Second thoughts on a rock-art date.

Nelson, D.E.

D.E. Nelson, radiocarbon scientist in the group which reported early dates for rock-art at two Australian sites in ANTIQUITY 64 (1990), reports a further study which leads him to withdraw the published date for one, at Laurie Creek (Northern Territory). As he notes, his co-authors in the 1990 paper are not all in agreement with his present view. The other site dated in 1990, Judds Cavern (Tasmania), is not addressed in this further report.

Introduction

A few years ago, as my part of a team effort, I measured a radiocarbon age of about 20 thousand years (RIDDL-1270; 20,300+3100-2300 b.p.) for rock art at the Laurie Creek site in northern Australia (Loy et al. 1990) for a sample (DMP-6) that had been collected during a first reconnaissance of the area. At the time of measurement, we believed that the substance extracted for dating was remnant human blood protein associated with artistic activity, as another team member had detected human blood on the sample and as the chemical extraction method itself selected only high molecular-weight material which was therefore probably proteinaceous (Loy et al. 1990). However, there was a nagging doubt, in that the carbon and nitrogen concentrations were atypical of protein. Given the great age recorded, repeat measurements seemed appropriate, and so more samples of the same material were obtained during a second visit to the site. The results from an examination of these new samples lead me to conclude that we have no demonstrable associative connection between the material dated and human activity, and so the original result can no longer stand as a reliable date for prehistoric art. As this view is not shared by all my colleagues, I report it here under my name only.

The site and the dated material

The Laurie Creek area contains an impressive display of Aboriginal painted rock art in varying states of preservation which suggest that some of the figures are very recent, while others are of considerable antiquity. It is clear that the area has been the focus of artistic activity for a long time. The original sample came from a red mineral 'skin' on a sandstone rock face without visible figurative art; this material was originally believed to be a result of human ritualistic activity associated with the local figurative art. The second suite of samples was collected from the same location on the rock face from which the original was taken. Some went to the Australian National University for further bio-chemical studies, and some to Simon Fraser University to test further the original extraction procedures. This report describes only the results of this latter study.

Sample description

A microscopic examination of the new samples at low magnifications (6-40 diameters) showed a layered mineral 'skin' of thickness about 1/2 mm on a very friable sandstone base. The red sandstone cement was weakening. It was soluble in HCl, as were the layers which could be removed a little at a time using the acid. The outermost layer was brownish-red; it was underlain in some subsamples by a thin red layer, which was in turn underlain by a thin yellow layer. These inner layers changed to a deep brown on heating to 200 |degrees~ C, and they are probably iron oxides.

While these coloured layers could be interpreted as human paints, such layers also occur naturally. Weathering processes on iron-containing rocks can yield coloured iron oxide minerals (generically called 'limonite') which may be found 'as varnish-like coatings' (Palache et al. 1966). As the cement of red sandstones contains limonite and haematite (Hurlbut 1976), and as the sandstone on which this particular 'skin' was deposited was disintegrating, the coloured layers may be nothing more than dissolved cement reprecipitated at the surface. As there is no visible figurative art on this rock surface, we thus have no evidence for artefactual material, and the only connection to human activity is the positive test for human blood on the surface of the original sample (Loy et al. 1990).

