Identified skeletal reference collections and the study of human variation

Abstract

<p>A comprehensive model that builds on cemetery studies theory and new biocultural synthesis theory is presented for investigating human variation using reference collections. This model is used to investigate several hypotheses related to the use of skeletal reference collections and the race concept in skeletal biology. Are racial categories or equivalent terms useful for investigating human variation? Have biases in reference collections resulted in a misinterpretation of human variation? Is it possible to identify and control for some of the biases in reference collections and develop identification methods that are still useful in the 21st century? Documentary, historical and skeletal data were collected from two different reference collections: the Terry and Coimbra Collections. Some data from the Forensic Anthropology Data Bank (FDB) were also used in conjunction with these two collections in analyses related to patterns of sexual dimorphism, sex determination methods, and assessment of secular changes using skeletal data. A critical review of the methodology for the study of secular change in femur length illustrates that there are methodological costs of ignoring bias in reference collections. In a combined sample from several collections, what appear to be significant secular changes occur at the points of transition between data sources. Dividing samples into racial sub-samples does not control for so-called population differences and only further confounds the analysis of secular changes using reference collections. An alternative methodology was followed for the development of a series of metric sex determination methods with allocation accuracies of 90-98.5%. Logistic regression was used to develop a series of models that use a new alternative measurement of the pubis along with traditional measurements of the hipbone and femur. Demographic and historical data were used to select a reference sample that included a wide range of human variation. Testing suggests that these metric models are widely applicable. Various skeletal, demographic, and historical analyses indicated that racial categories or equivalent terms are not useful for investigating human variation using reference collections, and that some of the biases in reference collections have resulted in a mis-interpretation of human variation in past research. It is possible to identify and control for some of the biases in reference collections and develop identification methods that are still useful in the 21st century.</p>