Begley ArikameduReconsidered 1983

Arikamedu Reconsidered
Author(s): Vimala Begley

Source: American Journal of Archaeology , Oct., 1983, Vol. 87, No. 4 (Oct., 1983), pp.
461-481
Published by: Archaeological Institute of America
Stable URL: https://www.jstor.org/stable/504104
REFERENCES
Linked references are available on JSTOR for this article:
https://www.jstor.org/stable/504104?seq=1&cid=pdf-
reference#references_tab_contents
You may need to log in to JSTOR to access the linked references.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms
Archaeological Institute of America is collaborating with JSTOR to digitize, preserve and

extend access to American Journal of Archaeology
This content downloaded from

164.100.151.123 on Fri, 22 Dec 2023 07:07:40 +00:00
All use subject to https://about.jstor.org/terms
Arikamedu Reconsidered
VIMALA BEGLEY
Abstract which flourished during the first two centuries A.C.

He dated the Arretine ware types found at Arikame-
Reconsidered here are the chronology and significance
of the important South Indian coastal site ofdu
Arikamedu,
between 20 and 50 A.C., and argued that the trad-
which was excavated by Sir Mortimer Wheeler in 1945
ing station itself was first founded during the time of
and Jean-Marie Casal in 1947-1950. Finds of fragmen-
Augustus. Between 1947 and 1950, further excava-
tary amphorae and Arretine ware in stratified contexts
tionstrading
led Wheeler to identify the site as an Indo-Roman were undertaken at Arikamedu by Jean-Marie
station, which he believed was founded during Casal2; since
the time ofthen several other coastal sites have been
Augustus and lasted for about two centuries. partially
Even though
excavated and their chronology tied to that of
subsequent investigations revealed conflicting evidence,
Arikamedu.
Wheeler's basic interpretation has gone unchallenged
Scholars in disciplines other than archaeology have
until now; as a result it has become increasingly difficult
also relied
to formulate a coherent picture of South Indian heavily on Wheeler's dates for Arikamedu
culture
sequences during the Early Historical period.in constructing their own chronological sequences. In
Reassessment of Wheeler's and Casal's material
light ofsug-
recent research, however, the dates of Arika-
gests that the date of the founding of Arikamedu should
medu can no longer be considered as "fixed" as was
be pushed back to the middle of the third century B.C.,
while its first trade contacts with the Mediterranean perceived by Wheeler, even on the basis of his own
excavated material. Granted that the excavated mate-
world may have been established by the late second cen-
rial was limited, he nevertheless left unexplored some
tury B.C. The Arretine ware phase can now be dated
more precisely to the first quarter of the first century
significant indications of a much earlier beginning for
A.C. (instead of the second quarter, as Wheeler had the
sug-site. This hypothesis is corroborated by material
gested); amphorae occur in substantially earlier levels,
from Casal's excavations at Arikamedu. He discov-
and Rouletted Ware in earlier levels still. Although the
Arikamedu Rouletted Ware was probably manufactured ered the existence of an earlier settlement dating ap-
proximately
locally, the technique of "rouletting" must have been in- from the second century B.C.; but, unfor-
troduced from the West, probably sometime in the second
tunately, Casal's material has been largely ignored by
century B.C. Since Arikamedu thus seems to have func- archaeologists and historians in India. A reassessment
tioned as a maritime trading center long before the time
of the evidence from both excavations suggests that
of Augustus, a new perspective emerges upon the ques-
tion of South Indian trade with the Mediterranean area. the ancient settlement of Arikamedu was first estab-
lished ca. 250 B.C. and lasted until ca. A.D. 200-a
much longer period than Wheeler supposed. Accord-
No excavation has had such a profound impact
ingly, the beginnings of the settlement and its function
upon archaeological and historical research in South
as a trading station should now be viewed within the
India as that at Arikamedu undertaken by Sir Morti-
context of historical developments which were taking
mer Wheeler in 1945.1 This was the first stratigraph-
place in South India from the time of the first contacts
with the Mauryan Empire of North India (ca.
ic excavation in the entire peninsula and was widely
322-185 B.C.)
hailed as providing a firm basis for constructing chro-
nological sequences for South Indian archaeologyTo demonstrate this conclusion, four major aspects
from the Iron Age to the Early Historical period. On
which have been illuminated by information gained
since Wheeler's excavations are examined: 1) the
the basis of his finds of amphorae and Arretine Ware,
and correlated data from Classical literary accounts
physical features of the settlement in relationship to
like the Periplus Maris Erythraei, Wheeler proposed
other contemporary sites; 2) its chronology and se-
that Arikamedu was an Indo-Roman trading station quential phases of development; 3) the dating of the
I R.E.M. Wheeler, A. Ghosh and Krishna Deva, "Arikamedu: tions are at Pondicherry, Hanoi and other places. The locations of
an Indo-Roman Trading-station on the East Coast of India," the An-
large number of surface collections have gone undocumented.
cient India 2 (1946) 17-125, hereafter cited as Wheeler et al.2The J.M. Casal, Fouilles de Virampatnam-Arikamedu (Paris
artifacts from the Arikamedu excavations are now dispersed 1949;
andhereafter cited as Casal, Virampatnam); see also J.M. and G.
can be found in several collections. The bulk of Wheeler's material Casal, Site urbain et sites funeraires des environs de Pondichery
is with the Archaeological Survey of India in Delhi, while some (Paris 1956; hereafter Casal, Site urbain).
sherds are on display in museums; finds from the French excava-
461

164.100.151.123 on Fri, 22 Dec 2023 07:07:40 +00:00
462 VIMALA BEGLEY [AJA 87
inscribed sherds; providedandevidence for 4) periodic the

sea-level fluctuations
questi
medu functioned in earlier
as times.aThe authors
trading suggest that the coastal
statio
settlements of this period may have lost their vitality
THE SITE IN CONTEXT
because of the silting of the rivers draining into the
The site of Arikamedu, also known Bay as of Bengal.6
Virampat-
nam and sometimes identified with the ancient Po- Location of ports along rivers must have facilitated
douke Emporion mentioned in the Periplus Maris commerce with the interior regions as well. Recent
Erythraei, is located on the Ariyankuppam river, surveys in the lower Cauvery valley have revealed the
about 3 km. south of Pondicherry, in a sheltered existence
la- of a series of interrelated sites; the cluster of
goon formed by it and the Gingee river on the south- sites in the lower Krishna valley also suggests a den-
eastern coast of India (ill. 1). The existence of the site
sity of contemporaneous settlements.7 If the same pat-
has been known since the eighteenth century when tern
its holds true for the Arikamedu area, future investi-
visible remains were described by Le Gentil in hisgations along the Gingee river may also reveal the ex-
Voyage dans les mers de l'Inde.3 From 1937 G. Jou-istence of a network of related settlements.
veau-Dubrevil started collecting surface finds from In addition to Arikamedu, several other probable
the site, which are said to have included a gem with
port sites along the eastern seaboard have now been
identified: Korkai, Kaveripattinam, Karaikadu, Va-
"head of Augustus in intaglio."4 Subsequently, other
savasamudram and perhaps Dharanikota/Amarava-
Mediterranean artifacts, including sherds of Arretine
ware, came to light when the French started excava-
ti, even though the last site is some 120 km. inland on
tions in 1941, under the direction of Fr. L. Fau- the river Krishna (ill. 1).8 That there was a well es-
cheux.5 Wheeler saw these artifacts in 1944 and rec- tablished communication network linking the entire
eastern coast of India, including northern Sri Lanka,
ognized the potential of the site for the archaeology of
can now be demonstrated on the basis of the distribu-
South India. He started the first stratigraphic excava-
tion of Rouletted Ware, the most distinctive ceramic
tion at the site in the following year, and subsequent
work was done by Casal along similar lines. associated with the Early Historical period (ill. 1).
The sheltered location of Arikamedu was undoubt- Rouletted Ware was first identified at Arikamedu
edly a crucial factor in its founding, since the south- where it was found in the same context with imported
eastern coast of India is almost devoid of natural har- Arretine ware, but it also occurs before and after. The
bors. Moreover, all the other port sites discovered site of Arikamedu therefore provides a significant link
since Wheeler's excavation are also situated upon es- between coastal/inland trade on the one hand and the
tuaries of rivers. A study of the formation of terraces overseas commerce with the West on the other.
along the eastern coast, undertaken by A.V.N. Sarma How Arikamedu relates to any of the known sites
and the Archaeological Survey of India in 1972, has in its immediate vicinity is still not clear. During Ca-
3 Cited in Wheeler et al. 21. brief summaries of the excavations which have appeared indicate
4 Wheeler refers to the gem in Jouveau-Dubrevil's collection, that
al- they are extremely important for the study of the period from
though it was not seen by him because the collection was sent to the
the 3rd century B.C. to the beginning of the 1st century A.C. For
French School Museum at Hanoi. See Wheeler et al. 21. Korkai, see "Excavations at Korkai, District Thirunelveli" in Da-
5 L. Faucheux, Une vieille cite indienne prks de Pondichery, milica 1 (1970) 50-54. For Kaveripattinam, see K.V. Raman, "Ex-
Virampatnam (Pondicherry 1945); see also P.Z. Pattabiramin, Les cavations at Pumpuhar" in The Handbook, published by the Exhi-
bition Committee of the II International Tamil Conference (Ma-
fouilles d'Arikamndu
6 Indian (Podoukd)
Archaeology, A Review (Pondicherry
(hereafter 1946).
cited as IndArch) dras 1968) 238-40; S.R. Rao, "Kaviripattinam Excavations," in
1972-1973, 30-32. For details of the survey, see A.V.N. Sarma, Archaeological Society of South India 7th Transactions, 1962-1965
"Upper Pleistocene and Holocene Ecology of East Central South (Madras 1969) 163-65; IndArch 1962-1963, 13; 1963-1964, 20;
India," in K.A.K. Kennedy and G.L. Possehl eds., Ecological Back- 1964-1965, 24-25; 1972-1973, 32-33. For Karaikadu, see
grounds of South Asian Prehistory (Ithaca 1976) 179-90; and also IndArch 1966-1967, 21. For Vasavasamudram see IndArch
"Upper Pleistocene and Holocene Ecology of Coastal Tamil 1970-1971, 33 and R. Nagaswamy and A. Abdul Majeed, Vasava-
Nadu," Journal of Tamil Studies 9-10 (1976) 59-86. samudram (Madras 1978). For relevant material on the sites of
7A brief reference to the Cauvery valley survey appears in Dharanikota/Amaravati, see IndArch 1953-1954, 38; 1958-1959,
IndArch 1961-1962, 36-37. In the lower Krishna valley there are 5; 1962-1963, 1; 1963-1964, 2-4; 1964-1965, 2-3; 1973-1974,
several well known Buddhist sites, such as Bhattiprolu and Jaggay- 4-5; 1974-1975, 2; 1975-1976, 79. For more recent evidence from
yapeta. Surface finds from Chebrolu include important Early His- Amaravati regarding an early stupa, see R. Subrahmanyam, "On
torical ceramic types; see IndArch 1960-1961, 1. A significant ex- the Nature of Utilization of Epigraphs for Art History," paper pre-
cavated site is Kesarapalli; see H. Sarkar, "Kesarapalli 1962," An- sented at the "Seminar on Indian Epigraphy: Its Bearing on Art
cient India 22 (1966) 37-74. History," held at the American Institute of Indian Studies, Vara-
8 These sites have not been published in detail as yet, but the nasi, December 1979.

