Baylissand Woodman 2009

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 25

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/236010719

A New Bayesian Chronology for Mesolithic Occupation at Mount Sandel,


Northern Ireland

Article  in  Proceedings of the Prehistoric Society · January 2009


DOI: 10.1017/S0079497X00000311

CITATIONS READS

36 1,989

2 authors:

Alex Bayliss Peter C. Woodman


Historic England University College Cork
224 PUBLICATIONS   26,428 CITATIONS    114 PUBLICATIONS   1,087 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Project Radiocarbon View project

High-precision radiocarbon wiggle-matching by AMS of wooden built heritage View project

All content following this page was uploaded by Alex Bayliss on 05 January 2014.

The user has requested enhancement of the downloaded file.


Bayliss and Woodman:Wessex Article Master Page 4/1/10 10:18 Page 101

Proceedings of the Prehistoric Society 75, 2009, pp. 101–123

A New Bayesian Chronology for Mesolithic Occupation at


Mount Sandel, Northern Ireland

By ALEX BAYLISS1 and PETER WOODMAN2

Mount Sandel has long been an iconic site for the Irish Mesolithic, having produced evidence for a sequence of
occupation huts and pits and the earliest radiocarbon dates for the Mesolithic on the island. This paper presents
details of a recent programme of redating whereby the application of Bayesian modelling has confirmed the early
date for the site but also helped to refine its internal chronology. The major phase of hut building at Mount Sandel
took place within a much shorter period of time than had previously been thought, perhaps only a generation or
two. The dating of pits of differing sizes suggests that many of them were created during other slightly later visits
to the area. The implications of the dating programme for the place of Mount Sandel in the Irish Mesolithic, and
for the chronology of the period and its relations with that on the British mainland, are discussed.

INTRODUCTION geometric microliths was recovered along with core


During the 1970s extensive excavations were and flake axes and other tools. This material became
undertaken adjacent to Mount Sandel Fort, Co. Derry, the type assemblage for what came to be called the
Northern Ireland. These centred on a series of Irish Early Mesolithic. Besides demonstrating the
Mesolithic settlement sites on a 30 m high bluff existence of substantial structures and, because of the
forming the eastern edge of the estuary of the River occurrence of concentrations of carbonised
Bann (Fig. 1; Irish National Grid, 28545 43075). The environmental evidence, Mount Sandel has been
site lay at least 40 m from the edge of the escarpment important in understanding the life-ways of the Irish
and its most significant component was the remnants Mesolithic. Even though a series of little pits, as well
of a number of circular structures defined by stake- as one of a group of larger deep pits, cut across the
holes and sunken hearths (Fig. 2). These were main hut area (Woodman 1985, 129, fig. 65), these
recovered upon the removal of areas of ‘occupation pits were interpreted as being somehow associated
soil’. A series of little and large pits that cut across the with phases of hut building and occupation that were
main area of huts was also found. Around that core not so well preserved. In other words, each hut was
area extensive agricultural activity had removed the thought to be associated with a number of little pits
upper levels of the B Horizon, but it was still possible and one larger pit which, it was suggested, was used
to excavate a series of heavily damaged pits and stake- for storage.
holes that lay up to 15 m from the main hut area (Fig. The site has also conventionally been regarded as
3; see Woodman 1985 for a full description of the producing the earliest unequivocal traces of human
site). A lithic assemblage dominated by so-called settlement in Ireland. After the excavation, a
substantial series of 15 conventional radiocarbon
determinations was obtained on bulk samples of
1 English Heritage, 1 Waterhouse Square, 138–142 charred plant remains. These measurements lay
Holborn, London, EC1N 2ST beyond the range of the tree-ring calibration data at
2 6 Brighton Villas, University College Cork, Cork, Ireland
the time of the site publication (eg, Pearson et al.
(pwoodman@ucc.ie)
1983), and so were interpreted on the basis of the
uncalibrated measurements. With the exception of
Received: October 2008. Accepted: February 2009 UB-2359 (F109), which was an outlier on both

101
Bayliss and Woodman:Wessex Article Master Page 4/1/10 10:18 Page 102

THE PREHISTORIC SOCIETY

Fig. 1.
Oblique aerial photograph showing location of Mount Sandel. Photograph courtesy of Fugro-BKS, Coleraine

chronological and spatial grounds, the dates obtained associated with the huts.
from the site lay within the period 9000–8400 In the intervening years since the Mount Sandel
radiocarbon years BP (Woodman 1985, 148; Fig. 4). excavation a number of other early Mesolithic sites
The major concentration of these dates lay between with the same general range of microliths have been
8700 and almost 9000 radiocarbon years BP and was found throughout Ireland. These sites, most notably

102
Bayliss and Woodman:Wessex Article Master Page 4/1/10 10:18 Page 103

A. Bayliss & P. Woodman. NEW BAYESIAN CHRONOLOGY FOR MESOLITHIC OCCUPATION AT MOUNT SANDEL, N. IRELAND

Fig. 2.
Photograph showing the area of the huts

the shore line settlement site at Lough Boora (Ryan Mellars 1974). Conventionally the appearance of this
1980) found in the midlands of Ireland and the technology marks the transition to the later
cremation burials at Hermitage in south-west Ireland Mesolithic of Britain. It was also noted that the
(Collins & Coyne 2006), clearly demonstrate that the earliest dates obtained from Mount Sandel were from
Irish Early Mesolithic existed throughout Ireland. charcoal samples (notably UB-952, -2357, -2362),
None of these assemblages pre-dates Mount Sandel, while those containing large quantities of hazelnuts
however (Woodman in press), because of the limited tended to be somewhat later. Although the slightly
number of dated sites from the Early Mesolithic, there earlier results from the charcoal samples raised the
is as yet no clear evidence about which part of Ireland possibility of an old-wood offset, several of the dates
was first settled. on hazelnut shell were stratigraphically later.
Despite such a coherent set of results, even in the Woodman (2004, 290–1) also discussed the issue of
original report some concerns were expressed about possible inter-laboratory bias in relation to the
the accuracy of the radiocarbon dating. Some of the accuracy of the Mount Sandel dating.
dates obtained appeared to be slightly earlier than The chronology of the site was recently placed in
expected for the so-called narrow blade or geometric sharp focus by the new excavation of huts at Howick
technology in Britain (op. cit. 146–7 and 171), which, in Northumbria (Waddington 2007) and East Barnes
at the time, was thought to lie, at its earliest, between in south-eastern Scotland (Gooder 2007). Both sites
8800 and 8600 radiocarbon years BP (Jacobi 1973; have produced dates that fall within the existing

103
Bayliss and Woodman:Wessex Article Master Page 4/1/10 10:18 Page 104

THE PREHISTORIC SOCIETY

Fig. 3.
Overall plan of excavated areas at Mount Sandel, showing the location of the outlying radiocarbon samples

radiocarbon chronology for Mount Sandel, although THE NEW RADIOCARBON DATING PROGRAMME
the greater precision now available from radiocarbon It was thus apparent that a number of separate, but
dating for these new sites highlighted the limitations related, issues needed to be investigated as part of a
of the chronology provided by measurements made new dating programme. First the concerns regarding
more than 20 years ago. Not only do many of the the accuracy of the initial dates from Mount Sandel
results from Mount Sandel have errors of 100 years or needed to be addressed (eg, the possibility of an old-
more (at one standard deviation), but sample choice wood effect or inter-laboratory bias). Second AMS
for the original programme was severely restricted by dating would enable the potential homogeneity of the
the sample-size demanded by conventional bulk samples originally dated to be assessed and new
radiocarbon dating. Indeed the availability of a single-entity samples of short-life material to be dated
sample of the required size often determined what was (Ashmore 1999). Another major impetus for the new
dated, rather than the situation today where the dating programme was the availability of Bayesian
availability of Accelerator Mass Spectrometry allows analysis to combine the relative dating evidence from
series of small samples to be dated that relate directly the site provided by stratigraphy with the radiocarbon
to specific archaeological questions (Lanting & van dates, and to overcome the statistical scatter inherent
der Plicht 1994). in any suite of determinations (Bayliss et al. 2007a).