Chemical analyses

The chemical nature of the original extracted material was re-examined by using exactly the same methods (Loy et al. 1990) to extract material from five new samples of similar dimensions to the original (a few square mm) and one sample of the underlying sandstone. As before, the new 'skin' samples each yielded a few hundred micrograms of a whitish material with an apparent molecular weight of |is greater than~ 30kD. This material was on the 'skin' itself, as only a negligible amount of extract was obtained from the sandstone. To test the assumption that the material extracted was proteinaceous, the concentrations of carbon, nitrogen and amino-acids in the extracts were determined. Measurements on two extracts yielded carbon concentrations of 17% and of 18%, and nitrogen concentrations below measurement sensitivity (|approximately equal to~ 1 ||micro~gram~ N). The N concentrations were thus |is less than or equal to~ 0.5% and the C/N ratios |is greater than or equal to~ 36. Within measurement uncertainty, these are the same as the values found for the original sample. Such carbon and nitrogen concentrations are not as they should be for protein, where typical values are 45-55% C, 15-20% N, with C/N ratios of about 3. From these measured values, one can infer that the protein concentration in the extracted sample is at maximum a few per cent of the total. This conclusion was confirmed by the measured concentrations of amino-acids. (As protein consists of linked amino-acids, this is in effect a test for protein.) Three extracts were hydrolysed in the standard manner (in 6N HCl for 24 hours under vacuum), and the amino acids detected quantitatively using the very sensitive ninhydrin reaction. Amino acids were found at concentrations of |is less than or equal to~ 2%, in entire accord with the nitrogen determinations. In round numbers, the protein concentration of these extracts is at maximum about 1-2% of the total.

Measures of the stable carbon isotope ratio were also made for two of the new extracts. The results for the original sample and for the new ones were very similar, with |Delta~13C values of -22.3, -21.8 and -22.6 % respectively. While these measures cannot provide definitive information, they are very puzzling if the extracted material is human protein. Carbon isotope determinations of 5 recent wallaby bones from the area (average value =-9.3%; author's unpublished data) show that plants must form a large part of the local food chain. Even for Pleistocene times, one might expect this signature to be reflected in the |Delta~13C value of local human protein.

What is the extracted substance? I can only speculate, but given the other problems with this specific material, I do not believe further work is justified. It will be more useful to start afresh on materials of undisputed artefactual origin.

Discussion and conclusions

These data lead me to conclude that for this Laurie Creek sample, we no longer have a demonstrable association linking the material dated with human actions. There is no figurative representation on the surface that provides direct indication of human activity, and the layered structure of the 'skin' could be due to natural processes. The material dated was not proteinaceous, and therefore not a remnant of human blood. There is thus no longer a reason to connect the date obtained with human activity, whether artistic or otherwise. It is not a date with any archaeological meaning.

Having thus questioned the date for the Laurie Creek sample, where does that leave the other dates we reported (Loy et al. 1990)? I have no answer, and moreover, I think we should extend this question to include other dates that have recently been reported for rock art. How strong are the associative links between human activity and material dated? In some cases, perhaps not as strong as one might wish. This may be a more serious problem than is presently recognized; on the second trip to Laurie Creek, we found (using a simple field chemistry method) that organic material was often present on the surface of natural rock faces. (Laurie Creek Project unpublished data). Will this aid or confuse the dating of rock art? We won't know until we understand the nature and history of both the carbon on the rock and that in the art.

Notes and acknowledgements. This study is part of the Laurie Creek Rock Art Project funded by the Australian Institute of Aboriginal and Torres Strait Islander Studies. Members of the 1990 field expedition were R. Jones, T. Loy and R. Gillespie (Australian National University), B. Meehan (Australian Heritage Commission), C. Chippindale (Cambridge University) and the author. Wagiman and Nanggiwumerri Aboriginal owners of the research area support this project. The SFU laboratory work was funded by NSERC (Canada) and aided by Cheryl Takahashi. I am very grateful to my colleagues for providing the opportunity to visit Laurie Creek. The trip was most useful, as it gave an opportunity to re-examine critically our own work and to consider seriously the broader problems of dating rock art.

References

HURLBUT, C.S. (ed.). 1976. The planet we live on (illustrated encyclopedia of the earth sciences). New York (NY): Harry N. Abrams.

LOY, T.H., R. JONES, D.E. NELSON, B. MEEHAN, J. VOGEL, J. SOUTHON & R. COSGROVE. 1990. Accelerator radiocarbon dating of human blood proteins in pigments from Late Pleistocene art sites in Australia, Antiquity 64: 110-16.

PALACHE, C., H. BERMAN & C. FRONDEL. 1966. The system of mineralogy of J.D. Dana and E.S. Dana I. 7th edition. New York (NY): John Wiley.

|T.H. Loy, another author of the 1990 report, plans to offer his view of the date's standing in the March 1994 issue -- Ed.~