In his comment, Charles Ward offers several criticisms of our (Carmack and Weeks 1981) attempt to isolate prehistoric social groups at Chisalin. He emphasizes two points: (1) the inapplicability of nearest neighbor analyses at the intrasite level, and (2) the replicability of our results. Since Ward's comments indicate a midunderstanding of both the objectives and analytical procedures of our work, it is worthwhile to elaborate and clarify these points. One of the important objectives of our work at Chisalin, and other highland Quiche Maya sites as well, has been to identify and define Prehispanic social groups in terms of architectural surface criteria and to investigate how these aggregates articulated within a larger site structure through predominately spatial relationships. It has been assumed that human occupation is essentially focal in character and that observable site form or structure will be related to the organization of the social system that created it (Abler et al. 1971; Fletcher 1981; Kuper 1972). Further, it is suggested that changes in one will necessarily affect transformations in the other. Since it is necessary to understand how one system functioned before suggesting how it changed, the immediate goal was to construct a specific structural model of the site using a variety of analytical techniques. The functional identification of architectural and nonarchitectural units was a principal consideration. Using observable morphological attributes of unexcavated mound height, summit surface area, and relative location, the nature of mound form variation was determined. On the basis of these criteria, mounds at Chisalin were collapsed for analysis into 12 formal categories; these comprised primary excavation sampling units. As stated in our paper, predominate mound functions were determined on the basis of architectural criteria, artifact inventories, and ethnohistoric/ethnographic inferences. Initially only two possibilities were considered: (1) permanent or semipermanent residential and (2) ritual constructions. Using these two functional categories, it was possible to construct a generalized land-use model. As Ward notes, even a cursory examination of the site plan would indicate that mound forms at Chisalin comprise a relatively compact cluster group. Furthermore, mounds and groups of mounds are not randomly distributed across the site. Thus, the essential problem is one of defining intrasite settlement distribution. There are no a priori reasons why at least some of the approaches and models used by modern geographers for analyzing urban settlement patterns cannot be used to investigate spatial distributions characterized by tinamit-type settlements in the Guatemala highlands. The behavioral premises upon which these models rest, as outlined by Garner (1967), apply as well to such settlement situations. The nearest neighbor analysis offers a convenient method of point pattern description that allows an objective comparison between observed and expected distances from a Poisson probability function. However, although our analysis indicates a clustered distribution, the factors that produced such a pattern are not indicated. At a regional level of analysis obviously there are correlations of site location with such environmental factors as topography, water accessibility, soil productivity, and so forth. Factors operating at intrasite and local levels of analysis are more difficult to define. A plot of all first nearest neighbor values indicates a range of values from 0 through 30 with most found in a mode less than 6 m. A series of boundaries imposed upon the site plan around unexcavated mounds which are 6 m or less from their first nearest neighbors are illustrated in