104
Bayliss and Woodman:Wessex Article Master Page 4/1/10 10:18 Page 105

A. Bayliss & P. Woodman. NEW BAYESIAN CHRONOLOGY FOR MESOLITHIC OCCUPATION AT MOUNT SANDEL, N. IRELAND

Fig. 4.
Uncalibrated radiocarbon determinations from Upper Mount Sandel available in 1985 (Woodman 1985, table 85)

Combined with a rigorous assessment of sample sampling strategy was also informed by a series of
taphonomy, this allows much more refined questions Bayesian simulation models which provided an idea of
of chronology to be addressed. the resolution that might be obtainable from different
New samples were therefore selected to address an dating strategies. Forty-five samples have now been
explicit series of objectives: dated from Mount Sandel, with six contexts being
dated in replicate. The locations of the dated contexts
• to provide more precise and robust dating for the are shown in Figures 3 and 5. Fifteen measurements
main hut, were obtained on samples of bulk charred plant
• to estimate the duration of the use of this remains in the late 1970s and early 1980s. Thirteen of
structure, these were dated by the Queen’s University, Belfast
• to check the chronological relationship between radiocarbon laboratory. All of these were pretreated
the various features found on the site, namely to as described by Smith et al. (1971, 103). Four samples
determine whether each hut was associated with (UB-912–3 and UB-951–2) were then converted to
a particular set of large and little pits, and methane and dated by gas proportional counting as
• to determine whether the outlying pits to the east described by Smith et al. (1970, 285–6). The
and west of the main hut area were part of the remaining nine samples processed by Belfast at this
main period of site use. time were converted to benzene and dated by liquid
scintillation spectrometry as described by Pearson
Answering these chronological questions would allow (1984). The other two samples were dated by the
Mount Sandel to be placed more firmly within the Centre for Isotope Research, Rijksuniversiteit,
early Mesolithic of Ireland, and allow a reassessment Groningen. These were pretreated using the
of the original interpretation of the activities acid/base/alkali method (Mook & Waterbolk 1985),
undertaken on the site. but were converted to carbon dioxide and dated by
Potential samples were assessed individually on the gas proportional counting (Mook & Streurman
basis of our archaeological judgement of their 1983). Between 2004 and 2007, 30 additional
association with these questions, but the overall measurements were obtained from the Queen’s

105
TABLE 1: RADIOCARBON DETERMINATIONS FROM UPPER MOUNT SANDEL (*MEASURED BEFORE 1985)
Lab. No. Sample Context & material Radiocarbon χ2 test δ13C Calibrated date range Posterior density estimate
Ref. Age (BP) (‰) cal BC (95% confidence) cal BC (95% probability)
Main Hut area
UB-2357* MtsH unident. charcoal, 8955±185 8610–7590 7730–7605
main occupation Hut 1
UB-7029 Mts15 hazelnut shell, main 8658±44 -24.0 7755–7585 7730–7605
occupation Hut 1
UB-2356* MtsG unident. charcoal, main 8765±135 8275–7550 7725–7605
occupation Hut 1
UB-7037 Mts23 hazelnut shell, main 8645±45 -24.0 7750–7580 7725–7605
occupation Hut 1
UB-2358* MtsI charcoal & hazelnut shells, 8795±135 T’=0.7; 7935–7590 730–7615
hearth 56/4 T’(5%)=3.8; ν=1
UB-6842 Mts1 hazelnut shell, hearth 56/4 8672±53 -24.0
UB-6844 Mts3 hazelnut shell, hearth 56/3 8697±52 -26.0 7940–7590 7720–7610
Bayliss and Woodman:Wessex Article Master Page

UB-6843 Mts2 hazelnut shell, hearth 56/2 8775±53 -26.0 8180–7605 7710–7605
UB-6847 Mts3 hazelnut shell, hearth 56/6 8724±53 -27.0 7960–7595 7745–7615
UB-952* MtsD charcoal & hazelnut shells, 8960±70 T’=8.9; 8295–7940 8295–7560
hearth 56/5 T’(5%)=3.8; ν=1
4/1/10

UB-6845 Mts4 hazelnut shell, hearth 58/5 8699±53 -26.0 7945–7590 7705–7600
GrN-10471* MtsO carbonised plant remains, 8430±60 T’=10.9; 7590–7350 7590–7505
layer 3, pit 56/1 T’(5%)=7.8; ν=3
UB-7035 Mts21 unident. charcoal, layer 3, 8509±43 -25.0 7600–7510 7595–7525
10:18

pit 56/1

106
UB-951* MtsC unidentified charcoal from 8790±185 8315–7515 8290–7565
layer 3 in pit 56/1
UB-7034 Mts20 hazelnut shell, layer 3, 8647±47 -28.0 7750–7580 7755–7580
pit 56/1
Page 106

THE PREHISTORIC SOCIETY

UB-2360* MtsK unident. charcoal & hazelnut 8670±100 8170–7535 8170–8110 (3%)
shells, layer 1, pit 56/1 or 7985–7540 (92%)
UB-913* MtsB hazelnut shells & unident. 8555±70 T’=0.4; 7650–7570 7645–7545
charcoal, hearth 31/0 T’(5%)=3.8; ν=1
UB-6848 Mts6 hazelnut shell, hearth 31/0 8608±52 -26.0
UB-6854 Mts12 hazelnut shell, small pit F23 8578±52 -25.0 7655–7535 7650–7570
UB-7036 Mts22 hazelnut shell, small pit F22 8636±45 -27.0 7745–7580 7650–7580
Little pits
UB-6849 Mts7 hazelnut shell, pit 31/3 8644±54 T’=1.1; -28.0 7790–7585 7680–7585
T’(5%)=6.0; ν=2
UB-2007* MtsF hazelnut shells & unident. 8795±135
charcoal, pit 31/3
UB-6852 Mts10 hazelnut shell, small pit F102 8674±51 -28.0 7820–7585 7680–7585
UB-6855 Mts13 hazelnut shell, small pit F33 8569±51 -24.0 7650–7535 7655–7565
UB-7032 Mts18 hazelnut shell, small pit 34 8631±50 -24.0 7750–7575 7675–7580
UB-7031 Mts17 hazelnut shell, small pit 8684±45 -27.0 7815–7590 7680–7590
F100/6
UB-6856 Mts14 hazelnut shell, small pit 8672±53 -25.0 7820–7585 7680–7585
F112
TABLE 1 : RADIOCARBON DETERMINATIONS FROM UPPER MOUNT SANDEL (*MEASURED BEFORE 1985)
Lab. No. Sample Context & material Radiocarbon χ2 test δ13C Calibrated date range Posterior density estimate
Ref. Age (BP) (‰) cal BC (95% confidence) cal BC (95% probability)
Other Features
UB-2362* MtsM unident. charcoal, 8990±80 T’=26.8; 8305–7950 8330–7935 (93%) or
layer 1, F99 T'(5%)=3.8; ν=2 7895–7835 (2%)
UB-6851 Mts9 hazelnut shell, 8502±52 -24.0 7600–7490 7600–7520
layer 1, F99
UB-6850 Mts8 hazelnut shell, hearth 83/b 8645±55 -27.0 7785–7575 7725–7580
UB-6853 Mts11 hazelnut shell, pit F91 8682±51 -25.0 7825–7585 7730–7590
UB-2008* MtsE hazelnut shells & charcoal, 8440±65 7590–7350 7595–7505
pit F27
UB-7033 Mts19 hazelnut shell, 8660±44 -24.0 7780–7585 7720–7585
possible hearth F36
GrN-10470* MtsN carbonised plant remains, 8380±50 7550–7330 7585–7500
layer 2, pit F100/2
UB-912* MtsA hazelnut & unident. 8725±115 8230–7550 7745–7565
charcoal, layer 4, pit F100/2
Bayliss and Woodman:Wessex Article Master Page

