Potential Errors in the Dating at the Jinmium Site, in Northwestern Australia

Timothy Owen (University of Leicester, United Kingdom)

Recent debate over the initial occupation of the Jinmium rock shelter, in the Kimberley region in far northwestern Australia, has highlighted some of the challenges involved in using thermoluminescence (TL) to date sediments in sandstone rockshelters. The original dating of the Jinmium site published by Fullagar, Price and Head (1996) suggested the possibility of initial site occupation over 100,000 years ago. However, Spooner’s (1998) alternative interpretation of data from this site and Roberts et al.’s (1998) new ‘single-grain’ optically stimulated luminescence (OSL) dates suggest an initial occupation date of less than 10,000 years BP.

The younger figure conforms with middle range theory (Flood 1995) and the conventional understanding of the Indigenous colonisation of Australia (e.g. Allen and Holdaway 1995). The Holocene OSL date is disputed by Price (1998), the TL expert who dated the original samples, owing to OSL sampling procedures and the interpretation of a single selected grain, the statistics used in OSL, and the apparent reliability and consistency of TL. Another indication of Jinmium’s antiquity is the scientifically-ascertained age (30,000 years) assigned to rock art in the site(Tacon et al. 1997), which was based upon dating bees-wax figures adhering to the walls of the rockshelter.

Jinmium’s TL was based on analysis of 28 aliquots for each sample removed from the ground; an aliquot comprised 2,900 quartz grains (Fullagar et al. 1996:756). However, there are problems involved with using TL to determine age for sediments in sandstone rockshelters. This paper will discuss how some of these problems may have biased the original TL dates.

Possible Errors

Fullagar et al. (1996:760) recognise the possibility that some of the samples at Jinmium may have been contaminated. In particular they point out that W1646 might “thus include grains from in situ disintegrated rubble and saprolite”. However, as argued by O’Connell and Allen (1998), examination of Jinmium’s stratigraphy reveals that other samples may also be open to ‘re-interpretation’ on the same grounds, namely W1645, W1648 and W1752, which were taken from trenches C1/III, C1/III and Auger hole 5 respectively.

Contamination may occur as part of the sampling procedure. Ideally, samples should be taken from a single, homogeneous stratigraphic layer. Wagner et al. (1983:19) state the “important consideration in this context is to ensure that the surroundings of each sample are homogeneous to a distance of 30cm so that a reliable assessment can be made of the soil gamma dose-rate reaching each sample.”

At Jinmium, however, the stratigraphy is such that some of the samples could not have been acquired under these ideal conditions. W1645 is directly adjacent to a large, ground, mudstone feature. W1646 was immediately on top of an un-excavated area of sandstone, below a different sandstone pavement. Sample W1648 was taken from an area located directly beneath the rockshelter, within 10cm of a sandstone pavement. In short, all of these samples were removed from inside the recommended 30cm buffer zone. It is therefore possible that they were all adversely affected by varying levels of alpha, beta or gamma radiation emanating from the non-homogeneous rockshelter walls, or the ground mudstone or sandstone pavement features.

Another possible difficulty with the dates concerns the vital ‘radiation-flux’ measurement (see Aitken 1990:143). Aitken (1985) states that gamma spectroscopy readings are vital for obtaining a correct sample age. It is essential that this reading be made for each and every individual sample hole owing to local gamma variation between all soil samples. Wagner et al. (1983:6) state that the problem of determining dose rate can lead to samples being discarded from a dating program. At Jinmium, though, level calibration for the TL sample holes was not measured (Fullagar et al. 1996:755).:

For TL samples taken from Auger hole A4 and sample numbers W1645-W1648 [that includes all the dates in question here] taken from excavation trench C1/III, in situ gamma spectrometer readings were made. Because of a technical problem that developed in the field, these values were not used in the final age calculations. Because of the sand sheet depth it was not feasible at the auger holes, to make in situ environmental radiation measurements using the gamma spectrometer.

On these grounds, it is possible that an inaccurate dose-rate value has been used in the calculation of these samples ages.

