or group of articulated specimens was quantified only once.
MNIs were calculated using Bökönyi’s (1970) method.
Similar quantification procedures were used for all other sites
in the region. No whole skeletons were found in the
assemblage.
Hajdučka Vodenica sample, while assemblages from con-
temporary sites in the gorge include many. Based on this
comparison, we can assume that the Hajdučka Vodenica
people probably exploited molluscs as well. The absence of
shell in the sample is a result of the previously described
sampling bias. The same is likely for fish and other typically
under-sampled taxa in haphazardly collected samples.

The sample — general considerations
Faunal sample and species diversity
With a total of 354 separate bone fragments recovered
(Appendix 1), the faunal sample from Hajdučka Vodenica is
the smallest collected from the Early Neolithic/Mesolithic
sites in the Iron Gates (Table 2). In contrast to other samples
from the area, the Hajdučka Vodenica sample lacks diversity,
although remains of at least 12 different taxa were identified.
This is undoubtedly a reflection of the recovery techniques
employed at Hajdučka Vodenica (Greenfield 1993, this
volume).
Identified species include wild and domestic cattle, wild
and domestic pigs, domestic dogs, red and roe deer, brown
bear, chamois, rodent, fish, and human. Only some of the
major domestic species are present in both the Mesolithic and
Early Neolithic assemblages from Hajdučka Vodenica (i.e.
cattle, pig and dog — all of which may have had a local
source of domestication: Bökönyi 1974, 1978). Notably
lacking are sheep and goat, which would have been domest-
icated in the Near East. The wild taxa at Hajdučka Vodenica
lack diversity when compared to the larger samples from the
gorge. For instance, there were no mollusc shells in the
Size categories
A small quantity of remains were identified only to more
general taxonomic size categories (Table 2), such as
medium-sized mammals (N=4) and large-sized mammals
(N=1). Most of the fragments relegated to these categories
were ribs, vertebrae, and other fragmentary remains that are
normally difficult to identify to a higher taxonomic level.
More commonly, remains were identifiable to a narrower, but
still relatively non-specific taxonomic grouping of genera
(e.g. Bos taurus/Cervus: N =40). Very few remains were
totally unidentifiable (N =14). The paucity of minimally
identifiable material is a reflection of the selective recovery
procedures employed during excavation.
Domestic:wild ratio
The Mesolithic has a much lower percentage of domestic
(N=38, 16.3% — Table 3) and higher percentage of wild
mammals (N=209, 83.7%) than the Early Neolithic levels
(domestic: N=14, 33.3%; wild: N=28, 66.7%). The Early
Neolithic percentages of domesticates are higher than those
from contemporary sites in the Iron Gates (Padina and
Lepenski Vir), but lower than most Early Neolithic as-
semblages outside of the Iron Gates (Greenfield, this
volume). In contrast to sites outside of the Iron Gates, all of

Table 2. Sum of number of fragments and NISP by taxon and period.

Period unknown Early Neolithic Mesolithic Grand Total

% of % of
Fragments NISP Fragments NISP NISP Fragments NISP NISP Fragments NISP

Taxon
Domestic Bos taurus 3 3 7.14 5 5 2.70 8 8
Sus scrofa dom. 4 3 8 8 19.05 23 22 11.89 35 33
Canis familiaris 1 1 3 3 7.14 6 6 3.24 10 10

Wild Bos primigenius 1 1 2.38 16 2 1.08 17 3

Sus scrofa fer. 2 2 4.76 12 12 6.49 14 14
Canis lupus 1 1 0.54 1 1
Cervus elaphus 22 22 23 23 54.76 139 130 70.27 184 175
Capreolus capreolus 1 1 0.54 1 1
Ursus arctos 1 1 2.38 1 1 0.54 2 2
Martes martes 1 1 0.54 1 1
Rupicapra rupicapra 1 1 0.54 1 1
Pisces (sp.) 1 1 2.38 13 3 1.62 4 4

Unknown Sus scrofa 2 2 2 2

Medium-sized mammal 4 2 4 2
Large-sized mammal 1 1 1 1
Bos/Cervus 4 4 8 4 28 22 40 30
Unknown 9 9 3 2 1 1 13 12
Sub-Total 15 15 12 7 33 25 60 47

N/A Homo sapiens 3 3 13 12 16 15

Grand Total 42 41 57 52 255 222 354 315

209
 The Iron Gates in Prehistory

Table 3. Summary of domestic and wild ratio by period.

Period unknown Early Neolithic Mesolithic Grand Total

% of % of
NISP NISP NISP NISP NISP NISP

Domestic Total 4 14 33.33 33 17.84 51

Wild Total 22 28 66.67 152 82.16 202
Total 26 42 100.00 185 100.00 253

Table 4. Temporal comparison of the NISP of Cervus elaphus with antler, without antler, and all elements by period.

Period unknown Early Neolithic Mesolithic Grand Total

% of % of
NISP NISP NISP NISP NISP NISP

Antler 4 14 33.33 33 17.84 51

All Elements (including antler) 22 28 66.67 152 82.16 202
All Elements (excluding antler) 26 42 100.00 185 100.00 253

Table 5. Number of burnt bone fragments by taxon and element.

Period: Unknown ? Mesolithic

Burning: Carbonized Slightly Partly

(black) burnt (red) carbonized

Taxa Element
Domestic Sus scrofa dom. Radius 1
Wild Cervus elaphus Antler 1
Astragalus 1
Bos primigenius Cranium – horn core, frontal and occipital 1
Grand Total 1 1 2

the Early Neolithic samples from the Iron Gates have much
higher percentages of wild than domestic fauna. The source
of the difference lies with the red deer antlers. The antlers
represent over half of the red deer remains (Table 4). When
the antler remains are removed from the bone counts, wild
fauna dominate the assemblage, albeit with a much smaller
percentage. The proportion of wild in the Mesolithic level is
much lower than in other Mesolithic assemblages (probably
because the antlers are counted in those assemblages). In
contrast, the proportions are much more similar to the Early
Neolithic assemblages from the region as a whole.
Sources of assemblage attrition
Burning
Most Mesolithic and Early Neolithic faunal reports do not
mention burnt bones at all. At Hajdučka Vodenica, only a
very small fraction of the bone sample was burnt (N=2 —
Table 5). In the indeterminate period sample, a subadult do-
mestic pig distal radius and a red deer subadult/adult antler
fragment were burnt. In the Mesolithic sample, a red deer
subadult/adult astragalus and an aurochs adult male horn core
were burnt. Three possible explanations why so few burnt
bones are found easily come to mind. First, if the low fraction
of burnt bones is a reflection of their original proportions in
the assemblage, we have to assume that most bones were
probably covered with meat when exposed to fire, were
boiled in a stew-like concoction, or had their meat removed
prior to cooking. In this case, there would have been limited
contact between the bone and the flame. A second possibility
is that the burnt bone may have simply disintegrated in the
soil, since burnt bone breaks up into fragments more readily
than uncooked bone (Bonfield & Li 1966; Shipman et al.
1984). A third reason that few pieces of burnt bone are found
could be due to differential disposal patterns. For example,
burnt bone is not necessarily discarded in a homogeneous
fashion across sites. Unless it is moved to a secondary de-
posit, most burnt bone would be expected to be found near
cooking areas (e.g. hearths). If they are redeposited, they
should be found in nearby middens. The depositional context
of the burnt bone is unclear. Based on the provenance in-
formation, none of the burnt fragments come from any of the
excavated hearths or features. While several hearths were
found, it is not clear if any burnt remains were recovered
from them (B. Jovanović, pers. comm. 1982). Did they come

210
 Haskel Greenfield: Reanalysis of the vertebrate fauna from Hajdučka Vodenica

Table 6. Distribution of butchered elements by period.

Period unknown Early Neolithic Mesolithic Grand Total

Butchering type: 4 5 9 5 7 9 4 5 7

Taxon Element Part

Bos taurus Metacarpus Distal 1 1
Cervus elaphus Cranium Frontal and antler 1 1 1 1 2 1 6 6 19
Astragalus Whole 3 3
Ursus arctos Astragalus Whole 1 1
Total 1 1 1 1 1 2 2 6 9 24

Butchering type
4 chop mark/severed lengthwise to main axis
5 chop mark/severed cross-wise to main axis
7 slice cut mark crosswise to main axis
9 slice cut mark and break in bone (bone broken subsequently)

Table 7. Distribution of NISP by weathering stages.

Period unknown Early Neolithic Mesolithic Grand Total

% of % of % of
NISP NISP NISP NISP NISP NISP NISP

Weathering stage:

Slight 24 27 48.21 123 54.42 174 53.87

Medium 17 29 51.79 101 44.69 147 45.51
Heavy 0.00 2 0.88 2 0.62
Grand Total 41 56 100.00 226 100.00 323 100.00

from food preparation areas or from middens where the burnt
bone would have broken up and disintegrated with relative
speed? Similarly, low proportions of burnt bone are found in
other Early and Late Neolithic assemblages (Greenfield
1991, 1993).
Butchering
A small sample of remains shows evidence of butchering
marks: the result of skinning or limb dismemberment (Table
6). Most butchered remains derive from the Mesolithic
(N=17 out of 185, 9.2%). These include a variety of taxa. A
domestic cattle metatarsus had its distal condyles chopped
off with a butchering instrument, probably in an attempt to
sever the foot bones from the rest of the carcass. Three red
deer astragali (subadult/adult) exhibit slicing marks perpen-
dicular to the long axis of the bone, which are also probably
by-products of severing the feet from the lower limbs. A
number (N=13) of red deer cranial remains (usually frontal
bones with antlers still attached) show signs of chopping,
cutting and/or slicing especially around the pedicles to either
remove the skin or the antlers (for tool preparation). While
the number of butchered remains during the Early Neolithic
declines (N=4 out of 42), the percentage is similar to the
Mesolithic (9.5%). Three red deer crania with antler remains
show evidence of chopping and slicing. A bear calcaneus has
a slice perpendicular to the long axis of the bone, on the lat-
eral side, as part of the severing of the muscles and ligaments
to detach the foot from the lower limb. There are no data from
other Early Neolithic sites with which to compare these data.
Weathering
While burning and butchering are more or less ambiguous
indicators of the degree of assemblage attrition, weathering is
less so. Table 7 shows the distribution of NISP elements by
weathering stage. The differences between either period and
the total are insignificant; the pattern is the same regardless of
the period. In general, the assemblage is more highly
weathered than it should be in order to have any expectation
that the more fragile specimens (based on age or morpho-
logy) would be found in the numbers according to which they
were deposited at the site (Lyman 1994).
Weathering was analyzed using three general stages fol-
lowing the system proposed by Meadow (1978). In general,
slightly more than half (53.87%) of the assemblage has slight
weathering (surface is discoloured, but unpitted). Slightly
less than half (45.51%) exhibited medium weathering (com-
pact surface pitted, lightly eroded and cracked). A small per-
centage (0.62%) also exhibited heavy weathering (major de-
struction of compact surface; cancellous bones extensively
exposed). While it is unclear to what extent weathering has
affected the other assemblages from the Iron Gates, the as-
semblage from Hajdučka Vodenica is only slightly more
heavily weathered than contemporary assemblages
(Greenfield 1993; Greenfield & Jongsma n.d.). For example,
at Blagotin 35.8% was lightly weathered, 57% was moder-
ately weathered and 7.1% was heavily weathered. The Early

211
 The Iron Gates in Prehistory

Table 8. Distribution of fragments by size class and taxon.

Period unknown Early Neolithic Mesolithic

Whole 3/4 1/4 ? Whole 3/4 1/2 1/4 ?<1/4 ? Whole 3/4 1/2 1/4 ?<1/4 ?

Taxon
Domestic
Bos taurus 1 1 1 2 3
Sus scrofa dom. 1 2 2 1 3 2 4 3 2 5 6 2
Canis familiaris 1 1 2 6

Wild
Bos primigenius 1 1 1
Sus scrofa fer. 1 1 1 2 1 3 5
Canis lupus 1
Cervus elaphus 1 1 1 19 1 1 5 4 12 7 6 2 8 14 93
Capreolus capreolus 1
Ursus arctos 1 1
Martes sp. 1
Rupicapra rupicapra 1
Pisces sp. 1 1 1 1
Grand Total 1 2 3 20 5 2 3 10 8 14 12 11 7 18 27 110

Neolithic assemblages, in general, are much more extens-
ively weathered than later prehistoric samples that have been
analyzed from outside of the Iron Gates (Greenfield 1986,
1991).
The species (Tables 2, 6 & 9)
Bos primigenius Bojanus, 1827 (aurochs) and Bos taurus
Linnæus, 1758 (domestic cattle)
Very small quantities of domestic and wild cattle bones were
identified. Transitional domestic–wild individuals (Bökönyi
1962) were not identified, although their presence would be
difficult to determine because of the small sample size and
fragmentary and eroded nature of the material. The paucity of
cattle remains in the sample is somewhat expected given the
rough terrain. Neither wild nor domestic cattle are good
climbers. Both can live in a variety of habitats, but usually
prefer the open fields or lightly forested areas. The environ-
ment surrounding Hajdučka Vodenica, today and in the past,
is characterized by broken terrain covered by dense decidu-
ous and evergreen forests (Mišic et al. 1970; Gigov 1972;
Cârciumaru 1978), and one not particularly favoured by
cattle. The small cattle frequencies are overshadowed by
those of red deer — a common situation among the other
Mesolithic and Early Neolithic sites in the Iron Gates
(Bökönyi 1970, 1978; Bolomey 1973; Clason 1980), and not
unexpected given the rough terrain and heavy forest cover.
Aurochs are represented during the Mesolithic by a NISP
of two and during the Early Neolithic by a NISP of 1. One of
the Mesolithic aurochs bones is a very large horn core with
extremely thick walls, which probably belonged to a male. It
was broken into several fragments, badly burnt, and not
measurable. The other two aurochs bones are calcanei, one of
which came from the Mesolithic and one from the Early
Neolithic levels. The Mesolithic calcaneus is complete and
measurable. The measured length of 152.0mm exceeds the
length of recognized prehistoric domestic varieties and falls
within the size range of wild cattle from other Mesolithic and
Neolithic collections from the region, such as Divostin, Obre
I and II, and Gomolava (Bökönyi 1974a, 1974b: 119, 1978;
1988; Clason 1979: 95). Bökönyi regards 148.0mm as a use-
ful and widely recognized lower limit for aurochs. The spe-
cific weight of the element and the developed state of the
muscle insertion areas are further support for its assignment
(cf. Bökönyi 1972b, 1978: 44). All of the bones came from
fully adult individuals.
Domestic cattle are represented in the Mesolithic by a
NISP of 5 and during the Early Neolithic by 3. It is highly
unlikely that any of the bones derive from the same indi-
viduals because of differences in size and age. Domestic
cattle are represented by a slightly wider variety of body parts
than wild cattle (Appendix 1). Cranial fragments are notably
absent. All of the bones, except a metatarsal and a calcaneus,
are fragmented (less than half their original size). Little can
be said about the age distribution. Three fragments are from
adults. The rest are completely fused but are from relatively
early fusing elements, and therefore could represent either
adults or subadults. This age distribution is heavily biased
towards adult remains, which is probably a result of heavy
attrition on the assemblage. Contemporary assemblages in-
dicate the presence of large numbers of immature individuals
(Bökönyi 1974b, 1976, 1978, 1988).
The withers height for a nearly complete Early Neolithic
metacarpus was determined in the following way: 218.0mm
(length) x 5.47 (Matolsci’s 1970 quotient) = 1192.46mm
(withers height). This value lies at the upper end of the dis-
tribution for early domestic cattle. Determination of the
Nobis (1954) index for sex was calculated as follows:
50.9mm (proximal width/218.0mm (greatest length) x
100 = 23.35. This value lies closer to the male range and
similar to the values for males from Obre II (Bökönyi 1974b:
69). Imposing the measurements from the distal metacarpal
(breadth versus depth) on the Obre cross-plot (Bökönyi
1974b: 73, fig. 5) places it within the upper size range of do-
mestic cattle.

