2015 GSA Annual Meeting in Baltimore, Maryland, USA (1-4 November 2015)

Paper No. 205-12
Presentation Time: 9:00 AM-6:30 PM


HARRINGTON, Lucy, Anthropology Department, Mercyhurst University, 501 E 38th St, Erie, PA 16546 and FARLEY, Michelle, Anthropology Department, Mercyhurst University, 501 E 38th St, Erie, PA 16546; Anthropology Department, The University of Chicago, 1126 East 59th Street, Chicago, IL 60637, lharri08@lakers.mercyhurst.edu

The distribution of specific types of chert and jasper in the northeastern United States is used to describe the range of Paleoindian mobility over 10,000 years ago. Archaeologists classify chert and jasper types based on their visual differences, which include color, texture, and fossil content. Visual identification of chert sources in the southeastern United States can have error rates as high as 70 percent (Parish and Durham 2015).

Previous geochemical studies indicate that cherts from different sources have statistically different elemental compositions. This study used a Bruker Tracer III-V portable X-ray fluorescence (p-XRF) device to discriminate the elemental compositions of a sample of cherts. The sample includes artifacts (n=6) and unmodified chert (n=27). The artifacts are from an unknown source, but were visually classified as Onondaga chert. The unmodified chert was collected from six different sources in the United States including the Onondaga Formation. The artifacts from an unknown source were included in the study to compare the visual identification method to the geochemical method.

Each sample was analyzed in five locations on its surface in areas of different color, and eleven of the samples were scanned ten times in a single location. After analysis by pXRF, Artax software was used to convert the spectrograph into relative elemental compositions. The composition analysis included the abundances of forty-five elements. Further analysis focused on rare earth elements and elements found in clay minerals. The elemental signatures were analyzed using bivariate and multivariate statistical methods including scatterplots, principal components analysis, and canonical variates analysis. Distinct elemental signatures were identified for the samples from six different sources. However, the six samples visually identified as Onondaga chert did not have the same elemental signature as those acquired directly from that source. Results indicate that while the pXRF can differentiate different chert sources, it conflicts with the visual classification.