Outlying features
UB-8872 Mts30 unident. charcoal, layer 4, 8655±37 -25.3 7750–7580 7740–7590
pit F74
UB-2361* MtsL unident. charcoal, layer 2, 8545±165 8170–7180 8180–8105 (4%) or
4/1/10

A. Bayliss & P. Woodman. NEW BAYESIAN CHRONOLOGY

F74 8065–8035 (1%) or


8000–7505 (90%)
FOR

107
10:18

UB-7030 Mts16 hazelnut shell, pit F210 8594±44 -24.0 7655–7570 7705–7695 (1%) or
7685–7545 (94%)
UB-7038 Mts24 hazelnut shell, pit F102 8678±41 -23.0 7790–7590 7725–7590
UB-8871 Mts29 hazelnut shell, pit F126 8681±49 -24.3 7820–7580 7730–7590
UB-8138 Mts26 hazelnut, layer 4, pit F209c 8566±52 -25.7 7640–7530 7675–7530
Page 107

UB-8139 Mts27 hazelnut, post-hole 991 8757±51 -25.4 8170–7600 7750–7595


UB-8137 Mts25 hazelnut, layer 3, pit F211 8680±44 -29.3 7800–7590 7725–7590
UB-8140 Mts28 unident. charcoal, 8660±44 -28.4 7780–7580 7760–7585
post-hole 992
MESOLITHIC OCCUPATION

UB-2359* MtsJ unident. charcoal, 7885±120 7080–6465 –


AT

hearth F109
MOUNT SANDEL, N. IRELAND
Bayliss and Woodman:Wessex Article Master Page 4/1/10 10:18 Page 108

THE PREHISTORIC SOCIETY

Fig. 5.
Site plan showing the locations of the dated samples in the central area of the excavation, the grey shading indicates the
location of the grid squares from which samples from the occupation layer were submitted

University, Belfast radiocarbon laboratory. These were The calibrated date ranges provided in Table 1 and
pretreated as described by Mook and Waterbolk given in normal type have been calculated using the
(1985) and converted to graphite using the method of maximum intercept method (Stuiver & Reimer 1986).
Slota et al. (1987). The samples were then dated at the The distributions of calibrated dates shown in outline
Oxford Radiocarbon Accelerator Unit by AMS using in the graphs (Figs 6–10) have been calculated using
methods outlined in Bronk Ramsey et al. (2004). the probability method (Stuiver & Reimer 1993).
Full details of all the radiocarbon measurements Replicate measurements from the same sample have
from Upper Mount Sandel are provided in Table 1. All been tested for statistically consistency and, where
are conventional radiocarbon ages (Stuiver & Polach appropriate, combined by taking a weighted mean
1977). They have been calibrated using the before calibration using the methods described by
program OxCal v3.10 (Bronk Ramsey 1995; 1998; Ward and Wilson (1978).
2001) and IntCal04 (Reimer et al. 2004).

108
Bayliss and Woodman:Wessex Article Master Page 4/1/10 10:18 Page 109

A. Bayliss & P. Woodman. NEW BAYESIAN CHRONOLOGY FOR MESOLITHIC OCCUPATION AT MOUNT SANDEL, N. IRELAND

Fig. 6.
Overall structure of the chronological model for Upper Mount Sandel. The component sections of this model are shown
in detail in Figs 7–10. The large square brackets down the left hand side of Figs 6–10 along with the OxCal keywords
define the overall model exactly

CHRONOLOGICAL MODELLING than the main phase of use of the site), UB-951-2, UB-
The Bayesian approach to the interpretation of 2360–1, UB-2362, UB-7035, and UB-8872 (which are
archaeological chronologies has been described by samples containing unidentified charcoal which
Buck et al. (1996). It is based on the principle that exhibit an old-wood offset), and UB-7034 (which is
although the calibrated age ranges of radiocarbon residual). All of these measurements, however (except
measurements accurately estimate the calendar ages of for UB-2359), are used to calculate the estimate of
the samples themselves, it is the dates of when the main phase of use of the site ended (‘end
archaeological events associated with those samples Mount Sandel’; Fig. 6), since they all provide termini
that are important. Bayesian techniques can provide post quem (that is maximum ages only) for the
realistic estimates of the dates of such events by contexts from which they were recovered. By
combining absolute dating evidence, such as comparing estimates such as these, it is possible to
radiocarbon results, with relative dating evidence, calculate new probability distributions which estimate
such as stratigraphic relationships between the durations of phases of activity (eg, ‘occupation
radiocarbon samples. These ‘posterior density Mount Sandel’; Fig. 11).
estimates’ (which, by convention, are always The technique used here is a form of Markov Chain
expressed in italics) are not absolute. They are Monte Carlo sampling, which has been applied using
interpretative estimates, which will change as the program OxCal v3.10. Details of the algorithms
additional data become available or as the existing employed by this program are available from the on-
data are modelled from different perspectives. line manual or in Bronk Ramsey (1995; 1998; 2001).
It is also possible to calculate distributions for the The algorithms used in the models described below
dates of events that have not been dated directly by can be derived from the structure shown in Figures
radiocarbon measurements, such as the beginning and 6–10. Full details of the archaeological information
end of a continuous phase of activity (which is which has been included in the model are provided
represented by a number of radiocarbon results). For below, and further explanation of the Bayesian
example, the posterior density estimate ‘start Mount approach to chronological modelling in archaeology
Sandel’ (Fig. 6) has been calculated using all the is provided in Bayliss et al. (2007a). Worked examples
radiocarbon dates from the site, except for UB-2359 demonstrating the application of this approach are
(which dates Mesolithic activity considerably later provided by the series of papers by Buck et al. (1991;

