The standardized analysis procedure followed for all samples approximates that presented by Yarnell (1974). Each sample was weighed and then sifted through a series of U.S. Standard geological sieves ranging in mesh size from 6.35 mm to 0.21 mm. All sorting and identification of plant remains was done with the aid of a binocular microscope. Material greater than 2.00 mm in diameter was completely sorted by genus and species if possible. Material passing through the 2.00 mm screen was searched for seeds only and for items not represented in the larger size category. Quantities of each component in the 1.41 mm to 0.71 mm size class were extrapolated on the basis of their representation in the 6.35 mm to 2.00 mm category (Table 18). It is assumed that components are represented equally in these two size classes. Even though this assumption is not always satisfied, the procedure provides a better estimate of actual quantities than quantification of only the largest size class. Material passing through the 0.71 mm screen was excluded from calculations, since it was composed mainly of fine soil particles.
Several techniques were used to compare plant remains data from different sites and contexts. Percentage of plant food remains by weight is especially useful for comparing remains of the same class (Table 19). However, since these classes differ in their durability and likelihood of preservation, between-class comparisons are sometimes misleading if based upon relative representation by weight. Ubiquity measures assess the frequency of occurrence of plant taxa without regard for their quantity (Table 20). Here, ubiquity was calculated by finding the percentage of flotation samples comprising both light and heavy fractions in which an item occurred (most of these samples represent 10 liters of fill; the few exceptions probably do not introduce significant bias).
A modified version of Simpson's diversity index (Peet 1974) also was used to compare data. This index, designed for ecological community studies, measures both richness (numbers of taxa represented) and equitability (evenness of representation of taxa), and its value can range from zero to one. Diversity (D) is computed as
D = 1- [[Sigma]] pi 2 ,
where p is the proportion of item i in the assemblage. Proportions of each plant food in a sample are determined using a ubiquity measure in which values sum to one and using equal numbers of samples from each site. It is assumed that these proportions reflect proportions in the diet to some extent, but the diversity of the plant food resource base for each community as a whole is of primary interest.
Seeds (except for common bean, corn, peach pit, and nut remains) were combined and their aggregate weight presented for each context. Numbers of seeds are presented in separate tables along with estimated corn cupule and kernel counts (Table 21, Table 22). In addition, actual counts were converted to number per gram of plant food remains, a procedure that takes into account differences between sample sizes. All comparative techniques are of course subject to the cautions presented above.
Most of the categories in the tables are self-explanatory. Non-botanical materials and uncarbonized plant parts were sorted completely for light fractions and weighed as an aggregate for heavy fractions, but only plant remains are reported in the tables. The "wood charcoal" category also includes stem charcoal and bark. "Residue" is the unsorted material less than 0.71 mm. Corn remains are presented as aggregate weights of cupules and kernels but are itemized in the seed count tables.