Typology is the method of general evolution. For Sahlins and Service, the ``taxonomic innovation that is required for the study of general evolution'' is to ``... arbitrarily rip cultures out of the context of time and history and place them, just as arbitrarily, in categories of lower and higher development'' ([Sahlins and Service 1960]: 32). Steward also used social typology, but of a different order. The social type, for Steward, does not necessarily represent or capture the essence of a society or culture. Rather, social types may be defined according to analytical convenience; they are not considered to be natural categories. This is very different from White's position that one social type was as different from another ``as reptile and mammal are'' ([White 1959]: 302). Steward himself defined two kinds of social types ([Steward 1955]: 88): area types, defined on the basis of uniformities and presumably resulting from direct inheritance or diffusion, and cross-cultural types, defined on the basis of ``regularities'' - or ``causally interrelated phenomena."
Steward's case studies typically involved relatively short time spans, on the order of a century or less, and employed historic and ethnographic data. However, there are many evolutionary processes which occurred in the past which may not be studied using Steward's methods. It may be the case, in fact, that there exist many societies in the archaeological record for which there are no true analogues in the ethnographic record. This is so on the one hand because all modern groups - and probably all ethnographic groups, as well - are integrated to a greater or lesser extent into the modern world system and have been transformed in significant ways as a result ([Wolf 1982]). Thus, they cannot be easily employed as analogues for prehistoric societies. On the other hand, the smaller-scale societies we know ethnographically - such as those which have been used to define the `chiefdom' social type - may in fact be very different from many prehistoric groups. The modern groups are ethnographic `chiefdoms' precisely because they were not transformed into larger-scale social formations, and because they did not disintegrate into simpler, more egalitarian communities. This alone is some indication that they differ in some significant way - particularly in terms of stability - from the Formative period societies which were the precursors of the great prehistoric states in the Americas and elsewhere ([Yoffee 1993]).
Finally, as Feinman and Neitzel put it, social typology and studies of ethnographic and historical cases ``can only demonstrate correlations and cannot reveal historical or causal processes responsible for societal variation'' ([Feinman and Neitzel 1984]: 78). In order to understand how social evolution actually took place - and is still taking place - over long time periods it is necessary to study long sequences of social change (cf. [Drennan 1987,Drennan 1991]). In the case of primary state formation, of course, these data may only be obtained archaeologically.
Settlement archaeologists working in the mode of general evolution developed their own set of correspondences to social typology. The analytical mode of much of settlement archaeology has been, ironically, predominantly synchronic. By this I mean to say that settlement analysis has often been undertaken in such a way as to constitute a regional sequence as a succession of synchronic forms, correlated with particular social types. Thus, a three-tiered site size hierarchy is taken to indicate a complex chiefdom ([Anderson 1994,Wright and Johnson 1975]; also see [Peebles and Kus 1977]), as opposed to the two-tiered hierarchy of simple chiefdoms. In this mode, the salient property of a regional sequence is the temporal sequence of forms which it reveals. The longitudinal study, by contrast, must consider much more carefully the manner in which the settlement system changes from one phase to the next. For this reason, much of the effort in my own analysis will be devoted to examining such measures as population growth, abandonment and site foundation rates, and the relative growth rates of specific sites within the region. These are inherently diachronic measures.
If social typology is the method of general evolution, then the longitudinal study is the method of multilinear evolution. A number of studies of such long regional archaeological sequences have been developed and presented for various parts of the world. It is my intention here to present one for the southern Lake Titicaca Basin, the heartland of the prehispanic Tiwanaku state.