Correlation as a tool for establishing functional variability within the site

This article discusses main patterns in the distribution of finds in the horizon of the Initial Upper Paleolithic at the Tolbor-21 site in Northern Mongolia. An attempt was made to define the functional areas in that site which was investigated by several excavation pits. The correlation method has made it possible to analyze cooccurrence of artifacts. Groups of artifacts would get the correlation coefficient 1 if they occurred in the same location. The correlation coefficient between two groups will be -1, if the occurrence of one group was accompanied by the absence of the other group. Cores, core preforms, small-sized debitage, tools, and faunal remains were taken as the artifact classes. Groups of tools were divided in accordance with their supposed functions. One more tool classification was intensity of modification, which reflected complexity of tool production. Correlation among these classes could be observed in different locations of the site and near anthropogenic structures (two areas with traces offre use). Separate occurrence of cores compared to tools and faunal remains has been detected. Cores tend to concentrate on the periphery of the site; the most sophisticated functions related to tool utilization (perforation/cutting, hafting) were associated with hearths. Bone processing activities (butchering?) were also observed in the areas of heat producing structures. High correlation of core preforms and situational tools, non-specialized and primitive in production, can be explained by the abundance of these classes in all areas of the site. Correlation analysis has made it possible to reconstruct the planigraphic structure of the site, which included peripheral areas/areas of non-specialized activities and areas of specialized tool use, meat processing, and using fire.