Robert J. Terry began collecting human skeletal remains in the area of St. Louis, Missouri for research and educational purposes in 1898. He continued collecting skeletal specimens in the Anatomy Department at Washington University until his retirement in 1941. Mildred Trotter succeeded Terry as anatomy professor and continued his collecting, and strove to balance the demographic distribution of the collection. In 1967, after her retirement, the collection was moved to the Smithsonian Institution's National Museum of Natural History. As with several other well-documented collections, the Terry Collection is widely used for a diverse range of anthropological and medical research. Despite its extensive use, there has been limited discussion of the collection's history and incomplete description of holdings and associated materials of this collection. In this paper, the historical background of the collection and the collection process is described; the demographic composition of the collection, and a description of the documentary and supporting data are presented; and the quality and of these data are assessed. The Terry Collection consists of 1,728 individuals. Age at death ranges from 14-102 years, with the majority of the individuals ranging from 20-80 years. Year of birth ranges from 1828-1943; the mean year of birth for males is 1880, and for females it is 1884. The mean age at death for males is 53 years, and for females it is 58 years. Terry's strict protocols for the processing of cadavers and the recording of documentary data make the Terry Collection a valuable resource for anthropological and medical research.

Since the earliest descriptions of the pubis length measurement, it has been recognized that the location of the key landmark in the acetabulum has to be estimated. Using samples from the Terry Collection (n = 324) and the Coimbra Collection (n = 232), the purpose of this research is to, first, test the reproducibility of a new alternative to the traditional measurement of the pubis, and second, to use the best measurement of the pubis along with other measurements of the hipbone and femur to develop a metric method that can be used with confidence to determine the sex of individuals of various geographic origins and time periods. In this study, it was found that, first, the alternative pubis measurement, known as the superior pubis ramus length (SPRL), can be collected more reliably with less mean intra-observer error (0.57%) than the more commonly used manner of measuring the pubis (2.7%). Second, a logistic regression sex determination method using the SPRL, along with other measurements of the hipbone and femur, has an allocation accuracy of 90% to 98.5% (depending on the model used and the manner of testing) across independent samples. Third, traditional racial categorization was irrelevant to the accuracy of the method. Fourth, measurement error greater than 2% in the measurement of the pubis can be the difference between a correct and an incorrect allocation of sex, particularly in borderline cases.

No abstract

The pubic bone is considered one of the best sources of information for determining sex using skeletal remains, but can be easily damaged postmortem. This problem has led to the development of nonpelvic methods for cases when the pubic bone is too damaged for analysis. We approached this problem from a different perspective. In this article, we present an approach using new measurements and angles of the proximal femur to recreate the variation in the pubic bone. With a sample from the Terry Collection (n > 300), we use these new variables along with other traditional measurements of the femur and hipbone to develop two logistic regression equations (femur and fragmentary hipbone, and femur only) that are not population specific. Tests on an independent sample (Grant Collection; n = 37-40) with a different pattern of sexual dimorphism resulted in an allocation accuracy of 95-97% with minimal difference by sex.

Sex estimation from skeletal remains can be an important part of preliminary identification. The best source of information for estimating sex is the pelvis but it is not always available for analysis. For these cases, a probabilistic sex estimation method is presented using combinations of standard and alternative measurements of the clavicle, humerus, radius, and ulna. Various equations are developed that are not population specific and that are applicable in various recovery scenarios. The equations were tested using four independent samples (n > 370), including a forensic sample. Allocation accuracies vary by test sample and equation and are consistently good (87.4-97.5%) except for a sample of very small males that show the extreme effects of poverty and mortality bias. For many of the cases where allocation was incorrect, the probabilistic approach indicated that no confidence should be placed in the incorrect allocation and the unknown should be classified as sex indeterminate.