212
 Haskel Greenfield: Reanalysis of the vertebrate fauna from Hajdučka Vodenica
Sus scrofa fer. Linnæus, 1758 (wild pig) and Sus scrofa
dom. (domestic pig)
Pig remains are the second most numerous in the total sample
(N=47). As with cattle, both domestic and wild forms are
present in each period. In each period, however, domestic
pigs are more common than wild. In the Mesolithic, domestic
pigs (11.89%) are twice as common as wild (6.49%). In the
Early Neolithic, this gap widens even more, with domestic
pigs representing 19.05% and wild pigs only 4.76% of the
assemblage (Table 2). Two pig bones whose stage of do-
mestication is uncertain were also identified.
Bones of domestic pigs in a late stage of domestication are
easily distinguishable from wild suids on the basis of size,
roughness around muscle insertion points, etc., as in cattle.
Therefore, clear evidence for the wide gap between the do-
mestic and wild forms should be found among the bone
measurements if there is no interbreeding between the popu-
lations or if the process of domestication has been completed.
All measurable bones except two fell into the clearly wild or
domestic categories (Appendix 2). In fact, most of the meas-
urable specimens fell within the wild and not the domestic
size range. A slightly larger percentage of the domestic bone
specimens were fragmented and unmeasurable than among
the wild specimens. While 15% of domestic pig bones were
complete, only 11.5% survived intact among wild suids
(Table 8). All the rest were fragmented to greater or lesser
degrees. The wild forms, because of their greater size exhib-
ited greater damage than the smaller domestic pig bones.
More whole specimens survived among the Mesolithic do-
mestic suid remains than in the Early Neolithic, however.
Two Mesolithic suid mandibles retained their third molars.
Both were from young adults, whose tooth wear was analog-
ous to Grant’s (1975) LM3 wear stages B and D, respectively.
Comparisons of the length measurements from both
Mesolithic and Early Neolithic remains (Appendix 2) with
those from Obre I and II, Vlasac, and Divostin (Bökönyi
1974b: 77, table 8, 1978: 59, 1988) places them within the
lower size range of wild swine. Measures of the distal end of
the scapula placed each of the specimens within the wild size
range, when compared with Obre I–II (Bökönyi 1974b: 77,
fig. 7) and larger than many of the wild specimens from
Vlasac (Bökönyi 1978: 60). Some of the wild individuals
were extremely large, as evidenced by the 56.8mm breadth of
a distal humerus. A few of the measured specimens fall
within the upper end of the domestic size range (i.e. a distal
tibia) or may even represent a transitional individual (i.e. a
proximal radius). The lack of agreement between two noted
authorities (Bökönyi 1974b, 1978; Clason 1979: 82) as to the
size limitations for the proximal radius between domestic and
wild forms lends an air of confusion to the latter
measurement.
Age profiles differ between the two species within each
period. Even with the obvious collection and survival biases
in the sample, the patterns somewhat mirror those from other
contemporaneous collections. The major difference is
between wild and domestic pigs in both periods. Wild pigs in
the Early Neolithic are entirely represented by adult remains,
and adults are 88.89% of the ageable sample in the
Mesolithic. The domestic pigs are more equitably divided
among the various age groups, although there is a fuller dis-
213
tribution in the Early Neolithic than the Mesolithic. In the
Mesolithic, the ageable remains are equally divided between
subadult and adult categories. In the Early Neolithic 28.57%
of the sample are in each of the juvenile and adult categories,
while 42.8% are subadults. The wild pig distribution vaguely
resembles Bökönyi’s (1978: 50, table 4) larger sample from
Mesolithic Vlasac. Assuming that it is representative (which
is highly questionable), it is consonant with a strategy of
hunting adults for meat and hides. It is a mortality distribu-
tion frequently associated with hunted populations
(Greenfield 1988, 1991). The well-watered and dense forests
throughout the region would have been an ideal habitat for
wild suids. The domestic suids’ age distribution is more
skewed toward younger individuals. This mortality profile is
relatively similar to that where young domestic pigs are
preferentially slaughtered for meat and a smaller adult
breeding stock is retained (Greenfield 1988). Domestic pigs
would have, in these early times, scavenged through the set-
tlement middens for sustenance or foraged in the surrounding
forests for wild tubers and foliage.
Canis lupus Linnæus, 1758 (wolf) and Canis familiaris
Linnæus, 1758 (domestic dog)
Eleven remains of canids were recovered — nine mandibular
and two maxillary fragments. Canids are the first animal to be
domesticated in Europe, with the earliest evidence coming
from the Upper Paleolithic (Musil 1984). However, Bökönyi
(1978: 38–44) has argued that the process of domestication
took place in the Iron Gates during the Mesolithic on the
basis of the identification of transitional individuals at
Vlasac. At slightly later sites, such as Lepenski Vir III and
Padina B (Bökönyi 1971; Clason 1980), there is a clear sep-
aration between wild and domestic forms. At Hajdučka
Vodenica only a single wolf bone (a mandible) was identified,
and it came from the Mesolithic level. The rest of the canid
bones were clearly from domestic forms. Domesticated can-
ids are present in both Mesolithic (N=6) and Early Neolithic
(N=3) levels supporting Bökönyi’s assertion that they were
already present in the Mesolithic levels of Lepenski Vir and
Vlasac. No post-cranial remains of canids were recovered,
although their absence is not a reflection of ancient human
behaviour. None of the specimens showed any evidence of
butchering or modification by humans. All the recovered in-
dividuals were adults.
Recognizing transitional individuals is always a difficult
task. The cross-plots from Vlasac (Bökönyi 1978: 39–40) are
quite ambiguous in this case. Instead, the clearest evidence
was found in the shortening of the mandible in the pre-molar
region, while the pre-molars remain relatively large in size,
causing a crowding of the teeth in the jaw. At Hajdučka
Vodenica, all except two of the specimens clearly belong to
domesticated individuals. They fall well within the size range
of domestic animals from contemporary and later Neolithic
sites of the region — i.e. Vlasac, Divostin, Gomolava, and
Padina (Bökönyi 1978, 1988; Clason 1979, 1980). One of the
two remaining specimens exhibits the classic characteristics
of transitional individuals — a carnassial length of 28.4mm
(well within the wolf size range), with a shortened pre-molar
region as evidenced by a mandible height before the PM1 of
21.3mm (well within the domestic dog size range). This in-
 The Iron Gates in Prehistory
dividual has been classified as a wolf here since it is much
more robust than most of the other specimens. It comes from
the Mesolithic horizon (Appendix 2). The second individual
comes from the Early Neolithic horizon and showed evid-
ence of the opposite nature — a short carnassial length
(23.2mm) and a high height before the LPM1 (26.8mm). This
latter specimen may represent either a relatively robust do-
mestic individual or an incorrect measurement. It has been
treated in the analysis as probably domestic. The presence of
a transitional individual in the Mesolithic assemblage sug-
gests that the process of domestication was not necessarily
completed by the Mesolithic in the region, supporting
Bökönyi’s (1978) hypothesis.
Cervus elaphus Linnæus, 1758 (red deer)
Red deer are the most common vertebrate species in the as-
semblage from Mesolithic and Early Neolithic Hajdučka
Vodenica (NISP=175). The frequency of red deer remains is
high in both periods. It is highest in the Mesolithic (70.27%)
and declines somewhat in the Early Neolithic (54.76%).
The frequency is fairly typical of Mesolithic and Early
Neolithic assemblages from the Iron Gates (when fish are not
counted), but is atypical of the Early Neolithic assemblages
from the surrounding regions (Greenfield 1993). Red deer are
the most common mammal in both periods at Iron Gates sites
such as Vlasac, Lepenski Vir, and Padina. At Vlasac and
Lepenski Vir I, only fish remains exceed those of red deer in
Iron Gates sites (Bökönyi 1970, 1978; Clason 1980), which is
a reflection of the differential recovery methods employed by
the different teams excavating Lepenski Vir and Vlasac (led
by Dragoslav Srejović — Srejović 1972; Srejović & Letica
1978) versus Padina and Hajdučka Vodenica (led by B.
Jovanović — Jovanović 1966a, 1966b, 1968a, 1968b, 1969a,
1971, 1983, 1984).
Antler fragments (N=122) represented, by far, the largest
category of body elements. This distribution strongly re-
sembles that from Padina (Clason 1980), the only other Iron
Gates locality from which an element distribution has been
published. The non-antler element distribution, however, is
probably a better measure of the relative frequency of red
deer in the assemblage. Red deer is the most common species
in the sample even if only the post-cranial distribution of
elements by species is considered. The preponderance of
antler fragments is partially traceable to the widespread use
of antler as one of the more important raw materials for tool
production before the advent of metallurgy. The vast majority
of tools from the site were made from red deer bone and
antler elements. However, comparisons with red deer fre-
quencies from other contemporary localities (e.g. Bukovačka
Česma near Svetozarevo/Jagodina — as cited in Greenfield
1993, 1994), where recovery was not preferential toward
bone tools or potential tools, indicate that the element distri-
bution at Hajdučka Vodenica is clearly a function of selective
recovery for bones that were actual or potential tools.
All of the tools, except two (which are from a tibia and
metacarpus) were made from antlers. Forty-three antler frag-
ments were probably used as tools, based upon the criteria of
shape and edge wear. The rest of the antler fragments,
however, are not tools, even though many were inventoried
as such during excavation. They represent little more than
214
battered stem and/or polished tine fragments that may or may
not achieve these conditions naturally. Deer, during the peri-
ods when the velvet and antler are shed, frequently rub their
antler against trees to remove the velvet and to relieve itch-
ing. Polish or the simple presence of edge wear on antler may
result from such rubbing and is not necessarily indicative of
their use by humans as tools. This is a common assumption
among prehistorians who have never examined modern spe-
cimens of newly shed deer antler. Therefore, caution must be
applied to any antler sample in the determination of human
modification. Several stem and tine fragments did show
evidence of human modification, resulting in systematic pat-
terns of polish and edge wear. This is easily spotted when it
is the result of repetitive movements against an object. Other
more obvious tools were those whose shape had been modi-
fied for some specific function. This group was classified us-
ing Bačkalov’s (1979) morphological typology for Early
Neolithic bone tools from the region (Table 10). It should be
noted that his morphological types do not necessarily repres-
ent uni-functional categories (Semenov 1964).
Chopping and slicing marks were observed on 13
Mesolithic and 3 Early Neolithic antler fragments, probably
the result of tool production. Parts of almost every section of
the post-cranial skeleton, including vertebrae, were found.
This would indicate the absence of any ‘schlepp’ effect
(Perkins & Daly 1968). The entire carcass of a kill was
brought back to the settlement for processing, as at Vlasac
(Bökönyi 1978: 45).
The age distribution of both periods is heavily weighted
toward adults (Early Neolithic, 89%; Mesolithic, 93%), with
the remainder belonging to subadults and juveniles (Table 9).
The age distribution resembles that from Vlasac, when the
antler data are included. The age profile resembles that of a
hunted population where hunters rely upon their prey for a
variety of products, such as meat, hide and antler. Under such
conditions, mature males would be the focus of hunting
activities. If the antler data are not included in the age estim-
ates, the preponderance of adults is significantly diminished,
but still dominates the distribution in both periods (66%
Early Neolithic; 56% Mesolithic). Bökönyi (1978: 50) con-
cluded that the presence of immature individuals in the
sample from Vlasac indicated that Mesolithic hunters were
‘uninterested’ in preserving the reproductive viability of local
herds by exploiting immature individuals too heavily. Jarman
(1972) argued, in contrast, that the presence of immature
specimens can be used as evidence for early domestication
among cervids on the basis of north Italian data. Alternative
explanations for such distributions may lie in the inherent
insecurity of hunting as a major food source during certain
times of the year. Time and energy are valued resources
among all human groups and these are often carefully alloc-
ated to maximize return for the energy invested and, when
pressed, hunters will take what they can. Faunal assemblages
from contemporaneous and later prehistoric localities outside
the Iron Gates (e.g. Greenfield 1986, 1991, 1993, 1994;
Bökönyi 1988) yielded smaller quantities of immature red
deer. The culling of immature individuals may simply be a
result of hunters catching some prey accidentally, without
any possibility of choice, such as in snares, shooting acci-
dents, or the finding of sick or disabled individuals who could
 Haskel Greenfield: Reanalysis of the vertebrate fauna from Hajdučka Vodenica

Table 9. Distribution of NISP by age groups.

Infant/Juvenile Juvenile Subadult Subadult/Adult* Adult Total

% of % of % of % of
NISP NISP NISP NISP NISP NISP NISP NISP
Early Neolithic
Bos taurus 2 1 100.00 1
Sus scrofa dom. 2 28.57 3 42.86 1 2 28.57 7
Canis familiaris 3 100.00 3
Bos primigenius 1 100.00 1
Sus scrofa fer. 2 100.00 2
Canis lupus 0
Cervus elaphus 2 10.53 4 17 89.47 19
Cervus elaphus (excluding antler) 2 33.33 4 4 66.67 6
Capreolus capreolus 0
Ursus arctos 1 100.00 1

Mesolithic
Bos taurus 3 2 100.00 2
Sus scrofa dom. 6 50.00 9 6 50.00 12
Canis familiaris 1 5 100.00 5
Bos primigenius 2 100.00 2
Sus scrofa fer. 1 11.11 2 8 88.89 9
Canis lupus 1 100.00 1
Cervus elaphus 4 3.88 3 2.91 24 96 93.20 103
Cervus elaphus (excluding antler) 4 25.00 3 18.75 23 9 56.25 16
Capreolus capreolus 1 100.00 1

Table 10. Distribution of bone tools by type, period and taxon (based on Bačkalov 1979).

Subadult/Adult*
Large-sized Unknown Cervus Ursus Bos taurus Total
Period Tool type mammal elaphus arctos

Unknown ? 3 3
1 – awl 1 1 2
4.2 – dagger, type 1 1 1 2
10.1 – axe, type 1 1 1 2
11 – hoe 1
Unknown Total 1 0 7 0 2 9

Early Neolithic ? 1 1 4 6
Awl? 1 1 2

Early Neolithic Total 1 2 5 0 0 8

Mesolithic ? 21 21
3.5 – puncher, type 3 1 1
4 – dagger 1 1
4.1 – dagger, type 1 1 1 2
5 – pick 1 1
9 – chisel 2 2
9.1 – chisel, type 1 1 1
9.2 – chisel, type 2 2 2
10 – axe 1 1
10.2 – axe, type 2 3 3
10.4 – axe, type 4 2 2
11.5 – hoe 1 1
10 or 11 – axe / hoe 4 4
10 or 11 – axe / hoe 2 2
14 or 15 – hammer / club 1 1
4 or 5 – dagger / pick 1 1
Mesolithic Total 0 0 43 1 2 46