164.100.151.123 on Fri, 22 Dec 2023 07:07:40 +00:00
1983] ARIKAMEDU RECONSIDERED 463
I. IRON
A SITES
AGEWITH
BURIAL ASOKAN INSCR
SITES Tamluk@
MAJOR ROULETTED WARE SITES /
X FINDS OF ROMAN COINS X x
x
0 100 200 300 /

S IKM
t Pauni . Sisupa garh
A rnoit_ n
C$ %
x
Am ravaiA* BENGAL
x
Kopbalal" .I
a Madras Mottra
Mask XPondic * Chebrouherry
Brahmclgir
Chandravall S Souttouoon MADRAS

AI
x X/
MdsMPu aleo
x Kanchipuram I AP
x
....c ~ *
A K xT
x/ 0 A
1
K inuzir is .-x / I
xx / Kantaroda Oi A R
S / 10KM TO
Korkal Mantaf CUDDALORE
Anuradhapura INSET MAP
Ill. 1. Southern and eastern India, showing

164.100.151.123 on Fri, 22 Dec 2023 07:07:40 +00:00
sal's1950 surveyother of the

more or less larger
contemporary Pond
settlements are still
located several burial
insufficient to definesites
a pattern that mayassociate
be applicable
to Arikamedu
lithic" complex.9 That as well. Nevertheless,
some the natureof
of the th
temporary with the
structures early
excavated phases
so far seems to indicate primarily of
tremely likely. an industrial-commercial section. Whether
According to the resi-
Casal,
Souttoukeny dential areas
date to were the
interspersed-a fairly common
second ce
so, some of the feature
Souttoukeny graves
in the Indian subcontinent-or separated re-
early phases of mains
the to be determined. The maximum extent of the
Arikamedu set
teraction, if any, existed
settlement between
seems to have been reached during the
to be determined, but
middle phases since
of its long the dist
period of occupation.
km. and they are located along the
some communication is almost inevitable. CHRONOLOGY AND SEQUENCE OF PHASES
The original settlement at Arikamedu appears to The sequence of occupational phases at Arikamedu
is a complex problem. Wheeler's excavation revealed
have been situated along the east bank of the Ariyan-
kuppam river. Surface distribution of artifacts an is essentially one-period site, although the stratigra-
densest closer to the river and the excavations have phy of architectural remains in the Southern Sector led
revealed that the focal point of the settlement, or rath-
him to divide this single-period site into three struc-
er the commercial and industrial center, was in facttural phases (with sub-periods)." In Casal's excava-
close to the river, probably for practical economic rea-tions, however, earlier material was discovered which
sons. The site, as it stands today, has suffered consid-
was absent from Wheeler's trenches. He attempted to
erable damage from a variety of factors, such as flood- resolve the problem by substituting a two-period clas-
ing of the river, cultivation, reutilization of bricks and
sification, designating the earlier period as "Mega-
lithic" and the later as "Roman," with an "interme-
the constant collection of artifacts by local residents.
diate" phase between them, but he still considered the
Along the river, flood waters continue to cut into large
occupation of the site as continuous.12 This termino-
portions of the ancient settlement, gradually destroy-
ing the most important areas. logy and the divisions are not entirely satisfactory, par-
ticularly for the early phases. Consequently, a revised
Wheeler's excavation divided the site into two sec-chronology of Arikamedu's continuous sequential
tors, the Northern and the Southern (ill. 2). Because phases is offered in the discussion and Table below,
of the higher elevation of the Southern Sector, itsbased upon a reassessment of all the available data.
lower levels are better preserved, while ca. 3 m. of oc-
The transition from one phase to another is determined
cupational strata in the Northern area are presently by the first occurrence of new traits in artifacts, pri-
below sea-level. Wheeler's excavations did not reveal marily ceramics and/or architectural remains.13
the full extent of the settlement; from Casal's excava-
tions, it appears that at its peak the site extended atPhase A
least 420 m. north-south along the river, while its The earliest phase of occupation at Arikamedu was
width was some 200 m. east-west in the Northern traced in the Southern Sector (see ill. 3). On the basis
Sector, and 100 m. or more in the Southern.1o The of the distinctive Black-and-Red pottery found here,
area occupied may have been greater, but excavations Casal related this phase to the so-called "Megalithic"
beyond these limits have so far not been undertaken Culture of South India, as it had been defined by
and a large part of the settlement to the west has been Wheeler in his 1947 excavations at Brahmagiri.14
permanently destroyed by the river. The data from The term "Megalithic" is now known to be mislead-
9 Casal, Site urbain 17-42. only. References to trenches, layers, etc., are according to the system
10 Casal, Site urbain 11, pl. 1. Casal's trenches in Ar. O, P, R, S,used by the excavators.
T show the known southwestern extent of the settlement. In the
'4 R.E.M. Wheeler, "Brahmagiri and Chandravalli 1947: Mega-
northeast, his excavation in Vi. C showed thinning of occupational lithic and Other Cultures in the Chitaldrug District, Mysore
debris, indicating perhaps that the outskirts of the major urban areaState," Ancient India 4 (1948) 199-207. For a comprehensive sur-
were reached.
vey and bibliography of the "Megalithic" material, see L.S. Lesh-
I1 Wheeler et al. 26-32.
nik, South Indian Megalithic Burials, The Pandukal Complex
12 Casal, Virampatnam 18-30. (Wiesbaden 1974).
13 The material utilized here is from stratigraphic excavations

164.100.151.123 on Fri, 22 Dec 2023 07:07:40 +00:00
AK11, III etc.

I
I
I Vi. D, 6
?4
L VGROUP III I GROUP IV
NORTHERN 5
SECTOR /
CASALS EXCAVATIONS (19478; 1950)
Vi C --
r. . P, RedST
Mission
House
to Vi. C
S AK I AAr.K
AK IVI WHEELER'S
GROUP Ar.C Ar.I EXCAVATIONS (1945)
";* Ar. B-HJ,7M

S I I . METERS
Ar. Q L GROUP II
SOUTHERN
SECTOR
Ar.4
AK I
GROUP I WHEELER S EXCAVATIONS (1945)

7. ~ CASAL'S EXCAVATIONS
0 50 (1947-48; 1950)
100
METERS
A0---P
Ar. P, R, S,T
Ill. 2. Site plan of Arikamedu. (Redrawn after Casal, Site urbain)

164.100.151.123 on Fri, 22 Dec 2023 07:07:40 +00:00
TABLE
REVISED CHRONOLOGICAL SEQUENCE OF ARIKAM
WHEELER'S AND CASAL'S EXCAVATIONS
B.C 20 200 150 100 50 B.C./ A.C. 50 10
PHASES i A B C D E
STRUCTURESI post-holes bricks "Reservoir" I ring-wells t cont. from "Warehouse" tanks ; corbelle
ring-wells I walls; floors I lined pits; Phase C floors i ringwells' wal
post-holes wharf (?) etc. ring-we
CERAMICS
Local Black-and-Red Black-and-Red; Black-and-Red(??

Rouletted Rouletted Rouletted; Rouletted; Roulette
black gray some red gray; red more red than gray same
(d
Imported Rouletng aphor amphorasa aphorse aamphors
stamped designs Arretine
Inscribed Sherds one no. uncertain one no. uncertain no. uncertain
OTHER ARTIFACTSs
Local worked ivory; terracottal terracotta terracotta;
stone celts and weightsl grinding
ear ornaments, etc.
Imported glass bowl bone stylus(?) glass bowl
CONCORDANCE
Wheeler (1945)
N. Sector AK II AK II AK II AK II
S. Sector AK I (?) AK IV AK IY AK IV
Casal (1947-48,
1950)
N. Sector Gr. IV (?) (unstratified) Gr. IV (?) Cr. III, IV (?) Cr. IV
S. Sector Gr. I Gr. I, II Gr. I, II Cr. I, I Gr. I. II Gr. I,