109
Bayliss and Woodman:Wessex Article Master Page 4/1/10 10:18 Page 110

THE PREHISTORIC SOCIETY

1992; 1994). Archaeological problems tackled by and UB-2358 from hearth 56/4). This approach is
integrating radiocarbon evidence with other forms of consistent with the assumptions inherent in dating
chronological information are becoming common in bulk samples. Where the results on such samples are
the literature (see for example Bayliss et al. 1997; consistent with those on short-life material, but where
Needham et al. 1998), although applications from the the dated deposit may actually form a period of
Mesolithic period are still rare (Blockley et al. 2004; activity, the dates have all been included in the model
Dark et al. 2006; Bayliss et al. 2007b). but a weighted mean has not been taken before
calibration (eg, UB-2356–7, UB-7029, and UB-7037
from the occupation layer in the main hut area).
Where results on unidentified or partially identified
samples are statistically inconsistent with those on
THE BAYESIAN MODEL short-life material from the same deposit, or where
The overall structure of the chronological model for repeat samples or other information that may suggest
Upper Mount Sandel is shown in Figure 6, with the that these results are reliable is not available, then
components relating to the main hut area, little pits, these measurements have been included in the model
and other and outlying features shown in as termini post quem only (eg, UB-952).
Figures 7–10. The brackets down the left hand side of Figures
We have interpreted the occupation at Mount 6–10, along with the chronological query language
Sandel as forming a single, continuous phase of commands, define the model for the chronology of
archaeological activity (Fig. 6; Buck et al. 1992). Mount Sandel exactly. Here, we discuss the
Additionally, the occupation of the main hut is archaeological interpretations which have led us to
interpreted as another single, continuous sub-phase of construct the model in this way.
activity within this period (Fig. 7), as is the period of
use of the little pits (Fig. 8). These statistical
distributions form the ‘uninformative’ prior Main hut area
information component of the model. The sequence of hearths at the centre of the main hut
We have only included direct stratigraphic was dated (Fig. 7). A single hazelnut fragment was
relationships between dated contexts as the dated from what appeared, stratigraphically, to be the
‘informative’ prior information component of the earliest hearth (56/6; UB-6847). Above this, two
model presented here. Critically, as stratigraphy statistically consistent radiocarbon determinations
provides relative dating for archaeological deposits were obtained on a single hazelnut shell (UB-6842)
and radiocarbon dates are obtained on samples of and a bulk sample of unidentified charcoal and
organic material from those deposits, our inclusion of hazelnut shell (UB-2358) from hearth 56/4. A
such information depends both on our understanding weighted mean has been taken of these measurements
of the site stratigraphy and our assessment of the (56/4). The consistency of these measurements
taphonomy of the dated material. suggests that the charcoal component of the bulk
These archaeological interpretations underpin our sample did not contain a significant proportion of old
choices of how to incorporate stratigraphic wood. Two single hazelnut shells were also dated
information into the model, and also of how to from further up this sequence of hearths – UB-6844
incorporate the radiocarbon dates (the ‘standardised from 56/3 and UB-6843 from 56/2. Later than these
likelihoods’ component of the model). We have to are two samples from hearth 56/5, which produced
deal both with dates on bulked, multiple-entity, two statistically inconsistent radiocarbon
samples and also with samples that may incorporate measurements (UB-6845 and UB-952). The bulk
an old-wood offset. Our approach has been sample of unidentified charcoal and hazelnuts
pragmatic. Where a result on an unidentified or produced a result that is significantly older than the
partially identified sample is statistically consistent result from the single fragment of hazelnut shell dated
with one made on short-life material from the same in the recent programme. UB-952 therefore
deposit, we have combined the results before probably contained an appreciable amount of old
calibration if the samples derive from what may be wood and so only provides a terminus post quem for
interpreted as a ‘single-event’ context (eg, UB-6842 the use of this hearth.

110
Bayliss and Woodman:Wessex Article Master Page 4/1/10 10:18 Page 111

A. Bayliss & P. Woodman. NEW BAYESIAN CHRONOLOGY FOR MESOLITHIC OCCUPATION AT MOUNT SANDEL, N. IRELAND

Fig. 7.
Probability distributions of dates from the main hut area at Upper Mount Sandel. Each distribution represents the
relative probability that an event occurred at a particular time. For each of the dates two distributions have been plotted,
one in outline which is the result produced by the scientific evidence alone, and a solid one which is based on the
chronological model used. The other distributions correspond to aspects of the model. For example, the distribution
‘start hut 1’ is the estimated date when hut 1 was constructed. Dates followed by ? have been excluded from the model
for reasons explained in the text. The figures followed by % are the individual indices of agreement which provide an
indication of the consistency of the radiocarbon dates with the prior information included in the model
(Bronk Ramsey 1995)

111
Bayliss and Woodman:Wessex Article Master Page 4/1/10 10:18 Page 112

THE PREHISTORIC SOCIETY

Fig. 8.
Probability distributions of dates from the little pits. The format is identical to that of Fig. 7

Four samples have been dated from the occupation or reworked charcoal from the underlying phases of
layer associated with the hollow that contained the occupation in the hut. UB-7034; a single fragment of
main hut, two single fragments of hazelnut shell (UB- hazelnut shell must be residual. UB-2360 from the
7029 and UB-7037) and two bulk samples of overlying layer 1 appears to have included reworked

statistically consistent (T’=3.3; T’(5%)=7.8; ν=3),


unidentified charcoal (UB-2356–7). The results are material or charcoal with a significant age-offset. It
therefore only provides a terminus post quem for the
suggesting that the charcoal samples did not contain end of occupation at the site.
an appreciable quantity of old wood. Feature 31/0 is a hearth that cuts across a line of
Some time after the hut went out of use a pit (56/1), stake-holes delimiting the eastern edge of the main hut
over a metre across and more than 0.6 m deep, was area. Two statistically consistent results were obtained
cut through the edge of the hearth sequence. Samples from a single fragment of hazelnut (UB-6848) and
from two of the fills of this pit have been dated. Four from a bulk sample of burnt hazelnut shells and some

obtained from layer 3 (T’=10.9; T’(5%)=7.8; ν=3).


statistically inconsistent measurement have been unidentified charcoal (UB-913). This unidentified
charcoal was probably not long-lived, and so the
GrN-10471, the latest determination made on weighted mean of these results (F31/0) provides the
carbonised plant remains, probably provides the most most reliable estimate for the date of this feature.
reliable indication of the date of this feature. This Single fragments of hazelnut shell were also dated

T’(5%)=3.8; ν=1) but, as this latter sample was


result is statistically consistent with UB-7035 (T’=1.1; from little pits F20 (UB-7036) and F23 (UB-6854),
which were cut through the occupation layer
unidentified charcoal and GrN-10471 has only been (Woodman 1985, fig. 65).
partially identified, we have erred on the side of
caution and included UB-7035 in the model only as a
terminus post quem for later activity. UB-951 was a Little pits
bulk sample of unidentified charcoal and appears to Samples from eight of the little pits that were found on
have contained a considerable proportion of old wood the site were dated (Fig. 8). These were little, shallow

112
Bayliss and Woodman:Wessex Article Master Page 4/1/10 10:18 Page 113

A. Bayliss & P. Woodman. NEW BAYESIAN CHRONOLOGY FOR MESOLITHIC OCCUPATION AT MOUNT SANDEL, N. IRELAND

Fig. 9.
Probability distributions of dates from other features. The format is identical to that of Fig. 7

Fig. 10.
Probability distributions of dates from outlying features. The format is identical to that of Fig. 7

113
Bayliss and Woodman:Wessex Article Master Page 4/1/10 10:18 Page 114

THE PREHISTORIC SOCIETY

Fig. 11.
Probability distributions for the number of years during which different activities occurred at Mount Sandel, derived
from the model defined in Figs 6–10