Grand Total 2 2 55 1 2 62

215
 The Iron Gates in Prehistory
no longer keep up with the herd. Deer shot while raiding vil-
lage fields may also account for some of the young speci-
mens. The two maxillae and two of the mandibles belong to
young adults. All the teeth had erupted through the bone, but
the tooth surface wear was very light. None of the data poin-
ted to the presence of very mature and/or senile individuals.
The third mandible was that of a subadult. This may indicate
the possibility that the hunters were selecting for older im-
mature or young adults. Such a strategy would have been
consonant with a long-term hunting mode of exploitation that
avoids the older reproductive core of the herd.
The red deer bone measurements reveal small, medium,
and large individuals (Appendix 3). They tend to be smaller
than those found during the Neolithic in the Hungarian Plain
and at Vlasac (Bökönyi 1974a, 1978). Geographic barriers,
such as mountains and rivers that cross the terrain between
Vlasac and Hajdučka Vodenica, at the opposite ends of the
Iron Gates, would have resulted in some interruptions in gene
flow between populations. In addition, a single specimen, a
complete radius of an adult, revealed progressive bone
growth typical of arthritis.
Capreolus capreolus Linnæus, 1758 (roe deer)
A single damaged antler stem from a roe deer was recovered
from the Mesolithic level. It is from an adult male and was
modified by humans for use as a tool. The very small fre-
quency for roe deer is not surprising. They are also present in
small numbers at Lepenski Vir, Vlasac, and Padina. The re-
mains of roe deer are invariably less common than those of
red deer on prehistoric sites of the region for a number of
reasons. The reasons for small frequencies of roe deer
include:
1) the presence of a dense forest, less suited to roe deer eco-
logical preferences (Boesseneck 1956, cited in Bökönyi
1978: 45). They favour more open country with good
ground cover (Clason 1980: 158);
2) a smaller return for energy invested in finding and killing
them as compared with a larger species. They are smaller
and lead a more solitary existence than red deer; and
3) their small bones are not easily visible and collected as
frequently as the larger deer forms.
Rupicapra rupicapra Linnæus, 1758 (chamois)
A single distal metacarpal of an adult chamois was found in
the Mesolithic level. Chamois are found at other Iron Gates
sites (Vlasac, Padina, Lepenski Vir) and at Obre. This speci-
men is larger than the Obre II specimens and larger than do-
mestic sheep at Neolithic sites from the region (Bökönyi
1974b: 142). The high mountains above the gorge are prefer-
ential habitats for chamois which live in medium to high
mountain ranges, between 800 and 2300 m in altitude. They
come down to the lower elevations during the winter but re-
main in either deciduous or coniferous forests, such as those
bordering the Danube either side of the gorge (Clason 1980:
163).
Ursus arctos Linnæus, 1758 (brown bear)
Two bones, a radius (Mesolithic) and a calcaneus (Early
Neolithic), of brown bear were found. Bears are frequently
found in prehistoric faunal assemblages of the region, but in-
216
variably in small quantities. Their large size does not usually
cause recovery problems. Both elements were from adults.
Martes sp. (marten)
A mandible of an adult marten was found in the Mesolithic
level. The geographical range of the two Holocene species of
marten indigenous to Europe overlap in this area. The speci-
men was not well enough preserved to determine which spe-
cies was present. Pine marten (Martes martes) has been
identified from Vlasac and Padina (Bökönyi 1978; Clason
1980).
Pisces sp. (fish)
One fish bone from the Early Neolithic and three from the
Mesolithic levels were found. The two from the Mesolithic
levels were vertebrae from large individuals. The other two
specimens were from indeterminate elements.2
Homo sapiens (human)
Fifteen specimens of human skeletal remains were mixed
with the Mesolithic (N=12) and Neolithic (N=3) fauna. It
was not clear if they derived from contemporary burials or
from the Iron Age necropolis that occasionally cut into the
Neolithic levels. They will not be discussed further.
Site seasonality
Only one category of data yields some hints as to settlement
seasonality. Two (one each from the Mesolithic and Early
Neolithic, respectively) red deer cranial frontal fragments
connected with antlers broken just above the pedicle and four
shed antler bases were recovered. Three of the shed antlers,
however, derive from indeterminate levels at the site, and
cannot be specifically assigned to the Mesolithic or Early
Neolithic. One of the shed antlers comes from the Mesolithic
level. The unshed antlers may indicate that the site was oc-
cupied at some time from late autumn into early spring — the
time of the year when red deer carry a full rack of antler
(Chaplin 1977: 106). Since no cranial fragments with shed
antlers were identified, the recovered shed antlers could have
been picked up in the forest at any time of the year and
brought to Hajdučka Vodenica. Red deer stags are easiest to
find in the forest and produce the best meat, hide, and antler
between January and March (Jochim 1976: 108–109).
Populations migrate in and out of regions, disperse and ag-
glomerate over the course of the year making them easy to
find only during a limited portion of the year. In mountainous
areas, red deer herds move seasonally between winter low-
land and summer upland grazing areas (Chaplin 1977).
Exploitation of deer most probably did not take place as a
major year-round activity, but only when body weight, antler
development, and herd behaviour (agglomeration of males)
made hunting more profitable an activity to pursue than other
resources.
Radovanović (1996: 58) proposed that the Mesolithic oc-
cupations from the gorge were year-round occupations. In the
absence of equivalent analysis for the Early Neolithic level, it
is impossible to conclude if the Iron Gates sites were occu-
pied year-round or for only a segment of the annual cycle.
 Haskel Greenfield: Reanalysis of the vertebrate fauna from Hajdučka Vodenica
Most explanations of the resource base of these Early
Neolithic localities have assumed, if only implicitly, that they
were occupied throughout the year and would, therefore,
have to cope with changes in the abundance and availability
of wild resources from season to season. Maintenance of do-
mestic stock at the site, thereby arranging for storage of food
on the hoof, would have helped, or so it is assumed, to even
out any fluctuations by compensating for lost food resources
(Flannery 1973). Even though the broad intermontane basins
beyond the gorge are agriculturally productive, the broader
river coves where the Early Neolithic settlements are found
are not. Each of the sites in the gorge is backed up against
steep inclines making agricultural activities difficult to pur-
sue. The likelihood, therefore, of some seasonal scheduling
of subsistence activities is high. Other methods for determ-
ining seasonality may still be used on the curated assemblage
(e.g. tooth cementum analysis). Only the faunal analysis from
Padina suggests that the site was permanently inhabited dur-
ing all the seasons of the year (Clason 1980: 171). The data
from the other sites are too ambiguous.
Evidence for pre-Neolithic domestication
Bolomey (1973) and Bökönyi (1974a) have both argued that
cattle and pigs were domesticated in Southeast Europe inde-
pendently of the centre of domestication in the Near East.
Bolomey (1973) argued on the basis of the age distribution
from the Mesolithic levels at Icoana on the Romanian side of
the gorge that pigs were domesticated in the gorge during the
Mesolithic. Clason (1980: 158) considers the domesticated
individuals from the Mesolithic levels at Padina to be prob-
ably intrusive. It is interesting that the age distribution of pigs
from Icoana resembles its contemporary site across the river
at Vlasac (Bökönyi 1978: 50). At Vlasac, no morphological
or age distribution evidence for domestication was found,
leading Bökönyi (1978) to conclude that the presence of im-
mature specimens was merely the outcome of occasional
kills of young wild specimens by local hunters. If we ex-
amine the data from Hajdučka Vodenica in light of this ques-
tion and use the same methodology as Bolomey (1973) and
Bökönyi (1969) above, in order to enhance comparability of
results, some interesting observations appear. Originally, the
entire faunal assemblage from the site was thought to derive
from the Early Neolithic, based upon the presence of ceram-
ics, the excavator’s notes and the early reports. However, as
a result of Radovanović’s (1996) recent work, I was able to
divide the assemblage between the Mesolithic and Early
Neolithic. Utilizing morphological criteria (such as size of
specimen and muscle attachment development in relation to
age class), most specimens could be classified as domestic or
wild. However, the results were somewhat surprising. Instead
of finding the expected pattern of domestic animals being
limited to the Early Neolithic, domestic animals were found
in both the Mesolithic and Early Neolithic levels. This is
similar to the pattern that Clason (1980) found at Padina and
Bolomey (1973) at Icoana. Clason (1980) and I (1984) ori-
ginally thought that the Padina and Icoana data were simply
later intrusions. However, this conclusion becomes less ten-
able now that there are three sites with similar patterns — the
217
presence of domestic fauna in the Mesolithic levels.
Radovanović (1996) has demonstrated that the Mesolithic
hunter-gatherers borrowed ceramics from their Starčevo
neighbours. My guess is that they are also borrowing do-
mestic animals. Domestic animal bones are far fewer in the
Mesolithic than in the Early Neolithic levels, but present
nonetheless. That would make sense since most of the later
Mesolithic sites in the Iron Gates are contemporary with the
earliest phases of the Early Neolithic outside of the Iron
Gates. Such a situation is not uncommon during the terminal
Mesolithic of Europe where hunter-gatherers lived for long
periods of time in close proximity to early farmers (e.g. in
Denmark — Price 1983).
The lack of evidence for transitional suid individuals at-
tests to the late stage of the domestication process among the
suid remains found at the site. Considering the evidence for
indigenous suid domestication in Southeast Europe (Bökönyi
1974a), the absence of transitional individuals and the pres-
ence of bones from pigs in a relatively later stage of domest-
ication is a cogent argument for borrowing of domestic fauna
by later Mesolithic hunter-gatherers from their contemporary
farming neighbours.
Conclusions
The appearance of domestic species on a site in a region that
has no previous history of domesticated animals has been
long used as one of the hallmarks of the Neolithic in Europe
(Price 1983: 762). Such is the case in Southeast Europe (the
central Balkans in particular), where the earliest domestic
animals have always been associated with the appearance of
the Early Neolithic archaeological cultures. There is no
evidence for indigenous domestic animals in the region, other
than the dog (Bökönyi 1978), prior to the appearance of the
Early Neolithic cultures.
The analysis of faunal assemblages from this region has
demonstrated that the advent of the Neolithic in Europe was
not a simple affair (Greenfield 1993). Domesticated eco-
nomies did not just move in and replace indigenous hunting-
gathering systems of resource procurement everywhere or at
once. In most Southeast European cases, there is evidence for
discontinuity in cultural data. Due to the relatively low fre-
quency of Mesolithic and Early Neolithic graves throughout
most of the region, it has been difficult to argue for human
population continuity or discontinuity between the two peri-
ods. However, in a few cases, such as the Iron Gates, there is
solid evidence for human biological continuity between the
late Pleistocene and early and middle Holocene populations
(e.g. Živanović 1976a; y’Edynak 1978; y’Edynak & Fleisch
1983; Mikić 1988) who adopt new subsistence strategies and
technologies as supplements, rather than replacements. This
is apparent from not only the faunal remains from the various
sites in the Iron Gates, but also from the stable isotope data
from Lepenski Vir and Vlasac. The stable isotope analysis
demonstrated a shift in the dietary regime from largely
aquatic to one based on terrestrial resources during the final
phase of the Iron Gates ‘Mesolithic’ (which was already
contemporary with early farming communities in the sur-
rounding region). The terrestrial resources probably included
 The Iron Gates in Prehistory
a major agricultural component (Bonsall et al. 2000).
In the Iron Gates, the obvious ‘advantages’ of an agricul-
tural lifestyle (apparent to so many prehistorians concerned
with early agriculture — e.g. Flannery 1973) were not ap-
parent to the indigenous early Holocene occupants of
Southeast Europe. The advantage of domestication lies in
minimizing the seasonal variations in available wild food re-
sources (Jochim 1976). But Mesolithic-type economies con-
tinue to exist for a short period of time in the Iron Gates even
after the appearance of food producers in surrounding re-
gions, as evidenced by the small frequencies of domestic an-
imals in faunal assemblages and the substantial cultural con-
tinuity between Mesolithic and Early Neolithic levels in the
Iron Gates sites. Their subsistence strategies were productive
and dependable enough to enable them to flourish (Tringham
1973). But they did not flourish in isolation since Starčevo-
type ceramics, domestic animals, and other aspects of the
material culture inventory of the surrounding societies even-
tually made their way into the Iron Gates. As a result, there
probably existed intergroup contact between indigenous
hunter-gatherers and early farmers during the Early Neolithic
in at least a few areas of Southeast Europe. They may have
made their way into the economies of the Mesolithic Iron
Gates communities through mechanisms, such as reciprocal
exchange relationships between hunting-gathering and agri-
cultural communities. Among modern hunter-gatherers in-
volved in exchange relationships with nearby farming com-
munities, meat from the former was exchanged for grain or
ceramics or metal tools from the latter (Turnbull 1965).
Are we really seeing the whole picture when we look at the
data from the Iron Gates? Or has our focus on the gorge itself,
with its spectacular view and remains, inhibited researchers
from turning their eyes upwards to the fields and forests of
the mountains towering above? What is the role of the Iron
Gates’ sites within a larger regional context? Could they have
been functionally specialized sites, where wild resources are
gathered, hunted or fished? The latter may be the case if sites
are fully contemporary with others from the larger region, but
located in a different setting. The resources exploited by the
inhabitants of sites in different ecological zones may be
complementary to one another. The exploitation of one set of
resources in a highland environment during the summer does
not preclude the exploitation of alternative resources in the
lowlands at other times of the year. We do not know about the
role of the Iron Gates sites within the larger region. It is pos-
sible and necessary to test the hypothesis that the Mesolithic
and Early Neolithic sites from the Iron Gates may represent
functionally specialized fishing-hunting localities for the ex-
ploitation of temporally limited abundance of a set of re-
sources. To test such a hypothesis would involve the system-
atic surface reconnaissance and excavation of contemporary
sites up in the mountains above the gorge, which is only now
being undertaken. Systematic survey of the gorge in the
1960s was limited to the area up to the flood line and not
beyond.
Notes
1. The prehistoric sections of the site were excavated under the
direction of Dr Borislav Jovanović of the Archaeological
Institute, Serbian Academy of Arts and Sciences, Belgrade. In
218
1982, Dr Jovanović reanalyzed the stratigraphic sequence and
artefactual assemblage from the site for publication (Jovanović
1984) and finally separated the fauna from the other curated ma-
terials. The fauna from the Early Neolithic layers was sub-
sequently made available to the author for analysis.
2. The fish specimens were sent to Dirk Brinkhuizen at the
Biologisch–Archaeologisch Instituut, Groningen in 1982 for
more specific identification. No information was received from
him as of the time of this report.
Acknowledgments
I should like to express my sincere gratitude to Dr Borislav
Jovanović for making the Hajdučka Vodenica material available for
study and encouraging its publication, to the late Prof. Dr Milutin
Garašanin, the late Prof. Dr Dragoslav Srejović, Dr Zagorka Letica,
Vesna Bogdanović Jeremenko, the late Svetozar Stanković,
Miodrag Sladić, and the countless other colleagues at the University
of Belgrade and Archaeological Institute (Belgrade) for aid and ad-
vice during the study of the material, to the late Dr Sandor Bökönyi
and Dr Anneke Clason for encouraging the investigation and com-
parison with their own studies of the Iron Gates fauna, to Valerie
McKinley for her critical insights during the writing of this paper,
and to my wife, Tina Jongsma, for urging me to re-evaluate my
earlier study. Financial support for the fieldwork was graciously
provided by the National Science Foundation, Fulbright-Hays
Foreign Language and Area Studies Program, International
Research and Exchanges Board, and Wenner-Gren Foundation for
Anthropological Research. The analysis was completed through the
support of The University of Manitoba and Social Science and
Humanities Research Council of Canada. Any errors in the analysis
are my own responsibility.
References
Bačkalov, A. 1979: Predmeti od Kosti i roga u preneolitu i neolitu
Srbije. Belgrade: Savez Arheoloških Društava Jugoslavije.
Bökönyi, S. 1962: Zur Naturgeschichte des Ures in Ungarn und das
Problem der Domestikation des Hausrindes. Acta Archaeologica
Scientarum Hungaricae 14: 175–214.
— 1969: Achaeological problems and methods of recognizing an-
imal domestication. In Ucko, P. & Dimbleby, G.W. (eds) The
Domestication and Exploitation of Plants and Animals. London:
Duckworth, 219–229.
— 1970: A new method of the determination of the number of indi-
viduals in animal bone material. American Journal of
Archaeology 74: 291–292.
— 1971: Animal Remains from Lepenski Vir. Science 167:
1702–1704.
— 1972: Aurochs (Bos primigenius Boj.) remains from the Orjeg
peat-bog between the Danube and Tisza Rivers. Cumania 1:
17–56 (Kecskemet, Hungary).
— 1974a: History of Domestic Mammals in Central and Eastern
Europe. Budapest: Akademiai Kiado.
— 1974b: The vertebrate fauna. In Gimbutas, M. (ed.) ‘Obre I and
II. Neolithic Sites in Bosnia’. Wissenschaftliche Mitteilungen des
Bosnisch-Herzegowinischen Landesmuseum, Arheologie 3, part
B: 55–154 (Sarajevo).
— 1976: The vertebrate fauna from Anza. In Gimbutas, M. (ed.)
Neolithic Macedonia. Los Angeles: University of California,
313–374.
— 1978: The vertebrate fauna of Vlasac. In Srejović, D. & Letica,
Z. (eds) Vlasac: A Mesolithic Settlement in the Iron Gates, vol. 2.
Belgrade: Serbian Academy of Sciences and Arts, 35–65.
— 1988: The Neolithic fauna of Divostin. In McPherron, A. &
Srejović, D. (eds) The Prehistoric Site of Divostin, Yugoslavia.
Pittsburgh: University of Pittsburgh Press, 419–445.
 Haskel Greenfield: Reanalysis of the vertebrate fauna from Hajdučka Vodenica
Bolomey, A. 1973: The present stage of knowledge of mammal ex-
ploitation during the Epipalaeolithic and earliest Neolithic on the
territory of Romania. In Matolcsi, J. (ed.)
Domestikationsforschung und Geschichte der Haustiere.
Budapest: Akadémiai Kiadó, 197–204.
Bonsall, C., Cook, G., Lennon, R., Harkness, D., ScottM.,
Bartosiewicz, L. & McSweeney, K. 2000: Stable isotopes, ra-
diocarbon and the Mesolithic–Neolithic transition in the Iron
Gates. Documenta Praehistorica 27, 119–132.
Cârciumaru, M. 1978: L’analyse pollinique des coprolithes de la
station archéologique de Vlasac. In Srejović, D. & Letica, Z.
(eds) Vlasac: A Mesolithic Settlement in the Iron Gates, vol. 2.
Belgrade: Serbian Academy of Sciences and Arts, 31–34.
Clason, A.T. 1979: The farmers of Gomolava in the Vinča and La
Tène periods. Vojvodjanskih Muzeja 25: 60–103 and
Palaeohistoria 21: 42–81.
— 1980: Padina and Starčevo: game, fish, and cattle. Palaeohistoria
22: 142–173.
Chaplin, R.E. 1977: Deer. Poole (Dorset): Blandford Press.
Flannery, K.V. 1973: Origins of agriculture. Annual Review of
Anthropology 2: 271–310.
Garašanin, M. 1983: The Stone Age in the Central Balkans.
Cambridge Ancient History, vol. 3, part 1. Cambridge:
Cambridge University Press, 75–135.
Gigov, A. 1972: Pollen analysis. In Srejović, D., Europe’s First
Monumental Sculpture: New Discoveries at Lepenski Vir.
London: Thames & Hudson, 185–186.
Greenfield, H.J. 1984: Analysis of the vertebrate fauna from
Hajdučka Vodenica: an Early Neolithic site in the Iron Gates,
Yugoslavia. Unpublished manuscript submitted to Borislav
Jovanović, Archaeological Institute, Belgrade.
— 1986: The Paleoeconomy of the Central Balkans (Serbia): A
Zooarchaeological Perspective on the Late Neolithic and Bronze
Age (4500–1000 BC). Oxford: British Archaeological Reports
International Series 304.
— 1988: The origins of milk and wool production in the Old World:
a zooarchaeological perspective from the central Balkans.
Current Anthropology 29: 573–593.
— 1991: Fauna from the Late Neolithic of the central Balkans: is-
sues in subsistence and land use. Journal of Field Archaeology
18: 161–186.
— 1993: Zooarchaeology, taphonomy, and the origins of food pro-
duction in the central Balkans. In Jamieson, R.W., Abonyi, S. &
Mirau, N.A. (eds) Culture and Environment: A Fragile Co-
Existence. Proceedings of the 24th Chacmool Conference.
Calgary (Alberta): University of Calgary Archaeological
Association, 111–117.
— 1994: Faunal remains from the Early Neolithic Starčevo settle-
ment at Bukovačka Česma, Starinar n.s. 43–44, 103–114.
Greenfield, H.J. & Jongsma, T.L. n.d.: Faunal remains from
Blagotin. Unpublished manuscript on file, Department of
Anthropology, University of Manitoba.
Grant, A. 1975: The use of tooth wear as a guide to the age of do-
mestic animals. In Cunliffe, B. (ed.) Excavations at Portchester
Castle, vol. 2. London: Society of Antiquaries of London,
245–279.
Jarman, M.R. 1972: European deer economies and the advent of the
Neolithic. In Higgs, E.S. (ed.) Papers in Economic Prehistory.
Cambridge: Cambridge University Press, 125–148.
Jochim, M.A. 1976: Hunter-Gatherer Subsistence and Settlement: A
Predictive Model. New York: Academic Press.
Jovanović, B. 1966a: Hajdučka Vodenica — praistorijsko nalazište.
Arheološki Pregled 8, 102–103.
— 1966b: Sculptures de la nécropole de l’âge du fer ancien à
Hajdučka Vodenica. Archaeologia Iugoslavica 7: 31–34.
— 1968a: Hajdučka Vodenica — naselje i nekropola starijeg
1
219
gvozdenog doba. Arheološki Pregled 10: 62–65.
— 1968b: Padina — naselje i starijeg neolita i starijeg gvozdenog
doba. Arheološki Pregled 11: 129–136.
— 1969a: Hajdučka Vodenica — naselje i nekropola starijeg
gvozdenog doba. Arheološki Pregled 11: 92–97.
— 1969b: Chronological frames of the Iron Gate group of the Early
Neolithic period. Archaeologia Iugoslavica 10: 23–38.
— 1971. Padina kod Gospodjinog Vira – istraživanja 1968, 1969
godine. Materijali 6: 65–71.
— 1974: Relativnohronološki odnos starijeg neolita Ðjerdapa i
Vojvodine. Materijali 10: 31–49.
— 1983: Padina, naselje mezolita i starijeg neolita. Starinar n.s.
33–34 (1982–1983): 156–167.
— 1984: Hajdučka Vodenica, praistorijska nekropola. Starinar n.s.
33–34 (1982–1983): 305–314.
Lyman, R.L. 1994: Vertebrate Taphonomy. Cambridge: Cambridge
University Press.
Manson, J. 1990: A Reanalysis of Starčevo Culture Ceramics:
Implications for Neolithic Development in the Balkans.
Unpublished Ph.D. dissertation. Southern Illinois University,
Carbondale, IL.
Matolsci, J. 1970: Historische Erforschung der Körpergrosse des
Rindes auf Grund von ungarischem Knochenmaterial. Zeitschrift
für Tierzüchtung und Züchtungsbiologie 87 (2): 89–137.
Meadow, R. 1978: Bonecode — a system of numerical coding for
faunal data from Middle Eastern sites. In Meadow, R.H. & Zeder,
M.A. (eds), Approaches to Faunal Analysis in the Middle East.
Cambridge, Massachusetts: Harvard University, Peabody
Museum Bulletin no. 2, 169–186.
Mikić, Ž. 1988: Anthropological remains from the Neolithic sites in
Serbia. In Srejović, D. (ed.) The Neolithic of Serbia. Belgrade:
The University of Belgrade, Faculty of Philosophy, Centre for
Archaeological Research, 20–23.
Mišić, V., Čolić, D. & Dinić, A. 1972: Ecological-phytocenological
investigation. In Srejović, D. (ed.), Europe’s First Monumental
Sculpture: New Discoveries at Lepenski Vir. London: Thames &
Hudson, 171–181.
Musil, R. 1984: The first known domestication of wolves in Central
Europe. In Clutton-Brock J. & Grigson, C. (eds) Animals in
Archaeology, vol. 4. Oxford: British Archaeological Reports
International Series 227, 23–26.
Nobis, G. 1954: Ur- und frühgeschichtliche Rinder Nord- und
Mittel-deutschlands. Zeitschrift fur Tierzüchtung und
Züchtungsbiologie 63, 155–194.
Payne, S. 1973: Kill-off patterns in sheep and goats: the mandibles
from Aşvan Kale. Anatolian Studies 23, 281–303.
Perkins, D. & Daly, P. 1968: A hunters’ village in Neolithic Turkey.
Scientific American 219 (11): 97–106.
Price, T.D. 1983: The European Mesolithic. American Antiquity 48:
761–778.
Prinz, B. 1987: Mesolithic Adaptations on the Lower Danube:
Vlasac and the Iron Gates Gorge. Oxford: British
Archaeological Reports International Series 330.
Radovanović, I. 1996: The Iron Gates Mesolithic. Ann Arbor:
International Monographs in Prehistory.
Semenov, S.A. 1964: Prehistoric Technology. London: Cory,
Adams & McKay.
Shipman, P., Foster, G. & Schoeninger, M. 1984. Burnt bones and
teeth: an experimental study of colour, morphology, crystal
structure and shrinkage. Journal of Archaeological Science 11:
307–325.
Srejović, D. 1972: Europe’s First Monumental Sculpture: New
Discoveries at Lepenski Vir. London: Thames & Hudson.
— 1975: The Odmut Cave — a new facet of the Mesolithic culture
of the Balkan Peninsula. Archaeologica Iugoslavica 15: 3–12.
Tringham, R. 1971: Hunters, Fishers and Farmers of Eastern
 The Iron Gates in Prehistory
Europe 6000–3000 BC. London: Hutchinson.
— 1973: The Mesolithic of southeastern Europe. In Kozłowski,
S.K. (ed.) The Mesolithic in Europe. Warsaw: University of
Warsaw Press, 551–572.
Turnbull, C.M. 1965: The Forest People: A Study of the Pygmies of
the Congo. New York: Simon and Schuster.
Vasić, R. 1983: Hajdučka vodenica, praistorijska nalazište. Starinar
n.s. 33–34 (1982–1983): 315–318.
Von Den Dreisch, A. 1976: Bone Measurement for Archaeologists.
Cambridge (MA): Peabody Museum Bulletin no. 1.
Voytek, B. & Tringham, R. 1989: Rethinking the Mesolithic: the
case of South-east Europe. In Bonsall, C. (ed.) The Mesolithic in
220
Europe. Edinburgh: John Donald, 492–499.
y’Edynak, G. 1978: Culture, diet, and dental reduction in Mesolithic
forager-fishers of Yugoslavia. Current Anthropology 19:
616–618.
y’Edynak, G. & Fleisch, S. 1983: Microevolution and biological
adaptability in the transition from food-collecting to food-
producing in the Iron Gates of Yugoslavia. Journal of Human
Evolution 12: 273–296.
Živanović, S. 1976a: Cromagnon in the Iron Gate Gorge of the
Danube. Nature 260: 518.
— 1976b: Ostaci ljudskih skeleta iz praistorijskog nalazišta na
Hajdučkoj vodenici. Starinar n.s. 26: 124–129.
 Haskel Greenfield: Reanalysis of the vertebrate fauna from Hajdučka Vodenica

Appendix 1. Sum of NISP of each taxon by element and period.

State of Period
Domestication Taxon Element Unknown Early Neolithic Mesolithic Grand Total
Unknown Sus scrofa Cranial 1 1

Mandible 1 1
Medium mammal Rib 1 1
Tibia 1 1
Large Mammal Long bone 1 1
Equus/Bos/Cervus Vertebra 1 1
Lumbar Vertebra 1 1
Innominate 1 1
Tibia 3 3
Long bone 3 4 12 19
Fragment 5 5
Bos/Cervus Thoracic Vertebra 2 1 3
Sacrum 2 2 4
Innominate 1 1
Unknown Fragment 9 2 1 12

Domestic Bos taurus Scapula 1 1

Humerus 1 1
Metacarpus 1 1 2
Femur 1 1
Metatarsus 1 2 3
Sus scrofa dom. Maxilla 1 1
Mandible 1 1
Atlas 1 1
Thoracic Vertebra 1 1
Scapula 3 3 6
Humerus 1 1 3 5
Radius 1 1 2
Ulna 1 1
Metacarpus 4 1 1
Innominate 1 1
Femur 1 1
Tibia 1 2 3
Astragalus 2 2
Calcaneus 1 4 5
Loose tooth-upper 1 1
Loose tooth-lower 1 1
Canis familiaris Maxilla 1 1
Mandible 1 2 5 8
Loose tooth-lower 1 1
N/A Homo sapiens Humerus 1 3 4
Ulna 1 1
Femur 1 1 2
Fragment 1 1
Miscellaneous 7 7
Wild Bos primigenius Cranial 1 1
Calcaneus 1 1 2
Sus scrofa fer. Cranial 3 3
Mandible 2 2
Scapula 1 1
Humerus 3 3
Ulna 2 2
Femur 1 1
Loose tooth-upper 1 1
Loose tooth-lower 1 1

221
 The Iron Gates in Prehistory

Appendix 1 (cont.). Sum of NISP of each taxon by element and period.

State of Period
Domestication Taxon Element Unknown Early Neolithic Mesolithic Grand Total

Wild Canis lupus Mandible 1 1

Cervus elaphus Cranial 19 12 89 120
Maxilla 1 1
Mandible 3 3
Axis 1 1
Lumbar Vertebra 1 1 2
Scapula 3 3 6
Humerus 5 5
Radius 2 4 6
Ulna 3 3
Metacarpus 1 2 3
Innominate 1 1
Femur 1 1
Tibia 1 2 3
Astragalus 1 1 8 10
Calcaneus 1 2 3
Metatarsus 1 5 6
Metapodia 1 1
Capreolus capreolus Cranial 1 1
Ursus arctos Radius 1 1
Calcaneus 1 1
Martes sp. Mandible 1 1
Rupicapra rupicapra Metacarpus 1 1
Pisces sp. Miscellaneous 1 1
Vertebra 2 2
Fragment 1 1

Grand total 323

222
 Haskel Greenfield: Reanalysis of the vertebrate fauna from Hajdučka Vodenica

Appendix 2. Bone measurements (mm).

Abbreviations refer to von den Driesch (1976), unless otherwise indicated below.

Measurements
Domestication
(Appendix 3)
Unit Code

Taxon

Cranium
39 40 41 Comments
2 Cervus (W) 260 Shed Antler
8 Cervus (W) 208 Shed Antler
9 Cervus (W) 277 Shed Antler
10 Cervus (W) 242 315 Shed Antler
21 Cervus (W) 220 155 205.5 Antler attached to cranium
Cervus (W) 243 Shed Antler

Cranium
16 18a 19 20
1 Canis (D) 21 9 19.8
1 Canis (D) 20.1 21.1 9 19.8

Cranium: Upper M3
30 31
41 Sus (D) 40.8 18.5

Mandible
7 8 9 10 11 12 13L 13B 14 15L 15b 17 20 HMP1 HMM1 CB
2 Canis (D) 73.4 57.7 63.6 33.1 36.7 32.4 20.5 10.6 18.5 22
2 Canis (D) 23.1 9.3
6 Canis (D) 63 69.1 64.4 33.7 36.4 31.7 21 20 18.1 23.2
11 Canis (D) 77.4 73.6 68.4 35.8 39 33 22.4 9.5 20.5 10 7.2 12 19.8
14 Canis (D) 80 75 71 37.5 39.8 35 23.6 9.1 22.7 12 23.8
15 Canis (D?) 23.2 8.6 26.8**
35 Canis (W) 28.4 11.3 21.3
35 Canis (D) 32.3 21.7 20.8 20.6
35 Canis (D) 26.3

Mandible
10L 10B
35 Sus (W) 41.3 18.5
35 Sus (W) 42.1 20.3
11 Cervus 32.8 15.3

Axis
BFcr SBV
12 Cervus 74 53.1

Scapula
SLC GLP LG BG
11 Sus (W) 32.4 49.3 37 34.8
15 Sus (W) 32.5 46 35 32.5
16 Sus (W) 35.3 53.5 41.7 34
14 Cervus (W) 44
15 Cervus (W) 33.6 42.3 38.7
16 Cervus (W) 37 59.5 44.9 40
26 Cervus (W) 40.5 63.5 49.3 44.8

223
 The Iron Gates in Prehistory

Appendix 2 (cont.). Bone measurements (mm).

