Discerning Empirical Relationships Between The Natural Environment and Prehistoric Site Location: An Example From the Watts Bar Reservoir, East Tennessee

The Watts Bar Reservoir study area is an artificially defined region of 13,815 hectares, demarcated by the resevoir boundary of the Tennessee Valley Authority. Following completion of the Watts Bar Dam in 1942, the reservoir impounded 95 river miles of the main Tennessee River, in addition to portions of the Clinch, Emory and Piney rivers, as well as several smaller tributaries. Since the mid-nineteenth century archaelolgical investigations have been conducted in the region. However, the sporadic nature of these research endeavors has created a somewhat fragmented picture of the regions prehistory.

Following Smith's (1978b) model of the linear bandinog of environmental zones adjacent to the course of meandering streams. this thesis addresses the site location in the reservoir. Specifically, the main river channels of the Tennessee and Clinch rivers were divided into one kilometer tracts in order to delineate the natural distribution of environmental variables. A comparison of tracts containing archaelolgical sites and those without sites was made using the Kolmogorov-Smirnov goodness of fit test. Although the use of random sampling methods to obtain negative information has been strongly advocated (i.e., Binford 1964;Thomas 1973; Kvamme 1985; Kellog 1987), I chose to use all the tracts to offset the biases in the archaeological record due to the sporadic nature of the region's research. A separate and additional test was conducted for the delineation of patterns of natural shelter selection.