Abstract
Controlled experimental reduction of a Georgetown chert nodule is used to evaluate the relationship of several discrete and continuous debitage attributes to the reduction sequence. The production of a bifacial blank using both hard and soft hammer percussion was the object of the experiment. Flake attributes are found to be strongly correlated with hammer type. The effects of switching hammer type during biface production complicate efforts to identify reduction stage from debitage assemblages. Discriminant analysis is used to remove the confounding effects of platform preparation flakes produced by small hard hammer percussion. Once these flakes have been removed from analysis, a discriminant function is defined which correctly classifies debitage at rates significantly greater than expected by chance assignment.