164.100.151.123 on Fri, 22 Dec 2023 07:07:40 +00:00
WEST Ar4 EAST

AEST
o 2m. 3m.
ENORTH SECTIONV 4m. 5.
TR.I dt.of
(2od
r .-o . -
r ). ..r
o ~ ~ ~ ~1J~~
r 9.-~a~*aaeN'l"~"
~ jl.?:tlltll r l r llltlllrcl"~
1105O
Hlll ad 11 wi llA f\/. I
ViitAvAilJlAllt ?U
tl lll ".'?-r"
ll
-.. p OfW, /B4 0: P'e a/i63c I 1
' l _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _lt__ _ _ _ _ _ _ _ _ _III_ _NeIIal 11 I ":XZ) '.
l* l ayers K
Ir ,,, " oIu0Mdd
..? t ' "/' " . " " / ' --ha
, 7 0 7 f'f="
'~'<%' . ??
TA..
1 o
"Co
I?A.-
o Of
o"Wo/ ME0
..str.ct..on III.
trench4
1/
. 1"lli"
:' 4 -.
Kc
" III.:
.:trench =. 5 III;.II".I.'.
igbars(I
~o&7~3 itoc C7 Cc; IL~
C=2 A 4 o
3? C
CD dl6e V 1%~raa
period zT, .
,LEVELS -? F?rr~Frrto renchofw o11A' c&Jdibi '
,.,.4 v
"czyer. -3903'
I "k//,
mdd le , 10 5 :. Nr\\ s e ~ .ca S ?eat .,? Tii 1f1~tiI ,l~: I111
pierr d o la " i ,, ".\\,L V ?f~l~ .1. Jill,
10/0 4"\A Oil W::-?-i I
" 0 , . .. ear~rhS,:
:"" O t, ' 4 O / : :. ...I... ':
. . . r`41L
. 'tirlu '" tlr '": "'"
~ ?-~'If megalithic\
)f1.p
\
layers
me A; C All 44 /
/a .5 4M.PC~C~_9 owl~l1?
71 op- V..- ..,,
/Z.-- ""/
?'' :~:2
.?. ?? ? ?.sod
Ill. 3. Section of Arikamedu trench Ar. 4, Southern S
ing, since it implies thatcause

the of disturbed
culture strata, its
isprecise
alwaysassociationcharac-
is un-
terized by burials marked certain.
with The only indications
large of building activity
stones, which in is
frequently not the case. this phase arethe
Since a few post-holes.
first An important find,
occurrence of
iron is a more consistent however,
feature, is a Black-and-Red
perhaps Ware a
sherd with a five-
better gen-
eral designation would be Iron
letter Age
inscription cultures.
in early Brahmi script fromAnother
the so-
diagnostic trait is the use called "Megalithic" layers
of inverted of Casal'sfor
firing Ar. R5the
(Group
pro-
duction of Black-and-Red Ware.
I).'6 Not These
only is this the earliestgroups were
inscription at Arika-
extensively distributed medu, in South but it is theIndia
only Brahmiand inscription
have from
been
assigned a wide range of dates
"Megalithic" spanning
layers almost
of any stratified site the
known so far.
entire first millennium B.C. At sites where stratified Brahmi and Tamil-Brahmi inscriptions are now re-
sequences are available, Iron Age cultures precede ported from Korkai, but their stratigraphic position is
cultures associated with Rouletted Ware, which not is yet clearly established."7
also the case at Arikamedu. The most important internal evidence for dating
Only a limited area in the Southern Sector (Casal's Phase A is this inscribed sherd (ill. 9a). Since Brahmi
Group I) can definitely be assigned to Phase A.'5 Inscript was apparently first introduced into South In-
the Northern Sector, "Megalithic" Black-and-Red dia during the reign of Asoka, the sherd would be no
Ware was also observed in Casal's Group IV but, be- earlier than the middle of the third century B.C. A
15 The clearest evidence for a separate phase is in Casal's Group I.16 Casal, Virampatnam 63, pl. 13D. The inscription was exam-
In Ar. 4 and 4N, layers 10, 10A and 11 contain mostly "Megalithic" ined by M.J. Filliozat, who remarks that the characters resemble
pottery followed by successive layers of later phases; see Casal, Vi-
those of the 1st century A.C. at Arikamedu, as well as those in use
rampatnam 20-21, fig. 3. Similar pottery is also reported from the from the 3rd century B.C.
undisturbed prestructural levels of Ar. O, P, R, S, T (p. 24) and is17 "Excavations at Korkai" (supra n. 8) 53-54.
distributed on the surface along the river in the Southern Sector.

164.100.151.123 on Fri, 22 Dec 2023 07:07:40 +00:00
third-second In ceramics, Rouletted

century B.C. Ware occurs
date for the first
for time, t
Age cultures but that
into in association area
with the "Megalithic"
is consistBlack-and-
dence from other Red Ware; insites-such as
architecture, the first bricks occur and K
which has a single radiocarbon date of B.C. perhaps also the first construction of terracotta ring-
315+100,18 or Souttoukeny, the jewelry from which wells. The clearest stratigraphic evidence, once again,
comes from Casal's excavations in the Southern Sec-
is dated by Casal to the second century B.C. on stylis-
tic grounds. The total duration of occupation during tor, with which some of Wheeler's material may be
Phase A is, however, difficult to determine. The aver-
collated.20 The area of the settlement had expanded
age depth of deposit in this phase is about 70 cm., but
considerably in the Southern Sector (Casal's Groups I
and II). What was happening in the Northern Sector
no satisfactory gauge for strata accumulation has been
established for Iron Age settlements. Wheeler, with is less clear. It is quite likely that the beginning of the
some reservation, had suggested about two centuries settlement in that area also dates from about the same
for the one-meter-plus deposit (3-4 feet) belonging to
time. But, unfortunately, because of the high water-
the "Megalithic" culture at Brahmagiri.19 If we apply
table, only very limited excavations in the early levels
this rough calculation to Arikamedu, Phase A might of the Northern Sector were carried out; it is therefore
represent about one hundred years of occupation--difficult to subdivide the approximately 2.44 m. (8
that is, approximately mid-third to mid-second cen-
feet) of pre-Arretine ware deposit reported by Wheel-
tury B.C. er in AK II (ill. 4).
At this point, the evidence provided by the strati-
Phase B graphic position of the amphora sherds should be con-
sidered. Wheeler mentions that Rouletted Ware oc-
Phase B, which corresponds to Casal's "Interme-
diate phase" (or "overlap layers"), demonstrates curs
the in all layers, and amphora sherds in all except
the lowest, in his trenches in the Northern Sector.21 In
beginning of new features in many spheres of activity.
NORTH SOUTH
A B
o
< P"" " P tis"h Ir... 2Io..-Pit
,Floor Level
II o of"Warehouse'-
idig I 6W Spoil 2
Trench
MEAN *. LtI N 7 S CTO

SEA
"A retine"?
LEVEL
S 1''' . .and."'
8A San CI
ose San
"ose.'Sdn . ,"ON
"
LaSlyerse 81 Green Partition Wall of
) Compact Cla "Warehou se
1;111 I Loose Clay1:lI' (Oblique Section)
-- I --ARIKAMEDU 1945
IlSandand Clay NORTHERN SECTOR AK II
I i I)'5 'll SECTION A-B
I 1 0o 1 2
Scale of Metres ,. ,
3
i . . .
4 5
aa 0Scale
1 2 3 4
of
5
Feet
10 : 15
? ,
20
Ill. 4. Section of Arikamed
18 D.P. Agrawal and

fig. 3. S. Kusumga
Although
Date list V," Radiocarbon
later 10
activities, (1
som
p. 135). This andto this
the phase
following (p.
da
lated using the show
5730 half-life.
ceramic T
ove
not been applied.8A of Ar. 4 with
19 Wheeler (supra
said n.
to 14) 201.
represent
20 In Group I, trenches
Ar. 4 and
are 4N,
notlar
Rouletted Ware 21 and Black-and-
Wheeler et al.

164.100.151.123 on Fri, 22 Dec 2023 07:07:40 +00:00
Wheeler's the with

absence view,
rouletted patternsof
on theamphora
inner side of the base.25sher
No spectrographic
only one level could be accidental, studies ofbecause
the Rouletted Ware the ar
excavation was limited.
fromBut Casal
Arikamedu also
or any other does
site, except not
Anuradha- r
any amphora sherdspura
in his
in Sri "overlap"
Lanka, have layers
been made,26 nor are most of i
Southern Sector, which
the sites
wouldpublished insuggest
detail, which limits
that
the scope the
of o
rence of Rouletted Ware precedes
any comparative study. the import of
diterranean amphorae.22 If this is the case, the
seems likely that the earliest layer of Whee
Northern Sector (AK II) also represents a pre
phora phase. The occurrence of "Megalithic" Bla
and-Red Ware in the Northern Sector (Casal's ?-.:-i ...,
""
Group IV) has been mentioned above. Some Black- :':;:
-~"'"'~
?.,
and-Red Ware is reported from Wheeler's pre-Arre- '.::~ -??:~ ~~? ?.. ..
1:~
=?'::i: g
?..-?.
tine ware layers as well, but at the time of excavation ?. ?;i;rz

'~;l??.l-li~r, ~,,,,,, ~~.-~j~;
it was not known to be diagnostic of the "Megalithic" :r._??r~?;~:.:=:::::;ii,: ;?
?. C~'?'
;\;~i\`?=;?.??=ri~?~~
period.23 The presence of Black-and-Red Ware sug-

gests that during Phase B the occupation area was ex-
panded in the Southern Sector, and perhaps extended
to parts of the Northern Sector as well.
With the increase in the area of occupation, growth
in population may also be assumed. The influx of out-
siders or outside influences seems to be borne out byIll. 5. Arikamedu Rouletted Ware, Type 1. (After Wheeler
the introduction of new traits. Brick architecture, theet al. fig. 12a)
construction of ring-wells and the use of Brahmi Wheeler and Casal maintained that the finer varie-
script are known elements of the Mauryan period. ties of Rouletted Ware were imported to Arikamedu
Archaeological evidence for their use and adaptationfrom the West. Wheeler even suggested that the pot-
is beginning to emerge for several areas of Andhra tery was derived from Arretine ware while Casal
Pradesh and Tamil Nadu between the third century points out similarities with earlier Mediterranean
B.C. and the first century A.C.24 Further discussiontypes-the latter appears to be a more likely source.27
on how these areas relate to each other must await the But it should be stressed that, except for the technique
publication of detailed reports. of "rouletting" or "chattering," no precise parallels for
The most significant artifact type of Phase B is fabric and shape can as yet be found. Therefore, the
Rouletted Ware, which is now known to have a very production centers of the ceramics remain uncertain.
wide distribution in Eastern India and which has be- The technique of "rouletting" seems to have been in-
come a major means of dating associated cultures (ill.troduced from the Mediterranean region, since it was
5). The problem of Rouletted Ware-its composition,not known to the cultures of South India at that time.
distribution and dating-is only briefly discussedAs for other features of the ceramics, the simple dish
here. At Arikamedu it is reported to be a fine ware, shape-although not with a pronounced beak rim-
wheel-turned, fired under reducing conditions and has prototypes in the Iron Age Black-and-Red Ware
frequently black-slipped, with the slip usually turn- as well as in Northern Black Polished Ware. Black-
ing gray, sepia or brown. The most common shape isand-Red Ware is present at Arikamedu itself in the
the dish, frequently with a beaked rim and decoratedearliest levels. Northern Black Polished Ware has a
Deva both noted the stratigraphically earlier occurrence of Rou- Indian Studies, New Delhi, in press).
letted Ware. 25 The description is by Krishna Deva in Wheeler et al. 46. For a
22 Unfortunately, Casal does not discuss amphora sherds from hismore detailed discussion of Rouletted Ware, see Begley (supra
n. 24).
excavations, except in general terms. Therefore it is difficult to de-
termine when and where they occur for the first time in his trenches. 26 The two analyzed sherds from Anuradhapura were not consid-
23 See, e.g., Wheeler et al. Types 6 and 7, p. 51. ered to have been produced locally; see S. Deraniyagala, "The Cita-
24 For a discussion of the material, see V. Begley, "From Iron Agedel of Anuradhapura 1969: Excavations in the Gedige Area," An-
to Early Historical in the Archaeology of South India," in J. Jacob-cient Ceylon 2 (1972) 163.
son, ed., Essays in South Asian Archaeology (American Institute of 27 Wheeler (supra n. 14) 200; Casal, Virampatnam 36-37.

