A study of multiple bitemarks inflicted in human skin by a single dentition using geometric morphometric analysis.

Criticisms of the forensic discipline of bitemark analysis state that the range of distortion in the shape of bitemark impressions in skin has not been scientifically established. No systematic statistical studies exist that explore this problem. As a preliminary investigation of this issue, a single dentition was mounted in a mechanical apparatus and used to create 89 bitemarks in human cadaver skin, both parallel and perpendicular to tension lines. Impressions of the same dentition were also created in wax. 2D scanned images of the biting dentition were obtained. Locations of incisal edges of all 6 anterior teeth as well as the midpoint of the canine were captured as landmarks in all specimens. This set of landmark data was then studied using established geometric morphometric methods. All specimen shapes were compared using Procrustes superimposition methods, and by a variation of Procrustes superimposition which preserves scale information. Match criteria were established by examining the range of variation produced by repeated measurements of the dentition for each class of specimen. The bitemarks were also compared to a population of 411 digitally scanned dentitions, again using the match criteria. Results showed that bitemarks in wax had lower measurement error than scanned images of the dentition, and both were substantially lower than measurement error as recorded in skin. None of the 89 bitemarks matched the measured shape of the biting dentition or bitemarks in wax, within the repeated measurements error level, despite the fact that all bitemarks were produced by this dentition. Comparison of the bitemarks to the collection of 411 dentitions showed that the closest match to the bitemarks was not always the same dentition that produced the bitemarks. Examination of Procrustes plots of matched shapes showed non-overlapping distributions of measurements of bitemarks in skin, wax, and the dentition. All had statistically significant differences in mean shape. Principal component analysis (PCA) and canonical variates analysis (CVA) both showed clear segregation of the three types of data. The patterns of variance revealed by PCA showed several distinct patterns produced by skin distortion; alteration of relative arch width, and varying displacement of non-aligned teeth in the dentition. These initial results indicate that when multiple suspects possess similar dentitions, bitemark analysis should be approached with caution.