RESEARCH

RESEARCHARTICLE
ARTICLE

Cultural sequence of Bet Dwarka island based

on thermoluminescence dating
K. H. Vora*, A. S. Gaur*,†, David Price‡ and Sundaresh*
*National Institute of Oceanography, Dona Paula, Goa 403 004, India
‡
Thermoluminescence Dating Laboratory, School of Geosciences, University of Wollongong, Northfields Avenue, Wollongong, NSW 2522, Australia

Bet Dwarka has rich marine resources such as the

Earlier studies on the onshore and inter-tidal zone conch shell, which possibly attracted early man (late
explorations in Bet Dwarka island had revealed the
Harappan) to establish a colony. Another important site
presence of a large number of potsherds and other
antiquities that correspond to the protohistoric (2000– of Harappan period at Nageshwar is located within 15 km
1500 BC), historical (3rd century BC to 5th century in the mainland of Okhamandal. This was one of the active
AD) and medieval (8th century to 15th century AD) shell industry centres of the Harappan period. Similarly,
periods. Until the present time there has been a lack Bet Dwarka was an important shell-working centre
of absolute chronology. This study therefore attempts during late Harappan and historical periods. Onshore and
to address this deficiency. Accordingly, nine potsherds offshore explorations in and around Bet Dwarka have
from four different sites of Bet Dwarka island were yielded a large number of antiquities of late Harappan
dated by thermoluminescence (TL). The ages obtained and historical periods which include pottery, a seal,
from site BDK-VI vary between 3870 and 2220 years coins, etc.2. To understand the cultural sequence and its
BP; those from BDK-I vary between 3160 and 830 continuity, a few potsherds were examined and were
years BP; those from BDK-II vary between 1780 and
scientifically dated using thermoluminescence (TL), when
960 years BP, and those from Khuda Dost Dargah
vary between 1240 and 880 years BP. When analysed found to be of historic importance the results of which
on the basis of their locations, they are in conformity are presented here.
with the expected period. Interestingly, at site BDK- It is important to prepare a chronology of an archaeo-
VI and BDK-I a few potsherds, believed to be of logical site to understand the sequence of events and
protohistoric period, are apparently considerably more correlate the same with different sites. Once the chrono-
recent (2000 years BP), which may suggest the conti- logy is established, it is possible to compare two sites. If
nuation of protohistoric habitation up to historical two sequences from different chronological events are
period at the same site. These TL ages assist in esta- found together, then that can form the basis to consider
blishing a cultural sequence for Bet Dwarka island. the occurrence of neotectonic activity. To obtain absolute
dates, generally sedimentary strata provide the proxies
BET Dwarka island (22°22′ to 22°28′N and 69°05′ to for obtaining dates either using 14C or U-series methods.
69°09′E) is situated in the Gulf of Kachchh about 5 km In archaeological explorations, however, antiquities are
north of the mainland of Okhamandal and to the east of used to obtain dates based upon cultural association as
Okha port. Bet Dwarka (also known as Bet Shankhodhar) standard procedure. The dates so obtained are generally
is famous for temples dedicated to Lord Krishna. The reliable and depended upon, but they are subjective.
island is a narrow strip of sand and rocks about 13 km However, at times scientific dates are very important. TL
long. The eastern part of the island is known as Hanuman dating of potsherd is now gaining ground to get absolute
point, and comprises sand-hills and bushes. The south- ages from them. An attempt is made here to discuss the
west half of the island is rocky table with relief of utility of TL ages in identifying cultural sequence from
+ 20 m. Recent geological studies have indicated faults Bet Dwarka.
and two prominent structures: the Padmteerth fault A number of datable artifacts such as seal, inscriptions,
running WNW-ESE and the Sonimiyar fault running coins and pottery have been reported from the island,
ENE-WSE, though the dates of the faults are not avai- which have helped to construct an archaeological cultural
lable at present. It is suggested that these faults caused sequence for the island3. To gain a better understanding
wedging of the area between them, thus causing sub- of the chronology of the periods, a programme of scien-
sidence. Evidence of subsequent in-filling and epi- tific analysis was undertaken using TL as a dating method.
sodes of uplifts during the recent past have been Potsherds from different periods were collected, exa-
reported1. mined and later dated at the School of Geo-sciences,
University of Wollongong, Australia. All potsherds were
collected from the surface and have been held in storage
†
For correspondence. (e-mail: asgaur@csnio.ren.nic.in) since collection (Figure 1). The present paper deals with
CURRENT SCIENCE, VOL. 82, NO. 11, 10 JUNE 2002 1351
RESEARCH ARTICLE