164.100.151.123 on Fri, 22 Dec 2023 07:07:40 +00:00
wide range of dates-it was

ments in favor of dating the used
first occurrence of Rou- in
riod and arrived
letted Ware
in at Arikamedu
theapproximately
lower to the sec-
Kris
ond century B.C. of
after. The technique may also firing
be briefly summarized und
tions was common
here.31 to both traditions
ing was normal in Black-and-Red
with a highly lustrous finish is a dis
Northern Black Polished Ware.
r M\a
At Arikamedu, Wheeler's report states that some of
the Rouletted Ware sherds have high polish, although
they are technically inferior to the Northern Black
Polished Ware. The black slip with metallic luster oc-
curs only on specimens from pre-Arretine and early
Arretine levels, and specimens from pre-Arretine lev-
els generally have brighter polish than those from
later deposits.28 These observations, in conjunction
with references to fine black slipped pottery, said to be
almost like Northern Black Polished Ware, at sites
along the eastern coast like Sisupalgarh, Kesarapalli
Ill. 6. Arikamedu stamped decoration on Wheeler's local
and Korkai, may prove significant in resolving the
wares Type 10. (After Wheeler et al. fig. 17a)
question of the genesis of the Rouletted Ware.29 The
wide distribution in eastern India of Rouletted Ware, First is the question of the interpretation of the
frequently in substantial quantities, reinforces the
stratigraphic evidence itself. In the Arikamedu report,
Wheeler assigns the first occurrence of Rouletted
possibility of local production with only the "rou-
letted" decoration being inspired by MediterraneanWare to the end of the first century B.C. or the begin-
types. Arikamedu has other ceramic types-suchning
as of the first century A.C., preferring the later
Wheeler's Types 10, 18 and 141--which were date.
no To be more precise, he attributes some 20 years
doubt produced locally, but were probably influenced to the occupation before the first occurrence of Arre-
by Mediterranean types. Wheeler's Types 74/75 may tine Ware, and 30 years to the Arretine ware levels.
also fall within this category, since it seems that they
The crucial evidence for the relative chronological se-
were manufactured to serve the same function as the quence of the early levels is considered to be at AK II
amphora.30 in the Northern Sector (ill. 4). Here the layers with
If we assume that the technique of "rouletting" was Arretine ware are from "8 Green" to "7AR" covering
introduced from the West and the stamped designs on some 0.91 m. (3 feet) of deposit. Below this, Rouletted
Wheeler's Type 10 were also similarly inspired, some Ware occurs in 2.44 m. (8 feet) of accumulation.32
contact with the Mediterranean region, direct or indi- Since the lower 3 m. or so (10 feet) of accumulation
rect, during Phase B must be postulated (ill. 6). Argu- in the Northern Sector is now below mean sea-level,
28 Krishna Deva in Wheeler et al. 46. For a description of the above types and illustrations, see Wheeler
29 For Sisupalgarh, see B.B. Lal, "Si'upalgarh 1948: AnetEarly
al. 55-91.
Historical Fort in Eastern India," Ancient India 5 (1949) 79;
31forDetailed arguments are presented in my article (supra n. 24).
Kesarapalli, see H. Sarkar, "Kesarapalli 1962," Ancient India 22 as a result of his own excavations, suggested the existence of
Casal,
(1966) 45; for Korkai, see "Excavations at Korkai," (supra n. 8) 52.
Indo-Mediterranean trade during the 2nd-1st centuries B.C.
30 Wheeler's Type 10 is a bowl with stamped decoration, and 32was
Wheeler et al. 24-26 and fig. 3. Even after Casal's excavations,
found in limited quantity throughout his excavations. Because of
Wheeler gave more or less the same dating to his material; see
the wide variety of "fabrics" in which this shape and decoration
R.E.M. Wheeler, "Roman Contact with India, Pakistan and Af-
appear, it can be presumed that the pottery was not imported but,
ghanistan," in W.E. Grimes ed., Aspects of Archaeology in Britain
like Rouletted Ware, its method of decoration could well have andbeen
Beyond (London 1951) 354-81; and Rome Beyond the Impe-
of Mediterranean origin. Type 18, however, imitates Arretine ware
rial Frontiers (London 1955) 137-53. Wheeler's interpretation of
shapes and is not found in pre-Arretine levels. The significance of
the evidence is based upon the premise that Roman trade with
these two types was first pointed out by Casal, Virampatnam India could not be earlier than the time of Augustus, consequently
36-37. Type 141 is described as a "dish with a flaring rimMediterranean
and a types of ceramics at Arikamedu could not be ear-
foot-ring, and is ornamented on the interior face with a row of nicks
lier. He does not seriously entertain the possibility that Mediter-
ranean
at the rim and a stamped floral pattern on the base." It occurs spo- trade with Arikamedu could have been established via in-
radically in the Arretine and post-Arretine ware levels. termediaries
Types before direct Roman involvement.
74/75 are the conical jars found throughout Wheeler's excavation.

164.100.151.123 on Fri, 22 Dec 2023 07:07:40 +00:00
excavations in these layers were

yur, a major findspot limited
of Rouletted and
Ware on the Cau- dif
to conduct. It is not very in Tamil Nadu,
entirely clearshould whether
also be re-examined.36
these
were submerged occupation layers,
Archaeological evidence from twoor estuarin
early coastal sites
into which artifacts were
in Tamil deposited-Wheele
Nadu, excavated since Wheeler's work at
more inclinedtoward the is
Arikamedu, latter explanation.
also consistent with an early date for H
ever, abundant pottery and
the first timber
appearance of Rouletted building
Ware on the south- ma
were recovered; Wheeler
eastern coast.mentions that the
Excavations at Kaveripattinam-lo-
Rouletted Ware came cated from theseonearly
south of Arikamedu, the mouth oflayers,
the river w
it was more profuse than in
Cauvery-have later
revealed layers.
the remains When
of an Iron Age
data are combined with settlement which existed before
evidence fortheaintroduction
definite of o
tion of Phase B from Casal's excavation in the South- Rouletted Ware.37 A large structure there has been
ern Sector, where Rouletted Ware occurs in the upper considered by the excavators to have been a wharf,
levels containing "Megalithic" material, it seems like- with a single radiocarbon date of B.C. 315?100.38
ly that the earliest layers of the Northern Sector (AK The other site is Korkai-farther south, originally
II) represent an actual settlement which has partially along the coast on the Tamraparni river-where ex-
submerged because of the rising sea-level. Its duration cavations have reportedly yielded a variety of arti-
is a matter of speculation because of limited excava- facts, including "imported" ceramics and inscribed
tions, although 20 years seem too short a time-span sherds.39 Unfortunately, the stratified sequence for
for an accumulation of 2.44 m. Other scholars, such as this site is still unpublished and I was unable to exam-
Siran Deraniyagala, have also commented on this ine the material. A radiocarbon date for the earliest
point and suggest revision of the chronology of the Iron Age material is B.C. 805+95.40
Early Historical period.33 Although the radiocarbon dates are for pre-Rou-
An earlier date than the first century A.C. is also letted Ware settlements, and single dates by them-
indicated by the script of the inscribed sherds from selves do not carry much weight, the cumulative evi-
Arikamedu itself and other sites along the southeast- dence intrinsically points to the existence of first mil-
ern coast. Regarding the Arikamedu graffiti, on epi- lennium B.C. settlements on the southeastern coast
graphical grounds, N.P. Chakravarti stated: "At first into which Rouletted Ware was introduced at some
sight the script found on the graffiti from Arikamedu point. The precise date or location for the first occur-
appears to belong to the first or second century B.C. rence of Rouletted Ware cannot be postulated as yet;
when compared with the script of other Brahmi in- but the available data point to the second century B.C.
scriptions, particularly those found in the North."34 Future research will test this hypothesis; for the pres-
But Wheeler maintained that the inscribed sherds be- ent, Phase B may be tentatively placed sometime in
longed to the first-second century A.C. because of the the second century B.C., as a transitional stage be-
"secure" archaeological dating of the site. According- tween Phases A and C.
ly, the dating of the inscribed sherds was modified by
him to bring it in line with his short chronology for Phase C
the early levels. Phase C, which corresponds to Wheeler's pre-Ar-
More recently, however, Subrahmanyam has dated retine ware layers in both the Northern and Southern
some inscribed Rouletted Ware sherds from Salihun- Sectors and to Casal's post-"overlap" layers in the
dam-a Buddhist site north of Arikamedu-also to Southern, is a period of rapid development and in
the second and first centuries B.C., and suggested thatmany ways the most significant stage in the history of
the lower dates of Rouletted Ware should be re- Arikamedu. For the first time, amphorae and other
vised.35 But the question was not pursued by otheritems of undoubted Mediterranean origin are encoun-
scholars. The dates of the inscribed sherds from Urai-
tered. Large scale building activities began, and both
33 Deraniyagala (supra n. 26) 104-105; for a discussion of his
37 See references under Kaveripattinam (supra n. 8).
dating, see Begley (supra n. 24). 38 See supra n. 18.
34 In Wheeler et al. 109.
39 See references under Korkai (supra n. 8).
35 R. Subrahmanyam, Salihundam, A Buddhist Site in Andhra 40 D.P. Agrawal, S.K. Gupta and S. Kusumgarh, "Tata Institute
Pradesh (Hyderabad 1964) 8-9 and 22; see also fig. 1. Radiocarbon Date List IX," Radiocarbon 13 (1971) 442-49 (sam-
36 IndArch 1964-1965, 25. Dates in the 1st-2nd centuries A.C.
ple no. TF 987, p. 447).
have been suggested because of the dates of Arikamedu.

