Academic journal article
By Bednarik, Robert G.
Antiquity , Vol. 72, No. 278
I am delighted that Phillips et al. (1997) have responded to my call for scientific attempts at refutation of the data so far offered concerning the age of some of the many petroglyphs in the Coa valley of northern Portugal. We have witnessed a great deal of non-scientific debate about this corpus of rock-art, and it is obvious that our understanding of the issue can be improved only through constructive scientific discussion, which may involve presentation of new data or logic. In continuing this productive mode of discourse I will respond to Phillips et al. (hereafter referred to as 'the Paper') and offer a critical examination of the use of their method in archaeology, and especially in rock art analysis.
The quantitative determination of cosmogenic nuclides has been used experimentally for dating geomorphic surfaces (Davis & Schaeffer 1955; Phillips et al. 1986; 1990; Zreda et al. 1991). The most interesting results have come from a meteor impact crater in Arizona, ranging from 36,500 to 50,400 BP, which appear to correspond with a single TL date (Phillips et al. 1991: 2697; Zreda et al. 1990). However, a study of 36Cl contents of boulders in a series of moraines of geologically known ages in California produced only ambiguous
results which disagreed with the moraine sequence (Phillips et al. 1990: 1530).
Cosmogenic radiation (CR) products are said to provide information of the relative age of exposures (rock surfaces), and some of the results offered by the Paper seem to confirm this. However, there are several reasons why the method cannot provide numerical ages for rock exposure, such as those relating to its severe dependence upon erosion rates. The CR products, including 36Cl, accumulate not only in the sampled surface layer, but also deep inside the rock. The amount of 36Cl that was present in a surface at the time it became freshly exposed to the atmosphere (e.g. through exfoliation or slabbing) can never be known, meaning that the relationship between the present concentration of the nuclide and the age of the present rock surface cannot be ascertained or even estimated.
The Paper demonstrates this problem very well: a sample taken from a depth of 16 m below the Coa river provided an 'age' of between 800 and 2800 years, instead of a zero age. Mindful of the exponential accumulation curve of the nuclides (Kurz 1986; Lal 1991; Zreda et al. 1991), we might reasonably expect to obtain an 'age' of perhaps 10,000 years from a sample taken 2 m below present surface, and 20,000 from a surface that was formed a few centuries ago. Similarly, Phillips' sample from Castelo Melhor, a structure erected in the Middle Ages, produced a 'date' of about 30,000. He tries to explain this away by suggesting that the building blocks were collected from the ground level. There are hundreds of schist quarries in the area from which thousands of tons of such building blocks have been quarried over the past couple of millennia, precisely because the stone has such excellent construction properties. Rock from near the surface is inevitably weathered and fractured, and quite unsuitable for masonry work. Thus the more plausible explanation for the high concentration of 38Cl in the Castelo Melhor sample is that it was quarried from a location where a stable geomorphological regime facilitated the accumulation of high contents of cosmogenic nuclides at some depth (from the plateau, i.e. the planation surface, from a depth of several metres).
There are two simple ways of testing this proposition. Firstly, blocks from historical masonry could be analysed, both at the surface exposed to the atmosphere and at a surface concealed in the masonry. If we made the reasonable assumption that the 36Cl concentration was uniform in the block at the time of quarrying, the difference in present concentrations might provide a valid measure of age for the structure. A second way of testing the proposition would be to take samples from recent exposures of known ages, such as the 100-year-old railway quarry sampled by Watchman in 1995, at depths some metres below the land surface. …