these archaeological discoveries and their associated TL Since 1969–70, onshore explorations at Bet Dwarka
dating. have been carried out by the Archaeological Survey of
TL dating has been used in various disciplines such as India (ASI), which established the existence of the late
earth sciences, archaeological science and space sciences4,5. Harappan, post Harappan, historical and medieval period
Recently, the use of TL dating has increased significantly settlements in different parts of the island10. Rao11 exp-
in archaeology6,7,. Potsherds manufactured by man during lored the island extensively and undertook underwater
the ancient period are used in TL dating. It has been exploration during 1983. During this exploration, many
observed that application of TL dating in historical pieces of pottery and sherds with graffiti marks, conch
archaeology is very effective. In India a few potsherds shell bangles, etc. were recovered. Explorations have
from Harappan settlements at Sanghol and Hulas and brought to light the following cultural sequence cons-
historical period settlements at Sringaverpura and Bagor tructed by comparative studies with contemporary archaeo-
were dated by this method6 to confirm their period. A logical sites of Gujarat.
few potsherds from Bet Dwarka were also dated and only
one date has been referred as 3520 years BP (ref. 8).
Unfortunately, TL dating has not been used extensively
Protohistoric period settlement
for archaeological sites.
The protohistoric period in the Indian subcontinent has
been bracketed between 5000 and 3000 years BP. Two
Archaeological background late Harappan sites have been discovered on the island,
these are located on the eastern and southern coast. The
Bet Dwarka island is located at a point from which boats
eastern site is shielded from tidal effects. However, recent
and ships sailing from south enter the Gulf of Kachchh,
agricultural activities have extensively damaged the site.
approach the mainland sailing in a northwesterly direc-
Important finds from this site include red-ware and grey-
tion. The area is also referred to as the Gulf of Barake by
ware bowls and sherds of perforated jars12. This type of
foreign travellers9.
pottery is typical of the Harappan (Figure 2) and late
Archaeological evidence testified by the discovery of
Harappan sites. A major portion of the southern coastal
seal inscriptions and pottery suggests that Bet Dwarka
island was inhabited during the Harappan period and
habitation has continued with hiatus between proto-
historic and historic period, until the present day.

Figure 1. Location of sites and samples. Figure 2. Protohistoric pottery from Bet Dwarka.

1352 CURRENT SCIENCE, VOL. 82, NO. 11, 10 JUNE 2002

 RESEARCH ARTICLE

site has been damaged by sea erosion (Figure 3). A large been identified on the island on the basis of findings like
number of potsherds and shell bangles lie scattered in the coins and pottery. The first site lies over a late Harappan
inter-tidal zone and are visible during periods of low tide. settlement near Nilakantha Mahadev temple. This site has
Land and underwater excavations at this site have yielded two Kushana period copper coins and one un-
brought to light a seal (Figure 4) made on a conch shell1 identified historical period lead coin3. The second site is
engraved with an animal with three heads. Similar types on the southwestern coast near Khuda Dost Dargah
of seals have also been reported from Mohenjodaro, (Figure 6). This site has natural protection from high sea
Harappa and Dholavira from Harappan level. Other imp- waves and storms but gets inundated during high tide,
ortant antiquities include potsherds with Indus period suggesting sea-level rise or shoreline retreat on this coast
graffiti (Figure 5), Harappan pottery3, shell bangles, chert since the time of habitation. The reason for this pheno-
blades, etc. menon has not been identified as yet. Important artifacts
found at this site include amphora sherds, pottery, conch
and shell bangles. Recent discovery of amphora and a
Historical period settlement lead ingot of the Roman period indicate that this site
probably was involved in trade with the Mediterranean
The historical period represents a time bracket between
world.
2600 and 1400 years BP. Two historical period sites have

Medieval period settlement

Medieval period settlement has been identified at a number

of places on the island. The majority of present-day
habitation is in fact built over medieval period ruins. A
number of coins, pottery, iron cannons and stone anchors

Figure 3. Habitational remains exposed in a cliff section in BDK-I.