164.100.151.123 on Fri, 22 Dec 2023 07:07:40 +00:00
source of
Sectors at the site the large number
were of finished and unfinished
extensively
beads found all over
tion, some idea of local industries the site, as well as of several un- a
traded commodities can also be obtained. finished shell bangles and two worked ivories-one
One of Casal's most significant discoveries was the
from this phase and the other later.45 Other important
remains of a brick wall (Wall A) in the Southern Sec-
finds include a bone stylus, perhaps of foreign origin,
and terracotta and bone ear-ornaments of local manu-
tor, traced up to a length of 27 m. from the river east-
ward.41 According to the excavator, this was the outerfacture. According to Casal, ornaments of the same
wall of an artificial water reservoir. If we accept this
type were found in excavations on the Palatine Hill in
interpretation, it would be logical to assume that
Rome in a second century B.C. context.46 If so, these
water was diverted from the river into the reservoir. finds would constitute additional evidence for early
In the bottom of the reservoir, Casal found some shal- trade between India and the Mediterranean area, and
low terracotta ring-wells. He suggested that these would also help corroborate the dates proposed here
for Phase C.
ring-wells were to assure a supply of water during the
dry season when the water-level in the reservoir Turning to the Northern Sector (AK II), much of
dropped to its lowest level, resolving thereby the prob-the 2.44-m. pre-Arretine deposit, with the exception
of the earliest layer, should be equated with this
lem of constant water supply for industrial and other
purposes. It should be noted that at Kaveripattinamphase. That there was a structure built substantially
remains of a water reservoir were also excavated inof timber, perhaps a wharf, is evident but unfortun-
layers associated with Rouletted Ware.42 Some otherately none of it can be reconstructed.47
structures-such as "lined pits," walls, floors and In ceramics, definite imports from the West are the
ring-wells-were also contemporary with the wall of amphorae (ill. 7), which contained wine and perhaps
the reservoir.43 oil as well.48 For the same purpose, but of local pro-
It appears that the reservoir served as the focal duction, were perhaps the conical jars (Wheeler's
point of the possible "industrial" complex in the
Types 74/75), sherds of which were found through-
Southern Sector. Casal's excavation indicates that
out Wheeler's trenches and several pointed bases in
situ
there was a cluster of small scale workshops along thein the early levels of Wheeler's AK I and Casal's
border of the reservoir, stratigraphically contempo-
ViB.49 The structures with which they are associated
could have been shops or storage areas. Similar coni-
rary with it.44 Abundant remains of working in metal,
glass, semiprecious stones, ivory and shell werecal
ob-jars, erroneously called amphorae, have also been
served by him. These workshops may have beenfound the in large numbers at Kanchipuram.50
41 Casal, Virampatnam 26, pl. 2. same type as no. 46. The rest of the illustrated shapes are from the
42 IndArch 1963-1964, 20: Raman (supra n. 8) 239. Arretine and post-Arretine phases and therefore belong to the first
43 Casal, Virampatnam 26. century A.C. or later. I am grateful to Professor E.L. Will for re-
44 Casal, Virampatnam 28-29. examining Wheeler's material for me. She informs me that the dou-
45 Wheeler et al. 108, fig. 43, pl. 39. ble-handled Coan-type pieces, generally speaking, are of the first
46 See Casal, Virampatnam 29. The objects are said to have comecentury A.C., but more precise identifications would be possible
from the "Casa Repubblicana" found during the course of excava-only upon an examination of the sherds, since Wheeler's descrip-
tions on the Palatine Hill. This material was apparently never pub-tion is not detailed enough. For other comments on Arikamedu am-
lished. Excavations were begun by S.M. Puglisi in 1948 but be-phorae see V.R. Grace, Standard Pottery Containers of the Ancient
cause of his death only the material from the early periods was pub- Greek World (Hesperia Suppl. 8, 1949) 175-89; and Amphoras
lished; see MonAnt 41 (1951) cols. 1-98. and the Ancient Wine Trade (rev. ed., Princeton 1979), comments
47 Wheeler maintains that this deposit accumulated quickly andon figs. 56-61. I am grateful to Professor V.R. Grace for these
references.
assigns some 20 years to it: Wheeler et al. 24-25. This interpre-
tation was based upon very limited and uncertain evidence, and was 49 Casal, Site urbain pl. 7; Wheeler et al. 32, fig. 29, pl. 33A.
advanced before the discovery of an earlier settlement at the site. Wheeler's
A AK I was a disturbed area with a deposit 2.7 m. in depth.
different interpretation is therefore offered here. Casal also excava- No detailed description or illustration of the upper levels was pub-
ted in the Northern Sector, but was confronted with the problems oflished but at a "Low level" 8 conical jars were found in situ. These
high level of sub-soil water and disturbed strata: cf. Casal, Site ur-are tentatively assigned to Phase C.
bain 9-11.
50 These jars are in the Museum of the Department of Archaeo-
48 The amphora shapes that could be reconstructed were pub- logy, Madras University, but do not include any Mediterranean
lished by Wheeler et al. (41-45). The three profiles from pre-Arre-
amphorae. In IndArch 1971-1972, 42 (see also pl. 44), there is a
tine layers are his nos. 46-48. No. 46 is described as a rim of pink-
reference to the discovery of "imitation amphorae." R. Subrahman-
ish buff ware; no. 47 as a carinated shoulder in pink ware with yam mentions that "Roman amphorae with floral designs and a
yellow slip; and no. 48 as a rim of yellow slipped buff ware, of the
dull pinkish fabric" were found at the site in addition to conical jars:

164.100.151.123 on Fri, 22 Dec 2023 07:07:40 +00:00
a i 'C ?
tf ,
I'
,"" /," I
ib f- -Q "
,/k
?r JI
Ill. 7. Arikamedu imported amphor

Arretine layers. (After Wheeler et a
Rouletted Phase D is
Ware abundant and
the pre-ArretineThe only layers
feature that distinguishes
ofPhase theD from N
Phase C is the presence of
Among the associated Arretine ware (ill. 8), an g
pottery,
inate over red. Especially
important import from the West fornotew
determining ab-
solute dates. No specific
stamped decoration changes can otherwise beT
(Wheeler's
above. Another artifact
seen in of
architecture, nor are there Medi
significant changes
fragmentary bowl of
in other ceramic blue
types. glass.52
Sherds of Arretine ware, some
with potter's
The date for the marks, are found in both Sectors.54
beginning of
course on theTheir
end of
precise number Phasearound
is uncertain-perhaps B-
50-but no more
curred in the late than 20 sherds are
second to reported
earlyfrom
stratified
according to the layers. The stratigraphic position
chronology of the
sugge
sherds is clearest
mination of Phase C in canthe Northern
be Sector (AK II),
place
ning ofthe Christian era,
where they can be related on
to other the
material.55 In the d
ware from the Southern
next phase.
Sector, Wheeler reports one sherd from a pit
"Kanchipuram Excavations," Journal of Andhra Historical 52Re-

Wheeler et al. fig. 42.
search Society 34 (1974-1975) 27. This reference apparently is precise dates may eventually be advanced if the amphora
53 More
mistaken, and may relate to sherds of Arretine ware mentioned
sherdsinare re-examined. For the three pre-Arretine ware shapes,
IndArch 1970-1971, 32. The sherds are not illustrated and
seeI supra
have n. 48.
not seen them. 54 Wheeler et al. 34-41.
51 Wheeler et al. 46. 11 Wheeler et al. fig. 2.

164.100.151.123 on Fri, 22 Dec 2023 07:07:40 +00:00
(Pottery Group Arretine

A) ware which contains
sherds found at Arikamedu,"5 for two a
other fragments,
reasons.including one
First, since Arikamedu does not show any with
inscription.56 discontinuity
The pit of occupation,
is sealed he believed that cessation
by stru
lowing phase and therefore
of import at the site could only be the result is
of stop- signif
purposes. page of production. Second, he considered the simple
Among the associated wares, according to Wheeler, types of Arretine ware found at Arikamedu (as
amphora sherds continue and Rouletted Ware re- against a wider range in the West) an indication of a
mains abundant and of high quality. The quantity of relatively late phase of production. Regarding the first
red wares seems to increase, so that they are almost hypothesis, it should be pointed out that, although the
equal in number with gray wares. A new shape is occupation
a at Arikamedu was continuous, significant
bowl (Wheeler's Type 18), somewhat similar to Ar- changes did take place in the post-Arretine ware
retine ware types," which may have inspired its phase. Consequently, there could be other reasons for
production. the stoppage of this particular import.59 As to the sec-
The dates of Phase D are linked to the dating of ond argument, much work has been done on Arretine
Arretine ware in the West. Wheeler had originally ware in the West since Wheeler's time, and more pre-
given 20-50 A.C. as an all-inclusive period for the cise dates for several sites are now available.60 On the
rJI..
-------------r-~
b
,,llJ,,-
.._ _ --- - .... --
ar
ci
/
Ill.
56wW
15 W
se
58HW
pub
fr
25-4
M
32)
De
prob
di
how
la
59 H
Li
may
m
for
fu
Tim
36
Disc
60 Th
It
of HA
tu
pert
"K
tine
tine
m
sites
Ha