Abbreviations refer to von den Driesch 1976, unless otherwise indicated below.

Measurements
Domestication
(Appendix 3)
Unit Code

Taxon

Humerus
Bd Bt Dd
1 Sus (W) 47.4 36.6 46.5
12 Sus (W) 47.7 40.8 51.5
23 Sus (W) 52.9 39.5 54.3
33 Sus (W) 56.8 47.3 53.8
40 Sus (W) 53.5 40.7 54.4
14 Cervus (W) 57.2 52.5 57.5
22 Cervus (W) 48.7
23 Cervus (W) 68.4
24 Cervus (W) 54.6 51.5

Radius
GL BP
BFp SD Bd BFd DP Dd Z
23 Sus (W)* 38.5 27.5
*Wild according to Bökönyi (1974b, 1976, 1978), but domestic according to Clason (1980).

1 Cervus (W) 48.9 47.6 34.5

16 Cervus (W) 55.6 53.4 29.8
23 Cervus (W) 300 58.6 53.5 33 52 49.5 32.1 40 18.5

Ulna
DPA SDO BPC
14 Sus (W) 51.4 39.4 27
26 Cervus (W) 35
23 Cervus (W) 31.2

Metacarpus
Bp Dp CD DD Bd Dd
15 Bos (D) 64 33.3
11 Cervus (W) 40.2 29.7
23 Cervus (W) 32.4 23.6 19
35 Rupicapra(W) 13.5 34 34.2

Pelvis
LA LAR SH SB
33 Sus (W) 45 38.5 36.1 17.8

Femur
DC
16 Cervus (W) 38.8

Tibia
Bd Dd
1 Sus (W) 39.9 33.7
23 Sus (D)* 32 31
*Domesticated according to Bökönyi (1974b, 1976, 1978), but transitional according to Clason (1979: 82).
16 Cervus (W) 51.3 43.6

224
 Haskel Greenfield: Reanalysis of the vertebrate fauna from Hajdučka Vodenica

Appendix 2 (cont.). Bone measurements (mm).

Abbreviations refer to von den Driesch 1976, unless otherwise indicated below.

Measurements
Domestication
(Appendix 3)
Unit Code

Taxon

Astragalus
GL1 Glm D1 Dm Bd
1 Sus (W) 49.5 46.6
23 Sus (W) 53.4 47.9
11 Cervus (W) 55.8 54.9 31.2 32.6 34.8
12 Cervus (W) 53.3 49.3 29.2 30 35.9
15 Cervus (W) 57.4 53.7 31.3 30.7 35.5
23 Cervus (W) 59.7 32.9 34.1
24 Cervus (W) 55.8 52.5 31.2 32.6 34.8
24 Cervus (W) 57.1 30.8 37.3
35 Cervus (W) 54.5 51.5 30.8 31.8 34.4
40 Cervus (W) 55.9 52.7 30 30.8 35.3
1 Cervus (W) 60.2 56.6 32.5 33.5 39.1

Calcaneus
GL GB
11 Sus (W) 99 27
26 Sus (W) 104.1 30.4
14 Bos (W) 152 55.5
24 Cervus (W) 120.4 38.3
15 Ursus (W) 73.1 46.7

Metatarsus
GL Bp Dp CD DD Bd Dd
30 Bos (D) 218 50.9 48.5 28.2 20 58 31.7
28 Cervus (W) 40.1 22.7 46.3 29.7

Key to non-von den Driesch (1976) measurements and other abbreviations:

Dd greatest depth of distal end
DP greatest depth of proximal end
HMP1 height of mandible before PM1
HMM1 height of mandible before M1
CB top of mandibular condyle to basal point of coronion
Z smallest depth of diaphysis
W wild
D domestic
** suspect measurement

225
 Appendix 3. List of excavation units with faunal remains analyzed in 1982.
Nivo 1 Early Iron Age (EIA)
Nivo 2 Early Neolithic (EN)
Nivo 3–4 Mesolithic (M)

Unit Period Possible Trench Context Level Profile Point Elevation Inventory # Date Bag
Code period (sonda) (nivo) (tačke) (kota) (IB) (kesa)

1 M Na nivou osnove 3 3 uz tačke 7–19 03-Sep-68 27

2 EN 17d 2 33–41 1969 127
3 ? M 15b tačka 8 dubina 0.4 133 01-Aug-67 42
4 M 17c Na kamenoj konstrukciji 3 42
5 EN 15f Nivo konstrukcije 1 1 11–21 10-Jul-69 6
6 M 15d Na nivou peska 03-Sep-68
7 M 15a Nalaz iz donjeg dela centralne 153 21-Aug-67
žrtvene konstrukcije
8 ? 15a B 137 01-Jul-67
9 ? 1969 69
10 ? 15a Profile made in Danube bank, away from B 01-Jul-67 48
main site near to necropolis
11 M 15c Nivo 3 konstrukcije 3 uz tačke 7–19 29-Aug-68 16
12 ? M 15c 18, 19 1968 15
13 EN 15c 9 8
14 M 15c Ispod nivoa 3 kamene konstrukcije 3 između 5–9 and 19–20 03-Sep-68 19
15 EN 15c 8 7
16 EN 15c 7 6

226
17 ? EN 15c Iz profila sonde 06-Jul-69 1
18 M 15c Kamena konstrukcija ispod nivoa 3 3 između 5–9 and 19–20 24
19 M 15b or 16d Nivo grobnih konstrukcije 1 15–16 29-Aug-68 11
površine spaljivanja
20 M 16d Ispod nivoa 2, horizon 4 4 154–156 29-Jul-69 61
The Iron Gates in Prehistory

21 M 16d Ispod kamene konstrukcije 4 12-Jul-69 9

22 M 16d 4 33–41 84
23 M 16d 33–41 02-Aug-69 89
24 M 16b Gornji nivo sloja nekropole 29-Aug-68 13
25 M 16c Ispod kamene konstrukcije 3 01-Aug-69 81
26 M 16d Ispod kamene konstrukcije 2 Horizont 4 01-Aug-69 86
27 M 16d 4 41–34 02-Aug-69 85
28 M 16c 3 13–33 15-Jul-69 12
29 EN 17c 2 13–33 19-Jul-69 25
30 EN 17b 31 14
31 M 17d 3 1969 90
32 M 18 2.495–2.186 1969 34
33 EN 15c 11 1968 10
34 M 16d 4 24–26 14-Jul-69 10
35 M 16–17c Ispod konstrukcije 3 146–148 57
36 M 17d Ispod kamene konstrukcije 41–34 1969 83
37 M 16c 4 93
38 M 16b Na nivou sahranjivanja 1969 82
39 M 16c Nivo konstrukcije 4 4 88
40 M 16c Ispod kamene konstrukcije 3 79
41 M 19 Kota izmedju tačaka 3 and 4 23-Jul-69 49
 Velesnica and the Lepenski Vir culture

Rastko Vasić

Abstract: Velesnica is a multi-period site situated on the right bank of the Danube, some 10 km south of Kladovo. It was settled during
the Early Neolithic (Starčevo), Eneolithic (Coţofeni–Kostolac), Middle Bronze Age (Žuto brdo–Gîrla Mare), Late Bronze/Early Iron Age
(Gava, Ostrov, Basarabi), and the Late Iron Age (‘Dacian’). There were also late Roman and Medieval habitations. Most important is the
Starčevo settlement with a cultural layer over one metre deep, which contained a large quantity of pottery as well as numerous bone and
stone objects. Three graves were found in the Starčevo settlement, two with single skeletons and one with seven skeletons. Two features in
particular distinguish the Starčevo settlement and connect it with the Lepenski Vir culture: two ‘altars’ of Lepenski Vir type without orna-
ments, and the lowest skeleton in grave 2 which is in a sitting position. Although a complete analysis of the excavated material has not been
carried out, the early Starčevo horizon at Velesnica can be paralleled with Padina B2, while the later horizon corresponds chronologically
to Padina B3 and other Starčevo sites in the lower Iron Gates.
Key words: Velesnica, Lepenski Vir, Mesolithic, Neolithic, Starčevo culture, settlement

Introduction plete picture was obtained. In 1981 seven larger cuttings

The village of Velesnica is situated in the Ključ area 10km to
the south of Kladovo on a small terrace on the right bank of
the Danube, downstream from the village Milutinovac where
the river takes a turn towards the south. According to the
consultant geologist, Slobodan Knežević, the river terrace at
Velesnica (Fig. 1) consists of alluvial deposits dating to the
end of the Pleistocene and beginning of the Holocene. Above
these is a layer of slopewash in which the archaeological re-
mains (artefacts and animal bones) were found. Most of the
archaeological material was found along the riverbank in the
southern part of the village (the locality known as Donja
strana — ‘lower side’) where the cultural layer was more
than 4 m deep in places. The remains of settlements of several
prehistoric periods were found — Early Neolithic (Starčevo
culture), Eneolithic (Coţofeni group), Middle and Late
Bronze Age as well as Early Iron Age (Žuto Brdo–Gîrla
Mare, Gava, Ostrov, Basarabi groups), then Late Iron Age
(Dacian), Roman and Medieval settlements.
History of research
Velesnica was already noted as an archaeological site at the
beginning of the twentieth century. Small-scale ‘excavations’
were carried out and on the basis of the pottery found the
locality was assigned to the Žuto brdo group (M. Vasić 1910).
A reconnaissance in 1971 confirmed the significance of the
site and the richness of the archaeological remains (Vasić &
Janković 1971: 110). Systematic excavations were carried
out between 1980 and 1982 and in 1984, during rescue work
on the Danube bank necessitated by the building of a hydro-
electric plant at Kusjak. At that time an area of c. 920 m2 was
opened (Fig. 2). During the first field season (1980) the work
did not last very long. Five small cuttings were opened,
which produced mainly Roman and Medieval material (Vasić
et al. 1984). The second and third campaigns were consider-
ably longer, more than two months. As a result, a more com-
(6–12) were opened. Of these the results from cuttings 7, 8,
9 and 11 are exceptionally important. Four blocks, A–D and
cutting 13 were opened in 1982, but because of the weather
and financial limitations only block A was fully investigated;
it also provided a wealth of material (Vasić 1986; Ercegović-
Pavlović & Minić 1986; cf. Srejović 1988: 67; Radovanović
1996: 345–347). During a relatively short campaign in 1984
several smaller cuttings (14–25) were opened but gave only
poor results. These are not discussed further in this paper.
The excavations at Velesnica confirmed that it was an ex-
tensive and significant site that had been inhabited at various
times between the Early Neolithic and the late Middle Ages.
Some periods were well represented, others were less so.
However, it was not possible to form a detailed picture of
cultural development through the different periods because
of extensive mixing of the layers owing to flooding by the
Danube and the digging of numerous pits mainly during the
Late Iron Age (Dacian) and Medieval periods, which had cut
through the Early Neolithic horizon. Also, erosion by the
Danube had removed part of the settlement together with a
large quantity of archaeological material. This was evident
from the trenches dug in 1981 adjacent to the river. These
provided much more varied and interesting material than
those dug in 1982, which were situated 10 m away from the
riverbank.
‘Mesolithic’ settlement
The existence of a Mesolithic layer was not confirmed, al-
though animal bones were found below the Starčevo horizon,
and many of these could be the remains of a Mesolithic oc-
cupation. In block A a large group of animal bones including
one that was worked (Fig. 22, no. 16; Vasić 1986: fig. 22, no.
1) were found in a depression in the subsoil, which may be a
natural feature or a deliberately excavated pit — it could even
be a primitive dwelling construction. The animal bones
comprise mainly red deer (Cervus elaphus) then aurochs

227
 The Iron Gates in Prehistory
Figure 1. The Danube terrace at Velesnica, photographed
in 1980.
(Bos primigenius) and wild boar (Sus scrofa ferus) (Mikić
1999). Otherwise, this layer of compact dark-brown earth
more than a metre deep lying beneath the Starčevo settlement
horizon contained only rare finds of bone and flint tools, oc-
casional fragments of pottery, ‘gravel’1 and animal bones
(Fig. 3, layer 1). It is interesting to note that the lowermost
cultural deposits in cutting 8, which lies to the west of block
A, also provided a large number of animal bones with very
few fragments of pottery and rare tools of bone or stone. The
small area uncovered provided no evidence of pits. The
lowermost cultural deposit in this part of the site may be as-
signed tentatively to the Mesolithic. If this interpretation is
correct, the hunters may be connected with the Mesolithic
population that has been confirmed in the area to the south of
Velesnica, at the sites of Kula and Ostrovul Mare
(Radovanović 1996: 313).
Early Neolithic settlement
The Early Neolithic is the best known of the cultural horizons
at Velesnica and has produced the greatest quantity of ar-
chaeological material. Remains of buildings extend from the
south part of cutting 7 to the north part of cutting 11.
Therefore, it may be supposed that the settlement extended
along the riverbank for some 80 metres. Its width was at least
40 m considering that in cutting 18, which lay 20 m northw-
est of block A, a Starčevo layer was uncovered above the
subsoil. It should not be forgotten that a large part of the site
Figure 2. Velesnica: central part of the excavated area.
228
had been washed away by the Danube, and the settlement
could have been wider originally. The thickness of the
Starčevo layer was not uniform. In cutting 8 it was 1.5 m, in
block A more than 1 m, in cuttings 7 and 9 nearly 1 m, while
in cutting 11 it was rather less. From this it may be concluded
that the settlement existed over a long period of time, but was
not always the same size.
The Starčevo layer was investigated most thoroughly in
cutting 8 and block A. In cutting 8 the cultural deposits had
a total thickness of about three metres, from 37.80 to 34.90 m
above sea level (Fig. 4). Starčevo material occurs between
c. 36.40 and 35.00 metres. Two occupation horizons are ap-
parent, an earlier horizon at 35.80 m, and a later horizon at
36.20–36.10 metres. Two hearths of packed earth with stone
bases of round irregular shape, about a metre in diameter, are
preserved in the earlier horizon in the northwest part of the
cutting (Fig. 5, B & C). A large concentration of stones,
‘gravel’ and pottery was uncovered in the southwest part
(Fig. 5, D). It is possible that the hearth adjacent to the west
profile (Fig. 5, C) was part of a larger floor considering its
shape, construction and proximity to the other hearth. In any
event it is further evidence of the former existence of houses
at this level. Remains of a hearth were also uncovered in the
later horizon, together with remains of a floor of packed earth
and a concentration of stones, ‘gravel’, daub and pottery, ad-
jacent to the west profile. Grave 1 was dug into this horizon
(Fig. 6). Remains of hearths and smaller stone constructions
of irregular shape appear between the two horizons, which
indicates that there was renovation of the buildings and con-
tinuity in the settlement throughout the entire period. On the
other hand, fragments of pottery, stone and bone tools,
‘gravel’ and stones found in the cultural deposits below the
earlier Starčevo horizon point to an even earlier phase of
Starčevo occupation, possibly associated with the first
dwellings to be constructed at Velesnica.
In block A, the cultural deposits had a total thickness of
4.5m, from 39.00 to 34.30 metres. Starčevo material is found
from 36.70 to 35.50 metres. At 36.50–36.40 m a concentra-
tion of stones, pebbles, ‘gravel’, shells and pottery was also
confirmed over the entire surface, which connects with the
later horizon in cutting 8 (Fig. 7). The height difference
between the two areas can be explained by the fall of the ter-
rain towards the river. A large pithos about 70 cm in diameter,
with a plastic band around the neck, was found in situ in this
horizon (Vasić 1986: fig. 20). Grave 3 is also connected with
this level.
 Rastko Vasić: Velesnica and the Lepenski Vir culture

Figure 3. Velesnica, block A, east profile: animal bones.

Figure 4. Velesnica, cutting 8, west profile.

Starčevo material in cutting 7 occurs between c. 36.50 and
35.30 m, while the concentration of stones, daub and pottery
at 36.40–36.30 m is connected with the later occupation ho-
rizon in cutting 8 and block A (Fig. 8). Grave 2, with seven
skeletons, dug at 35.30–34.80 m, is possibly connected with
the earliest dwelling activity in cutting 8.
In cutting 9, Starčevo pottery occurs between 36.50 and
35.70 m, and the existence of a settlement horizon can be es-
tablished with certainty at 36.30–36.20 m on account of the
concentration of stones and pottery as well as the remains of a
hearth or floor surface in the north profile of the cutting. A
large quantity of archaeological material was found beneath
this horizon but there were no traces of any construction.
In cutting 11 Starčevo pottery occurs at 36.30–36.20 m
while the remains of a dwelling horizon, represented by
traces of daub in the northwest part of the cutting, were con-
firmed at 35.90–35.85 metres. On the basis of this and other
elements it can be suggested that the Starčevo settlement
confirmed in cuttings 7 and 8 was relatively small in the be-
ginning and over time expanded to an area of 80 x 40 metres.
Architecture
Little can be said about the architectural elements of the
Starčevo settlement. The features confirmed are circular
hearths of packed earth with a substructure of stone, which
appear in both older and later dwelling horizons. Nothing
more precise can be said of the concentrations of stones,
pebbles, daub, shells and pottery that are particularly notice-
able in the west profile of cutting 8 and in block A, but also
present to a certain extent in cuttings 7, 9 and 11, and which
appear to belong to the later phase of the settlement. It is
possible that they served as a foundation for a floor with a
thin layer of mud or daub. Given that these features are
widespread in the settlement, it may be supposed that the

229
 The Iron Gates in Prehistory

Figure 5. Velesnica, cutting 8: earlier Starčevo settlement horizon.

Figure 6. Velesnica, cutting 8: later Starčevo settlement horizon.

buildings were above ground and rectangular in plan. The
abundant remains of soot indicate that fires were frequent. It
is possible that flooding of the site due to periodic rises in the
level of the Danube was a reason for the frequent renewal of
the settlement.
Two large stones with hemispherical hollows on the upper
surface, found in cutting 8, can be considered as part of the
architecture (Fig. 9). The first, 45 cm wide and 38 cm long
with a hollow 12 cm in diameter, was found adjacent to the
west profile of the cutting above grave 1, at a level of 36.36 m
(Fig. 6). From the position of the stone in relation to the
grave, it is not clear if the two are connected. The second,
43 cm long and 23 cm wide with a hollow 10 cm in diameter,
was uncovered in the north part of the cutting at 35.87 m, one
metre to the east of a hearth belonging to the earlier Starčevo
horizon (Fig. 5, B). On the basis of parallels at sites in the
upper Iron Gates — Lepenski Vir, Padina and Hajdučka
Vodenica (Jovanović 1969: 31–34, 1987; Srejović 1969: 131;
Srejović & Babović 1983: 160–180) — these objects can be
interpreted as ‘altars’, but because of their locations they

230
 Rastko Vasić: Velesnica and the Lepenski Vir culture

Figure 7. Velesnica, block A: later Starčevo settlement horizon, disturbed by later pits (1–3 = Medieval; 4–5 = Late Bronze
Age; 6 = Dacian).

Figure 9. Velesnica, cutting 8: stone ‘altars’. The one on

the left was found adjacent to the west side of the trench
above grave 1 (cf. Fig. 5). The one on the right was found in
the north part of the trench about a metre from a hearth be-
longing to the earlier Starčevo horizon (cf. Fig. 4).

Figure 8. Velesnica, cutting 7, west profile.

