Toward disentangling stages in mixed assemblages of flake debris from biface reduction: An experimental approach

Their abundance in the archaeological record makes assemblages of lithic debris popular subjects of analysis. Archaeologists study them to identify both kinds and stages of reduction, using approaches that range in scale from attribute analysis to typology and size distributions. Experiments demonstrate meaningful pattern in flake assemblages by both kind and stage. Yet empirical assemblages often are mixtures of flake debris from various reduction kinds and stages. Mixing confounds the patterns often found in controlled experimental data. To address this problem, we use Stahle and Dunn's (1982) data and a similar constrained-regression method, QUADPROG in R, to allocate hypothetical mixed assemblages of biface-reduction stages to their constituent stages. In combining their Stages 2 and 3, a measure that Stahle and Dunn themselves advocated, QUADPROG improves their allocation results. Methods like these deserve further testing on a wider range of flake assemblages to address the challenge that mixing poses.