164.100.151.123 on Fri, 22 Dec 2023 07:07:40 +00:00
basis of these studies,er's

theexcavations,
first or in aquarter
modest way even
ofearlier
thedur- f
tury A.C. seems to be ing the
Phase C (Lined
most Pits inappropriate
Casal's Group II) must re- ti
for the Arretine ware found at Arikamedu. Accord- mained unresolved until a larger area is cleared.
ingly, Phase D should also be placed within the same The end of Phase E was marked by extensive brick-
time-span. robbing in the Southern Sector; obviously some of the
structures had gone out of use or outlived their utility.
Phase E The Northern Sector, on the other hand, may have
The sequential development following Phase beenD abandoned; at least that is what the present evi-
finds clearest evidence from Wheeler's excavations, dence suggests. Since Phase E postdates the importa-
with which Casal's material may also be correlated. tion of Arretine ware, its beginning should be placed
Judging from Wheeler's data, Phase E, which imme- in the second quarter of the first century A.C. Its du-
diately follows the cessation of Arretine ware imports,
ration, and that of the following two phases, can only
exhibits some distinctive changes in ceramic prefer-be tentatively estimated on the basis of repeated build-
ences. Although amphorae-and therefore amphora- ing activity. Wheeler had calculated between one to
related trade commodities, wine and oil-continue twotocenturies for the structural stages. Accordingly, in
be imported, Rouletted Ware decreases and is said
the to
absence of any other precisely datable material,
be inferior in quality. Wheeler mentions that red
approximately 50-plus years have been assigned to
wares outnumber gray wares, and spouts and lug- E and some 100 years to Phases F and G com-
Phase
bined in the Table above.
handles appear for the first time. What is puzzling,
however, is that finer wares do not seem to have been
Phase F
in great demand, although the architectural activity
points to economic prosperity. During Phase F, which corresponds to Wheeler's
During Phase E, both Sectors remained in occupa-
Middle Phase (Sub-periods 1, 2 and 3), building and
tion, to judge from the stratigraphic position of Arre-
rebuilding activity continues in the Southern Sector.63
tine ware. Structures of Wheeler's "Early Phase,Whether
Sub- or not new locations were sought to compen-
period 2" in the Southern Sector should be more sate
orfor the loss of the Northern Sector remains to be
less contemporary with the "Warehouse" of the
examined. An innovation of this phase was the cor-
Northern Sector, which must have been built around
belled drain, a marked improvement over the drains
the time when the importation of Arretineof ware
the earlier phases. Since the drains were very much
stopped.61 Evidence for continued building activity in of the presumed textile industrial complex, it is
a part
both Sectors is also present in Casal's excavations
obvious that this particular industry continued to
(Groups II, III and IV).62 thrive. In ceramics there is no distinctive change. The
Among the structures, the most distinctive are the of amphorae continues, implying continuity of
supply
trade with the West.
"Warehouse" in the Northern Sector, and the "Dyeing
Tanks" with their walled quadrangles and the "Lined
Pits" in the Southern. The construction of the "Ware-
Phase G
house" indicates increased commerce; if the other Phase G, Wheeler's Late Phase, represents the last
buildings pertain to the textile industry as suggested
stage in the history of the ancient settlement of Arika-
by Wheeler, then the whole complex could indicatemedu, during which its prosperity seems to have de-
the emergence of an important industry of some clined
mag- considerably. Stratigraphically, it is not entire-
nitude at Arikamedu. Whether it started duringly Phase
clear whether there was any break in the continuity
E, as is indicated by structural stratigraphy in Wheel-
of the occupation between Phases F and G.64 Building
For the study of stamps see A. Oxe and H. Comfort, Corpus Vaso-
mania 20 (1952) 389-92. Since Ohlenroth's dating is no longer ac-
rum Arretinorum (Bonn 1968) 91-96, 518-35. cepted, the dates proposed here supersede those mentioned in my
earlier work.
I am greatly indebted to Professor H. Comfort for his comments
and suggestions. He has also drawn my attention to the two "mar- 61 Wheeler et al. 24.
bled" sherds excavated by Wheeler. Wheeler et al. 36, had classified
62 Casal has correlated the structures from his excavations with
them under Arretine ware. These sherds, if re-examined, may
Wheeler's Sub-periods on the basis of brick sizes and ceramics: Site
prove to be significant for determining the date for the end of Phase
urbain 9-10. However, he does not give the details of either. There-
D at Arikamedu. In my earlier article (supra n. 24) I had men-
fore, a detailed analysis of the material is not possible.
tioned L. Ohlenroth's dating of Arikamedu Arretine ware as dis-
63 Wheeler et al. 29-31.
64 Wheeler et al. 32.
cussed by him in "Zur Datierung der Funde von Arikamedu," Ger-

164.100.151.123 on Fri, 22 Dec 2023 07:07:40 +00:00
ab
80 !%LC
Ill. 9. Arikamedu pottery inscribed in Brahmi script: a) from Ar. R5, Southern Sector (
b) AK IV 198 (Wheeler et al. fig. 46.3); c) AK II 571 (Wheeler et al. fig. 46.9); d) unstr
excavations.65
continued during this phase, but the structures As mentioned above, an addit
are not
aligned with those of the earlier phases. General
scribed sherddete-was excavated by Casal. More r
rioration is evident in local ceramics asanother
yet well, asinscription
re- has been published b
devan.66 A few
ported by Wheeler and Casal. Nevertheless, inscribed sherds have been found at
Wheeler
points out that amphora sherds continue,
other Earlyimplying
Historical sites as well, such as Salihun-
dam, Kanchipuram and Uraiyur in South India, and
that the source of supply was still open.
Kantarodai and
With no precise means of dating available, Anuradhapura in Sri Lanka.67
Phases
F and G may be tentatively placed in the second
Therefore, cen-
the practice seems to have been fairly com-
tury A.C., as stated above. The end ofmonthe
and ancient set-
may have been a legacy of the Iron Age, for
tlement may be related to the shift ingraffito
trade, markings
presumingwere quite common on the pottery
that its economy was primarily dependent upon"Megalithic"
of the so-called it. As cultures and also exist at
stated earlier, some scholars believeArikamedu.68
that the fate of Brahmi script was intro-
But when
several sites along the eastern coastduced
was in conditioned
South India in the middle of the third cen-
by the drop in sea-level, or perhaps byB.C.,
tury the theshifting
earlier marking system was gradually
focus of trade due to political or other factors.
replaced by the new scripts that developed from it.
The Arikamedu inscriptions are brief (the longest
THE INSCRIBED SHERDS
has 19 characters) and frequently contain no more
The revised dates proposed above forthanthe Arika-
a name. The pots on which the inscriptions occur
medu sequence have a direct bearing onshowthe study
a wide varietyof
in types and do not form a homo-
the Tamil-Brahmi inscriptions from the sitegroup.
geneous and Obviously
from it was not a specific ceramic
elsewhere (ill. 9). The Arikamedu inscriptions
type that was being are
labelled.
graffiti on pottery; 20 inscriptions were
Forpublished
the study of by
the development of early Tamil-
Wheeler, 18 from his and two from Brahmi
earlier French
only two dates have been considered "se-
65 Wheeler et al. 109-14. (supra n. 26) 122-30. At Kantarodai one sherd inscribed in Brahmi
66 I. Mahadevan, "Arikamedu Graffiti: A Second Look," Dami-characters was found by myself and other colleagues in 1970, dur-
lica 2.3 (1973) 63, fig. 5. ing the course of a brief excavation undertaken by the University of
67 No relative study of the inscriptions from these sites has yet
Pennsylvania Museum. The sherd is presently in the collection of
been done. For brief references to the inscriptions from Kanchi- the Department of Archaeology, Colombo.
puram and Korkai, see supra n. 8. For Salihundam, see Subrah- 68 Casal, Virampatnam fig. 21.
manyam, (supra n. 35) 84. For Anuradhapura, see Deraniyagala

164.100.151.123 on Fri, 22 Dec 2023 07:07:40 +00:00
cure"-the dates of Asokan epigraphs

19 have been identified and
as being in Tamil, two in Pra- Wheele
dates for Arikamedu. krit,
Evenand the lastthough
is uncertain.72 It maythetherefore inscripti
be
are few and brief, much work
assumed that has
Tamil was the languagebeen
in common use.done by
graphists on defining the
The writersearly
of the Prakrit forms of
inscriptions probably camethe lang
and determining the typology of
from a different linguistic the
background and script.69
may have Bu
mentioned above under been Phase B,
traders. Regarding because
the first Prakrit inscriptionthe arch
logical context of the Arikamedu graffiti
(AK IV 198), because of the similarity of the charac- was con
ered to be securely dated toof the
ters to those the Kushan first-second
period, Mahadevan states centu
A.C. by Wheeler, inscriptions which
that "the pottery was probably onNorth
imported from palaeogra
ic grounds appeared to be
India, or, atearlier
any rate, the scribewere sometimes
was a Northerner."73
signed later dates. In addition, allis on
The other Prakrit inscription the graffiti f
an unstratified
Arikamedu were collectively
sherd, presumably fromattributed
earlier French excavations,to that
span.70 now in the collection of the French Institute, Pondi-
The longer chronology and the more detailed cherry (no. 1-13). The inscription was published by
breakdown of the sequential phases outlined here ob- Mahadevan,74 and probably is the same one alluded
viously imply that all the graffiti from the site were to by Filliozat.75 On the basis of the reading of this
not from one specific period. This fact is also clearly inscription, ... yya de dhamo ("pious gift of ..."),
evident from the forms of the characters in the inscrip- which is quite common on votive Buddhist inscrip-
tions. Whether or not it will help in refining the typo- tions, Mahadevan also raises the possibility that a
logy of the script, the relative archaeological dating of Buddhist place of worship existed at Arikamedu. This
the inscribed sherds should be attempted nevertheless. would not be surprising, for glass beads with Bud-
The sherd from Phase A, cited above, should be at dhist symbols have been found there76; other Early
least as early as the second century B.C., while the one Historical sites, such as Dharanikota/Amaravati,
from Phase D (AK IV 198), associated with Arretine Kanchipuram and Kaveripattinam, demonstrate the
ware, should date from the first quarter of the first existence of Buddhist religious structures at or near
century A.C. (ill. 9b). Six sherds from Wheeler's the urban center. It should be pointed out, however,
trenches in the Northern Sector could also be consid- that there is still no archaeological evidence for a reli-
ered no later than the first century A.C.71 Unfortun- gious edifice of any kind at Arikamedu.
ately the precise stratigraphic position of most of the
ARIKAMEDU AS A TRADING SETTLEMENT
sherds is not published. Therefore, at present, a com-
plete correlation with phases cannot be worked out, That Arikamedu was a coastal trading settlement is
but it should eventually be feasible to do so from the indisputable. In fact the site is unique since it is the
excavation records. This approach would provide only one in all South India which has yielded evidence
more precise dating for individual inscriptions and for continuous trade with the West over an extended
perhaps take us a step further in tracing the chrono- period of time. Other sites are occasionally said to
logical development of the Tamil-Brahmi script. have items of Mediterranean origin, but the evidence
We may also ask the question whether the inscrip- so far is not adequate to vouch for extensive or direct
tions, brief as they are, tell us anything about the peo- trade. As a result of his excavations, Wheeler con-
ple who wrote them. First of all, of the 22 inscriptions, cluded that Arikamedu was an Indo-Roman trading
69 See Casal, Virampatnam 63; as well as comments in Wheeler would appear that these sherds were no later.
et al. 109-14. Relevant are also two articles by M.J. Filliozat, 72
"Les
See Mahadevan's revised reading of some of the inscriptions in
Inscriptions de Virampatnam," CRAI Jan. 1947, 107-18; "Les 2.3 (supra n. 66). The inscription from Casal's excava-
Damilica
echanges de l'Inde et l'Empire Romain aux premiers siecles de l'Are
tion is not included in Mahadevan's list.
Chr6tienne," La Revue Historique 201 (1949) 1-29. Among 73
more
Mahadevan, Damilica 2.3 (supra n. 66) 60. Filliozat, in "Les
recent studies are three important ones by Mahadevan, Inscriptions
Tamil- de Virampatnam," (supra n. 69) 117, considered this
Brahmi Inscriptions (Madras 1970); "Tamil-Brahmi Inscriptions
inscription to be of the 2nd-3rd century A.C., but Mahadevan's
of the Sangam Age," Proceedings of the II International Conference
dating in the 1st century A.C. is closer to the dates assigned here to
Seminar of Tamil Studies, January 1968 (Madras 1971) 73-106;
Phase D, in which this inscription occurs.
and Damilica 2.3 (supra n. 66) 60-64. 74 Mahadevan, Damilica 2.3 (supra n. 66) 63.
70 See N.P. Chakravarti's comment in Wheeler et al. 109.
75 Filliozat, "Les Inscriptions" (supra n. 69) 117-18.
71 The six sherds from AK II are nos. 5-9 in Wheeler et 76 al.
See Casal, Virampatnam pl. 13C.
111-12. Since the Northern Sector was abandoned after Phase E, it