231
 The Iron Gates in Prehistory
Figure 10. Velesnica, cutting 8: grave 1.
cannot be connected with certainty to any particular con-
struction. Thus, their purpose may have been different at
Velesnica. It is interesting that near to the first stone, at the
same level, a fragment of a stone slab with a hollow was
Figure 11. Velesnica, cutting 7, burial 2A.
232
found — part of a millstone (Fig. 19, no. 5). A working plate
was uncovered in the vicinity, which indicates activity of a
domestic nature in this area. On the other hand, two working
plates were found in block A below the later horizon of the
Starčevo settlement (Fig. 19, no. 6). In view of the fact that
the first ‘altar’ belongs to the later Starčevo horizon and the
second can be assigned to the earlier horizon, these objects,
regardless of their purpose, occur at Velesnica over a long
period.
Burials
Three graves were uncovered within the Starčevo settlement
at Velesnica. Grave 1, in cutting 8, contained one skeleton.
Grave 2, in cutting 7, had five complete and parts of two
skeletons. Grave 3, in block A, contained one skeleton.
Grave 1 contained the flexed skeleton of a child, orientated
north–south. It was found adjacent to the west profile of cut-
ting 8 at 36.10 m, underneath the first ‘altar’ described above
(Fig. 10). The feet and spine were discovered in situ, but the
head was missing. Associated with the skeleton, and also in
situ, was a small ball-shaped vessel with four small vertically
perforated handles on the belly. This was probably a grave
offering (Fig. 16, no. 1). The grave belongs to the later
Starčevo horizon.
Grave 2 was a pit containing five complete skeletons and
parts of perhaps two other skeletons. All were in the flexed
position, placed one above the other, so that the bones were
mixed and their separation and identification as individuals
was difficult. The pit was more or less circular in shape, with
a diameter of c. 1.20 metres.
Skeleton 2A, a female, was found adjacent to the north
edge of the pit, orientated ENE–WSW. The skull rested on
the right shoulder while the leg bones lay across the rib cage
so that the femur reached to the left shoulder. The skull was
found at 35.10 m, while the remaining parts of the skeleton
were found at the same level or slightly lower (Fig. 11).
Skeleton 2B, a female, was found in the middle of the pit,
orientated south–north. The skeleton was lying on the left
side. The arm was flexed at the elbow and the upper arm was
found at the level of the shoulder. The legs were in a flexed
position. The poorly-preserved pelvis and legs lay across
skeleton 2D, while the skull was found on the legs of skeleton
2C. The skull was found at 35.30 m and the remaining parts
of the skeleton at 35.20–35.15 metres.
Skeleton 2C, a child, was found adjacent to the south side
of the pit, orientated east–west. It lay on the right side in a
flexed position. The right arm was slightly bent at the elbow
and some of the hand bones were preserved. Only the upper
arm of the left side was found. The skull was found at
35.13 m (Fig. 12).
Skeleton 2D, a female, was found in the middle of the pit,
orientated WSW–ENE. The skull was turned face down. The
skeleton lay on the left side. The legs were flexed towards the
north west, while the left femur was folded under the pelvis.
Both arms, facing south east, were flexed at the elbow and lay
under the rib cage to its right side. The skull was found at
35.09 metres.
Skeleton 2E was found in the northeast part of the pit, un-
der the pelvis and legs of skeleton 2A. Only the remains of
the spine, ribs and upper arm were preserved.
 Rastko Vasić: Velesnica and the Lepenski Vir culture
Figure 12 Velesnica, cutting 7, burials 2C–F.
Skeleton 2F was found between 2D and 2A. Only the pel-
vis and femora were preserved.
Skeleton 2G, a child, was found in the middle of the pit at
the lowest level, 34.84 metres. It was orientated more or less
west–east, but with a slight deviation towards the south east.
It rested on its back with the face turned toward the south.
The right arm was flexed at the elbow with the hand under the
head, while the left arm was folded across the chest so that
the hand rested on the right shoulder. The legs were flexed in
the so-called ‘à la turque’ position (Fig. 13).
In his analysis of the anthropological material, S. Živan-
ović kept to this division of bones in relation to the skeletons
(Živanović 1986). However, a new analysis by Mirjana
Roksandic (this volume) has shown that the bones of skeleton
2A and 2E belong to the same individual, marked as 2A, and
the bones of skeletons 2B and 2F likewise belong to one in-
dividual, recorded as 2B. Among the bones she discovered
pieces that could have belonged to other skeletons, and these
she marked as 2A-1 and 2G-1.
In the burial pit, especially its upper part, were found
fragments of several Starčevo vessels (Fig. 15), lumps of red
fired clay, shells and some animal bones, which may indicate
certain burial rituals. These finds, some regularity in the pos-
itioning of the skeletons, and the fact that there are no finds
underneath skeleton 2G, suggest that this was a grave pit, and
not a pit for waste that was used secondarily for interment —
as is characteristic of Starčevo sites. The grave was dug at
c. 35.40 m, i.e. beneath the earlier dwelling horizon in cutting
233
8, so that it can be considered older than it and placed in the
initial phase of the Starčevo settlement in this locality (see
above).
Grave 3 lay in the middle of block A at 36.50 metres. It
contained the remains of a female skeleton in a flexed posi-
tion, orientated north–south. The skeleton lay on the right
side; the arm and leg bones are preserved, together with parts
of the pelvis, the lower jaw and teeth (Fig. 14). Even though
it was disturbed, it can be assigned to the Starčevo settlement
on the basis of fragments of pottery around it. There were no
grave offerings. In character, it was similar to grave 1 in cut-
ting 8. This grave probably belonged to the later Starčevo
horizon.
Pottery
Among the archaeological finds, there is a large amount of
pottery that has parallels in the Starčevo sites of the lower and
upper Ðerdap, as well as further afield — in Serbia, Romania
and elsewhere. Coarse pottery is the most numerous. Fine
monochrome red pottery is also common, but forms a smaller
proportion of the ceramic assemblage. No painted Starčevo
pottery was found at Velesnica.
Shapes are simple and repeated in both fine and coarse
pottery. They comprise deep and shallow dishes, and various
forms of bowls and jars. An interesting find is the miniature
vessel in the shape of a barrel from cutting 11, which has all
the characteristics of Starčevo pottery (Fig. 16, no. 2; Vasić
1986: fig. 22, 2; cf. Stalio 1986: fig. 23). Fragments of large
 The Iron Gates in Prehistory
Figure 13. Velesnica, cutting 7, burial 2G.
Figure 14. Velesnica, block A, burial 3.
vessels with four protrusions on the rim are found in block A
and belong probably to the later Starčevo horizon (Fig. 17,
no. 1; cf. Srejović 1969: fig. 89; Jovanović 1987: Abb. 11).
The vessels frequently have vertical, perforated handles at-
tached to the belly, while horizontal, perforated handles and
lug handles are less frequently seen. Globular and hemi-
spherical vessels often have a circular base, but occasionally
they are 4- or 5-leaved (Fig. 16, nos 11–12).
The ornamentation on coarse pottery consists of impres-
sions made with fingers, nails, or pieces of wood on the damp
surface of the vessel (impresso), and done more-or-less sys-
tematically. Engraving appears on a large number of vessels,
most frequently in the form of groups of parallel lines that
intersect. Barbotine ornamentation is rare. The use of plastic
bands decorated with incisions made by fingernails or pieces
of wood was more frequent. Horizontal plastic bands appear
on the neck and rim relatively frequently, particularly in the
case of larger vessels. Plastic bands, variously combined, are
also a frequent decoration on the bellies of the pots (Fig. 17,
nos 6–7).
There were also several legs belonging to sacrificial ves-
sels (‘altars’) — small tables with four legs supporting a
round recipient — but no intact examples were found
(Fig. 16, nos 8–10; Fig. 17, no. 3). The legs were varied: they
usually have lugs on the upper part and there could have been
decorative and plastic bands as well as other relief features.
Fragments of portable ceramic hearths were also found (these
234
cm
cm
Figure 15. Velesnica, cutting 7, pottery from burial 2.
were spherical with flat bottoms), as well as fragments of
small tables (Fig. 17, nos 4–6).
No sculptures were uncovered at Velesnica, with the ex-
ception of a fragmented bovine figure from cutting 11
(Fig. 16, no. 3; Vasić 1986: fig.23, 8; cf. Stanković 1986a:
fig.4, nos 5 & 7; Stanković 1986b: fig. 3, no. 9; Srejović
1988: 77). Some ceramic amulets in the shape of stylized
bulls’ heads and horns, found in both the older and later
Starčevo horizons, should also be mentioned (Fig. 16, nos
4–7). They are of unusual shape and have parallels in many
of the sites in the Ðerdap as well as in Serbia and Vojvodina
(Srejović 1969: fig. 43a, nos 1–5; Stanković 1986a: fig.4, nos
8–11; Stanković 1986b: fig. 3, nos 10–12; Srejović 1988: 71,
no. 77).
No significant differences can be identified between the
pottery shapes of the upper and lower Starčevo horizons, al-
though a more detailed study may yet reveal some differ-
ences, with the predominance of certain shapes and orna-
ments in the earlier or later horizons. Coarse pottery
dominates in all levels but there is a larger percentage of fine
wares in the lower horizon compared to the upper horizon.
Chipped stone tools
The Starčevo layer produced a large collection of stone and
bone tools, although the quantity is considerably less than the
pottery. The raw materials for the manufacture of chipped
stone tools at Velesnica were quartz and flint. An analysis of
 Rastko Vasić: Velesnica and the Lepenski Vir culture

cm

Figure 16. Velesnica: Starčevo pottery. 1, vessel from burial 1; 2, miniature vessel; 3, bovine figure: 4–7, amulets; 8–10,
legs of ‘altars’; 11–12, vessel bases.

the material from cutting 8 and block A was carried out by
Josip Šarić (1987). Of the 524 pieces examined, 420
(80.15%) were quartz. These included pieces of quartz that
may not have been used as tools. Among the 104 pieces of
flint, 56 are unretouched blades and flakes, chips and cores,
while 48 are retouched pieces. Among these, retouched
blades (35 pieces) dominate (Fig. 18, nos 1–5), with not a
single retouched flake. Compared to the two other Starčevo
localities in the lower Ðerdap analyzed by Šarić (1987) —
Ušće Kameničkog potoka (at the mouth of the Kamenica
stream) and Knjepište — Velesnica is distinguished by a high
frequency of quartz tools, which indicates a strong
Mesolithic tradition, and by a preponderance of retouched
blades and a lack of retouched flakes.

235
 The Iron Gates in Prehistory

cm

cm

Figure 17. Velesnica: Starčevo pottery. 1, fragments of a vessel with four protrusions on the rim; 2, fragmented ceramic
plate; 3 fragment of an ‘altar’; 4–5, fragments of portable hearths; 6–7, plastic ornamentation.

Coarse and ground stone tools
These were analyzed by Dragana Antonović (Antonović
1998: 124–132), but only from the lower Starčevo horizon in
block A. However, it is possible that some of her conclusions
apply to the entire Starčevo layer. The coarse and ground
stone tools from Velesnica are made from a variety of mater-
ials, mainly sandstone, magmatic rocks and granite gneiss.
Ground edge tools are relatively rare, while tools made by
minimal working of waterworn stones are much more nu-
merous (Fig. 19, nos 1–4). So-called ‘fishing weights’ or
amulet-pendants are common. These are small, flat stones
‘pinched’ at the narrower end, which involved a small
amount of working; some fortuitously-shaped pieces were
used without any preparation (Fig. 20, nos 1–9; Antonović
1998: 72–73). Bigger and heavier ‘weights’ or hammers
(punches) were also used with no or minimal workmanship
(Fig. 21, nos 1–5; Antonović 1998: 69–72). Both forms were
found in the earlier and later Starčevo horizons, although the
pieces from the earlier horizon appear to be more primitively
worked. These two forms obviously served different pur-
poses, although among the older material a small ‘weight’ in
the shape of a hammer was found, which points to an original

236
 Rastko Vasić: Velesnica and the Lepenski Vir culture

cm cm

cm

Figure 18. Velesnica: retouched blades. 1–5, Starčevo; Figure 19. Velesnica: stone artefacts. 1–4, hammers and
6, Eneolithic. axes of polished stone; 5 , millstone; 6, working plate.

cm

cm

Figure 20. Velesnica: ‘weights’ for fishing nets. 1–7, later Figure 21. Velesnica: hammers (1–4, later Starčevo hori-
Starčevo horizon; 8–9, earlier Starčevo horizon. zon; 5, earlier Starčevo horizon).

