On the systematic bias in skeletal sexing.

Comparison of a large series of sexed adult skeletal populations and a similar series of adult pre-industrial peoples shows that there is a regular and systematic bias in the sexing of adult skeletons. This bias, which is about 12% in favor of males, is due to the nature of secondary sex characteristics in bone. It should be corrected in skeletal series before demographic analysis is made of them. Application of this knowledge and the same data to problems of age-specific male and female mortality rates is inconclusive, but points to an area for important future investigation. Application to the fossil record confirms some ecological ideas about human evolution. For many purposes, the analysis of a series of excavated skeletons requires that the adult specimens be sexed. On the accuracy of the sexing procedures hinge the results of, for example, analysis of mortality patterns or metric analysis of sexual dimorphism. This paper will look at the sexing procedures as they have been used and will demonstrate that they have resulted in a significant bias which must be considered when sex-specific analysis of skeletal material is attempted. There are many works which discuss the morphological features of the bones of the skeleton by which sexing can in theory be performed (HrdliEka, ’52; Krogman, ’62; Bass, ’71). The pelvis seems to be the most reliable single bone for this, and it receives the greatest treatment in the literature. Pelves are often not well preserved, however, and in such cases sexing must be performed on the other, less reliable, bones. Several studies have stressed the increased accuracy in sexing skeletal material which results from the combination of many measurements from a single bone or a series of measurements from different bones; these studies are multivariate in nature (Hanna and Washburn, ‘53; Pons, ’55; Thieme and Schull, ’57; Giles, ’63). They recognize that as secondary sex characteristics on bone vary from individual to individual and from AM. J. PHYS. ANTHROP., 37: 23S250. group to group, one must sex a newfound individual only in comparison to a well analyzed population. These studies derive methods for determining population-specific cutoff points between males and females for the particular array of traits they include. The multivariate methods which consider populational variation are clearly superior to subjective morphological criteria in that they provide rigorously-defined, objective methods for determining sex and their results seem to be such that fewer specimens are “doubtful” than is the case with the morphological criteria. Unfortunately, however, the more sophisticated methods require better data on which to work; usually, one needs to have the same bones from each individual in the population, and often the measurements to be taken demand a good state of preservation as well. In practice, the data are usually somewhat fragmentary, and the rule seems to be that subjective anatomical criteria are used which differ with each specimen sexed, depending on the bones represented. The vast majority of sexed skeIetal populations in the literature have been sexed in this way. Stewart (‘54) and HrdliEka (‘52) in fact support subjective methods when performed by experienced professionals. In today’s computer parlance they could argue that the trained human brain can