164.100.151.123 on Fri, 22 Dec 2023 07:07:40 +00:00
station which was of thefirst

Mannar Gulf,79 established
which was probably the great-du
Augustus and continued est attraction of the southeastern
to flourish coast at that time. fo
turies thereafter. He believed that earlier trade with Archaeological research has now revealed the exis-
the Mediterranean, if any, could only have been tence of the coastal site of Korkai, with a radiocarbon
"spasmodic."77 Casal, proceeding on similar lines,date early in the first millennium B.C. Evidence from
found evidence for the earlier intermittent trade from other coastal sites for pre-first century A.C. settle-
his "intermediate" levels. It is important to pursue thements and trade networks has also been set out in the
matter of trade further; to do so, it is necessary at thisdiscussion of Phase B.
point to separate the commerce of the first few cen- At Arikamedu, the amphora sherds in Phase C are
turies A.C. from that of the preceding period. certainly of Mediterranean origin, and the stamped
decorations on some of the other pottery (Wheeler's
Pre-Roman Phase Type 10) also seem to have been influenced by West-
The archaeological evidence at Arikamedu for pos-ern ceramic traditions. Still earlier, in Phase B, the
sible pre-Roman contacts with the Mediterranean technique of "rouletting" also appears to have been
area, discussed above under Phases B and C, suggests introduced from the West. Rouletted Ware was most
that there was some systematic trade between the probably manufactured locally, but whether some ac-
southeastern coast of India and the West-probably tual imported prototypes exist at Arikamedu will re-
through Arab or other intermediaries-prior to main the uncertain until the ceramics are better analyzed.
beginning of Indo-Roman trade in the first century The trade network for Rouletted Ware was vast
A.C. Furthermore, the evidence from Casal's excava- (ill. 1), although its exact extent during the pre-Ro-
tions indicates that Arikamedu was first settled after man phase remains to be worked out precisely. It did
the period of Mauryan contact with South India. Aso- extend at least as far north as Salihundam in Andhra
kan inscriptions clearly point to diplomatic connec-Pradesh, since some of the inscribed Rouletted Ware
tions between the Mauryans and the Mediterraneansherds from that site date from the second and first
area during the third century B.C., for Asoka sent cul-centuries B.C. Thus, the cumulative evidence implies
tural missions to the kingdoms of contemporary Hel- that there was a series of interrelated southeastern
lenistic rulers-specifically Antiochus II Theos ofcoastal settlements during the second and first cen-
Syria, Ptolemy II Philadelphus of Egypt, Antigonusturies B.C. Arikamedu was already an established
Gonatas of Macedonia, Magas of Cyrene and Alex- trading center during the pre-Roman phase, but wit-
ander of Epirus.78 nessed intensification of trade with the beginning of
Asokan missionaries most probably traveled over-
the Roman phase.
land, for the northern route of communications was
well established from the time of Alexander's invasion Roman Phase
of the northwestern regions of the Indian subcontinent Many scholarly studies exist on the question of Ro-
man trade with India; it is not the intent here to re-
in 327/6 B.C. It is quite likely that extensive contacts
with the Hellenistic world would have extended to view this extensive literature.8" The discussion below
South India as well, especially to the Coromandel
focuses primarily on the archaeological evidence for
that trade at Arikamedu, and how it relates to the
coast which was known for its pearl industry. As early
a writer as Megasthenes speaks of the pearl industry
South Indian peninsula. The evidence from early Ta-
77 Wheeler et al. 22. 1928); J.I. Miller, The Spice Trade of the Roman Empire (Oxford
78 Major Rock Edict XIII of Asoka; for text and translation see Important sources dealing more specifically with the Indo-
1969).
Roman trade are: P.L. Gupta, "Roman Trade in India," in S.S.
E. Hultzsch, Inscriptions of Asoka (Corpus Inscriptionum Indica-
Mookerji
rum 1, Delhi 1969 reprint) 22-25; for identifications of names see Felicitation Volume (Varanasi 1969) 169-80; W.
R. Thapar, Asoka and the Decline of the Mauryas (Oxford 1961)Schmitthenner, "Rome and India: Aspects of Universal History
40-41. During the Principate," JRS 69 (1979) 90-106; H.G. Rawlinson,
79 J.W. McCrindle, Ancient India as Described by Megasthenes
Intercourse Between India and the Western World (Cambridge
and Arrian (Calcutta 1960) 60-61, 115. 1916); J. Thorley, "The Development of Trade Between the Ro-
80 For Roman trade with the East, the most comprehensive man
gen- Empire and the East under Augustus," Greece and Rome 16
eral works are: M.P. Charlesworth, Trade-routes and Commerce of 219-23; Wheeler, Rome Beyond the Imperial Frontiers (su-
(1969)
the Roman Empire (Cambridge 1926); E.H. Warmington,pra Then. 32).
Commerce between the Roman Empire and India (Cambridge

164.100.151.123 on Fri, 22 Dec 2023 07:07:40 +00:00
mil literature is limited and

in beryl ofand
deposits uncertain date;
could also have been co
the supply
source
quently it is useful only for for other commodities, suchthe
reinforcing as pepper and cot
premi
ton. Thus,and
a prosperous overseas trade the geophysical
the arrivalfeatures ofofthe region
for
mercantile vessels intowould
the harbors
partially of
account for theSouth
accumulationIndia
of early
finds in that
Prior to the excavations at area.
Arikamedu, the mo
The locations of the for
important artifactual evidence coin findsRoman
in the Coimbatoretrade
South India was the large number
region, combined with referencesof Roman
to numerous ports den
and aurei found in "hoards"
on the westernsince
coast in the 1775,
Periplus Mariseach
Erythraei cont
ing from one to several hundred
led Wheeler coins.82
to hypothesize that Some
early Roman trade
enty-five finds are known,
(i.e., from the totaling
time of Augustus to several thou
Nero) was with the
coins-the vast majoritysouthwestern
of which (Malabar) coast
wereand not the southeast-
minted du
the time from Augustus ern (Coromandel).
to Nero. We have no certain evidence
Many of astheto fi
spots for these early coins arewhat
when and under clustered
circumstances Roman inships
the C
batore region of peninsular India,
rounded the peninsula and direct
and established consequen
trading
seem to have significance
emporiafor the Arikamedu
on the southeastern coast. Despite the pres-trad
ence of Arretine
10). Surprisingly, no Roman ware in hishave
coins excavations been
at Arika- foun
Arikamedu, nor are they reported
medu, Wheeler from
maintained that during the other
first part co
sites as a general rule.83
of the first century A.C. trade was most probably car-
The Coimbatore region is
ried on on
from the
the ports main
of the route
Malabar coast. From of
land communication between the western and the there commodities were transported overland via the
eastern coasts via the Ponnani valley. The area is richCoimbatore gap to Roman emporia like Arikamedu
on the east coast-the eastern emporia being neces-
sary for controlling the indirect trade with the North
.:::':':' PONDICHERRY and Sri Lanka.
In 1951, Wheeler claimed that "down to the third
COIMBTOREt TRANQUEBAR
quarter of the 1st century A.D., therefore, there was
c %o,
PALG . -.. 'c ve
d PT NJORE
CRANGANUC
O a no regular direct communication between Rome and
Ceylon. It follows that Roman vessels were not at that
time in the habit of rounding Cape Comorin; and this
' MADURA OF
inference confirms and explains the importance of the
CIO,
o, " ."oBENGAL
overland route from west to east through the Coimba-
tore gap during the period of the Coimbatore hoards,
i.e. from Augustus to Nero."84 In 1954, Wheeler reit-
4110CEYLON erated this belief: "It is a fair inference that Roman
CAPE COMORIN
agencies established in the east coast ports under Au-
SS/NGLE COINS .- HEIGHTS Scale of Miles gustus and Tiberius were, so far as the Westerners
SHOARDS .ABOVE 20 0 20 40 60 80 100
were concerned, the termini of trans-peninsular
O PLACES .- 2000 FEET routes, and that only towards the end of the 1st cen-
Ill. 10. Map of South India showing distribution tury
of Roman
were the western and eastern ports linked also by
coins of first century A.C. (After Wheeler, Rome regular circumpeninsular traffic."85 However, in the
Beyond
the Imperial Frontiers fig. 17) light of his and Casal's excavations at Arikamedu, it is
81 K.V. Zvelebil lists ten references to the Yavanas in the Tamil
dhra Pradesh (Hyderabad 1965). Five relatively recent finds have
Sangam texts: The Smile of Murugan (Leiden 1973) 35, n. 1.been For reported in IndArch 1956-1957, 38; 1970-1971, 62;
the dating of the literature see his Tamil Literature (Leiden 1975).
1972-1973, 55; 1975-1976, 72; 1976-1977, 71.
82 A list of Roman coins found in India, compiled from earlier 83 One coin is reported from the excavations at Nagarjunikonda:
sources, was published by Wheeler et al., 116-21, along with seethe
IndArch 1956-1957, 38; one coin is said to have been found at
Arikamedu excavation report. A revised version of the same list ap-
Kaveripattinam, but no other details have been furnished by Ra-
pears in Wheeler, "Roman Contact" (supra n. 32) 375-81. For a (supra n. 8) 239.
man,
more up-to-date listing, see C. Rodewald, Money in the Age84ofWheeler, "Roman Contact" (supra n. 32) 370.
Tiberius (Manchester 1976) table 5, n. 378. P.L. Gupta made
85 aWheeler, Rome Beyond the Imperial Frontiers (supra n. 32)
144-45.
study of a large number of coin collections: Roman Coins from An-

