Results of a large-scale humeri pair matching study.

During the analysis of a commingled human remains assemblage, pair-matching is often conducted as part of the inventory to inform the DNA sampling strategy and to calculate the minimum or most likely number of individuals. As commingled assemblages become larger, pair-matching becomes more difficult, and it is unknown whether accuracy declines. Therefore, a study to determine the accuracy rates of visual pair-matching for multiple observers was conducted on a large, commingled human assemblage. The sample consisted of 580 left and right humeri (n = 287 and n = 293, respectively) from the commingled remains of the USS Oklahoma, which were undergoing analysis at the Defense POW/MIA Accounting Agency. Five anthropologists completed this study, determining humeri pairs with varying degrees of confidence and non-pairs. The overall precision was 81.0% and ranged from 74.1% to 95.9% for participants, including all confidence levels. When considering only confident matches, the overall precision for all participants increased to 91.4%.