pits that were normally between 0.8 m and 0.5 m in unlikely that the small charcoal component in this
diameter. Sometimes these were found on their own, sample (UB-2008) created a significant old-wood
but sometimes in little intercutting groups (eg, effect, especially as this result is one of the youngest
F31/2–5). Eight single fragments of hazelnut shell associated with the period of occupation on the site.
were dated, one from each of the sampled pits. One In approximately the same area, pit F100/2, produced
feature has two statistically consistent radiocarbon two samples from a sequence of fills. UB-912, another
determinations (F31/3; UB-2007 and UB-6849). This bulk sample dominated by hazelnut shell, from layer
consistency confirms that the bulk sample of hazelnut 4 is stratigraphically earlier than GrN-10470, a bulk
shells and charcoal dated by UB-2007 did not contain sample of unspecified carbonised plant remains, from
a high proportion of old wood, and so a weighted layer 2 above. The results are in good agreement with
mean has been taken of these measurements. relative dating suggested by the excavated sequence,
which may suggest that neither sample contained a
large proportion of old wood. Finally, another
Other features hazelnut shell was dated from a heavily truncated
Some of the other samples dated were from a group of feature, F36 (UB-7033). It is just possible that this
features that lay some metres north of the main hut feature was the last remnants of a hearth.
area (Figs 3 & 9). These included two samples from
F99, a large and deep pit which was just over a metre
in diameter. These results are statistically inconsistent, Outlying features
with the bulk sample of unidentified charcoal (UB- Nine features on the periphery of the main area of
2362) providing a substantially older date than the activity have been dated (Figs 3 & 10). F74 was
single hazelnut shell fragment from the same layer another large, deep pit, just over a metre in diameter,
(UB-6851). For this reason, UB-2362 only provides a containing charcoal-rich layers separated by dumps of
terminus post quem for the fill. Two other, potentially clean gravel. Two samples of unidentified charcoal
hearth-like features, were dated from this area, both from this feature (UB-2361 and UB-8872) provide
using samples of single hazelnut shells (UB-6850 and termini post quem for its filling. Three dates were
UB-6853; Fig. 9). It should be noted that a partial arc obtained from charred hazelnuts shells recovered
of stake-holes appeared to be centred on these two from pits that lay roughly 10 m away from the centre
features, which may suggest the presence of further of the main focus of activity. F102 lay to the east (UB-
huts in this area. 7038), F210 to the west (UB-7030), and F126 to the
The remaining dated features include a series of pits north (UB-8871).
that vary in character. F27, a partially truncated pit to Four features have been dated from an area to the
the west of the main hut area, is dated by a north-east of the main area: hazelnuts were dated
measurement on a bulk sample in which hazelnut shell from pits F209c (UB-8138), post-hole 991 (UB-8139),
predominated (although some unidentified charcoal and pit F211 (UB-8137). A date on a fragment of
was also included in the sample). It is, however, unidentified charcoal from post-hole 992 (UB-8140)

114
Bayliss and Woodman:Wessex Article Master Page 4/1/10 10:18 Page 115

A. Bayliss & P. Woodman. NEW BAYESIAN CHRONOLOGY FOR MESOLITHIC OCCUPATION AT MOUNT SANDEL, N. IRELAND

Fig. 12.
Probability distributions of key parameters from Mount Sandel, derived from the model defined in Figs 6–10

provides a terminus post quem for pit F209c, as post- 7570–7480 cal BC (95% probability; end Mount
hole 992 is stratigraphically earlier than this feature. Sandel; Fig. 6), probably in 7560–7510 cal BC (68%
One radiocarbon result was obtained from a bulk probability). The difference between these date
sample of unidentified charcoal from F109, a hearth estimates suggests that the main occupation at Mount
that lay roughly 15 m north of the main hut (UB- Sandel occurred over a period of 80–290 years (95%
2359). This provides a terminus post quem for the use probability; occupation Mount Sandel; Fig. 11),
of this hearth. It is significantly later than all the other probably over a period of 125–240 years (68%
Mesolithic activity dated from the site, and probability).
must represent a separate episode of human activity in The first hut in the main hut area was constructed
the area. For this reason, it has not been included in 7755–7615 cal BC (95% probability; start main
in the model for the chronology of the main hut; Fig. 7), probably in 7725–7650 cal BC (68%
period of activity. probability). Occupation associated with huts in this
area ended in 7700–7595 cal BC (95% probability;
end main hut; Fig. 7), probably in 7660–7600 cal BC
(68% probability). The series of structures associated
A CHRONOLOGY with the hearths in this area were in use for 1–120
The chronological model shown in Figures 6–10 years (95% probability; use main hut; Fig. 11),
suggests that the Mesolithic activity at Mount Sandel probably for 1–65 years (68% probability) – perhaps
started in 7790–7635 cal BC (95% probability; start for one or two generations.
Mount Sandel; Fig. 6), probably in 7750–7670 cal BC The radiocarbon dates from the little pits form a
(68% probability). This period of activity ended in coherent group, suggesting that the first was dug in

115
Bayliss and Woodman:Wessex Article Master Page 4/1/10 10:18 Page 116

THE PREHISTORIC SOCIETY

7720–7595 cal BC (95% probability; start little pits; bulk samples do provide maximum ages for activity
Fig. 8), probably in 7665–7600 cal BC (68% on the site and so should contribute to our estimate
probability). The last of these pits was dug in for date when occupation ended.
7630–7535 cal BC (95% probability; end little pits; Following the chronological trend visible in Figure
Fig. 8), probably in 7605–7570 cal BC (68% 12, a second alternative model was constructed. In
probability). Overall these pits seem to have been dug this case, the conventional radiocarbon dates were
over a period of 1–160 years (95% probability; use incorporated into the model as shown in Figures
little pits; Fig.11), probably of 1–85 years (68% 6–10, but additional ‘informative’ prior information
probability). was included that was not based simply on vertical
Figure 12 shows the probability distributions of stratigraphy. Here, we incorporated a sequence of
some key parameters from Mount Sandel. A archaeological phasing based on different types of
chronological trend can be discerned, which is feature: that the main hut was earlier than all the little
confirmed by statistical comparison of the posterior pits, and that all the little pits were earlier than all the
density estimates. The main hut was constructed larger pits. Although this interpretation is consistent
before the first of the little pits (80% probable), with the stratigraphic information where it exists,
although it is unclear whether the earliest of these pits many of these features are not in fact related by
was cut during the time when the hut was in use. It is stratigraphy. This model has good overall agreement
apparent, however, that the little pits continued to be (Aoverall=124.2%), which means that the radiocarbon
dug after the main hut had gone out of use (97% dates do not contradict this interpretation. The
probable). The cutting of a series of large pits in the posterior density estimates produced by this reading
76th century cal BC seems to be the latest activity are very similar to those from the main model (Table
during the principal phase of use of the site. F99 and 2). These analyses suggest that our modelling of these
F56 are both likely to have been dug after the last of data is reliable, probably to within a few decades.
the little pits (79% probable and 87% probable
respectively). The dating of a similar feature, F74,
dated only by samples of unidentified charcoal, is too
insecure to contribute to this discussion.
DISCUSSION
This new chronology has implications for both the
interpretation of Mount Sandel itself and its
chronological position within the earlier part of the
SENSITIVITY ANALYSES Mesolithic of the British Isles. The original concern
In order to investigate the reliability of this model, a expressed about the early dates provided by some of
number of alternative models were constructed the Mount Sandel samples were justified. At the time
incorporating different archaeological interpretations it was suggested that the old-wood factor might have
of these data. First, all the conventional radiocarbon only had a very limited influence on the age of the
dates on samples of bulk charred plant remains samples, as it was assumed that in the early Boreal
(which might contain fragments of material which forests there would have been few trees that lived for
derived from different plants which ceased several hundred years. It is now apparent that the
exchanging carbon with the atmosphere in different three oldest dates from the original series (UB-952,
years) were excluded from the analysis. Otherwise this UB-2362, and UB-2357) contained old wood. There
model has precisely the same form as that defined in is, however, no evidence that there is any systematic
Figures 6–10. The posterior density estimates for the laboratory bias between the original series and the
start and end of occupation at Mount Sandel are new AMS dates.
shown in Table 2. It is apparent that this alternative The suggested commencement of occupation at
approach to modelling the results from the Mount Sandel, sometime just after 7700 cal BC, still
conventional samples shifts the start date for the site leaves the site as the earliest known evidence for
later by c. 15 years and brings the end date c. 25 years Mesolithic settlement in Ireland (so far, no indications
earlier. In our view, this is not the most plausible of an equivalent of the English so-called Early/Non
model for the site chronology, however, because these Geometric Mesolithic has been found in Ireland).

