Introduction

Discussions among archaeologists concerning the dates and direction of ancient cultural influences during periods of Southeast Asian prehistory have gradually shifted their emphasis from theories based on mass migration waves to questions concerning smaller groups of people who were apparently able to effect a shift in the material culture of the populations with whom they were in contact. Unfortunately, as yet, no distinctive chemical marker has been identified that will unambiguously indicate nature of these contacts, their scale, their date and the identity of the peoples undertaking these trading activities. In their recent paper An Indian trader in ancient Bali (Lansing et al. 2004), Lansing et al. seemingly broke the impasse, providing both the first earliest direct evidence and a datable marker for individual contact between ancient Bali and India. Since the work presented by Lansing et al. points to a biomarker capable of isolating migrants from the indigenous populations of Southeast Asia, the following study was undertaken in order to assess the wider applicability of this discovery, in particular for our own work on the 'Indianised' culture of Viet Nam. Unfortunately, our results indicate that the scientific evidence for an Indian trader may have been overstated and that much further work is required before the date, forms and directions of 'foreign' influence within Ancient Bali can be isolated and identified.

The argument

The evidence for early trading contact between Bali and India advanced by Lansing et al. rests on four pillars. These are that:

1. 1. The tooth which forms the focus of the study was found in association with imported objects of Indian origin;

2. 2. The single radiocarbon determination on the tooth yielded a date consistent with dates for Rouletted Wares and the Kharaoshthi graffito found at the site, both said to be of Indian origin;

3. 3. Stable carbon isotopic analysis of the tooth yielded a result indicating the individual's primary food source was of terrestrial rather than marine origin (indicative of a 'foreigner' rather than a native Balinese);

4. 4. Mitochondrial DNA analysis of the tooth combined with computer-based modelling of the Ancient Balinese population, confirmed the foreign extraction and likely north-east Indian ancestry of its original owner.

Unfortunately, on the basis of the evidence published to date, this argument does not seem to be well-founded.

Origins of the sample

The first problem with the proposition advanced by Lansing et al. is that no trace of the tooth or its archaeological context in a trench PCN III is to be found in the publications relating to Ardika's excavations at the site of Pacung. Four fragments of animal bones were recovered from below 3.5m within the single trench opened at Pacung, PCN I, amongst them a tooth of uncertain identification, labelled 'possibly bovine'. If the tooth analysed was indeed that recovered from PCN I then doubts must exist as to the integrity of the archaeological context, the veracity of the date of the sample and the stated association with the Rouletted Ware ceramics. Details of the Pacung site provided by Ardika indicate the tooth was unassociated with any other material indicative of a burial, raising questions as to how the tooth was lost by its original owner. The nature of the stratigraphy within PCN I, makes it impossible to determine whether this tooth was deposited in higher layers, arriving in its final resting place either through natural bioturbidation or via subsequent well digging/infilling operations. It is, for instance, notable that a plastic bag and ballpoint pen were recovered from the in-filled well of PCN I at the same level at which the archaeological material was found within Layer 7!

Dating the sample

Dating the age of the tooth by its association with Rouletted Ware ceramic fragments of Indian origin and glass beads seems doubtful, as only one sherd of Rouletted Ware was found at the PCN I site and no glass beads. The Rouletted Ware, glass beads, the radiocarbon dated pottery sherd and the sherd with Kharoshthi script to which Lansing et al. refer appear to be those recovered from a site at Sembiran, some 300 metres west of Pacung. As noted by the excavators, Pacung and Sembiran are independent sites and were subject to different depositional histories and transformational influences.

Although two human inhumations were indeed discovered in a Sembiran trench (SBN VII), the associations necessary to support the conclusions drawn by Lansing et al. are absent. The human inhumations within SBN VII were not associated with any finds, including Rouletted Wares. Instead, it seems all Rouletted Ware fragments were originally derived from Layer 7 and those found within Layer 6 had been redistributed when the burial pit was cut though into the lower layer. This conclusion seems to be borne out in the results of the AMS radiocarbon dating, which, contra Lansing et al., do not indicate any chronometric relationship between the Rouletted Ware and the inhumations. The dating ranges for two radiocarbon determinations reported for Sembiran indicate that the earliest possible date of manufacture for the Arikamedu type 10 pottery fell within the range 1100 BC to 450 BC. In contrast, AMS radiocarbon dating of the Pacung tooth indicates that the tooth ceased its association with a living person at an indeterminate date somewhere between 340 BC - 20 BC. The only other ¹⁴C analysis from Sembiran comes from a sample of unspecified material within Layer 6 located some 0.5 metres above that of the Rouletted Ware. This material ceased actively exchanging carbon with its reservoir at a date somewhere between AD 770 and AD 1240. The dates suggest that the Sembiran burials belong to an entirely separate temporal horizon from that of the Rouletted Wares and throw into question both the chronometric integrity of the Layer 6 and the ¹⁴C-date obtained from the tooth (if it is indeed the date of this inhumation that was measured). There is no prima fasciae evidence for any cultural association between a tooth, recovered from either Pacung or Sembiran, and Rouletted Wares. The calibrated date ranges of the Rouletted Wares and the tooth do not overlap and in fact are separated by at least 110 and anywhere up to 1080 years.

