Kristen M. Livingston* and Matthieu Baudelet | University of Central Florida Jonathan Bethard | University of South Florida Katie Zedjlik-Passalacqua | Western Carolina University Abstract: The commingling of human remains poses an obstacle for death investigations in both modern and archaeological forensic contexts. After recovering a mixed assemblage, anthropologists face the challenge of sorting each skeletal element to its proper individual. Using physical features and osteometric methods, the reassociation process can be tedious, especially if bones have undergone fragmentation or taphonomic changes. However, in addition to specific physical traits, bones also have chemical profiles representative of the individual. This information provides useful discriminatory data for sorting. This study proposes that the elemental signatures obtained from bones in commingled assemblages can be used as a preliminary sorting technique. Laser-induced breakdown spectroscopy (LIBS) is an analytical technique well-suited for acquiring chemical information from bones. It requires no sample preparation and provides an emission spectrum within seconds that is representative of the sample surface composition. LIBS is also a quasi-nondestructive method, showing no noticeable indication that material has been removed from the bone during analysis. Furthermore, LIBS technology is available in portable, field-deployable instruments. Because much of the casework for forensic anthropology begins out in the field, handheld instrumentation conveniently aids in efficient analysis. To simulate data collection from a mass grave, the skeletal remains of 12 individuals were obtained from the Forensic Osteology Research Station (FOREST) decomposition facility at West Carolina University. A dataset was created by acquiring LIBS spectra from multiple locations on 28 bones for each individual, providing more than 2,000 chemical signatures for classification. Following data reduction and optimization, supervised learning algorithms were used to build discriminant models for the classification of each individual. These models were able to correctly match unclassified bones to their corresponding individuals with greater than 90% accuracy. Further statistical analysis of the spectral dataset provided insight on the significance of some trace elements responsible for the variation between each set of skeletal remains as well as the minimum number of bones required to classify individuals. The results of this study illustrate how the chemical profiles of bones help expedite the sorting process for skeletal assemblages and demonstrate the usefulness of portable LIBS as a potential tool to help forensic anthropologists reassociate commingled remains directly in the field.