237
 The Iron Gates in Prehistory
connection (Fig. 20, no. 9). Many other different shapes were
found, but in smaller numbers.
Bone tools
Awls are the most numerous of the bone tools from the
Starčevo layer. There were fewer needles, projectiles and
spatulas (Fig. 22, nos 1–19). No hooks or harpoon heads were
discovered at Velesnica. There were no obvious chronolo-
gical differences in the forms of the bone tools, although
certain types were found in larger numbers in the later hori-
zon (e.g. awls–spatulas in cutting 8).
All these elements indicate that Velesnica was probably a
large rural settlement where different activities took place —
working of stone and bone, and pottery making.
Archaeobotanical analyses were not undertaken. Analysis of
the animal bones from the lower levels of block A (Mikić
1999) shows that wild animals dominated over domestic
(95.62%), which indicates that hunting was an important as-
pect of the economy in the earlier Starčevo phase. In spite of
the absence of harpoons and hooks, if the numerous ‘weights’
in the form of very stylized human figures served as weights
for fishing nets, then fishing was also an important activity.
cm
Figure 22. Velesnica: bone tools. 1–4, projectiles; 5–7,
awls; 8–10, awls-Spatulas; 11–12, spatulas; 13–15, needles;
16, tranchet; 17, pendant; 18–19, awls or projectiles from
lower levels in block A; 20, point of horn found together with
animal bones in the Mesolithic layer in block A.
238
External relations of the Starčevo occupation
There are many features that connect the Starčevo settlement
at Velesnica with the Lepenski Vir culture.
Skeleton 2G, from grave 2 in cutting 7, which was the first
burial placed in the grave pit, lay in the ‘à la turque’ position
(Fig. 13). This position is known from several localities in the
Ðerdap — Padina, Lepenski Vir, Vlasac, Ostrovul Corbului
and Kula — and chronologically it is placed in the time range
from the later Mesolithic to the Early Neolithic
(Radovanović 1996: 160–224). Skeleton 2G at Velesnica
clearly belongs to this tradition and can be assigned to the
same period. The fact that the lowermost skeleton in grave 2
is placed in the ‘à la turque’ position, while those above it are
crouched may have some symbolic significance. It belongs to
the earliest phase of occupation at Velesnica and, perhaps, is
connected with the foundation of the settlement. This is an
important feature that links Velesnica with the Early
Neolithic of the upper Iron Gates.
The two stone ‘altars’ from cutting 8 are a second link in
this direction (Fig. 9). The ‘altar’ from the earlier Starčevo
horizon was definitely found in situ, while the second ‘altar’
uncovered in the later Starčevo horizon (above grave 1) was
probably not in its original position. However, it is possible
that it belonged originally to this or a slightly lower level. In
any case, it should be dated later than the first ‘altar’. Similar
objects were found at Lepenski Vir, Padina, Hajdučka
Vodenica and Cuina Turcului (Radovanović 1996: 138–159)
and are dated to the Mesolithic and Early Neolithic. In the
lower Ðerdap such objects have been discovered so far only
at Velesnica and Schela Cladovei (cf. Boroneanţ et al. 1999).
Regardless of their purpose, whether cult or practical, un-
doubtedly they represent a specific tradition of continuing
form that links Velesnica with the Early Neolithic of the up-
per Ðerdap.
This connection is manifest also in other aspects of the
material culture. Parallels in pottery exist but cannot be
drawn precisely before a detailed analysis of the ceramic as-
semblages from Velesnica, Padina and Lepenski Vir has been
undertaken. The characteristics of the chipped stone as-
semblage from Velesnica, viz. the large percentage of quartz
in relation to flint and the dominance of retouched blades, is
indicative of a cultural connection, but is not necessarily
conclusive. It could also be explained in terms of incomplete
collection of flint and quartz artefacts during excavations at
Velesnica and the sites used for comparison. It is character-
istic of all the Ðerdap sites, however, that among the ground
stone tools there is a complete dominance of tools without
blades, i.e. tools made on waterworn stones that show min-
imal working.
All these elements demonstrate a connection between
Velesnica and the Early Neolithic of the upper Ðerdap, with
the Starčevo culture appearing at Velesnica perhaps at a time
equivalent to Padina phase B2 (Jovanović 1987) or
Radovanović’s phase 5 of the Ðerdap Mesolithic
(Radovanović 1996: 285–290). The later phase of the
Starčevo settlement at Velesnica probably coincided with
Radovanović’s Iron Gates Mesolithic phase 6 (Radovanović
1996: 285), equivalent to Padina B3 of Jovanović (1987).
The finds from Velesnica show numerous parallels with those
 Rastko Vasić: Velesnica and the Lepenski Vir culture
of the other Starčevo localities in the lower Ðerdap —
Knjepište (Stanković 1986a) and Ušće Kameničkog potoka
(Stanković 1986b) as well as with Ajmana near Kladovo
(Stalio 1986). These sites probably overlap chronologically
with the later Starčevo phase of Velesnica. It must be em-
phasized, however, that these chronological interpretations
are tentative, based on excavation reports, and that none of
these lower Ðerdap localities has been fully studied or
published.
Post-Starčevo occupation
There is no Late Neolithic horizon at Velesnica. One should,
thus, expect that there was a hiatus in settlement during that
Figure 23. Velesnica: pottery from later periods. 1, fragmented Vučedol bowl; 2–4, Žuto brdo pottery; 5–7 & 9, Basarabi
pottery; 8, Gava miniature vessel.
239
period. However, this is not evident from the Velesnica
stratigraphy. There is no sterile layer overlying the Starčevo
deposits. Starčevo pottery is often found mixed with
Eneolithic and Bronze Age material, and even with later ma-
terial in those cuttings where Dacian and Medieval pits were
found. So, it is quite possible that the Starčevo settlement
lasted longer and was partly contemporaneous with the Late
Neolithic.
The Eneolithic period is not represented by a clearly
defined cultural layer, but the locality was certainly inhabited
at that time as can be seen from many pottery fragments that
can be attributed to different Eneolithic groups. Fragments of
pottery from the Coţofeni group — jars, deep bowls, beakers
with one handle and decorated with incision, pricking, cuts
and plastic bands — which at one time dominated in the Iron
cm
cm
 The Iron Gates in Prehistory
Gates, were uncovered in several cuttings (Jevtić 1987; cf.
Tasić 1995: 65–69). Some chipped stone tools belong to this
culture (Fig. 18, no. 6). On the other hand, the rim of a shal-
low vessel richly ornamented with incised motifs (Fig, 23,
no. 1) found in cutting 6, is of particular interest and could be
connected with the incursion of the Vučedol culture along the
Danube (Garašanin 1973: 237; Srejović 1984). An accidental
find on the riverbank of a ceramic foot, also with incised
decoration, may have belonged to the Vučedol culture (Vasić
1986: fig. 22, no. 10).
Considering the amount of material uncovered, the period
from the Middle Bronze Age to the Early Iron Age was much
more abundantly represented (Žuto brdo–Gîrla Mare, Gava,
Ostrov and Basarabi groups: Jevtić & Vukmanović 1996).
The Žuto brdo–Gîrla Mare group is dated to the later phase of
the Middle Bronze Age, so there was probably a hiatus
between the late Eneolithic and Middle Bronze Age settle-
ments at Velesnica. Because of disturbance, no definite
dwellings could be attributed to this long period but remains
of daub, packed earth, baked earth, floor surfaces and soot in
almost all the cuttings, usually at 36.80–36.60 m, suggest the
continuous presence of settlements throughout this period.
The poor preservation of the structures may be explained by
the use of less durable materials for building which have not
survived, as well as the numerous Dacian and Medieval pits
that destroyed the Bronze and Early Iron Age levels.
Bronze Age finds comprise mainly pottery of the Žuto
brdo–Gîrla Mare group — beakers, urns, dishes and plates,
mostly in fragments (Fig. 23, no. 4). Ornamentation is typical
for this group and consists of spirals, meanders, garlands and
concentric rings in different combinations, all filled with
white incrustation. About twenty fragmented anthropo-
morphic statuettes were also found, which belong to the type
with arms on the abdomen and elbows pointing outwards
(type A2 according to M. Garašanin 1973: 346f.). Several
‘double axes’ and two bird heads that derive from some ves-
sels (Fig. 23, nos 2–3) should also be mentioned. The finds
have numerous parallels from Serbian and Romanian sites of
this group (Jevtić & Vukmanović 1996: 285, map 1).
A period of black channelled pottery follows — one of the
variants of the Gava culture, which is widely represented in
the central Danube area where it is dated to the Late Bronze
Age. In the Ðerdap it is known as the Mala
Vrbica–Hinova–Balta Verde type (Jevtić & Vukmanović
1996: 287, map 3). The pottery is black, polished on the out-
side, and the main shapes are biconical urns, jars, dishes,
beakers, cups with high handles, and salt cellars (Fig. 23,
no. 8). The rare bronze finds from Velesnica — pins, hooks
and rings — very probably belong to this time horizon. This
variant of the Gava group is later than the Žuto brdo–Gîrla
Mare group, but it is stratigraphically impossible to separate
two phenomena which must have been to some extent con-
temporary, particularly as the last phase of the Žuto
brdo–Gîrla Mare group, without rich ornamentation, was in-
corporated into a new group.
Pottery of the Early Iron Age in the Ðerdap is not very
abundant at Velesnica. It belongs to the Ostrov or Insula
Banului group (Jevtić & Vukmanović 1996: 289, map 4) and
comprises vessels decorated with circular impressions, lines
of impressed ‘S’ motifs, rows of false cords, etc., and to the
240
Basarabi complex (Fig. 23, nos 5–7, 9; Garašanin 1996)
which spread over a wide area of Southeast Europe, com-
prising urns, beakers, dishes and plates with grooves and or-
naments with white incrustation (rows of ‘S’ motifs, edged
with false cords, hatched triangles, Maltese crosses, etc.).
The small quantity of Early Iron Age artefacts may mean that
the settlements were small. Alternatively, they may be the
remains of larger settlements that were right beside the river
and were subsequently largely destroyed by the Danube.
After the Basarabi phase there appears to have been an-
other hiatus in settlement. The next phase that is clearly dis-
tinguishable is the Late Iron Age, which according to the
finds belongs to the Dacian period. This ‘Dacian’ material 2
does not appear in all cuttings, which may be explained by
assuming a greater distance between individual dwellings. It
is present in quantity in profile 2, adjacent to the riverbank to
the north of the village, in cutting 6 together with the remains
of plaster floors between 37.50 and 37.00 m, in cutting 7, and
in blocks A and C. It is especially well represented in cutting
9 where two occupation horizons can be confirmed between
37.50 and 37.00 m, particularly in the older horizon where a
well-preserved section of flooring measuring 4 x 1m along
the west profile indicates that a large rectangular construction
may have existed here. One metre to the south of the floor a
circular pit was uncovered, containing soot, ashes, cinders,
pottery and bones, which could have served for various pur-
poses. It is dated by the pottery to the Late Iron Age.
The finds from the Late Iron Age are mainly of pottery,
such as Dacian cups with one handle below the rim, smaller
pots with tongue- and button-shaped handles and plastic
bands around the neck, lids with button-shaped handles,
strainers, and dishes with everted rims — so called ‘fruit
dishes’. A ‘Thracian’ fibula with the catchplate twisted in the
shape of an ‘S’ was found in block B (Vasić 1986: fig. 22, no.
2). Even though the Dacian settlement may have been foun-
ded earlier, the height of its development can be dated to the
first century BC, which would correspond with the historical
events of this time — viz. greater and more widespread
power of the Dacian rulers (Popović 1991).
Remains of a Roman settlement, whose peak of develop-
ment was in the fourth century BC, were also uncovered at
Velesnica, as well as a large Medieval settlement that existed
continuously from the seventh to the fifteenth century AD
(Ercegović-Pavlović & Minić 1986).
Conclusions
Prehistoric settlements existed at Velesnica, with some
breaks in occupation, from the end of the Mesolithic up to the
arrival of the Romans in the area. Some time periods are bet-
ter represented than others; while at certain times there was a
hiatus in settlement. The Early Neolithic (Starčevo culture)
was a period of particularly intensive settlement, after which
there was a hiatus corresponding to the later Neolithic. New,
smaller settlements were founded in the Eneolithic period by
the bearers of the Coţofeni group. After another hiatus during
the Early Bronze Age, there was a renewed phase of intensive
settlement in this area during the Middle Bronze Age, which
continued without a break right up to the beginning of the
 Rastko Vasić: Velesnica and the Lepenski Vir culture
Early Iron Age. There followed another break in occupation
during the Iron Age. Finally, a large new settlement sprang up
during the last phase of the Late Iron Age, in the time of the
Dacians. These periods of development and hiatus that char-
acterized Velesnica are typical of prehistoric sites in the
lower Ðerdap.
Why the Iron Gates were densely inhabited in some peri-
ods and not in others will probably be better understood after
further analysis and final publication of the excavated mater-
ial. On present evidence, it seems that all the groups that
flourished here originated to some extent in this region,
where they found suitable conditions for their further devel-
opment. On the other hand, alien groups or cultures, which
penetrated the region from time to time, never dominated.
The Starčevo culture in the Ðerdap was certainly partially
connected and intermingled with the latest Mesolithic inhab-
itants of the region. The Coţofeni culture, widespread in
Oltenia, intermingled in this region with the Kostolac culture,
which was widespread in Serbia and probably formed a local
variant characteristic of the Ðerdap. The Ðerdap was the
centre of the Žuto brdo–Gîrla Mare group as well as that of
the Ostrov group. Even the Basarabi complex is connected to
some extent with this region — according to some research-
ers, its nucleus was in the Iron Gates. This course of events in
the prehistoric Ðerdap was caused by various historical, eco-
nomic and social factors, which will probably be clarified
through more detailed and thorough analysis of the archae-
ological evidence.
Notes
1. Throughout this paper, the term ‘gravel’ is used to refer to (ap-
parently man-made) concentrations of small stones.
2. It is uncertain whether the settlement belonged to the Dacians or
to another Late Iron Age tribe.
Acknowledgments
I wish to express my gratitude to the colleagues who helped me in
preparing this paper, either by studying various aspects of the
Velesnica material or by discussing with me various topics
concerning the prehistory of the Iron Gates: Dragana Antonović,
Aleksandar Durman, Borislav Jovanović, Bojana Mihailović,
Gordana Mikić, Dubravka Nikolić, Slaviša Perić, Viktorija Podboj,
Mirjana Roksandic, Josip Šarić, Nikola Tasić and Mirjana
Vukmanović.
References
Antonović, D. 1998: Nastanak i razvoj industrije glačnog kamena u
neolitu Srbije (Origin and Development of the Ground Stone
Industry in the Neolithic of Serbia). Unpublished PhD disserta-
tion, Department of Archaeology, Faculty of Philosophy,
University of Belgrade.
Boroneanţ, V., Bonsall, C., McSweeney, K., Payton, R. & Macklin,
M. 1999. A Mesolithic burial area at Schela Cladovei, Romania.
In Thévenin, A. (ed.) L’Europe des Derniers Chasseurs:
Épipaléolithique et Mésolithique. Actes du 5e colloque interna-
tional UISPP, commission XII, Grenoble, 18–23 septembre 1995.
Paris: Éditions du Comité des Travaux Historiques et
Scientifiques, 385–390.
241
Ercegović-Pavlović, S. & Minić, D. 1986: Site d’habitation
mediéval à Velesnica. Đerdapske sveske 3: 289–301.
Garašanin, M. 1973: Praistorija na tlu SR Srbije. Belgrade: Srpska
knjizevna zadruga.
Garašanin, M. (ed.) 1996: Der Basarabi-Komplex in Mittel- und
Südosteuropa. Bucharest: Rumänisch–Jugoslawische
Kommission für die Erforschung der Region des Eisernen Tores
I.
Jevtić, M. 1987: Les stations énéolithiques dans le secteur de
Ðerdap I–II (Portes de Fer). In Tasić, N. (ed.) Hügelbestattung in
der Karpaten-Donau-Balkan Zone wahrend der äneolithischen
Periode. Belgrade: Institute for Balkan Studies, 21–26.
Jevtić, M. & Vukmanović, M. 1996: Late Bronze and Early Iron
Ages in the Danube Valley from V. Gradište down to Prahovo. In
Tasić, N. (ed.) The Yugoslav Danube Basin and the
Neighbouring Regions in the 2nd Millennium B.C. Belgrade-
Vršac: Institute for Balkan Studies, 283–293.
Jovanović, B. 1969: Chronological frames of the Iron Gate group of
the Early Neolithic period. Archaeologia Iugoslavica 10: 23–38.
— 1987: Die Architektur und Keramik der Siedlung Padina B am
Eisernen Tor, Jugoslawien. Germania 65: 1–16.
Mikić, G. 1999: Fauna mezolitskog sloja Velesnice. Unpublished
B.Phil. dissertation, Faculty of Mining and Geology, University
of Belgrade.
Popović, P. 1991: Mladje gvozdeno doba Ðerdapa. Starinar n.s.
40/41: 165–176.
Radovanović, I. 1996: The Iron Gates Mesolithic. Ann Arbor:
International Monographs in Prehistory.
Srejović, D. 1969: Lepenski Vir, Nova praistorijska kultura u
Podunavlju. Belgrade: Srpska književna zadruga.
— 1984: Katarinine Livade. Agglomeration de l’âge du Bronze.
Starinar n.s. 33/34: 209–210.
— 1988: The Neolithic of Serbia: a review of the research. In
Srejović, D. (ed.) The Neolithic of Serbia. Archaeological
Research 1948–1988. Belgrade: Centre for Archaeological
Research, Faculty of Philosophy, University of Belgrade, 5–19.
Srejović, D. & Babović, Lj. 1983: Umetnost Lepenskog Vira.
Belgrade: Jugoslavija.
Stalio, B. 1986: Le site préhistorique Ajmana à Mala Vrbica. Đer-
dapske sveske 3: 27–50.
Stanković, S. 1986a: Localité Knjepište — une station du groupe de
Starčevo; fouilles de 1982–1983. Đerdapske sveske 3: 447–452.
— 1986b: Embouchure du ruisseau Kamenički potok — site du
groupe de Starčevo: compte-rendu des fouilles de 1981. Đer-
dapske sveske 3: 467–471.
Šarić, J. 1987: Starčevačka kremena industrija sa lokaliteta Ušće
Kameničkog potoka, Knjepište i Velesnica. In Lazić, M. (ed.)
Arheologija istočne Srbije. Belgrade: University of Belgrade,
Faculty of Philosophy, Centre for Archaeological Research
vol.18, 177–187.
Tasić, N. 1995: Eneolithic Cultures of Central and West Balkans.
Belgrade: Institute for Balkan Studies, Special Monograph 61.
Vasić, M.M. 1910: Žuto brdo. Nova nalazista Žutobrdskog tipa u
Srbiji. Starinar, novi red 5: 1–207.
Vasić, R. 1986: Compte-rendu des fouilles du site préhistorique à
Velesnica 1981–1982. Đerdapske sveske 3: 264–285.
Vasić, R., Ercegović-Pavlović, S. & Minić, D. 1984: Velesnica.
Prospection par sondage 1980. Đerdapske sveske 2: 125–132.
Vasić, R. & Janković, Dj. 1971: Rekognosciranje desne obale
Dunava od Kladova do Prahova. Arheoloski pregled 13:
107–113.
Živanović, S. 1986. Restes des ossements humains à Velesnica.
Đerdapske sveske 3: 286–288.
 The human osteological material from Velesnica
Mirjana Roksandic
Abstract: This paper is a report on the human osteological material from the Neolithic site of Velesnica. A revision of basic demographic
information together with postcranial measurements and non-metric traits is provided. A comparison of the postcranial traits of the Velesnica
population (measurable females only) is presented in a PCA scatterplot. According to this analysis, the Velesnica material does not show any
clear distinction from the other Iron Gates Mesolithic/Early Neolithic sites.
Key words: Early Neolithic, Mesolithic, Iron Gates, non-metric traits, robusticity
Introduction
The first analysis of the osteological material from Velesnica
was undertaken by Srboljub Živanović (1986). In my ana-
lyses of the material from other Iron Gates Gorge sites re-
ported on by Živanović, many discrepancies were found in
MNI, sex, age and the degree of preservation of the skeletons.
Therefore, comparisons with his report from 1986 will be
made throughout this text. Apart from the revision of the ba-
sic demographic information the paper looks into non-metric
traits and postcranial robusticity in order to bring the material
in line with already published results for the Mesolithic/Early
Neolithic sites in the Iron Gates (Roksandic 2000). For the
description of non-metric variables see Roksandic (2000:
44–54), and for the metric variables Buikstra and Ubelaker
(1994). The measurements are summarized in Table 1, and
compared with available metrics of the rest of the Iron Gates
Mesolithic/Neolithic series. Non-metric traits, where ob-
served, are summarized in Table 2.
In addition, to the extent that it is possible from the pub-
lished photographs, I discuss the dispositional taphonomy
(sensu Roksandic 2002) of the burials, in order to ascertain
descernible aspects of the mortuary ritual.
Grave 1
This grave contained the bones of a small child. Based of the
size of the bone fragments, the child was between 3 and 7
years old. Bones present at the time of analysis were two
fragments of ribs, the distal half of the right humerus, the
second and third right metacarpals, the proximal half of the
left ulna, and one lumbar vertebra. A septal aperture is
present on the right humerus in the form of multiple
pinprick-like small perforations. As can be seen from the
photograph (Fig.1a), the bones originally present were much
more numerous. Since Živanović noted that “leur datation
précise sur la base du collagène extrait de la profondeur de
l’os sera connue après la mesure du carbone 14 dans la
laboratoire du Musée britanique” (Živanović 1986: 286), it is
possible that the missing bones were sent to the British
Museum for dating. Some bones of the lower member can be
identified. Based on a study of the photographs (Fig. 1a;
243
Vasić 1986: fig. 9, this volume: fig. 10) the following obser-
vations can be made:
The grave represents a primary burial as a number of ele-
ments are in proper anatomical position (vertebrae, ribs, right
coxal and femur, etc.). It seems to have been deposited on the
left side of the body with legs contracted. The position of the
ribs and the lateral aspect of the vertebrae are consistent with
this explanation. However, there are some minor indications
that the primary position could have been different, notably
the position of the right leg and that of the skull. The femur
is most probably presented in its lateral aspect, while the tibia
and fibula present their anterior surfaces. It would have been
valuable to have information on the relative depth of the
distal femur and proximal tibia and fibula in order to recon-
struct whether the knees were originally elevated or not. This
information is lacking, but the anterior aspect of the tibia and
fibula suggests this possibility. It is hard to ascertain the as-
pect of the left femur, but the left tibia is shown by its medial
surface, indicated also by the fact that the left fibula, most
probably underneath the tibia in its proper anatomical posi-
tion, is not visible on the photo. The right clavicle is dis-
placed and the medial articulation is closer to the lateral part
of the body than, as would be expected, the lateral articula-
tion. This abrupt movement could have been induced by the
‘rolling off’ of the skull that is presented on the photograph
by the superior aspect of the calotte. The position of the ribs
and of the vertebral column indicates the ‘effect of the wall’
or the existence of a burial construction that prevented a more
decisive movement of the ribs once the soft tissue decom-
posed. The original position could have been that of dorsal
rather than lateral decubitus with knees contracted and
slightly raised. Although elements of the superior member
were found among the remains, they cannot be recognized on
the photograph. They could have been removed before the
photo was taken, or they could have been under the exposed
bones. Both the position of the skeletal elements and the
stones around the skeleton indicate that decomposition took
place in an open delimited space (for explanation of terms,
see Roksandic 2002), namely in a pit covered with some
perishable material. Direct covering of the skeleton with
earth is unlikely given the displacement of the skull.
However, a large stone ‘altar’ placed above the head of the
deceased (Vasic 1986: fig. 9) would have crushed the skull if
 The Iron Gates in Prehistory

Figure 1. Velesnica burials: a. grave 1; b. grave 2A; c. graves 2C–2F; d. grave 2G; e. grave 3.

it collapsed together with the sediment. More direct field ob-
servations and information on the relative depths of these
elements is needed in order to facilitate a more definite ex-
planation of the burial architecture and the ritual.
Although Živanović (1986: 286) noted cranial fragments
and gave a description of them, none were found during the
present analysis, conducted in 1998. According to Živanović
(1986) this was due to the bones disintegrating as soon as
they were exposed. It seems that his recognition of the
“naissance des arcades orbitaires marquées ... et bourelets qui
donneraient plus tard le torus auditorius” is guided by his
expectations and assertions that all the buried individuals be-
longed to the population that he identified as being of the
‘Padina type’ (for a discussion of the validity of this ‘type’,
see Roksandic 2000: 79–80).
Grave 2
At least five individuals (2A, 2B, 2C, 2D, 2G), probably a
sixth (2A-1), and possibly a seventh (2G-1) were buried in
this grave (Figs 2b–2d). Although Živanović also claims

244
 Mirjana Roksandic: The human osteological material from Velesnica

Table 1. Postcranial measurements in millimetres, *reconstructed; ‘R’ taken on the right bone.

Grave
2A 2B 2D

maximal length; clavicle 112*R 13

clavicle anterior-posterior diameter at midshaft 10 11 8
clavicle superior-inferior diameter at midshaft 9 8 8
humerus epicondylar breadth 52R 53
humerus vertical diameter of the head 37
humerus maximum diameter at midshaft 18 18 21
humerus minimum diameter at midshaft 15 15R 14
radius maximum length 201 205*
radius anterior-posterior diameter at midshaft 8 9 10
radius medial-lateral diameter at midshaft 11 11 12
ulna maximum length 234
ulna minimum circumference 30 30 31
femur maximal length 384 387
femur bicondylar length 381 384
femur epicondylar breadth 67* 70
femur maximum head diameter 37R 39 39
femur anterior-posterior subtrochanteric diameter 21 21 30
femur medial-lateral subtrochanteric diameter 29 29 23
femur anterior-posterior midshaft diameter 24 25 21
femur medial-lateral midshaft diameter 21 23 22
femur midshaft circumference 75 74 70
tibia length 311 312
tibia maximum proximal epiphyseal breadth 57R 60
tibia maximum distal epiphyseal breadth 28 38R 41
tibia maximum diameter at the nutrient foramen 28 28R 28
tibia transverse or medial-lateral diameter at nutrient foramen 18 20R 19
tibia circumference at the nutrient foramen 75 77R 75
calcaneus maximal length 70
calcaneus maximal breadth 36R 34

seven, he distinguished 2E and 2F that in the present study
were found to match in every detail 2A and 2B, respectively.
Moreover, he did not note 2A-1 (possibly the only adult male
in the burial) and 2G-1 (several bones of a newborn). It is
hard to ascertain the synchronous or diachronous nature of
the burials within this grave based exclusively on the pub-
lished photographs, more detailed information is needed.
However, several positions can be noted on the published
photo (Vasic 1986: figs 14, 15 & 16): position ‘à la turque’ for
individual 2G at the bottom of the burial pit with no notice-
able disturbances caused by the superimposed burials. The
posterior aspect of the scapulae, pelvis, skull and vertebrae of
the directly superimposed skeleton 2D indicates frontal
decubitus, with flexed legs in their primary position. It seems
that decomposition of 2G had already taken place before in-
dividual 2D was placed in the grave, and therefore a dia-
chronous burial with subsequent interments is most likely.
For the rest of the skeletons it is impossible to advance any
explanation based on the published photographs.
Individual 2A(+2E)
Based on the preserved portion of the pelvis, this individual
is a female. From the auricular surface, tooth abrasion and
suture closure it is a mature adult female. For measurements
of the postcranial skeleton and for non-metric traits refer to
Tables 1 and 2, respectively. As with all other individuals in
this burial, when analyzed in 1998, the skull was still par-
tially encrusted with the fine sediment from the grave and it
is unlikely that the measurements provided by Živanović
(1986: 287) are correct. Since there is no duplication of bones
between 2A and 2E, and because it was possible to re-
assemble the right radius from the proximal (2A) and distal
(2E) fragments, there is little doubt that they represent the
same individual. In addition, the degree of robustness of the

245
 The Iron Gates in Prehistory

Table 2. Non-metric traits. Both left and right occurrences were noted for the traits that occur
on the paired bones. They are recorded as L/R side respectively; 0=absent; 1,2 or 3 =present
(with specific degree or form of expression); and 9=unobservable. For detailed description of
traits and scoring procedures see Roksandic (1999, 2000).

