As is common in archaeological research, some of the most intriguing patterns identified in this study were not predicted by preexisting models. Of particular interest were patterns revealed through the generation of the cost-paths used to evaluate road function. These unexpected patterns include the close association between so-called "special sites" and intercommunity cost-paths, the appearance of small hamlets at points where several cost-paths crossed, and the placement of small isolated Great Houses on routes between communities.

In the GIS database, the "special sites" category included features identified as either shrines or herraduras. Shrines were defined as small stone circles with diameters of 3 - 5 m. that were associated with Anasazi ceramics (see also Windes 1978; Kincaid et al. 1983:9-20 - 9-23). Some circles were not completely closed, while others surrounded boulders. The clear majority of these features were within a few hundred meters of cost-paths between communities (Figure 8). This pattern was especially prominent north of Lobo Mesa, where the majority of intercommunity cost-paths were associated with shrines (Figure 9). Interestingly, few roads were associated with shrines. Instead, almost all of the herraduras, defined as especially large (6 - 10 m.) masonry or rubble circles (Kincaid et al. 1983:9-14 - 9-16), were located along road segments.

The network of intercommunity cost-paths (Figure 2) includes a few points where several paths cross. In many of these locations, small clusters of habitations appear (Figure 10). All of these hamlets are considerably smaller than the named communities, and none exhibit specialized architecture such as Great Houses or Great Kivas. All appear late in the Chacoan prehistory of the area, after A.D. 1050, and most occur on Lobo Mesa at elevations that would have made agriculture challenging.

The final unexpected pattern is the association of two isolated Great Houses on prominent cost-paths between Dalton Pass, Muddy Water, and Kin Ya'a (e.g., Figure 4). Unlike most Great Houses, which appear within the heart of communities, these two structures are located some distance from habitation areas, and both are exceptionally small in size. The purpose of these isolated structures has been a mystery.

These unexpected patterns suggest that the cost-paths align closely with the actual routes used for travelling between communities in the study area. At the risk of overinterpreting the data, the shrines may have served as boundaries or stopping places along footpaths between communities (Robertson 1983). However, despite this evidence of intercommunity interaction, apparently it was not formalized by augmenting footpaths and shrines to road and herradura proportions. This pattern provides insight into which spatial networks surrounding Chaco Anasazi communities warranted increased investments in labor and material.

The late appearance of hamlets at important points in the regional network of paths corresponds with the appearance of two path-related Great Houses north of Lobo Mesa. The impression is that the intensity of regional interaction was increasing after A.D. 1050, stimulating the construction of new path-related features. These new patterns warrant further attention, and will undoubtedly provide new insight into the development of Chaco Anasazi communities.