Blind in a Cloud of Data: Problems with the Chronology of Neanderthal Extinction and Anatomically Modern Human Expansion

Article excerpt

Recently, radiocarbon chronology has played the central role in debates over the nature of the Middle to Upper Palaeolithic transition in Europe. Here, we raise some concerns about both the efficacy of radiocarbon dating to these areas, and the way in which it is employed to reconstruct human palaeodemography. We centre our concerns on the study by Bocquet-Appel & Demars, published in the last issue of ANTIQUITY (74 (September 2000): 544-52).

Using the earliest modern human and latest Neanderthal radiocarbon dates kriged in a 2 [degrees] x2 [degrees] grid representing Europe, Bocquet-Appel & Demars present a time series of the proposed distribution of Neanderthals and modern humans. The kriged results apparently reveal the contraction and disappearance of Neanderthals between 40,000 and 27,500 years BP, alongside the expansion of modern humans from eastern Europe. Here, we raise concerns in four main areas; the problems of using chronological data in inferring human palaeodemographics; refinements to Bocquet-Appel & Demars' data; the precision of the statistical methods they employ; and general problems of radiocarbon accuracy in the period concerned.

Inferring range contractions (extinctions) and expansions (colonizations) from radiocarbon data: some assumptions and problems

Bocquet-Appel & Demars (2000) use radiocarbon chronology to model the regional extinction of a human species -- European Neanderthals, and the potentially related range extension of another species -- anatomically modern Homo sapiens. The goal of reconstructing the specifics of such processes is notoriously difficult, and it is disappointing that the authors ignore recent problematizations of this procedure both for Plio-Pleistocene mammals (Vrba 1995) and for Neanderthals (Pettitt 1999a; 2000a). These are not events but complicated processes that may occur over brief periods in restricted regions and may therefore be chronologically `invisible', in the context of continent-wide processes occurring over some 10 millennia for which chronological precision is not homogeneous, with radiocarbon errors at c. 30,000 BP half that of those at 40,000 BP (Pettitt 1999a). Such limitations feed directly into the nature of the chronological databases used in statistical modelling. In addition to these general effects the act of `reconstructing' Neanderthal extinction requires several nested assumptions that turn individual dates into inferences about continent-scale phenomena, which need to be addressed before bold statements can be forwarded about complicated biogeographic processes. Bocquet-Appel & Demars have not problematized their data and worked within its limitations.

Refinement of the database of Bocquet-Appel & Demars

With the advent of AMS radiocarbon dating in the mid 1980s the potential problems of the stratigraphic mobility of the small samples routinely measured using this technique were appreciated. In no area is this more clear than the apparent age of fossil human material, whereby in recent studies specimens believed to be of early Upper Palaeolithic age -- on the basis of dates on materials with which they were apparently associated -- have been demonstrated to be of Holocene age, e.g. at Velika Pecina (Smith et al. 1999) and the Krems Hundssteig `Gravettian' remains which are Late Bronze Age (Trinkaus & Pettitt in press). The one clear lesson archaeologists have learnt is never to accept a radiocarbon date at face value.

A glance through the database of radiocarbon measurements apparently relating to Middle Palaeolithic/transitional industries (one assumes Neanderthal) and Early Upper Palaeolithic industries (one assumes anatomically modern human) reveals that Bocquet-Appel & Demars' data is flawed in two ways. First, we are not convinced that they have effectively evaluated their raw data on a date-by-date level, resulting in their including a number of dates which we regard as highly questionable and/ or irrelevant; this is notably apparent in questions of the latest Neanderthal and earliest modern human appearance at Coygan and Paviland caves (Wales), the Grotte du Renne (France), Le Trou Magrite (Belgium), Zafarraya (Spain), Vogelherd and Geissenklosterle (Germany), Willendorf (Austria) and Vindija (Croatia). …