116
Bayliss and Woodman:Wessex Article Master Page 4/1/10 10:18 Page 117

A. Bayliss & P. Woodman. NEW BAYESIAN CHRONOLOGY FOR MESOLITHIC OCCUPATION AT MOUNT SANDEL, N. IRELAND

TABLE 2: POSTERIOR DENSITY ESTIMATES FOR THE START AND END OF OCCUPATION AT MOUNT SANDEL,
FROM SENSITIVITY ANALYSES (ALTERNATIVE MODELS)
Parameter Posterior density estimate cal Posterior density estimate cal
BC (95% probability) BC (68% probability)
Main model (Figs 6–10)
start Mount Sandel 7790–7635 7750–7670
end Mount Sandel 7570–7480 7560–7510

Main model (no bulk samples)


start Mount Sandel 7775–7620 7740–7655
end Mount Sandel 7590–7525 7585–7550

Sequence (main hut > little pits > big pits: other features)
start Mount Sandel 7790–7635 7750–7670
end Mount Sandel 7575–7480 7560–7510

However, it should also be noted that certain local structure/start main hut, Fig. 13), while the
elements occur in the Mount Sandel assemblage. radiocarbon dates at East Barnes (Gooder 2007, 53)
These include the quadrilateral sectioned and surface suggest that occupation may have occurred slightly
retouched needlepoint microliths (Fig. 14: 15–16) earlier again (Fig. 13). Both these sites appear to
which, although they have a limited resemblance to contain simple assemblages dominated by scalene
the basally modified points forms found within Honey triangles and rods. The needle points from Howick
Hill assemblages (Saville 1981), are different from any illustrated in Waddington (2007, 7: 2) differ quite
forms found in Britain. Perhaps the local significantly from the Irish forms referred to earlier.
developments are most striking when amongst the The other Irish local implement types also referred to
axes especially the development of local forms of flake earlier, in particular the flake axes, are certainly
axes (Fig. 14: 23) as well as ground/polished stone totally lacking from Howick.
axes. (For a broader selection of these artefacts types There is some doubt about when the so-called
see also Woodman 1985, figs 24–9). This suggests scalene triangle dominated assemblages came into
that Mount Sandel, with its local forms of artefacts, existence in Britain. This uncertainty is based on the
was occupied some time after an initial phase of assemblage from Cramond near Edinburgh in
settlement in Ireland (ie, when sufficient time had Scotland, where six radiocarbon dates on hazelnut
elapsed to allow these local types to develop). shell suggest activity in the 85th and/or 84th centuries
Therefore, it is probable that some marginally earlier cal BC (Fig. 13), but where the microlith component is
traces of Mesolithic settlement remain to be dominated by scalene triangles (Saville 2008,
found in Ireland. 210–11). As these dates seem to be out of step with
The re-dating of Mount Sandel, combined with the the appearance of similar assemblages in adjacent
recent excavations at Howick in north-east England parts of Europe, Saville urges that some caution is
(Waddington 2007) and East Barnes (Gooder 2007) in necessary in using Cramond as an indication of a very
south-east Scotland, has helped resolve the problem early presence of scalene triangle dominated
posed by the fact that a geometric assemblage at assemblages in the British Isles. However, even if this
Mount Sandel in Ireland seemed to be considerably type of assemblage did not emerge in Britain until
older than equivalent assemblages in Britain. It would after 8000 cal BC it is still probable that a simpler
now appear that both these latter sites are older than assemblage, that pre-dates Mount Sandel and has
Mount Sandel. In the case of Howick it is clear that more in common with Howick and other sites in
the first occupation occurred 100 years or more Britain, will be discovered in Ireland.
before the first documented settlement at Mount In the case of Ireland, although Mount Sandel is
Sandel (30–260 years (95% probability) or 75–190 slightly later in date than the British sites, it still has
years (68% probability); Difference start produced the earliest unequivocal evidence for human

117
Bayliss and Woodman:Wessex Article Master Page 4/1/10 10:19 Page 118

THE PREHISTORIC SOCIETY

Fig. 13.
Probability distributions of key parameters from dated Mesolithic huts in Britain and Ireland and from activity associated
with an early narrow blade assemblage from Cramond, Scotland. Distributions are derived from the models defined in
Figs 6–10 (Mount Sandel), Bayliss et al. (2007b, fig. 6.2, Howick), and Waddington et al. (2007, fig. 15.12 Cramond)

settlement in Ireland (see Woodman in press for a 1985, chapter 11) was predicated on two
fuller discussion of problematic early dates). It is not assumptions. First, given the spread of the
possible to refine the dating of Castleroe (Woodman uncalibrated radiocarbon dates over a period of at
1985, 199–200) as suitable material does not appear least 600 years, it was felt that the area had been used
to survive in archive. Lough Borra (Ryan pers. regularly for more than half a millennium. As noted
comm.), Hermitage (Collins & Coyne 2006), and earlier, it would now appear that the site was actually
perhaps Killuragh Cave (Woodman 1997), all of used for less than half that period of time. The use of
which lie further south, are later in date. As noted by this place persisted, but for a century or two, not for
Woodman (in press), well-dated assemblages with many hundreds of years.
diagnostic artefacts such as microliths are still The new chronology has also led to a new phasing
remarkably rare in Ireland. Therefore, given the for Mount Sandel. It suggests that the site was used in
paucity of dated early Mesolithic assemblages whose a different manner. Instead of envisaging a sequence of
locations are recorded in Figure 15, some caution is huts, each one of which would have had several little
should be exercised in assuming that Mount Sandel pits and one large (storage?) pit associated with its
represents the earliest phase of settlement in Ireland. use, there would now appear to be a better case for
Earlier sites may well remain to be discovered, as
arguing that the huts from the main hut area, centred
hinted by insular elements in the make-up of the
on hearth complex 56, as well as the huts associated
Mount Sandel flint assemblage. These putative sites
could be found anywhere within the island of Ireland. with other potential hearths such as Features 31/0, 36,
Besides re-interpreting the site in its broader and 91, belong to an initial period of activity spanning
context, it is also necessary to re-evaluate the phasing only one or two generations (Figs 9 & 11). As can be
of Mount Sandel itself. The new dating programme seen from Woodman 1985 (figs 64 & 65), many of the
has helped immeasurably in providing a more precise series of little pits were noted to be cutting through
history of the occupation of the site. Much of our the initial hut complex. The new AMS dates from the
current interpretation of site organisation (Woodman little pits confirm this stratigraphic sequence as they

118
Bayliss and Woodman:Wessex Article Master Page 4/1/10 10:19 Page 119

A. Bayliss & P. Woodman. NEW BAYESIAN CHRONOLOGY FOR MESOLITHIC OCCUPATION AT MOUNT SANDEL, N. IRELAND

post date those obtained from the main hearth and endured for some time, as is best illustrated by the
hut complex. sequence of intercutting of large pits that made up the
The dates for the larger pits overlap those for the complex F209/F211. Both the little and larger pits
hut complex but a number are also later in date. It is may have been created at a later date when, instead of
therefore probable that some of the larger pits, such as building circular dwellings, very different types of
Features 56/1, 99, and possibly 74, post-date the activity were taking place at Mount Sandel. While the
earlier phases of occupation associated with the main case for over-winter storage of hazelnuts still remains
hut complex. This period of activity seems to have (ie, due to their association in the same hearths as the

