Can Lithic Attribute Analyses Identify Discrete Reduction Trajectories? A Quantitative Study Using Refitted Lithic Sets
- First Online:
- Cite this article as:
- Scerri, E.M.L., Gravina, B., Blinkhorn, J. et al. J Archaeol Method Theory (2016) 23: 669. doi:10.1007/s10816-015-9255-x
- 391 Downloads
Quantitative, attribute-based analyses of stone tools (lithics) have been frequently used to facilitate large-scale comparative studies, attempt to mitigate problems of assemblage completeness and address interpretations of the co-occurrence of unrelated technological processes. However, a major barrier to the widespread acceptance of such methods has been the lack of quantified experiments that can be externally validated by theoretically distinct approaches in order to guide analysis and confidence in results. Given that quantitative, attribute-based studies now underpin several major interpretations of the archaeological record, the requirement to test the accuracy of such methods has become critical. In this paper, we test the utility of 31 commonly used flake attribute measurements for identifying discrete reduction trajectories through three refitted lithic sets from the Middle Palaeolithic open-air site of Le Pucheuil, in northern France. The experiment had three aims: (1) to determine which, if any, attribute measurements could be used to separate individual refitted sets, (2) to determine whether variability inherent in the assemblage was primarily driven by different reduction trajectories, as represented by the refitted sets, or other factors, and (3) to determine which multivariate tests were most suitable for these analyses. In order to test the sensitivity of the sample, we ran all analyses twice, the first time with all the available lithics pertaining to each refitted set and the second time with randomly generated 75 % subsamples of each set. All results revealed the consistent accuracy of 16 attribute measurements in quadratic and linear discriminant analyses, principal component analyses and dissimilarity matrices. These results therefore provide the first quantified attribute formula for comparative analyses of Levallois reduction methods and a basis from which further experiments testing core and retouch attributes may be conducted.