Technological Characterization of Lithic Waste-Flake Assemblages: Multivariate Analysis of Experimental and Archaeological Data

Citation: Robert J. Austin (1999) Technological Characterization of Lithic Waste-Flake Assemblages: Multivariate Analysis of Experimental and Archaeological Data. Lithic Technology (RSS)
Internet Archive Scholar (search for fulltext): Technological Characterization of Lithic Waste-Flake Assemblages: Multivariate Analysis of Experimental and Archaeological Data
Tagged:

Summary

Context: This paper is looking into the Sullivan and Rozen typology. Sullivan and Rozen (1985), introduced typology, where sorting flake debitage into four different categories could help to decipher what kind of flake it came from: complete flake, broken flake, flake fragments, or no flake (debris). There has been criticism about this typology, but during this experiment, the author sorts assemblages using the Sullivan and Rozen typology to test the accuracy of this method.

Methods and Materials: The experimental assemblage used in this experiment was derived from 15 replications. The results produced stemmed bifaces, unifaces, and halfted flake tools. Reduction strategies used a variety of methods, including both hard hammer and soft hammer percussion. The debitage was collected from each assemblage and bagged and labeled separately. The flakes were subjected to discriminant analysis in order to evaluate objectively the Sullivan-Rozen typology. The assemblages were separated into three groups: patterned tool, large core, and small core based on prior criteria. It was hoped that the differences between the three groups would provide multivariate discrimination. There was another test conducted to determine whether flakes from bipolar reduction strategies could be distinguished from the assemblages. A three group classification was applied in order to compare flake assemblages from patterned tool, conventional core, and bipolar core reduction. The results from these reflected single assemblages. In an authentic archaeological site, assemblages may be mixed. So three simulated assemblages were created and subjected to discriminant analysis.

Results: The results from the discriminant analysis were 86.67% accurately classified. The results from the final test were 95.83% accurate. And the simulated assemblages suggested that archaeological assemblages composed of a mix of flakes from patterned tool and core reduction strategies are likely to be classified as part of the patterned tool group if the core reduction debitage is 50% or less. The accuracy of the analyses (86.67% and 95.83%) suggest that the Sullivan-Rozen typology is effective. However, the author notes that this study did not take into account nontechnological (such as trampling) factors that can alter the results of the Sullivan-Rozen typology, but this method can still be applied effectively in a preliminary fashion.

Theoretical and Practical Relevance

The author here tests the ideas that Sullivan and Rozen put forth in their article Debitage Analysis and Archaeological Interpretation. However, their article sparked a fairly large debate among archaeologists and the typology was eventually deemed inadequate. Their ideas are still used today, but their exact methods and conclusions are not, which is something that the author of this article suggests. This debate, however, did emphasize the importance of examining flake debitage.