Redefining lithic microdebitage with experimental archaeology

Lithic microdebitage is “invisible” to the naked eye. Since ancient knappers struggled to remove it from activity areas, its presence can reveal stone tool production and maintenance, as well as site-formation processes. So far, scholars have examined soil samples under a microscope. More angular forms, higher transparency, and other characteristics set microdebitage apart from other soil particles. While intuitive, this definition is subjective and difficult to apply consistently. In our alternative approach, we combine experimental archaeology with dynamic image analysis, statistics, and machine learning. A modern stone knapper who uses traditional raw materials and tools produced 101,298 microdebitage flakes. We compare 38,787 chert and 62,511 obsidian flakes with 23,980 sand and gravel particles. A dynamic image particle analyzer measured each of the 125,278 particles. We discuss seven variables as proxies for Fladmark’s definition of microdebitage. Since they distribute non-normally, we use non-parametric statistical tests to show that microdebitage differs highly significantly from sand and gravel for all seven proxies. Particle-specific data allow us to quantify microdebitage characteristics, propose 0.5–4 mm as size limits, and discuss the robustness of our definition. We conclude that our approach enables an objective analysis of microdebitage.