Figure 5. Harappan graffiti over a potsherd.

Figure 6. Archaeological site near Khuda Dost Dargah exposed

Figure 4. Harappan seal made on conch shell. during low tide.

CURRENT SCIENCE, VOL. 82, NO. 11, 10 JUNE 2002 1353

RESEARCH ARTICLE

belonging to this period have been found on the island. fraction was settled out of a 6-cm column of acetone onto
The Island had good contact with mainland settlement in a series of 1-cm aluminium planchettes. Sixteen such
the Okhamandal area. In fact, most of the modern sample aliquots were prepared for each sample analysed.
villages came into existence during the medieval period. Five aliquots were used for the determination of natural
The discovery of five iron cannons suggests that the TL accumulated since the time of firing. These aliquots
island was protected with a defence system during the were further utilized to determine the supralinearity cor-
late medieval period. A large number of stone anchors rection for each sherd14. Seven aliquots were serially and
found near present-day Bet Dwarka jetty indicate that incrementally irradiated by means of a calibrated (90)Sr
this location was used for anchoring boats during the plaque radiation source and the remaining four using a
medieval period. calibrated (241)Am radiation source, and the subsequent
data used to determine the k-factor as described by
Aitken14. All sample aliquots were heated to 500°C at a
Evidences on sea level/shoreline changes rate of 20°C/s in a high-purity nitrogen atmosphere. The
natural and laboratory-induced TL emitted was detected
As stated above, two archaeological sites have been
with an EMI 9635QA photomultiplier fitted with a Corn-
affected by the shoreline changes. One historical period
ing 9–59 blue transmission filter. All first-glow TL
site (Figure 6) becomes submerged during high tide near
outputs were corrected for aliquot-to-aliquot variation
Khuda Dost Dargah13. Another protohistoric site has been
using a second-glow procedure involving a secondary
severely damaged by the sea and a large number of
irradiation of approximately 28 greys. In a single case
pieces of pottery and other artifacts can be found in the
only (W2879) did there appear to be a suggestion of TL
inter-tidal zone. This evidence suggests that some kind of
sensitivity change due to the laboratory procedure fol-
shoreline change has taken place since 2000–3000 years
lowed. Finally, the archaeological radiation dose accu-
BP. A geological study by Shakhawala and Shringarpure1
mulated since the time of the last firing was determined
indicates two faults on the island located near Pancha-
from the natural and beta-irradiated data and the alpha-
tirtha and also Soni Miyar. At least one of the historical
radiation efficiency factor (k) from the natural and
period sites is located upon the Soni Miyar fault. Tec-
alpha-induced TL data. The k value was used in the
tonic activity therefore provides the explanation as to
determination of the amount of radiation received annu-
why this site becomes inundated. The submergence of
ally by the potsherd under study.
sites near Nilkanth Mahadev Temple, however, requires an
The annual radiation dose received by the sherds was
alternative explanation and must be further investigated.
determined by means of thick-source alpha counting to
determine the radiation contribution from uranium and
TL dating thorium, and flame spectroscopy to determine the potas-
sium content of each of the sherds. Smaller, compa-
At the time of firing, any previously acquired TL stored ratively negligible, contributions from rubidium were
within the crystalline lattice of TL-sensitive minerals included assuming a value of 100 ppm, and an annual
contained within a piece of pottery is reset to zero. From cosmic contribution of 185 microgreys was also assumed.
that moment onwards the TL signal begins to rebuild at a As the radiation flux-levels from within the sherds is
rate dependent upon the amount of the long-lived iso- relatively high (average approximately 5000 microgreys
topes of uranium, thorium and potassium present. Rubi- per year) these contributions are comparatively insigni-
dium and cosmic radiation play a smaller contributory ficant. The environmental (gamma) radiation dose derived
role and the entire radiation flux level is modified by the from the soil, etc. in the nearby vicinity of the sherds
presence of water. Thus, if the amount of stored TL was computed assuming that the uranium, thorium and
energy induced since firing (archaeological dose) and the potassium levels were the same as those measured for the
radiation flux level are measured, the age of the potsherd sherd itself. It was not possible to refine this measure-
may be computed from the basic equation: ment as soil samples were not available for this purpose.
Archaeological dose The final annual radiation flux-level computed for each
TL age = . sherd was corrected for an assumed average water
Annual radiation dose
content of 10%. Extremely crude, porous pottery may
have a maximum water content as high as 25% (ref. 14),
TL dating methodology but the pottery studied here is not of a porous nature.
However, an underestimation of the pottery moisture
Each of the nine potsherds analysed and reported in the content of 1% will result in underestimation in age of
study were dated by means of the fine-grain additive approximately 1%. Hence the final age of sample W2879
method as described by Aitken14,15. This technique uti- becomes 1460 years BP assuming a 15% moisture con-
lizes the 1–8 micrometre polymineral grain size fraction tent, as opposed 1380 years BP (uncorrected for fading)
extracted from an internal portion of the sherd. This assuming a 10% water content.
1354 CURRENT SCIENCE, VOL. 82, NO. 11, 10 JUNE 2002
 RESEARCH ARTICLE