Fig. 14.
Range of Mesolithic flint tools from Mount Sandel; 1–20 microliths and related forms; 21–3 & 26 axes and related
forms; 24 micro awl; 24 scraper (after Woodman 1981)

119
Bayliss and Woodman:Wessex Article Master Page 4/1/10 10:19 Page 120

THE PREHISTORIC SOCIETY

Fig. 15.
Map showing locations of key Mesolithic sites from Ireland and Britain dating to the 8th millennium cal BC

120
Bayliss and Woodman:Wessex Article Master Page 4/1/10 10:19 Page 121

A. Bayliss & P. Woodman. NEW BAYESIAN CHRONOLOGY FOR MESOLITHIC OCCUPATION AT MOUNT SANDEL, N. IRELAND

remains of piglets that would have been born in the dated to just before 6000 cal BC, few other large
spring), it certainly seems less likely that many of the structures have been found. Staoisnaig (Mithen 2000),
large pits were used for storage in association with the which is smaller and may have served a different
main group of excavated huts. The evidence that the purpose, and Cass ny Hawin (Woodman 1987),
Mount Sandel huts were used at different seasons of which is a much smaller hollow, are both later. The
the year, however, remains. nature of Broomhill is, as noted by Gooder (2007,
It is of course always tempting to assume that the 55), rather unclear.
little and larger pits were associated with other It is of interest that, as was shown during a hut
undocumented huts. However the excavation building experiment carried out by local primary
explored an area approximately 30 x 40 m across. schools from the Coleraine area in 2006, quite
These other parts of the site, as well as fieldwalking substantial structures of up to 4 m in diameter, can be
and geochemical analysis of the area around the built and will leave little or no trace in the form of
excavation, revealed no other traces of settlement. stake-holes. Therefore it would appear that these
Within the excavated area it is probable that many larger circular huts found at the three sites referred to
post- and stake-holes were partially obliterated but all above represent, at the very least, a significant
traces of potential hearths were found in one small investment in time. Whether these were, as could be
area (Woodman 1985, fig. 86). Therefore there is little argued at Mount Sandel, at key locations chosen to
reason to believe that the pits that post-date the main maximise access to a broad range of resources
hut complex were associated with other undiscovered throughout much of the year (Woodman 1985,
or destroyed traces of huts 167–8) or because they would have had more ritual
Perhaps, too much attention is paid to the search purposes is still unclear. Only further research and
for dwellings and not enough is being paid to the other excavations will show whether these distinctive
possibility that other forms of structures were created types of huts are really confined to a very restricted
and used throughout the Mesolithic. Therefore it is of period of time.
interest that at Crathes in Aberdeenshire (British
Archaeology 2007 March/April, 7), a line of large pits
has ‘been confidently dated to c. 8000–7500 BC’. The
same article notes that these are not unique, as large
pits that have produced dates of a Mesolithic age have
been found at Stonehenge, Thornborough in SUMMARY
Yorkshire, and at Bryn Celli Ddu in Anglesey (see also The Mount Sandel project demonstrates that Bayesian
Allen and Gardiner 2002). modelling, combined with the careful use of AMS
There are many other aspects of Mount Sandel radiocarbon dating of well contextualised samples
which are worthy of reconsideration but lie outside from older excavations, can provide the basis for a
the implications of the re-dating programme and are critical re-evaluation of the original interpretations of
therefore outside the remit of this paper. However, this the excavation. In this case, it has been possible to
project combined with the publication of Howick show that the major phase of hut building at Mount
(Waddington 2007) and the fuller publication of East Sandel took place within a short period of time and
Barnes highlights one curious aspect of the settlement that this may have lasted well under 100 years,
archaeology of the British Isles. These three sites, perhaps only a generation or two. This, in itself,
which are characterised by the presence of substantial suggests a more intensive use of the site at one time
huts up to 6 m in diameter with large stake-holes set rather than occasional re-uses of the location in the
into artificially created or enhanced depressions of same manner for a period of perhaps more than half a
varying depths, were all built during one short period millennium. Similarly the dating of other features,
of time between c. 8000 and 7600 cal BC. Other forms namely a range of pits of differing sizes, suggests that
of structures and dwellings from the Mesolithic have many of them were created during other slightly later
been documented by Wickham Jones (2004) and visits to the Mount Sandel area. It is possible that
Waddington (2007, fig. 15:1) but, with the possible either the manner in which the site was used had
exception of the two large circular structures from changed or that the reasons for the use of this location
Silvercrest near Elgin (Suddaby 2007), one of which is had changed.

