The study of prehistoric Maya settlements has been hampered by simplistic views of cultural ecology, over generalized ethnographic analogy, and a lack of attention to both natural and cultural site formation processes. As a result, Mayanists have tended to expect very little variety in archaeological features and have assumed cultural uniformity over wide ranges of time and space. Traditional research designs support these assumptions. Current knowledge of Maya social organization suggests that more structural variety may occur in Maya archaeological sites than is ordinarily discovered. Some of this variation is evidenced by features not currently visible on the ground-surface. The Nohmul Settlement pattern project employed a "purposive" sampling design to search for settlement variation over time and space. Several assumptions about surface-subsurface relationships were tested. Surface indications were not found to outline subsurface variety. Excavating at intervals from site center in both visible and "invisible" features, showed that the Nohmul community was affected by both centralizing and decentralizing influences and grouped into residential clusters resembling neighborhoods. The degree of centralization and the location of the clusters, as well as some of their characteristics, changed notably over Nohmul's 2500 year occupation.