The problem of classification and distance between two populations

Abstract

The problem of specifying an individual as a member of one of many populations, and the classification of a number of populations themselves in some significant system based on the configuration of various characteristics, are of great importance in anthropological and biological research. We may find a collection of skulls with unspecified sexes, and the problem faced by an anthropologist is assigning proper sex to a skull. Judgment based on mere anatomical appreciation of a skull may not be altogether wrong, but is subject to criticism especially when objective methods are available. This problem has been solved by Fisher's discriminant function. If we have a collection of skulls grouped according to specified populations, the problem is to arrive at constellations of populations such that any two members of a constellation are closer to one another than any two belonging to different constellations. This problem can be solved by Mahalanobis's generalized distance.