164.100.151.123 on Fri, 22 Dec 2023 07:07:40 +00:00
through
difficult to accept Phases E, Fproposition
the and G. The imports that survive th
trade to and from fromArikamedu did
these phases are primarily ceramics: not
approxi-
third quarter of mately 150 first
the sherds of amphorae (including those from A.C
century
At present, the Phase SouthC) and about 50 Indian
or so sherds of Arretine ware,
cerami
does not support Wheeler's claim fo representing an uncertain number of vessels. Other
peninsular trade artifacts route of Mediterranean
from origin consist the
of fragmentsMal
mandel coast. Rouletted Ware occurs on the Coro- of glass bowls, two fragmentary Roman lamps, a crys-
mandel coast and Sri Lanka, but its distribution on
tal gem and perhaps a stylus. The total number of
the Malabar coast is not yet known, while Russet- imported artifacts is not impressive per se; in assess-
coated Painted Ware-a distinctive ceramic of the ing the quantitative value of these finds for determin-
Early Historical period-is quite common in the ing the volume of trade, it should, however, be re-
Coimbatore region and occurs on the Malabar coast
membered that only a mere fraction of the original site
has been excavated, and that the most vital area of the
as well. Russet-coated Painted Ware is represented
by only one sherd at Arikamedu, and as a rule is settlement
not in the Northern Sector has been perma-
found on the Coromandel coast.86 The two ceramic nently destroyed by the river.
Items of export from Arikamedu were probably
types do occur simultaneously in the interior, for in-
jewelry, specifically beads of semi-precious stones and
stance at Uraiyur on the river Cauvery.87 The distinc-
tive distribution of these two important and wide-shell bangles; worked ivories; textiles; and perhaps
spread ceramic types suggests that trade betweenleather
the or leather-products.89 Spices, incense and oth-
eastern and western coasts was not direct, but waserin-
perishables were most probably exported, but no
archaeological
stead through intermediate centers such as Uraiyur. If documentation exists. Much of the in-
this hypothesis proves to be correct, then it would dustrial production must have been local, but except
seem doubtful that "Roman" merchants regularly for tra-shell and perhaps amethyst, raw materials must
versed the inland route from the western to the east- have been brought from the inland regions.
ern coast, as Wheeler proposed. It is puzzling, however, that Arikamedu appears to
Returning to the question of coins, P.L. Gupta in- be the only site of its kind so far, for one could rea-
terprets the numismatic evidence slightly differentlysonably expect other comparable trading centers on
from Wheeler. In trying to reconstruct the chronologythe southern part of the Coromandel coast, simply be-
of the trade on the basis of the coin types and theircause of geography. Some of the other sites where
distribution, he concludes that in the first centuryRouletted Ware occurs may also have been engaged in
A.C. trade was carried on from ports like Tondi andoverseas trade with the West, but more extensive re-
Muziris on the Malabar coast, and Korkai and Kave- search is needed before definite conclusions can be
drawn.
ripattinam on the Coromandel; later the trade shifted
to ports farther north."8 Although this is quite pos- Wheeler's hypothesis that Arikamedu may have
sible if we include Arikamedu on the list, nevertheless
been a supply center for other settlements on the east-
the ports of the Malabar coast have not yet been ar-ern seaboard (and those in Sri Lanka should also be
chaeologically identified; Korkai and Kaveripattinam,included) has considerable validity. In fact, the hypo-
although partially excavated, do not have the same thesis is strengthened by more recent evidence per-
kind of precise documentation for Mediterraneantaining to the distribution of Rouletted Ware along
trade as Arikamedu during the first century A.C. the eastern coast. Since Wheeler's excavations, some
At Arikamedu, the period of Roman trade starts inarchaeological work has been done in Sri Lanka as
Phase D-i.e., toward the beginning of the first cen-well, and three sites yielding Rouletted Ware have
tury A.C.-and continues for about two centuries,been excavated. At Kantarodai, in northern Sri
86 One sherd from Arikamedu, illustrated on pl. 31.C1 of Wheel- pear to fall within this category: see R. Nagaswamy, "Roman Sites
er's report, is clearly of the Russet-coated Painted Ware type, both in Tamil Nad: Recent Discoveries," in Madhu: Recent Researches
in shape and decoration, although it is not labelled as such. Since it in Indian Archaeology and Art History (Delhi 1981) 337-39.
is from Wheeler's "Southern Sector, Early Phase," it should belong 88 Gupta (supra n. 80) 172.
to either Phase C or D of the present classification. 89 Leather-working has been suggested by J.C. Nagpall on the
87 For summaries of the Uraiyur excavations, see IndArch basis of chemical analysis of limeshell incrustation on some of the
1964-1965, 25-26; 1965-1966, 26; 1967-1968, 30-31. Two other pots from Arikamedu. See Nagaswamy and Abdul Majeed (supra
sites, Karur and Kodumanal, published in a recent article, also ap- n. 8) 25-27.

164.100.151.123 on Fri, 22 Dec 2023 07:07:40 +00:00
Lanka, a Roman-type
A largecarnelian
volume of the trade of thiscameo
period must have of t
century A.C. was found in
come from the the
ports period
of the Roman of Ro
Empire. Yet evi-
Ware during the course
dence thatof athelimited
some of explorato
merchant-sailors trading with
vation.90 Two sherds
the of Rouletted
Mediterranean Ware
via the Red Sea may have beenwere
at Anuradhapura, andTamil speakers
some now comes
are from two short
also Tamil-
reported
the current excavations at Mantai.91
Brahmi inscriptions on potsherds found in recent ex-
The cumulative evidencecavations at theseems
Egyptian coastalto imply
site of Quseir al- th
tence of an extensive Rouletted Ware trade network. Qadim, in contexts associated with the period of Ro-
Many of the Rouletted Ware sites have only occasion- man trade.92 Both inscriptions have been identified as
al items of Roman imports, the exception being Ari- parts of proper names, and dated on epigraphical
kamedu (and perhaps Vasavasamudram), which has grounds to the first or second century A.C. This is
extensive material of Western origin. Therefore the precisely the period when Arikamedu flourished as an
archaeological evidence at present indicates two trade Indo-Roman emporium, in continuation of what ap-
networks on the Coromandel coast: one along the pears to have been the long standing function of the
eastern seaboard including northwestern Sri Lanka, site as a trading center.
the other trading with the Mediterranean directly-- R.R. 6, BOX 247A
Arikamedu is part of both networks. IOWA CITY, IOWA 52240
90 The cameo was found in the 1970 excavations conducted by the Chicago), who is presently excavating the site.
University of Pennsylvania Museum, and publication of the mate- 92 For the inscriptions and Mahadevan's reading, see D.S. Whit-
rial by Bennet Bronson is expected in the near future. For an ear- comb and J.H. Johnson, Quseir al-Qadim 1980 (American Re-
lier survey see V. Begley, "Archaeological Exploration in Northern search Center in Egypt Reports, Malibu 1982) 263-64. See also
Ceylon," Expedition 9.4 (1967) 20-29. D.S. Whitcomb and J.H. Johnson, Quseir al-Qadim 1978 (Cairo
91 For Anuradhapura, see supra n. 26. The information on Man- 1979) pl. 27.j.
tai is from John Carswell (The Oriental Institute, University of

164.100.151.123 on Fri, 22 Dec 2023 07:07:40 +00:00

Begley ArikameduReconsidered 1983

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Begley ArikameduReconsidered 1983

Uploaded by

Copyright:

Available Formats

Arikamedu Reconsidered

Author(s): Vimala Begley

Stable URL: https://www.jstor.org/stable/504104

Archaeological Institute of America is collaborating with JSTOR to digitize, preserve and

This content downloaded from

Abstract which flourished during the first two centuries A.C.

This content downloaded from

inscribed sherds; providedandevidence for 4) periodic the

This content downloaded from

0 100 200 300 /

t Pauni . Sisupa garh

Chandravall S Souttouoon MADRAS

Ill. 1. Southern and eastern India, showing

This content downloaded from

sal's1950 surveyother of the

This content downloaded from

AK11, III etc.

CASALS EXCAVATIONS (19478; 1950)

";* Ar. B-HJ,7M

GROUP I WHEELER S EXCAVATIONS (1945)

Ill. 2. Site plan of Arikamedu. (Redrawn after Casal, Site urbain)

This content downloaded from

REVISED CHRONOLOGICAL SEQUENCE OF ARIKAM

WHEELER'S AND CASAL'S EXCAVATIONS

B.C 20 200 150 100 50 B.C./ A.C. 50 10

Local Black-and-Red Black-and-Red; Black-and-Red(??

Imported glass bowl bone stylus(?) glass bowl

This content downloaded from

WEST Ar4 EAST

Ill. 3. Section of Arikamedu trench Ar. 4, Southern S

ing, since it implies thatcause

This content downloaded from

third-second In ceramics, Rouletted

MEAN *. LtI N 7 S CTO

Ill. 4. Section of Arikamed

18 D.P. Agrawal and

This content downloaded from

Wheeler's the with

tine ware layers as well, but at the time of excavation ?. ?;i;rz

period.23 The presence of Black-and-Red Ware sug-

This content downloaded from

wide range of dates-it was

This content downloaded from

excavations in these layers were

This content downloaded from

This content downloaded from

Ill. 7. Arikamedu imported amphor

"Kanchipuram Excavations," Journal of Andhra Historical 52Re-

This content downloaded from

(Pottery Group Arretine

.._ _ --- - .... --

This content downloaded from

basis of these studies,er's

This content downloaded from

This content downloaded from

cure"-the dates of Asokan epigraphs

This content downloaded from

station which was of thefirst

This content downloaded from

mil literature is limited and

This content downloaded from

This content downloaded from