The Morphological and Osteometric Sorting of Human Commingled Assemblages: Tali and Calcanei
MetadataShow full metadata
The accurate sorting of elements from a commingled assemblage can be important to estimating the number of individuals represented by those particular elements, and to gain a better understanding of the actual assemblage. Many methods of sorting utilize visual matching, particularly in smaller assemblages, while others use osteometric sorting methods specific to the re-association of elements of the same side or pair-matching elements of opposite sides. The purpose of this study was to create a method utilizing morphological variation of the clinical subtalar joint (CSTJ) of tali and calcanei, in combination with using simple and multiple regression models to make predictions of matching elements in order to accurately sort elements of a larger commingled assemblage.
Facet patterns were categorized into four types, with varying anterior and middle articular facets, though a fifth type is acknowledged. Type A consisted of a missing anterior facet, but a separate middle facet. Type B had separate anterior and middle facets. Type C had a continuous anterior and middle facet, but with a significant angle between them. Type D also consisted of a continuous anterior and middle facet, but this facet was flat. Type E typically consists of the anterior, middle, and posterior articular facets all being fused together, but was not observed in this assemblage.
Five measurements from each element were utilized in the creation of simple and multiple regression equations and include the length and width of the posterior articular facet, the breadth of the element, and the fraction and subtense of the most superior point of an arc on the posterior articular facet. These regression equations were created using 120 individuals from the Texas State Donated Skeletal Collection. These equations utilized the left talus to predict measurements of the calcaneus from the left side, as well as the right talus to predict measurements of the calcaneus from the right side for reassociation. The left talus was then used to predict measurements of the right talus, and measurements of the left calcaneus was used to predict measures of the right calcaneus for pair-matching. The absolute value of the differences between element measurements was taken, and then all measurement differences were combined into an average to predict element matches, rather than matching elements from measurement to measurement. Elements were then ranked by lowest average difference up to 1.5mm average difference.
A holdout test consisted of 30 individuals with varying degrees of element presence. Out of 120 elements, only 85 elements were present. Using the predictions generated via the regression equations as well as elimination of potential matches via facet pattern the left elements were re-associated with 84.6% accuracy, the right elements were re-associated with 93.8% accuracy, the calcanei were pair-matched with 93.8% accuracy, the tali were pair-matched with 93.3% accuracy, and the solo elements were classified as solo with 66.7% accuracy. These results are similar to other metric methods of re-associating tali and calcanei, but are less accurate than analyses for pair-matching using two- or three-dimensional shape comparisons. Overall this method worked well in predicting ranked matches, as well as eliminating potential matches using facet patterns.