Palaeodietary evidence for a 'foreign' tooth

Analytical evidence that the tooth formerly belonged to an individual who was not an indigenous Balinese 'but rather to a foreigner' stems from analysis of the ¹³C/¹²C ratio within the tooth. For Lansing et al. (2004) the value of d¹³C = -21.1‰ indicates that the individual's diet was primarily terrestrial. In contrast, the proximity of the sea to the Sembiran/Pacung sites suggests to the authors that the ancient Balinese inhabitants existed of a diet composed primarily of marine foods, said to be typical of coastal Indonesian peoples.

From the details given of the ¹³C/¹²C analyses, such a conclusion would seem to be premature. Any archaeological study based on results obtained from a single tooth and on a comparison with global ¹³C/¹²C ranges of various food sources would seem to oversimplify:

• • the dietary choices available to individuals and the possible sources of food;

• • variability of ¹³C/¹²C within the local drinking water;

• • the degree of variability of ¹³C/¹²C between the different fractions of the tooth structure;

• • spatial and temporal variability of ¹³C/¹²C values within species; as well as

• • the degree of post-depositional diagenetic turnover that has taken place.

As it stands, the single figure of d¹³C = -21.1‰ is meaningless. Its significance should be re-evaluated, after adjustment using appropriate offsets, against the range of possible dietary d¹³C values for the specified area in Bali and dietary d¹³C values applicable to 'foreign traders'. As it stands, even if the tooth belonged to a foreigner trader, it is unclear why that person's food intake would have been significantly different to a 'typical' Balinese diet?

Mitochondrial DNA analysis

The study presented by Lansing et al. ultimately focuses upon the results of their study of the mitochondrial DNA within the tooth. Bearing in mind the uncertain origins of the sample and the length of time since excavation, it is regrettable that Lansing et al. provide no information regarding the protocols employed to assess the authenticity of their mtDNA analyses. As the matter stands, their results and conclusions drawn should be regarded with extreme caution.

Several researchers have pointed to the dangers of assaying prehistoric populations using a restricted set of polymorphisms culled from modern populations since these may carry much less genetic variation than their ancestors, or may be more distantly related to ancient populations than is currently recognised. To assay ancient specimens for only a few diagnostic markers is to invite incorrect haplogroup assignments. Furthermore, whilst it has been demonstrated that in cold dry environments mtDNA has the potential to survive in the burial environment for thousands of years these environmental conditions are certainly not those of the Sembiran/Pacung sites where temperature, site hydrology, redox potential, pH and microbial infestation would all be conducive to oxidative and hydrolytic chemical attack and enzymatic attack on the tooth's mtDNA. Post-mortem hydrolytic damage results in deamination and depurination of nucleotide sites in the hypervariable and coding regions of the tooth's mtDNA. This will result in amplified sequences containing artefacts, particularly if a small number of DNA templates initiate the polymerase chain reaction (PCR). These artefacts, which are particularly troublesome if the quantity of pre-PCR mtDNA is low, will, if undetected, lead to incorrect mtDNA typing of the tooth and to subsequent errors in any phylogenetic analysis. In fact, the ancient mtDNA tooth's hypervariable region sites 16261 and those that are associated with haplogroup A (16223, 16290 and 16319) are sites which are susceptible to diagenetic alteration. Given the undoubtedly degraded nature of any existing endogenous tooth mtDNA and that the only reliable indication for a haplogroup A classification is that obtained from the single A-G transition at point 663 in the coding region, any conclusion that the Sembiran tooth necessarily derived from its north-east Indian ancestry would seem to be premature. Even supposing that the mtDNA derived from the tooth is authentic and its mtDNA analysis reliable, the mtDNA haplotype within the 'Pacung tooth' is not unique. Apart from the two private substitutions detected, the haplotype reported for the tooth is present in current Asian populations and is not exclusive to a particular geographic area.

Summary

Unfortunately, too little detail is given regarding the tooth, the circumstances of its discovery and the conditions of its storage to enable anyone to evaluate the validity of the claims now being made for it. On the basis of the currently available information, there is insufficient evidence to substantiate any of the conclusions made for the tooth found at Pacung/Sembiran and in particular the assertion that the tooth belonged to an Indian (foreign) trader present on Bali c. 150-200 BC. It is important to note that this study does not preclude the possibility that the presence of Rouletted Ware and Kharaoshthi graffito may still indicate direct trading links with India. Unfortunately, this conclusion cannot be reached via the evidence presented in the article. It would also seem that major obstacles exist between reliable mtDNA determinations and specifying the provenance of ancient Southeast Asian individuals, at least until some other more distinctive biomarker becomes available.