The latest of the last?

Through the magic of the information superhighway, I managed to get a copy of Finlayson et al.'s (2006) paper that I mentioned in my last post. It's pretty short, and I've had time to go through it in depth.

At its most basic, this is a schematic site report focused on chronology as derived from AMS 14C dates, of which the authors present fully 30. There's unfortunately little to no discussion of the lithic or faunal assemblages (presumably because that information will be included in the project's final report/publication), so there is almost no information about the patterns of mollusk and sea mammal exploitation in the Mousterian levels of Gorham's Cave, which I was originally so excited about, based on comments by Finlayson on the Nature news report.

The main of the paper is an argument for an age of ca. 28,000 radiocarbon years (that is, uncalibrated) for the Late Mousterian at Gorham's, and that there is no evidence for infiltration from the overlying Upper Paleolithic (Solutrean and Magdalenian) level III into Mousterian level IV. That these two form discrete and well separated depositional units is well demonstrated by a combination of two measurements of “geochemical detrital ratios”, in this case K/Al and Mg/Al, which indicate radically different depositional contexts for levels III and IV (Finlayson et al. 2006: Figure 1c). So, the late MP and UP levels are clearly distinct, no argument there, and this is further suggested by a photograph (Figure 1d) of a section at the site which leaves little doubt as to the dramatically distinct sediment color in each.

As for the dates themselves, the 22 that concern the Late MP level (samples 9-30) are a bit hard to make sense of at first glance, especially since the authors present doubled, rather than single error ranges. Now there's absolutely nothing wrong with doing this; in fact, given their large corpus of dates, it's probably better to do so, since it means that all the radiocarbon determinations thus presented are ca. 95% likely to correspond to the range within which an individual sample's actual age falls. However, it does lead to a bit of confusion since dates are usually presented with single error ranges.

Regardless, even at second glance, the chronological data are very messy, and as a whole they display no obvious logical stratigraphic coherence, although they all fall somewhere between ca. 24-32,500 uncal. BP, distibuted over about 50 cm of sediment. Finlayson et al. explain this general lack of coherence as resulting from

“repeated use [of the site] confirmed by the stratigraphic distribution of the dates within level IV that indicate localized alterations due to use and reuse (for example trampling and cleaning) in the area around the position of the hearths but dates in stratigraphic sequence within the location of the hearths themselves. Thus, three samples (16, 17 and 20; Fig. 1) came from Mousterian superimposed hearths. These three dates provide a stratigraphic sequence from 24,0106320 to 30,5606720 yr BP. Taken together, all the dates show that Neanderthals occupied the site until 28 kyr BP and possibly as recently as 24 kyr BP.” (2006:p.1).

In other words, they explain away 19 of their dates that are not in coherent stratigraphic order by focusing on only three of them that are coehrent and suggest a very young age for the Mousterian deposits. Their argument for this (i.e., focus at the center of the heart which remained in a fixed position throughout the Mousterian at the site), while certainly valid is also somewhat problematic. For one thing, the majority of the bottommost dates (i.e., samples 24-27 and 30) display a coherent age of roughly 31-32.5 kya uncal for the base of the deposit, and this even well away (i.e., roughly 4 meters) from the hearth which is supposed to be the epicenter of disturbance. Embedded between them are samples 28-29, which date to (ironically) 28 and 29 kya. If anything, given that these two samples come from immediately next to each other (as best as can be gauged from Figure 1c), their unexpected recent ages are perhaps best explained by some unspecified localized form of contamination. So, overall, the very base of the hearth would seem to date to somewhere between 31-32.5 kya, thus providing a terminus post quem (or ante quem, depenging on your take) for the age of overlying samples. The four stratigraphically highest samples (9-12) give ages ranging from 26-30 kya (with an average of around 28.5 kya), which is coherent with the minimum age of the Late MP deposits just outlined. I think that this is basically where the data presented leaves us at this point, however. And what is known so far should, in the absence of evidence to the contrary encourage us to lean towards an older age for the top of the Mousterian (i.e., closer to 30 kya than 26 kya).

I think that the argument that the dates should all be considered valid but in secondary context because people continued using the site for prolonged periods and moved stuff around is not very good for two main reasons: First, this is likely to have been the norm at many cave/rockshelter sites whose physical characteristics (e.g., the presence of a high vault, or ceiling openings to enable smoke to clear out) would have constrained the positioning of certain features such as hearths. This is certainly something I've observed in my own field work, where Mousterian and Aurignacian hearths were located at the same spot in the rockshelter. Should we suppose that Italian Pleistocene hominins were more mindful about kicking stuff around and/or digging around than those from Gibraltar? Or that they operated with more of a concern for future archaeologists? I don't think so. This being the case, the argument that people lived in sites and must have disturbed sediments and the position of dating samples is unconvincing and basically a post hoc argument that has little inherent merit. There are ways to demonstrate the integrity (or lack thereof) of given layers within archaeological sites (e.g., micromorphology) and in their absence, it is simply impossible to accept the argument for anthropogenic (or any other source) sedimentary disturbance at face value. Maybe this information will be provided in the final report, but at the moment, it's lacking.

The second line of evidence against the explanation for mixing has to do with the intensity of occupation at the site. The authors report a total of 103 lithics for level IV, which was excavated over an area of 29 square meters. The site stratigraphy presented in the paper indicates a thickness of about 1m for level IV, which means that we're dealing with a density of about 3.55 lithic artifacts per cubic meter of excavated sediment! Even if we take only the top half of the deposits (ca. 50 cm thick), that still only gives about 7 lithics per cubic meter. Practically, that means that, digging a 1x1m unit in 10 cm spits, one would encounter one lithic per spit, 7 times out 10! In my view at least, this is hardly the kind of evidence of very intensive use and reuse of an area, especially around a hearth, where lithic concentrations tend to be relatively dense in general. So the argument for heavy human action on the sediment is, as a result, severely weakened. A caveat to this argument, however, is that it is unclear from this paper what is considered an artifact. Are we only talking about retouched tools here? Or are we (as I assume) including debitage as well? What's the frequency of retouched pieces relative to the total? This is information that could significantly clarify the discussion about site occupation intensity. Again, I suppose we'll have to wait for the final report to be sure.

So, in the end, is this site really 28 ky or even 24 ky old? As is usually the case in paleoanthropology, it's not impossible. However, in light of the reasoning just outlined, I think that the possibility of sample contamination should be dealt with in a more thorough fashion before being discarded out of hand. At the moment, I'm not unwilling to consider that the top of the sequence may be as old as 28.5 ky old, but my money's still on a slightly older age, probably along the lines of 30 kya. What would have considerably clarified the issue here would have been the presentation of the results of complementary dating methods (e.g., TL, OSL, ESR). This would have enabled an objective discussion of whether or not the sediments were really disturbed or if the charcoal samples might have been contaminated by younger material at some point.

Check out also John Hawks' blog for some additional comments on the paper.

References:

Finlaysn, C. L., et al. 2006. Late survival of Neanderthals at the southernmost extreme of Europe. Nature: in press. doi:10.1038/nature05195