Toward a Behavioral Ecology of Lithic Technology: Cases from Paleoindian Archaeology

4. The Reoccupation Problem

99 The Reoccupation Problem Archaeologists have long recognized that single archaeological components could be composed of multiple occupations. For example, when Emil Haury (1959:27) was faced with a Clovis site containing remains of at least nine individual mammoths, he asked, “Were they all killed at once?” In this case, the possibility of reoccupation in the Clovis level at the Lehner site was clear. To Haury, it seemed only a remote possibility that Clovis hunter-gatherers could have succeeded in killing nine mammoths in one hunt and in having those mammoths fall in close proximity to one another (but see Saunders 1977). Rarely, though, are we faced with such scenarios where the archaeological record provides us with a clear suggestion of reoccupation when no stratigraphic separation of occupations is evident. Reoccupation of archaeological sites is a pattern established in the earliest of all archaeological sites (Semaw 2000), so reoccupation of sites must always be considered a possibility. For Haury, it was the assumption that all of the mammoths at Lehner fell prey to Clovis hunters that led him to question whether a single event was represented. For Ingbar (1992, 1994), I would argue, it was the assumption that Northern Plains Folsom sites should look like Southern Plains Folsom sites that eventually led him to the conclusion that Hanson was the product of “many occupation events” (Ingbar 1992:186). Of course, all models, whether formal or informal, are built from a set of assumptions. In this chapter, working from a set of assumptions and building on the measures of mean per capita occupation span developed in chapter 3, I construct a simple model of lithic accumulation that shows how one might distinguish single occupations from reoccupied sites. This model follows from those of Lightfoot and Jewett (1984, 1986) and Gallivan (2002), which were discussed in the previous chapter. Detecting Reoccupation in Archaeological Sites For any set of positive numbers, its mean is less than its sum (go ahead, try it). For a single number, its “mean” is equal to its sum. It is this simple 4 100 Chapter 4 mathematical fact that can be used as a basis for detecting the reoccupation of archaeological sites, because for reoccupied sites (n occupations .1) the mean occupation span will be less than the cumulative occupation span. Consider three occupations lasting 3, 9, and 15 days. The cumulative (summed) occupation span is 27 days. The mean occupation span is 9 days. Contrast this with a single occupation of 27 days. The mean and sum are 27. If one were to plot cumulative versus mean occupation span for a set of single occupation sites (fig. 4.1a), the results would fall on a line with a slope of one and a y-intercept of zero (y5x). Reoccupied sites would fall to the right of that line (y,x). As argued in chapter 3, ratios of local:nonlocal raw materials and debitage:nonlocal tools can provide proxy measures of mean per capita occupation span. Because measures of mean per capita occupation span monitor the average length of stay for individuals, deviations from the line should indicate reoccupation by individuals or groups. Fundamentally , such events are identical but operate at slightly different scales. Unless group membership is extremely variable, such deviations, I would argue, are likely a product of multiple occupations by groups rather than individuals since the impact on an assemblage by the coming or going of one or two individuals should be slight compared to that of multiple occupation by large numbers of people. Horizontal artifact densities can be used as a proxy for cumulative occupation span since the occurrence figure 4.1. (a) A model for distinguishing between single occupations and multiple occupations using the relationship between cumulative occupation span (x-axis) and mean occupation span (y-axis). (b) Same with proxy variables substituted for generalized variables. The Reoccupation Problem 101 of multiple occupations leads to greater numbers of artifacts per unit area (fig. 4.1b). Generally speaking, reoccupied sites should be characterized by high artifact densities coupled with relatively low frequencies of local raw materials and debitage. Applying the model archaeologically is not simple because mean and cumulative occupation spans are not known and can be estimated only using proxy variables, meaning that the exact expected relationship between the variables for a set of single occupations is unknown. Therefore, a correlation between cumulative and mean occupation span could indicate a set...