121
Bayliss and Woodman:Wessex Article Master Page 4/1/10 10:19 Page 122

THE PREHISTORIC SOCIETY

Acknowledgements: The authors wish to thank the Buck, C.E., Kenworthy, J.B., Litton, C.D. & Smith, A.F.M.
following for financial support: Environment and Heritage 1991. Combining archaeological and radiocarbon
Service (DOE NI), UCC Faculty of Arts research fund and information: a Bayesian approach to calibration.
the Dean of Arts research fund, and finally the Royal Irish Antiquity 65, 808–21
Academy/Chrono radiocarbon fund. We would also wish to Buck, C.E., Litton, C.D. & Smith, A.F.M. 1992. Calibration
express our thanks to Sinead McCartan, Antiquities of radiocarbon results pertaining to related archaeological
Department, Ulster Museum for facilitating the re- events. Journal of Archaeological Science 19, 497–512
examination of the Mount Sandel collections. Finally we Collins, T. & Coyne, F. 2006. As old as we felt. Archaeology
would wish to thank Hugh Kavanagh for preparing the Ireland 20(4), 21
illustrations and Sam Barr of Fugro-BKS for his help in Dark, P., Higham, T.F.G., Jacobi, R. & Lord, T.C. 2006.
identifying and providing a suitable image of the Mount New radiocarbon accelerator dates on artefacts from the
Sandel area. early Mesolithic site of Star Carr, North Yorkshire.
Archaeometry 48, 185–200
Gooder, J. 2007. Excavation of a Mesolithic house at East
Barnes, East Lothian, Scotland: an interim report. In C.
Waddington & K. Pedersen (eds), Mesolithic Studies in
the North Sea Basin and Beyond, 49–59. Oxford: Oxbow
Jacobi, R.M. 1973. Aspects of the Mesolithic age in Britain.
BIBLIOGRAPHY In S. Kozlowski (ed.) The Mesolithic in Europe, 237–65.
Allen, M.J. & Gardiner, J. 2002. A sense of time: cultural Warsaw: University Press
markers in the Mesolithic of southern England. In B. Lanting, J.N. & Plicht, J. van der. 1994. 14C AMS: pros and
David and M. Wilson (eds), Inscribed Landscapes: cons for archaeology. Palaeohistoria 35/36, 1–12
making and marking place, 139–53. Honolulu: University Mellars, P.A. 1974. The Palaeolithic and the Mesolithic. In
of Hawaii Press C. Renfrew (ed.) British prehistory: a new outline, 41–99.
Ashmore, P. 1999. Radiocarbon dating: avoiding errors by London: Duckworth
avoiding mixed samples, Antiquity 73, 124–30 Mithen, S.J. 2000. Hunter-Gatherer Landscape
Bayliss, A., Bronk Ramsey, C. & McCormac, F.G. 1997. Archaeology: the Southern Hebrides Project 1988–98.
Dating Stonehenge. In B. Cunliffe & C. Renfrew (eds), Cambridge: MacDonald Institute Monograph
Science and Stonehenge, 39–59. London: Proceedings of Mook, W.G. & Streurman, H.J. 1983. Physical and
the British Academy 92 chemical aspects of radiocarbon dating. In W.G.
Bayliss, A., Bronk Ramsey, C., Plicht, J, van der & Whittle, Mook & H.T. Waterbolk, Proceedings of the First
A. 2007a. Bradshaw and Bayes: towards a timetable for International Symposium 14C and Archaeology,
the Neolithic. Cambridge Archaeological Journal 17, PACT 8, 31–55
1–28 Mook, W.G. & Waterbolk, H.T. 1985. Radiocarbon
Bayliss, A., Waddington, C., Bronk Ramsey, C., Boomer, I. Dating. Strasbourg: European Science Foundation
& Hamilton, W.D., 2007b. Absolute Dating. In Handbook for Archaeologists 3
Waddington 2007, 65–74 Needham, S., Bronk Ramsey, C., Coombs, D., Cartwright,
Blockley, S.P.E., Lowe, J.J., Walker, M.J.C., Asioli, A., C. & Pettitt, P.B. 1998. An independent chronology for
Trincardi, F., Coope, G.R., Donahue, R.E. & Pollard, British Bronze Age metalwork: the results of the Oxford
A.M. 2004. Bayesian analysis of radiocarbon Radiocarbon Accelerator Programme. Archaeological
chronologies: examples from the European Late-glacial. Journal 154, 55–107
Journal of Quaternary Science 19, 159–75 Pearson, G.W. 1984. The Development of High-precision
Bronk Ramsey, C. 1995. Radiocarbon calibration and 14C Measurements and its Application to Archaeological
analysis of stratigraphy: The OxCal Program. Timescale Problems. Unpublished PhD thesis, Queen’s
Radiocarbon 37, 425–30 University Belfast
Bronk Ramsey, C. 1998. Probability and dating. Pearson, G.W., Pilcher, J.R. & Baillie, M.G.L. 1983. High-
Radiocarbon 40, 461–74 precision 14C measurement of Irish oaks to show the
Bronk Ramsey, C. 2001. Development of the radiocarbon natural 14C variations from 200 BC to 4000 BC.
calibration program. Radiocarbon 43, 355–63 Radiocarbon 25, 179–86
Bronk Ramsey, C., Higham, T. & Leach, P. 2004. Towards Reimer, P.J., Baillie, M.G.L., Bard, E., Bayliss, A., Beck,
high precision AMS; progress and limitations. J.W., Bertrand, C.J.H., Blackwell, P.G., Buck, C.E., Burr,
Radiocarbon 46, 17–24 G.S., Cutler, K.B., Damon, P.E., Edwards, R.L.,
Buck, C.E., Cavanaugh, W.G. & Litton, C.D. 1996. Fairbanks, R.G., Friedrich, M., Guilderson, T.P., Hogg,
Bayesian Approach to Interpreting Archaeological Data. A.G., Hughen, K.A., Kromer, B., McCormac, F.G.,
Chichester: Wiley Manning, S., Bronk Ramsey, C., Reimer, R.W., Remmele,
Buck, C.E., Christen, J.A., Kenworthy, J.B. & Litton, C.D. S., Southon, J.R., Stuiver, M., Talamo, S., Taylor, F.W.,
1994. Estimating the duration of archaeological activity Plicht, J. van der & Weyhenmeyer, C.E. 2004. IntCal04
using 14C determinations. Oxford Journal of Archaeology terrestrial radiocarbon age calibration, 0–26 cal kyr BP.
13, 229–40 Radiocarbon 46, 1029–58

122
Bayliss and Woodman:Wessex Article Master Page 4/1/10 10:19 Page 123

A. Bayliss & P. Woodman. NEW BAYESIAN CHRONOLOGY FOR MESOLITHIC OCCUPATION AT MOUNT SANDEL, N. IRELAND

Ryan, M. 1980. An Early Mesolithic site in the Irish Ward, G.K. & Wilson, S.R. 1978. Procedures for comparing
midlands. Antiquity 54, 46–7 and combining radiocarbon age determinations: a
Saville, A. 1981. Mesolithic industries in central England an critique. Archaeometry 20, 19–31
exploratory investigation using microlith typologies. Wickham-Jones, C.R. 2004. Structural evidence in the
Archaeological Journal 138, 49–71
Scottish Mesolithic. In A. Saville (ed.), Mesolithic
Saville, A. 2008. The beginning of the later Mesolithic in
Scotland. In Z. Sulgostowska & A.J. Tomaszewski (eds), Scotland: the early Holocene prehistory of Scotland and
Man–Millennia–Environment: studies in honour of its European context, 243–60. Edinburgh: Society of
Romauld Schild, 207–14. Warsaw: Polish Academy of Antiquaries of Scotland
Sciences Woodman, P.C. 1981. A Mesolithic camp in Northern
Slota Jr, P.J., Jull, A.J.T., Linick, T.W. & Toolin, L.J. 1987. Ireland. Scientific American 245, 120–32
Preparation of small samples for 14C accelerator targets Woodman, P.C. 1985. Excavations at Mount Sandel,
by catalytic reduction of CO. Radiocarbon 29, 303–6
1973–77. Belfast: HMSO, Northern Ireland
Smith, A.G., Pearson, G.W. & Pilcher, J.R. 1970. Belfast
radiocarbon dates I. Radiocarbon 12, 285–90 Archaeological Monograph 2
Smith, A.G., Pearson, G.W. & Pilcher, J.R. 1971. Belfast Woodman, P.C. 1987. Excavations at Cass ny Hawin, a
radiocarbon dates III. Radiocarbon 13, 103–25 Manx Mesolithic site, and the position of the Manx
Stuiver, M. & Polach, H.A. 1977. Reporting of 14C data. microlithic industries. Proceedings of the Prehistoric
Radiocarbon 19, 355–63 Society 53, 1–22
Stuiver, M. & Reimer, P.J. 1986. A computer program for Woodman, P.C. 1997. Killuragh cave. In I. Bennett (ed.),
radiocarbon age calculation. Radiocarbon 28, 1022–30 Excavations 1966: summary accounts of archaeological
Stuiver, M. & Reimer, P.J. 1993. Extended 14C data base
excavations in Ireland, 67–8. Bray, Co Wicklow:
and revised CALIB 3.0 14C age calibration program.
Radiocarbon 35, 215–30 Wordwell
Suddaby, I. 2007. Downsizing in the Mesolithic? The Woodman, P.C. 2004. Retrospect and prospect. In A.
discovery of two associated post-circles at Silvercrest, Saville. Mesolithic Scotland: the early Holocene
Lesmurdie Rd. Elgin, Scotland. In C. Waddington & K. prehistory of Scotland and its European context, 285–99.
Pedersen, Mesolithic Studies in the North Sea Basin and Edinburgh: Society of Antiquaries of Scotland
Beyond, 60–8. Oxford: Oxbow Woodman P.C. in press. Challenging times: reviewing Irish
Waddington, C. (ed.). 2007. Mesolithic Settlement in the
Mesolithic chronologies. In P. Crombé, M. Van
North Sea basin: a case study from Howick, North-East
England. Oxford: Oxbow Strydonck, J. Sergant, M. Bats & M. Boudin (eds),
Waddington, C., Bailey, G., Bayliss, A. & Milner, N. 2007. Proceedings of the international congress ‘Chronology
Howick in its North Sea context. In Waddington 2007, and Evolution in the Mesolithic of NW Europe’,
201–24 Cambridge: Cambridge Scholar Publishing

123
Bayliss and Woodman:Wessex Article Master Page 4/1/10 10:19 Page 124

View publication stats

You might also like