Abstract
From a collection of accessions of Asian greater yam, Dioscorea alata L., 235 different cultivars were rated for 28 previously selected cultivar-differentiating characteristics. From these data and from geographical data, groups were developed and compared. Various technique of numerical taxonomy were used. The best group were developed first by correlation clustering and then by clustering based on K-means. Fifteen groups were delineated, each of which was described in terms of some of the 28 characters. These groups differed highly significantly with respect to 27 characters. The groups corresponded in some cases to geographical regions, and in others the distribution of the group could be explained by intervention of man. They can be thought of as main branches of an evolutionary tree, not entirely separated and frequently anastamosing.