Because of the polymineral nature of these samples, it lected with the sole purpose of TL analysis. The sherds
is possible that the TL-sensitive minerals may not store were properly identified and labelled accordingly before
TL in a stable fashion over long time periods. Aitken14 storage under dry, clean, ambient conditions. Given the
suggests that the first defence against such an effect is the relatively brief storage period compared to the age of the
so-called temperature plateau comparison, where the sherds, the environmental radiation dose delivered to the
shape of the first TL glow-curves are compared with that sherds during this time is relatively negligible. More
of the recently, laboratory-induced glow-curves over a appreciable is the environmental radiation dose delivered
given temperature range (200–500°C). If the trapped prior to sample collection, as this is totally unknown and
electrons are stored in a stable fashion, a plot of the ratio can only be assumed to be as that delivered to by each of
of the natural TL divided by the natural TL plus the sherds internally. This uncertainty is acknowledged
laboratory-induced TL against temperature should result and any future TL analysis should include measurement
in a plateau region between 300 and 500°C. Aitken14 also of the environmental radiation dose levels delivered from
suggests that all laboratory irradiations should be the burial soil along with the burial depth, to enable the
performed 24 h before TL measurement. In order to correct cosmic radiation contribution to be made. These
further guard against the effect of TL fading, all sample sherds should therefore be collected with the purpose of
aliquots were preheated to a temperature of 150°C for TL analysis in mind. Mindful of the limitations of these
5 min prior to TL measurement. Finally, and in addition present determinations, we are of the opinion that this
to the precautions described, the amount of TL fading study forms, at the very least, a first step in the under-
over a minimum 21-day period at 70°C was actually standing of the chronology of pottery in the study area.
measured for each potsherd. A series of sample aliquots
were irradiated (c. 56 greys) and stored at the conditions
specified. Following storage, the TL output exhibited by TL dating results
these aliquots was compared with the output displayed by
similar newly irradiated sample aliquots. This procedure Nine potsherds have been analysed using TL and with the
is described by Aitken14 and takes advantage of the fact exception of three, all samples are more or less similar to
that most TL fading takes place over the initial storage our assumption (Table 1). Two dates, i.e. W2876 and
period. The results of this exercise are shown in Table 1, W2877 are very close to, or fall into the protohistoric
where sample ages are given in both uncorrected and period if the laboratory TL-fading loss is taken into
corrected form. account. The age of the third potsherd (W2878) is also
The analytical technique followed for this chrono- close to the protohistoric period at the one standard devi-
logical exercise differs from that generally adopted in the ation level.
TL authentication of pottery. Here, annual radiation flux Sample W2876 is from a perforated jar collected from
levels are measured using full samples for determination site BDK-VI, which is popularly accepted as Harappan
of sample-specific activity and potassium content. In the pottery that continued through until late Harappan. The
case of TL authentication, the amount of sample avai- final TL age for this potsherd is 3260 ± 470 years BP.
lable is prohibited and these measurements must gene- However, after the fading correction this age becomes
rally be conducted on lesser representative sample amounts. 3380 ± 490 years BP. Late Harappan culture anywhere in
TL fading tests were also conducted, which is generally the Indian subcontinent closes around 1500 BC when
not the case for TL authenticity analysis. perforated jars disappear from the archaeological sites1.
The sherds analysed in this study were collected 10 A second potsherd represents a rim portion of red-ware
years prior to the actual TL analysis and were not col- bowl. Similar types of bowls have been found at Rang-

