Abstract

Two solutions are offered to the problem of distinguishing "historical" from "functional" associations in cross-cultural surveys. The underlying logic of the mathematical model is discussed and three kinds of association distinguished: hyperdiffusional or purely "historical" association, undiffusional or purely "functional" association, and semidiffusional or mixed "historical-functional" association. Two overland diffusion arcs constitute the test sample; the relationship of social stratification to political complexity constitutes the test problem. A sifting test establishes a bimodal distribution of interval lengths between like types and sifts out repetitions with a lesser interval length than the second mode. A cluster test shows that for the test problem, the "hits" cluster more than the "misses"