The natural stratigraphy at Jinmium prevented a large number of samples being removed from each stratigraphic layer. The content of the samples may also have placed an upper limit on the dateable range of sample ages. Wagner et al. (1983:19) state that while “some of the factors that cause error are general to a site, others vary from sample to sample and it is advantageous to test half a dozen or more samples from each context”. Although dating accuracy was increased by multiple analysis of each sample (Fullagar et al. 1996:756), it was only possible to remove one sample from each stratigraphic layer, thereby not addressing the problem of sample variation within each layer.

With reference to the limitations of TL, Aitken (1990:173-174) states that the “furthest age that can be reached with quartz is limited by the onset of saturation. In round terms the limit should be put at about 50,000 years BP, more if the annual dose [of gamma radiation levels] is low.” The Jinmium TL samples contained quartz. All of the dates under consideration have values greater than 50,000 years BP, but the specific gamma level of each sample remains unknown. Therefore, it is possible that the dates have been biased owing to the limitations inherent in TL dating.


A combination of the difficulties described above may account for the differences obtained at Jinmium using various dating techniques. New OSL evidence obtained by Roberts et al. (1998) could provide a date younger than 10,000 BP, but this only indicates the most recent exposure of the sand grains to light. OSL is also a relatively new technique which has yet to be substantiated as a solid dating procedure. While Roberts and his colleagues’ initial results are interesting, it is still possible that initial human occupation at Jinmium dates well back into the Pleistocene.

The Jinmium debate questions the importance and reliability pertaining to different methods of analytical dating. The relentless pursuit of scientific authentication in archaeology has greatly expanded the discipline over the past fifty years. However education of archaeologists regarding the theory behind dating techniques and exact sampling procedures is currently lacking. Advancements in teaching and an increase in the number of ‘on-site’ specialists (such as Price) will allow future research to combine accurate scientific methodology, with precise excavation, observation and ethnography. This should hopefully reduce the chances that future projects will become mired in the sort of ‘heated’ debate and disagreement that arose over Jinmium.

References cited

Aitken, M.J. 1985. Thermoluminescence dating. Academic Press, London.

Aitken, M.J. 1990. Science-based dating in archaeology. Longman, London and New York.

Allen, J. and Holdaway, S. 1995. The contamination of Pleistocene radiocarbon determination in Australia. Antiquity 69:101-112.

Flood, J. 1995. Archaeology of the Dreamtime. The Story of Prehistory and its People. (3rd edition). Angus and Robertson. Sydney.

Fullagar, R.L.K., Price, D.M. and Head, L.M. 1996. Early human occupation of northern Australia: archaeology and thermoluminescence dating of Jinmium rockshelter, Northern Territory. Antiquity 70:751-773.

O’Connell, J., and Allen, J. 1998. When did humans first arrive in greater Australia and why is it important to know? Evolutionary Anthropology 6: 132-146.

Price, D.M. 1998. Monday viewpoint: Dating the past from 100,000 years to 10,000 years. The Sydney Morning Herald June 1, page 19.

Roberts, R.G. 1998. Monday viewpoint: Dating the past from 100,000 years to 10,000 years. The Sydney Morning Herald June 1, page 19.

Roberts, R.G. and Jones, R. 1992. Luminescence dating of sediments: new light on the human colonisation of Australia. Australian Aboriginal Studies 1992/2:2-17.

Roberts, R.G, Bird, M., Olley, J., Galbraith, R., Lawson, E., Laslett, G., Yoshida, H., Jones, R., Fullagar, R.L.K., Jacobsen, G., and Hua, Q. 1998. Optical and radiocarbon dating at Jinmium rock shelter in northern Australia. Nature 393: 358-362.

Spooner, N.A. 1998. Human occupation at Jinmium, northern Australia: 116,000 years ago or much less? Antiquity 72:173-178.

Tacon, P.S.C., Garde, M., Nelson, E., and Southon, J.R. 1997. Dating Beeswax rock-art: the view from central Arnhem land. In Ward, G., and C. Tuniz, (eds) Proceedings of the first Australian workshop on rock picture dating, Lucas Heights, February 1996. Australian Rock-art Research Association. Melbourne.

Wagner, G.A., Aitken, M.J. and Mejdahl, V. 1983. Handbooks for Archaeologists. No 1. Thermoluminescence Dating. European Science Foundation, Strasbourg.