Table 1. Thermoluminescence ages of pottery from Bet Dwarka island

Sl. Laboratory Anticipated age Uncorrected Corrected

no. Site reference Pottery type (years BP) for fading for fading

1 BDK-VI W2876 Perforated jar 4500–3500 3260 ± 470 3380 ± 490

2 BDK-I W2877 Red-ware carinated dish 4500–3500 2460 ± 330 2790 ± 370
3 BDK-VI W2878 Red-ware bowl 4000–3500 2300 ± 200 2430 ± 210
4 BDK-I W2879 Rim portion of a jar Unknown 1380 ± 150 1440 ± 160
5 BDK-I W2880 Grey-ware bowl 4000–3500 1960 ± 160 2140 ± 170
6 BDK-I W2881* Red-ware lid 2000 740 ± 50 890 ± 60
7 BDK-Khuda Dost W2882 Red-ware bowl 2000 1000 ± 170 1060 ± 180
8 BDK-II W2883 Red-ware bowl 2000 1270 ± 380 1370 ± 410
9 BDK-I W2884 Red-ware bowl 4000 970 ± 120 1060 ± 130

*This sample (W2881) exhibited a TL-fading factor of more than 20% over the period of the test. This casts a degree of doubt
upon the reliability of the final TL age computed.

CURRENT SCIENCE, VOL. 82, NO. 11, 10 JUNE 2002 1355

RESEARCH ARTICLE

pur, Prabhsa Patana, Lothal and other Harappan and late Conclusions
Harappan sites in Gujarat. This sherd was collected from a
site near Balapur which has been designated as BDK-VI. To some extent, the TL ages of the pottery from Bet
A few sherds belonging to perforated jars and carinated Dwarka island here endorse the hypothesis on the cultural
dishes suggest that this was the late Harappan settle- sequence of Bet Dwarka island established earlier. Accor-
ment16. The TL age for this potsherd is 2460 ± 330 years dingly, habitation record at the island begins from proto-
BP and the corrected date is 2790 ± 370 years BP. At the historic period (4000 years BP) to medieval period (500
one standard deviation level, this date falls into the proto- years BP). When seen in conjunction with other finds
historic period. However, two sherds (W2878 and W2880) from the island, i.e. seal inscriptions, large number of
appear to be associated with the historical period, although anchors, shell working sites, etc., we can suggest that the
our archaeological records suggest that these should be island of Bet Dwarka was an important active port since
from the protohistoric period. One of these, a potsherd ancient times and remained in the limelight for a long
from a grey-ware bowl again belongs to the late Harap- time. Bet Dwarka has provided a continuous habitation,
pan period and was recovered from BDK-VI, and the which suggests that this island played a vital role in
other from BDK-I. The remaining pots could have their maritime activities since Harappan times. The apparent
origin in the historical and medieval periods as jars, continuation of the protohistoric pottery into the histori-
basins and bowls continue until the present day. To cal period needs careful study. When confirmed, this will
some extent therefore, this allays concern regarding the help in unravelling the problem of Okhamandal archaeo-
results of the dating of other potsherds. However the logy, and these present TL ages will provide the initial
important point regarding the dating of other sherds is evidence for a solution to this conundrum.
that the results suggest a continuous habitation on the
island.
1. Shakhawala, M. V. and Shringarpure, D. M., Mem. Geol. Surv.
India, 1990, 16, 111–116.
2. Rao, S. R., J. Mar. Archaeol., 1990, 1, 59–98.
3. Sundaresh and Gaur, A. S., Man Environ., 1998, XXIII, 77–86.
Discussion
