Quantifying differences in hominin flaking technologies with 3D shape analysis

Genetic and climate-driven estimates of past population dynamics are increasingly influential in broader models of hominin migration and adaptation, yet the contribution of stone artifact variability remains more contentious. Scientists are increasingly recognizing the potential of unretouched stone flakes ('flakes') in exploring existing models of hominin behavioral evolution. This is because flakes (1) were produced by all stone tool manufacturing groups in the past, (2) are abundant from the inception of the archaeological record up into the ethnographic present, and (3) preserve under most conditions. The statistical tools of 3D geometric morphometrics capture detailed approximations of flake form that are challenging to document with conventional artifact analyses. We analyze a collection of 717 3D scans of experimentally produced flakes from 5 production strategies that were practiced by hominins through large parts of the Pleistocene and that scientists have drawn on also to make demographic arguments about past human behavior (n = 45 reduction sequences, n = 3 knappers naive toward the study objectives). First, as a proof of concept, we demonstrate that we can estimate the strategies used to produce these flakes at a high success rate even when flakes from early stages of core reduction are included. We frame the significance of this finding against archaeological classifications from several key Middle Paleolithic assemblages in France (n = 4 sites, n = 28 layers, n = 16,467 flakes). Second, we show that 3D geometric morphometrics captures subtle differences in these strategies that influence flake formation on a flake-by-flake basis and that reflect decisions made by knappers about platform selection, preparation, and core-surface management. We explore the broader potential of our model with a cross-validation approach, and we describe a means of assessing flake form on a continuum wherein variability among assemblages separated by large expanses of space and time can be meaningfully explored. (C) 2020 Elsevier Ltd. All rights reserved.