Published online by Cambridge University Press: 20 January 2017
One kind of R-mode analysis, the examination of the co-occurrence of particular artifact classes across a number of collections, facilitates the identification of recurrent artifact clusters. Its use with proportional data, however, requires the safeguard of equalizing collection sizes, which in turn may necessitate the use of some statistic other than the product-moment coefficient of correlation. Two examples of R-mode analysis are given, both using an average-linkage clustering procedure. The first uses proportional data from unmixed collections of a southwestern Alaskan phase of the Arctic Small Tool tradition, permitting a definition of artifact groups that may represent specialized activities, and highlighting 1 cluster that may have specifically temporal significance; some results of factor analysis are compared. The second example employs presence-absence information from surface-collected ceramics from Tlaxcala, Mexico, and makes possible a refinement of the ceramic sequence.