4. Singhvi, A. K. and Wagner, G. A., Dating Young Sediments (eds
Hurford, A. J., Jager, E. and Tencate, I. A. M.), Bangkok, 1986,
The TL analysis of the Bet Dwarka potsherds has pro- pp. 159–197.
vided an opportunity to give a proper time bracket for 5. Singhvi, A. K., Bluszcz, A., Bateman, M. D. and Rao, M. S.,
Earth-Sci. Rev., 2001, 54, 1–3, 193–211.
Bet Dwarka archaeology, without dispute (Table 1). The 6. Chawla, S. and Singhvi, A. K., Naturwissenschaften, 1989, 76,
TL age of one of these potsherds, W2876 (3380 years 416–418.
BP, corrected) falls within the protohistoric time period 7. Kuzmin, Y. V., Hall, S., Tite, M. S., Bailey, R., O’Malley, J. M.
(5000–3000 years BP), which substantiates a settlement and Medvedev, V. E., Quat. Sci. Rev., 2001, 20, 945–948.
during the late Harappan period. Similarly, the age of 8. Rao, S. R. (ed.), Marine Archaeology of Indian Countries, NIO,
Goa, 1988, pp. 47–53.
another sherd (2790 years BP, corrected) is quite close to 9. Schoff, V. (ed.), The Periplus of the Erythrean Sea, London, 1912,
the end of the protohistoric period, certainly within one (reprinted in 1974, New Delhi).
standard deviation (370 years). However, at the lower 10. IAR-Indian Archaeology-A Review, Archaeological Survey of
one standard deviation level, the age of this sherd falls India, New Delhi, 1969–70.
into the historical period. Further, the ages of two sherds, 11. Rao, S. R., Progress and Prospects of Marine Archaeology in
India. NIO, Goa, 1987.
W2878 and W2880, were expected to be within the 12. Gaur, A. S., Sundaresh, Gudigar, P. and Sila, Tripati., Proceedings
protohistoric period but they fall into the early historical of Third Marine Archaeology Conference on Indian Ocean Coun-
period (2430 and 2140 years BP). This anomaly could tries (ed. Rao, S. R.), 1994, pp. 165–172.
either be due to the fact that this type of pottery conti- 13. Gaur, A. S. and Vora, K. H., Curr. Sci., 1999, 77, 180–185.
nued from the protohistoric period till the historical 14. Aitken, M. J., Thermoluminescence Dating, Academic Press,
London, 1985.
period, or as the sample was collected from surface it 15. Aitken, M. J., Science-based Dating in Archaeology, Longman,
could actually be of a later period. Furthermore, it is also London, 1990.
unlikely that the assumptions made regarding the environ- 16. Rao, S. R. and Gaur, A. S., J. Mar. Archaeol., 1992, 3, 42–47.
mental radiation dose received by these sherds could
account for this discrepancy, as this forms only 25% of ACKNOWLEDGEMENTS. We are grateful to the Director, National
the total annual radiation dose in each case. In our opi- Institute of Oceanography, Goa without whose encouragement and
support this work would not have been possible. Thanks are due to Dr
nion the second option is also eliminated, as nowhere in Ashok Singhvi for suggesting valuable improvements in the manu-
the vicinity have we found continuance of pottery of this script. Thanks also due to Shri P. Gudigar, Dr Sila Tripati and Shri
type from the protohistoric period to the historic period. S. N. Bandodkar for help during field work. Shri S. B. Chitari prepared
If the first option is taken into consideration, then it will the maps.
provide evidence of the continuation of Harappan pottery
up to historical period in Saurashtra region. Received 24 September 2001; revised accepted 2 February 2002

1356 CURRENT SCIENCE, VOL. 82, NO. 11, 10 JUNE 2002