Grave
2A 2B 2D

metopic suture 0 0 0
supraorbital notch 1/0 9/9 1/2
supraorbital foramen 0/0 0/0
infraorbital suture 9/9 0/9
multiple infraorbital foramina 9/9 0/9
zygomatico-facial foramina 9/9 0/0
parietal foramen 0 2 1
bregmatic bone 0 0 9
sagittal ossicle 0 9 9
apical bone 0 9 9
lambdoid ossicle 0/1 0/0 1/1
asterionic bone 0/1 9/9 0/0
ossicle in occipito-mastoid suture 0/1 9/9 0/0
parietal notch bone 0/ 9/9 0/0
inca bone 0 0 0
condylar canal 9 9 1
divided hypoglosal canal 0/0 9/9 0
flexure of superior sagittal sulcus 3 2 2
tympanic dehiscence 0/0 0/0 0/0
auditory exostosis 1/0 0/0 0/0
mastoid foramen location 2/0 9/1 0/0
mastoid foramen number 1/0 9/2 0/0
mental foramen 1/1 1/1 1/1
mandibular torus 0/0 0/0 0/0
maxillary torus 0/0 0/0 9/9
palatine torus 0 9 9/9
mylohyoid bridge location 2/0 9/9 2/2
mylohyoid bridge degree 2/0 9/9 2/2
ligament teres in cavitas glenoidalis 0/9 0/0 0/0
perforatio fossae olecrani 9/4 9/4 5/3
supratrochlear spur 0/0 9/0 0/0
fossa bicipitis radii 0/0 0/0 0/0
unfused processus olecrani 0/0 0/9
fossa faciei lunatae 0/0 0/0
Allen’s fossa 0/1 0/9 1/9
third trochanter 1/1 0/0 0/0
Poirier’s facet of extension 9/9 1/9 1/9
Vastus notch 1/1 1/9
Squatting facets on distal tibia 9/0 3/1
Squatting facets (talus) tibia 0/9 0/0 0/0
Shape of the talar articular surface on calcaneus 1/1 1/1

246
 Mirjana Roksandic: The human osteological material from Velesnica
bones is comparable. It is therefore difficult to understand
Živanović’s identification of 2E as a male, especially as the
bones in general seem to be very gracile.
Individual 2A-1
Only the fragment of a frontal bone including the glabella
catalogued as 2A-1 seems to be robust and could be very
tentatively identified as male. This fragment is catalogued as
a separate individual rather than ignored as a possible earlier
grave disturbance (Meiklejohn & Denston 1987), because of
the common occurrence of these ‘extra’ individuals repres-
ented by small fragments in many Iron Gates graves
(Roksandic, in preparation — and see below).
Individual 2B(+2F)
There is no duplication of elements between 2B and 2F and
pairing of the bones that are present shows the degree of
similarity between the auricular surfaces and the humeri,
which indicates that they belonged to the same individual.
Only the left temporal fragment of 2F belongs to a different
individual (it is comparable in size and form to the 2C tem-
poral fragment). Based on the form of the great sciatic notch
and the presence of a deep auricular sulcus, the individual is
a female. Advanced abrasion of the teeth and closure of skull
sutures, as well as the aspect of the auricular surface features,
indicate a mature adult. All the teeth show an advanced de-
gree of attrition but also a degree of dental pathology not
commonly encountered in the Iron Gates Mesolithic popula-
tion. There is an abscess above the upper first left molar (26)
(Fédération Dentaire Internationale 1971), while excessive
carious lesions can be noted on the upper right premolars (14
and 15). Caries are also present on the lower left canine (43)
and lower right first molar (36). Measurements and non-
metric traits are presented in Tables 1 and 2, respectively.
Individual 2C
This skeleton is that of a pre-puberal child approximately 9
years of age (±24 months) based on tooth formation and
eruption stage. While the lateral segments of the lambda su-
ture are completely open, the sagittal part of it and the sagittal
suture in its two posterior segments are almost completely
obliterated. A circular depression on the occipital bone,
slightly above and to the right of the exoinion, approximately
2.5 cm in diameter is the only pathological change observed
on the skeleton. Given its position and the age of the indi-
vidual, coupled with the fact that the lesion is healed, it is
most probably a healed depressed fracture. Depressed frac-
tures of the skull are 3.5 times more common in children than
in adults (Galloway 1999: 68) and can cause both infection
and brain damage. However, due to the generally thinner and
more flexible bones of the cranial vault in young children,
they are often limited to a depression, and may not involve a
fracture site. Their common name of ping pong fractures is
due to this specific feature (Galloway 1999: 68). Pitting, ob-
served on the floor of the lesion and in its immediate sur-
rounding, indicates a post-traumatic infection. Non-metric
traits are presented in Table 2.
As mentioned above, a fragment of a left temporal, com-
parable to the right temporal of individual 2C was found with
individual 2F. Some fragments of an adult individual were
247
also separated from 2C. They most probably belong to one of
the adult individuals present in this grave.
Individual 2D
Except for the axial skeleton almost all the bones were
present, including the hyoid, a strong indication of a primary
burial. Based on the preauricular sulcus and the form of the
sciatic notch, the individual is a female. Excessive abrasion
and suture closure suggest that it was most probably a mature
or senile adult. Apart from intra vitam loss of the upper left
first molar (26) there are no other recorded pathologies.
Measurements and non-metric traits are presented in Tables 1
and 2, respectively.
Individual 2G
According to the tooth formation and eruption pattern
(Ubelaker 1989) this individual is a pre-puberal child, 7±2
years old at death. Almost all of the bones are present in a
more or less fragmentary state including the hyoid in its ana-
tomical position (held together with the mandible and the rest
of the skull by the sediment), which is a strong indication of
a primary burial. There are no macroscopically observable
pathologies on the skeleton. A circular pit hypoplasia is
present on the upper left deciduous canine (63). In all ar-
chaeological or modern populations where the defect has
been observed it was limited to the primary canine (Skinner
& Hung 1986, 1989: 199) and has been associated with in-
adequate maternal nutrition during gestation, and premature
birth. The location indicates that it starts to form at birth or
shortly thereafter and can serve as an easy observed marker
of nutritional inadequacy during gestation (Skinner et al.,
1994). In addition, very thin hypoplastic lines are present on
the upper permanent incisor, indicative of mild non-specified
systematic stress that occurred in early childhood between 2
and 4 years of age (Skinner & Goodman 1992: 164–165, figs
6 & 7).
Individual 2G-1
A proximal fragment of a right femur, a left femur (78 mm
long) and a proximal left tibia of a newborn were separated
from the bones assigned to individual 2G. As with individual
2A-1, the bones were treated as an extra individual rather
than accidental grave filling, because of the relatively com-
mon occurrence of neonatal bones in other Iron Gates burials.
As many as 33 out of 84 newborns in the Iron Gates gorge
(almost 40%) are found within the context of another burial,
sometimes represented by only one or few bones or bone
fragments. (Roksandic, in preparation). Therefore we con-
clude that it represents most probably the secondary burial of
a newborn and is indicative — together with the disposition
‘à la turque’ of individual 2G — of continuity in mortuary
practices between Mesolithic and Neolithic in the region.
Grave 3
Unfortunately, the bones of the individual from grave 3 at
Velesnica (Fig. 1e) could not be found in the storage facility
at Karataš, as well as some of the bones from the other burials
described by Živanović that were not present at the time of
 The Iron Gates in Prehistory
Figure 2. Scatterplot of PCA scores for Iron Gates gorge sites. Based on postcranial measurements in Table 2 for Velesnica;
(for other sites, see Roksandic 1999: appendix II).
my analysis. A possible explanation for this may lie in
Živanović’s claim that “… l’analyse paléopathologique de
matériel osseux a été confiée au laboratoire anthropologique,
Department of Anatomy, The Medical College of St.
Bartholomew’s Hospital à Londres” (Živanović 1986: 286).
It is conceivable that individual bones from some of the
skeletons and the entire skeleton from grave 3 were removed
to London. Živanović left no indication of the procedures
used, in either his notes or his published report. Živanović
(1986: 287) identified it as an adult female whose epiphyses
were not yet fused and third molars were erupting (?). From
figure 21 in Vasić (1986), it is possible to conclude that the
individual is a primary burial, contracted, laying on the right
side, with most of the preserved bones in the expected ana-
tomical position.
General observations
The number of measurable bones from Velesnica is insuffi-
cient for statistical analysis. Therefore some basic comparis-
ons were performed with the material from other Iron Gates
sites. Only the general features of the postcranial skeletons
are presented here. As indicated in the introduction, I have
attempted to incorporate the metric information into the
scatterplot of the factor scores for a combination of metric
variables grouped according to chronology and sex of the in-
dividuals from the other Iron Gates gorge sites (Fig. 2). The
three measurable individuals (all females) are plotted in a
principal components analysis (PCA) with the measurable
248
adults from Lepenski Vir, Padina, Hajdučka Vodenica, and
Vlasac. As expected, the most evident feature of the graph is
the clear separation of female and male skeletons. A method-
ological bias is excluded since the sex determination was
based on the pelvic remains only. Another important feature
is the clustering of females from all of the sites in a tight
group, while males seem to be more dispersed. This strong
indication of more pronounced heterogeneity in the male
group and more homogenous female group accords well with
the quantitative analysis of Vlasac cranial remains by
Nemeskéri and Szathmary (1978: 178), who propose greater
mobility for males and possible matrilocal pattern of resid-
ence. Vlasac females also tend to be towards the lower end of
the female values indicating even more gracile features than
the rest of the female sample. It is important to note that we
do not have any measurable males from the Neolithic period
on any of the sites, so that interpretation must remain ex-
tremely tentative. However, no deviation from the general
Iron Gates Mesolithic and Contact periods can be noted.
Conclusion
Human remains from Velesnica show substantial continuity
with the Mesolithic population of the Iron Gates Gorge, both
in their burial rituals and in their bio-metric data. Indicative
in this respect are the position ‘à la turque’ of burial 2G, and
the inclusion of the secondary burial of bones from both an
adult and a newborn. In terms of continuity within the popu-
lation, it should be noted that the three measurable females
 Mirjana Roksandic: The human osteological material from Velesnica
from the site show the same tight clustering in terms of the
postcranial metrics, and a trend toward gracilization of the
postcranial skeleton.
Acknowledgements
This research was made possible by a Wenner-Gren Small Research
Grant no. Gr. 6250. I am indebted to Dr Rastko Vasić of the Institute
of Archaeology in Belgrade for permission to study the material, to
my colleagues Mirjana Glumac of the National Museum in Belgrade
for logistical support during the analysis, and Sylvia Leek of the
Max Planck Institute for Demographic Research for technical help
with the graph.
References
Buikstra, J. & Ubelaker, D. 1994: Standards for Data Collection
from Human Skeletal Remains. Fayetteville: Arkansas
Archaeological Survey.
Fédération Dentaire Internationale. 1971: Two-digit system of des-
ignating teeth. International Dental Journal 21: 104–106.
Galloway, A. 1999: Fracture patterns and skeletal morphology: in-
troduction and the skull. In Galloway, A. (ed.) Broken Bones:
Anthropological Analyses of Blunt Force Trauma. Springfield:
Charles C. Thomas, 63–80.
Meiklejohn, C. & Denston, B.. 1987: The human skeletal material:
inventory and initial interpretation. In Mellars, P.A. (ed.)
Excavations on Oronsay: Prehistoric Human Ecology on a Small
Island. Edinburgh: Edinburgh University Press, 290–300.
Nemeskéri, J. & Szathmary, L. 1978: Analysis of the variations of
quantitative traits. In Srejović, D. & Letica, Z. (eds) Vlasac. A
249
Mesolithic Settlement in the Iron Gates, vol. 2. Belgrade: Serbian
Academy of Sciences and Arts, 157–176.
Roksandic, M. 2000: Between foragers and farmers in the Iron
Gates gorge: physical anthropology perspective. Documenta
Praehistorica 27: 1–100.
— 2002: Position of skeletal remains as a key to understanding
mortuary behavior. In Haglund, W.D. & Sorg, M.H. (eds)
Advances in Forensic Taphonomy. Boca Raton: CRC Press,
99–117.
— In preparation: Skeletal representation and secondary burials in
the Iron Gates gorge.
Skinner, M. & Goodman, A.H. 1992: Anthropological uses of de-
velopmental defects of enamel. In Saunders, S.R. & Katzenberg,
M.A. (eds) Skeletal Biology of Past Peoples: Research Methods.
New York: Wiley–Liss, 153–174.
Skinner, M., Hadaway, W. & Dickie, J. 1994: Effects of ethnicity,
nutrition, and birth month on localized enamel hypoplasia of the
primary canine. Journal of Dentistry for Children 61: 109–113.
Skinner, M.F. & Hung, J.T.W. 1986: Localized enamel hypoplasia
of the primary canine. Journal of Dentistry for Children 53:
197–200.
Skinner, M.F. & Hung, J.T.W. 1989: Social and biological correlates
of localized enamel hypoplasia of the human deciduous canine
tooth. American Journal of Physical Anthropology 79: 159–175.
Ubelaker, D.H. 1989: The estimation of age at death from immature
human bone. In Yscan, M.Y. (ed.) Age Markers in the Human
Skeleton. Springfield, IL: Charles C. Thomas.
Vasić, R. 1986: Compte-rendu des fouilles du site préhistorique à
Velesnica 1981–1982. Đerdapske sveske 3: 271–285.
Živanović, S. 1986: Restes des ossements humains à Velesnica.
Đerdapske sveske 3: 286–288.
 The Mesolithic–Neolithic transition in the Trieste Karst
(north-eastern Italy) as seen from the excavations at the
Edera Cave

Paolo Biagi, Elisabetta Starnini & Barbara A. Voytek

Abstract: This paper is a preliminary report of excavations carried out between 1990 and 2000 at Edera Cave in the Trieste Karst. The cave is
3km from the present coast. It was first occupied during the Boreal period by Mesolithic (Sauveterrian) hunter-gatherers. The remains of
this occupation comprise pits and hearths, abundant faunal remains, and typical chipped stone and bone tools, many of which were recovered
in situ on an almost intact palaeosurface (layer 3c). A hearth in layer 3a above belongs to the Late Mesolithic Castelnovian culture, and has
yielded a few potsherds of non-local production, as well as bones of domesticated animals. Layer 2a consists of several superimposed
charcoal lenses attributed to the local Early Neolithic Vlaška group on the basis of characteristic vessel shapes and a series of four ra-
diocarbon ages that date this horizon to c. 6500 BP (5500 cal BC). Although the cave continued to be used sporadically until the Migration
Period, this paper is concerned mainly with the problem of the relationship between the last hunter-gatherers and the first food producers
who used the cave during the mid-seventh millennium BP. So far, layer 3a of Edera Cave is the only evidence from this part of the Adriatic of
interaction between the last Castelnovian bands and the first Neolithic farmers.
Key words: Edera Cave, Italy, Trieste Karst, Mesolithic, Neolithic, transition

Introduction
The excavations carried out in the caves of the Trieste Karst
during the last forty years have produced particularly inter-
esting evidence of both Mesolithic and Early Neolithic occu-
pation. Nevertheless, all the sequences so far excavated are
very incomplete, mainly as regards the beginning of the
Atlantic, Mesolithic Castelnovian. This period is currently
known only from the Cavernetta della Trincea in the
Rosandra Valley (Andreolotti & Stradi 1963), from the upper
layers of the Grotta della Tartaruga (Cremonesi 1984) and
from the Benussi Cave (Broglio 1971). At Benussi Cave, the
Castelnovian levels have been dated between 7050±60 BP
(R-1043) and 7620±150 BP (R-1044). Other sites, such as the
Grotta dei Ciclami and the Grotta Azzurra di Samatorza
(Ciccone 1992), yielded Mesolithic sequences that are ab-
ruptly interrupted between the end of the Boreal and the be-
ginning of the Atlantic, according to the flint assemblages
(Cremonesi et al. 1984a, 1984b; Montagnari Kokelj 1993).
The Trieste Karst is a limestone ellipsoid whose Italian
territory, after the Second World War, has been reduced to
some 850km2. The Karst upland that rises gradually from the
north west to the south east, is a hilly landscape the highest
peak of which is Mt Concusso at 672 m above sea level. The
coast that borders it to the south west is characterized by high
cliffs that drop vertically into the Gulf of Trieste. The cliff-
line is interrupted by two alluvial plains, those of Grignano
and Trieste (Biagi & Voytek 1994).
Edera Cave
The cave opens at the bottom of a doline close to the Aurisina
marble quarries (Fig. 1), some 15km north west of Trieste
and 3km from the present coastline. The opening, which
faces north east, is at 230 m above sea level (Fig. 2). Local
amateurs discovered the cave in 1969 and carried out pre-
liminary excavations between 1969 and 1975. These re-
vealed the importance of the archaeological sequence, which
extends from the beginning of the Mesolithic to the Medieval
period (Marzolini 1970; Boschian & Pitti 1984).
The excavations were resumed in 1992 by the Department
of Historical–Archaeological and Oriental Sciences of the
University of Venice (I) and by the Archaeological Research
Facility of the University of California, Berkeley (USA) and
continued until 2002. The new research revealed a strati-
graphic sequence some 4 m thick with occupation layers at-
tributed to the Migration and Roman periods, the Bronze and
Copper Ages, the Middle and Early Neolithic, as well as the
Sauveterrian and Castelnovian Mesolithic. Even though the
sequence is discontinuous, it is the most detailed so far dis-
covered in the upper Adriatic region (Figs 3 & 4).
In order to retrieve as complete and accurate a record as
possible, we adopted wet-sieving, using a 2 mm mesh screen,
as part of our recovery technique. As a result, we have had
exceptional success in collecting archaeological evidence,
including specimens of small dimensions such as bioarchae-
ological remains (microfauna, fish bones, charred seeds, and
land snails) and artefacts (Mesolithic stone tools, débitage
and shatter). Edera is the first archaeological project in the
Trieste Karst to be conducted with complete wet-sieving of
all excavated soil. It will allow us to evaluate the degree of
excavation technique bias in comparison with the previous
research during which only dry sieving was employed. The
excavation was carried out following the natural stratigraphic
layers, which are designated by numbers and letters. Each
layer has been further subdivided into horizontal unit levels
(‘spits’) of 10 cm thickness, in order to be able to better ap-
preciate the variability of human activity within each layer.
The earliest deposits excavated so far belong to the Early

251
 The Iron Gates in Prehistory
Figure 1. Location of the Edera Cave, north east of the Fornaci di Aurisina, in the Trieste Karst.
Neolithic and the Mesolithic. The Early Neolithic of the
Trieste Karst is characterized by the Vlaška group, which has
always been considered a local, impoverished variant of the
Danilo culture. Layer 2a can be attributed to this phase. It is
a silty layer, some 0.90 m thick, with evidence of six super-
imposed ash and charcoal lenses. Three of these have been
radiocarbon dated. In stratigraphic order, from the lowermost
upwards, they produced the following results: 6615±390 BP
(GX-19568) and 6590±100 BP (GrN-23129) from the same
charcoal sample, 6445±210 BP (GX-19567) and 6305±285
BP (GX-19022) (Fig. 6).
Although the results of palaeopedological and soil thin
section analyses are still in progress, the variations in texture
and nature of the sediment are easily observed from layer 3
downwards. The deposit assigned to layer 2 is essentially a
fine-grained grey-buff silt, while the deposit of layer 3 is a
compact reddish clay. Layer 3 is c. 20 cm thick, within which
we uncovered a quantity of very small potsherds as well as a
few very small flint flakes obtained from local raw material
(rounded river pebbles).
A hearth about one metre in diameter was discovered in
layer 3a (Fig. 5A), associated with a typical, though very
small, Late Mesolithic, Castelnovian flint assemblage
chipped from local pebbles. The assemblage comprises 538
artefacts among which are 2 trapezes, 1 denticulated bladelet,
252
2 truncations, 1 short end-scraper, 1 core fragment, and 61
microburins (Fig. 8: 1–5). A total of 17 potsherds, most of
which are very small, come from this layer 3a. Two of the
potsherds are diagnostic pieces that can be reconstructed.
Thin-section analyses of the latter indicated that they had
been manufactured of clay with different mineral inclusions
from those detected in the potsherds of layer 2a above, sug-
gesting that they had been imported from outside the karstic
area (Spataro 2001: 95). A charcoal sample from this hearth
has been 14C dated to 6700±130 BP (GX-19569), which is
very similar to the age of the lowermost Vlaška group char-
coal lens of the overlying layer 2a.
Layer 3b has three radiocarbon dates, two on charcoal
(GrN-25137: 8060±70 BP and GrN-25138: 8110±90 BP) and
one on an animal long bone fragment (GrA-14106: 8045±40
BP). On the basis of both the 14 C ages and the flint as-
semblage this layer may be attributed to the very beginning
of the Atlantic/end of the Boreal period. Only a small per-
centage of the flint assemblage from layer 3b has been stud-
ied so far.
Layer 3c is an almost intact palaeosurface with in situ
lenses of ash and charcoal as well as faunal remains, flint
artefacts, and a single human tooth. Four square metres of
this palaeosurface have been investigated (Fig. 5B). Two ra-
diocarbon dates of 8350±120 BP (GrN-25139) on a charcoal
 Paolo Biagi et al.: The Mesolithic–Neolithic Transition in the Trieste Karst
Figure 2. Aerial view of the doline where the Edera Cave is
located (the cave entrance is visible in the upper part of the
picture, below the white roof of the protection gate). The
picture was taken before the spontaneous reforestation of
the Karst caused by the progressive abandonment of land
management for agricultural purposes (photograph by G.
Marzolini).
sample and 8250±50 BP (GrA-11818) on a charred hazelnut
shell indicate that this feature belongs to the Boreal period.
Layer 3d below seems to mark the beginning of the
Holocene sequence of the cave. Radiocarbon ages of
9930±50 BP (GrA-14108) and 9810±70 BP (GrN-23130)
from pine charcoal date this layer to the Pre-Boreal period
(Fig. 6).
The material culture assemblages
The finds from layer 2a are clearly attributable to the Vlaška
group, whose appearance in the Trieste Karst is dated c. 6500
BP (5500 cal BC). Similar dates are known for the earliest
sites of the Danilo culture along the Dalmatian coast (Müller
1994) as well as for the Early Neolithic of the Friuli Plain
(Improta & Pessina 1999) west of the Trieste Karst. For ex-
ample, at Sammardenchia di Pozzuolo in Friuli, Danilo cul-
ture materials have been recovered in the shape of vase frag-
ments, a rhyton and a clay phallus (Pessina et al. 1999).
The archaeological material from layer 2a comprises typ-
ical Vlaška potsherds including large pedestalled vessels
253
with restricted mouth and incised linear patterns below the
rim, open black burnished bowls, fragments of a character-
istic four-footed rhyton, and a small pipe-spoon (Biagi et al.
1993: 50). The number of potsherds and their weight, cut by
cut, is presented in Figure 7, which shows that the state of
fragmentation and the quantity of ceramic sherds vary
throughout the entire Vlaška sequence. Given the small
quantity of pottery and its very fragmented state, it is at
present impossible to recognize any variation in the pottery
assemblage through the c. 300 radiocarbon years that this
part of the Neolithic sequence appears to represent.
Nevertheless, it should be noted that most of the pottery from
this layer, including the rhyton and the pipe-spoon, is most
probably of local manufacture. This is demonstrated by thin
section and x-ray diffraction analyses of 20 potsherds, which
revealed the presence of spathic calcite in the fabric — a
characteristic filler used in pottery production in karstic areas
(Spataro 2001).
Layer 3a produced a very similar radiocarbon date. It
contained a hearth with a typical Late Mesolithic
Castelnovian but very specialized lithic assemblage (Fig. 8),
a few potsherds (Biagi et al. 1993: 50) and bones of both wild
and domesticated animals. The faunal assemblage also in-
cludes a significant quantity of shells of edible marine mol-
luscs, including topshell (Monodonta sp.) and limpet (Patella
sp.). The lithic industry, which is produced from pebbles of
good quality local flint, may indicate that this was an area
where a specialized activity of manufacturing trapezoidal ar-
rowheads was undertaken. This is suggested by the very high
frequency of microburins and small flake débitage. Of great
interest is the presence of three very small perforated beads
made of sandstone, a raw material that is unknown in the
area. The occurrence of both potsherds of non-local origin
and domestic animals, as well as the radiocarbon date of
6700±130 BP, point to a very late phase in the development
of the Castelnovian culture. Similar, and sometimes more re-
cent, dates are known from some of the Impressed Ware cul-
ture open-air settlements of the Istrian and Dalmatian coasts
(Müller 1994, 2000).
The lithic assemblages from layers 3c and 3b show strong
similarities both in the tool inventory and in the raw material
used. The materials from just a few square metres have been
examined so far. From layer 3c comes an assemblage mainly
on poor quality material, that is, on bands of black chert
found in the local limestone. Cores are mainly tabular, but
there are also discoid and pebble cores from which very small
flakes had been struck. The most common tools are long,
scalene triangles with straight or concave truncation.
Smaller, isosceles triangles with three retouched sides are
also represented. Other typical tools are the circular and
short, rarely nosed, end-scrapers, as well as the long, very
narrow bilaterally backed points. The microburin technique
is represented by a few specimens. The bone assemblage in-
cludes one long, narrow bone point polished all over its sur-
face (Fig. 8).
The flints from layer 3b comprise the same types as de-
scribed for the palaeosurface of layer 3c, except for the ab-
sence of isosceles triangles and the occurrence of a few tools
on straight bladelets, namely three trapezes and one trunca-
tion. It is important to note that two bone points polished all
 The Iron Gates in Prehistory
Figure 3. Plan of the cave with the position of the excava-
tion grid, the principal profiles (a–b, b–c) and the squares
excavated in each campaign. The dashed line represents the
limit of the modern drip line (drawn by E. Starnini).
over their surface come from this layer (Fig. 9).
The length–width–thickness diagrams of the triangles
from the two layers based on the finds from six square metres
excavated during the 1999 field season, do not suggest any
great difference between the two assemblages. In fact the
measurements taken from 43 triangles of layer 3b and 52 of
layer 3c show an almost identical elongation index, while the
triangles of layer 3b are slightly thicker than those from layer
3c (Fig. 10). The blanks from which these armatures had been
manufactured consist mainly of fine burin spalls (i.e. very
thin blades with triangular section that had been struck using
a burin technique) and, to a lesser extent, microblades
(defined as between 12.5 and 25 mm long). The average
length of the triangular armatures ranges in both cases
between 10 and 20 mm. Some of the microburins are even
smaller. This fact highlights once more the importance of the
retrieval methods, above all the wet sieving technique. No
artefacts have so far been analyzed from layer 3d.
The archaeobotanical and archaeozoological
assemblages
The identification of some 700 charcoals from different lay-
ers of the cave recently published by Nisbet (2000) helps re-
254
Figure 4. Edera Cave: section a–b through the deposits in
the northern sector with the indication of the radiocarbon
dates (drawn by N. Ilić, V. Muncan and P. Biagi).
construct the history of the woodland environment of the
Karst plateau during the last 10,000 years. Charcoals are re-
corded since the Pre-Boreal layer 3d, from which a pine
(Pinus sylvestris/montana) forest is documented with rare
birch (Betula sp.). A very different situation is represented in
the Sauveterrian layers 3c and 3b, where a pine forest is still
dominant accompanied by other species such as oak
(Quercus sp.) and hazel (Corylus sp.). A rapid change seems
to have occurred from layer 3 and the subsequent Early
Neolithic Vlaška occupation where the main species repres-
ented is oak, followed by ash (Fraxinus ornus). Of special
interest is the first appearance of Cotinus coggygria, a shrub
that is common in the local vegetation today.
So far, identification of the animal bones has been limited
to the Late Mesolithic, Neolithic and Copper Age as-
semblages (Boschin & Riedel 2000). An exception is the
analysis of the amphibian and reptilian remains for which a
 Paolo Biagi et al.: The Mesolithic–Neolithic Transition in the Trieste Karst

Figure 5. Edera Cave: A. plan of the Castelnovian hearth of layer 3a; B. plan of the Sauveterrian palaeosurface uncovered
in the Boreal layer 3c. Key: a – yellowish ash, b – ash and charcoals, c – ash, d – black and yellowish soil, e – rocks, f –
marine shells, g – mammalian bones, h – potsherd, i – flints, l – tortoise shell, m – burnt rocks, n – limit of the palaeosurface
(drawn by N. Ilić and E. Starnini).

detailed report is already available (Delfino & Bressi 2000),

Figure 6. Edera Cave: graph of the radiocarbon dates
calibrated using the OxCal (version 3.1) calibration pro-
gram. The dates plot into three distinct groups: Pre-Boreal
(ED16, ED10), Boreal (ED15–ED11) and Early Atlantic
(ED9–ED5).
while the study of the fish bones, which are extremely com-
mon from the Sauveterrian levels upwards, is still in pro-
gress. The identification of the reptiles has shown the relative
importance of the freshwater tortoise (Emys orbicularis)
which was hunted exclusively during the Mesolithic. The
presence of many, sometimes burnt, shell fragments suggests
utilization of this small reptile for dietary purposes. Of great
interest are the mammal bones from the late Castelnovian
hearth in layer 3a. Here, hunting of red deer, pig and roe deer
is attested together with the presence of caprines, cattle and
badger. The occurrence of bones of domestic animals in this
layer is of great interest, even though represented almost ex-
clusively by caprines (6 individuals).
The marine shells identified so far come from the upper
part of the sequence down to layer 3b excavated between
1992 and 1996 in grid squares 3–4 and 8–9. They are partic-
ularly concentrated in layer 3a (between 350 and 360 cm
depth), whilst a smaller amount is recorded in the hearth and
in the underlying layer 3b. As regards the land snails, a very
high number of specimens per square metre have been re-
corded between 360 and 390 cm depth.
Analysis of the marine shells from layer 3a reveals the
importance of marine resources in the diet of this
Castelnovian community. In fact, shells of Patella caerulea
and Monodonta turbinata are extremely common. They are
represented by hundreds of specimens, while other species
such as Gibbula divaricata seem to have been collected only
sporadically. The subsistence economy of the Early Neolithic
Vlaška group community of layer 2a, on the other hand, was

255
 The Iron Gates in Prehistory
Figure 7. Edera Cave: histograms of the number and weight of the potsherds according to layer and depth.
almost exclusively based on the rearing of caprines (24 indi-
viduals) and, to a much smaller extent, cattle (2 individuals).
The collection of marine shellfish was practised sporadically
and played a very minor role in the diet of the Early Neolithic
occupants of the Edera Cave.
Discussion
The main aim of the new excavations at Edera Cave was to
acquire data to enhance our understanding of the environ-
mental changes and human adaptations that characterized the
Trieste Karst/upper Adriatic region between the
Pleistocene/Holocene transition and the Migration Period.
Although the sequence represented in the cave reflects inter-
mittent human occupation rather than continuous and sys-
tematic habitation of the site, the stratigraphic succession re-
vealed by excavation has provided interesting new data on
the Mesolithic and Neolithic periods especially.
A plot of the radiocarbon dates obtained so far from these
layers clearly groups them into three main periods of occu-
pation which can be referred to the Pre-Boreal, Boreal and
Early Atlantic climatic phases, respectively (Fig. 6). Rather
than showing a continuous occupation throughout these
periods, the cave was consistently settled during a well-
defined period of the Boreal, Sauveterrian culture. This is in-
dicated by a horizontal palaeosurface in layer 3c with eph-
emeral charcoal lenses, mammalian bones and lithic artefacts.
Layer 3b, above, had no distinct palaeosurface but con-
tained a few charcoal and ash lenses, one of which has been
radiocarbon dated. Both the flint assemblage and the 14C date
suggest a Boreal/Atlantic transition age for this layer. Of
particular importance is the presence of a limited number of
256
trapezes with retouched truncations, which seem to mark the
start of the Atlantic period at other sites in northern Italy
(Baroni & Biagi 1997) and the Tusco-Apennine chain
(Castelletti et al. 1994).
Although most of the faunal remains from these two layers
are still awaiting identification, the occurrence of several
bones of freshwater fishes as well as of water tortoise (Emys
orbicularis) is remarkable. In fact, both are good indicators
of the presence of abundant freshwater resources in the vi-
cinity of the cave, such as a stream or lake basin which is
absent from the present Karst landscape.
After a long chronological gap the cave was reoccupied
during the beginning of the Atlantic as shown by a well-
defined hearth with a stone base. A charcoal sample from this
hearth gave a date of 6700±130 BP. This is very similar to
dates obtained from the lowermost Early Neolithic Vlaška
group charcoal lens of layer 2a. This hearth seems to repres-
ent an ephemeral occupation of the last hunter-gatherers who
inhabited the Karst region just before the Neolithic.
Nevertheless, some kind of interrelationship may already
have existed between the last hunter-gatherers and the first
farmers who settled the surrounding territory. This is indic-
ated by the presence of a few potsherds which have no typo-
logical link with those of the Vlaška group and which do not
seem to belong to the Impressed Ware culture, at least as
represented in the Istrian sites (Müller 1994). In addition, the
presence of domesticated fauna remains also points to con-
nections between hunter-gatherers and food-producers. The
occurrence of a large number of Patella caerulea and
Monodonta turbinata marine shells confirms the presence of
a rocky coastline not too far from the site, most probably al-
most in the same position as the present coastline.
The Early Neolithic use of the cave seems to have oc-
 Paolo Biagi et al.: The Mesolithic–Neolithic Transition in the Trieste Karst

Figure 8. Edera Cave: flint and bone tools from layer 3a (1–5) and layer 3c (6–17). End-scraper (1), truncations (2, 3),
trapeze (4), denticulated bladelet (5), end-scrapers (6–9), backed blades (10, 11), scalene triangles (12, 13), isosceles tri-
angle (14), microburin (15), core (16) and bone point (17) (drawn by E. Starnini).

curred repeatedly at the cave entrance. Six charcoal lenses
representing a period of some 300 years have been partially
excavated. They all contained a variable quantity of typical
Vlaška group pottery and just a few atypical flints. The great
majority of the pottery is of local manufacture, with the ex-
ception of one sherd of black-burnished ware, very similar to
the Danilo ceramics, which seems to have been imported
from elsewhere.
Although the site distribution pattern is poorly known in
the Karst, and considering that we are presently hampered by
a lack of open-air Mesolithic and Neolithic sites, there appear
to be aspects of the use of the cave which differ through time,
and which can be studied in detail. In fact, studies of the us-
age of the Edera Cave throughout the early Holocene should
provide insights into changes in human adaptation to the en-
vironment, resource usage and exploitation of the environ-
ment surrounding the cave and, to a certain extent, of the
coastline of the Trieste Gulf.
A difference was noted in terms of hearth structures and
attendant fuels between the two main Mesolithic horizons. In
fact, in the Boreal occupation layer hearth constructions are
absent from the excavated area, although there is a palaeo-

257
 The Iron Gates in Prehistory

Figure 9. Edera Cave: flint and bone tools from layer 3b. End-scraper (1), truncation (2), trapezes (3–5), backed points
(6–9), scalene triangles (10, 11), isosceles triangle (12), microburins (13–15), cores (16, 17) and bone points (18, 19) (drawn
by E. Starnini).

surface, mentioned above, that is characterized by ashy areas, Conclusions

remains of red deer and wild boar, as well as debris from
stone tool manufacture (Fig. 5B). In contrast, the early
Atlantic layer had a structured hearth with a stone base and a
probable cooking area (Fig. 5A). The hearth was created by
an apparently smaller group of foragers, in what appears to
have been a single and shorter episode of occupation.
The abundance of the bones of caprines (mainly sheep,
then goat), the presence of charcoal lenses and of dropped
deciduous teeth are evidence that the cave was utilized as a
shelter for both humans and animals at the beginning of the
Neolithic.
The northern Adriatic region (more specifically the Trieste
Karst) and the Iron Gates, are two territories which, though
culturally unrelated and markedly different from geograph-
ical, environmental and climatic points of view, can be seen
as a type of laboratory in which one can examine the trans-
ition to agriculture and its attendant themes. Both regions
have evidence for the active and effective role of Mesolithic
hunter-gatherers who experienced gradual changes in their
subsistence economy through time. What is to be stressed is
the role of the last hunter-gatherers of the Early Atlantic

258
 Paolo Biagi et al.: The Mesolithic–Neolithic Transition in the Trieste Karst

Figure 10. Edera Cave: scatterplots of length:width and length:thickness of the triangles from layers 3b and 3c.

period. Their distribution patterns, technology and subsist-
ence strategies differed from those of the preceding
Sauveterrian (Boreal or Pre-Boreal) foragers who inhabited
the area prior to the beginning of the eighth millennium BP.
Castelnovian sites are far less abundant in the Karst region
than those of the preceding Sauveterrian. Furthermore, the
known Castelnovian sites seem to belong to early stages in
the development of the culture (Montagnari Kokelj 1993) or
rather brief occupations, as in the case of the Edera Cave
hearth which is the only feature of this type and age ever re-
corded from the Karst caves.
There is little doubt that interaction occurred between the
last hunter-gathers of the Castelnovian culture and the first
Neolithic farmers that might have led to an acculturation of
the former or at least reduced their visibility in the archae-
ological record. We still face the main question of when, why
and how such interaction is represented archaeologically.
That is, can we recognize the possible introduction (or ac-
ceptance) of domesticated plants and animals as well as the
presence of (single) specific objects reflecting new technolo-
gies acquired from elsewhere?
This question is complicated by the apparent fact that
three distinct archaeological cultures were present in the re-
gion of Edera Cave around the middle of the seventh millen-
nium BP (Biagi 2003): the late Castelnovian culture, the
Vlaška group, and the Impressed Ware culture. Generally, the
radiocarbon dates and site distributions of the various cul-
tures overlap in this region, but specific archaeological asso-
ciations are elusive. The occurrence of potsherds of non-local
production and bones of domesticated animals in the late
Castelnovian (layer 3a) of Edera Cave is such an association.
Further analysis of these finds should provide a clearer inter-
pretation and understanding of the human interactions in the
prehistoric karst.

259
 The Iron Gates in Prehistory
References
Andreolotti, S. & Stradi, F. 1963: L’industria mesolitica della
Cavernetta della Trincea in Val Rosandra. Atti e Memorie della
Commissione Grotte E. Boegan (Trieste) 3: 71–85.
Baroni, C. & Biagi, P. (eds) 1997: Excavations at the High Altitude
Mesolithic site of Laghetti del Crestoso (Bovegno, Brescia –
Northern Italy). Brescia: Ateneo di Brescia, Accademia di
Scienze, Lettere ed Arti.
Biagi, P. 2003: New data on the Early Neolithic of the Upper
Adriatic. In Nikolova, L. (ed.) Early Symbolic Systems for
Communication in Southeast Europe, Volume 2. Oxford:
Hadrian Books (BAR S1139), 337–346.
Biagi, P. & Voytek, B.A. 1994: The neolithization of the Trieste
Karst in north-eastern Italy and its neighbouring countries. A
Nyíregyházi Jósa András Múzeum Évkönyve (Nyíregyháza,
Hungary) 36: 63–73.
Biagi, P., Starnini, E. & Voytek, B.A. 1993: The Late Mesolithic and
Early Neolithic settlement of northern Italy: recent considera-
tions. Poročilo o raziskovanju paleolita, neolita in eneolita v
Sloveniji (Ljubljana) 21: 45–67.
Boschian, G. & Pitti, C. 1984: I livelli mesolitici della Grotta
dell’Edera. In Il Mesolitico sul Carso Triestino. Società per la
Preistoria e Protostoria della Regione Friuli-Venezia Giulia
(Trieste) Quaderno 5: 143–210.
Boschin, F. & Riedel, A. 2000: The Neolithic and Late Mesolithic
fauna of the Edera Cave: a preliminary report. In Biagi, P. (ed.)
‘Studi sul Paleolitico, Mesolitico e Neolitico del bacino
dell’Adriatico in ricordo di Antonio M. Radmilli’. Società per la
Preistoria e Protostoria della Regione Friuli-Venezia Giulia
(Trieste) Quaderno 8: 73–90.
Broglio, A. 1971: Risultati preliminari delle ricerche sui complessi
epipaleolitici della Valle dell’Adige. Preistoria Alpina (Trento)
7: 135–241.
Castelletti, L., Maspero, A. & Tozzi, C. 1994: Il popolamento della
Valle del Serchio (Toscana settentrionale) durante il
Tardiglaciale Würmiano e l’Olocene antico. In Biagi, P. &
Nandris, J. (eds) Highland Zone Exploitation in Southern
Europe. Brescia: Monografie di Natura Bresciana (Brescia) 20,
189–204.
Ciccone, A. 1992: L’industria litica della Grotta Azzurra di
Samatorza: scavi 1982. Atti della Società per la Preistoria e
Protostoria della Regione Friuli-Venezia Giulia (Trieste) 7:
13–45.
Cremonesi, G. 1984: I livelli mesolitici della Grotta della Tartaruga.
Il Mesolitico sul Carso Triestino. Società per la Preistoria e
Protostoria della Regione Friuli-Venezia Giulia (Trieste)
Quaderno 5: 65–107.
260
Cremonesi, G., Meluzzi, C., Pitti, C. & Wilkens, B. 1984a: Grotta
Azzurra: Scavi 1982: Il Mesolitico sul Carso Triestino. Società
per la Preistoria e Protostoria della Regione Friuli-Venezia
Giulia (Trieste) Quaderno 5: 21–64.
Cremonesi, G., Pitti, C. & Radmilli, A.M. 1984b: Considerazioni
sul Mesolitico del Carso Triestino. Il Mesolitico sul Carso
Triestino. Società per la Preistoria e Protostoria della Regione
Friuli-Venezia Giulia (Trieste) Quaderno 5: 229–240.
Delfino, M. & Bressi, N. 2000: L’erpetofauna olocenica proveniente
dalla Grotta dell’Edera (Trieste). In Biagi, P. (ed.) ‘Studi sul
Paleolitico, Mesolitico e Neolitico del bacino dell’Adriatico in
ricordo di Antonio M. Radmilli’. Società per la Preistoria e
Protostoria della Regione Friuli-Venezia Giulia (Trieste)
Quaderno 8: 115–125.
Improta, S. & Pessina, A. 1999: La neolitizzazione dell’Italia
settentrionale. Il nuovo quadro cronologico. In Pessina, A. &
Muscio, G. (eds) Settemila anni fa il primo pane. Ambienti e
culture delle società neolitiche. Udine: Museo Friulano di Storia
Naturale, 107–116.
Marzolini, G. 1970: La Grotta dell’Edera. Annali del Gruppo Grotte
Associazione XXX Ottobre (Trieste) 4: 19–35.
Montagnari Kokelj, E. 1993: The transition from Mesolithic to
Neolithic in the Trieste Karst. Poročilo o raziskovanju paleolita,
neolita in eneolita v Sloveniji (Ljubljana) 21: 69–84.
Müller, J. 1994: Das Ostadriatische Frühneolithikum. Die
Impresso-Kultur und die Neolithisierung des Adriaraumes
(Prähistorische Archäologie in Südosteuropa 9). Berlin: Volker
Spiess.
— 2000: ‘Earplugs’ ceramics and sheep: examples of communica-
tion and boundaries in the Adriatic Early Neolithic. In Biagi, P.
(ed.) ‘Studi sul Paleolitico, Mesolitico e Neolitico del bacino
dell’Adriatico in ricordo di Antonio M. Radmilli’. Società per la
Preistoria e Protostoria della Regione Friuli-Venezia Giulia
(Trieste) Quaderno 8: 151–159.
Nisbet, R. 2000: Nota preliminare sull’antracologia dei depositi
olocenici della Grotta dell’Edera, Carso Triestino (scavi
1990–1999). In Biagi, P. (ed.) ‘Studi sul Paleolitico, Mesolitico e
Neolitico del bacino dell’Adriatico in ricordo di Antonio M.
Radmilli’. Società per la Preistoria e Protostoria della Regione
Friuli-Venezia Giulia (Trieste) Quaderno 8: 161–170.
Pessina, A., Ferrari, A. & Fontana, A. 1999: Le prime popolazioni
agricole del Friuli. In Pessina, A. & Muscio, G. (eds) Settemila
anni fa il primo pane. Ambienti e culture delle società neolitiche.
Udine: Museo Friulano di Storia Naturale, 133–146.
Spataro, M. 2001: An interpretative approach to the prehistory of the
Edera Cave in the Trieste Karst (northeastern Italy): the archae-
ometry of the ceramic assemblage. Accordia Research Papers 8:
83–99.