Dry bone and virtual modality interchangeability for the estimation of sex on the human pelvis and skull

Background: Traditional osteological methods are typically based on visual and tactile observational approaches. However, the tactile sensation is excluded when investigating skeletal remains in virtual reality. This raises the question of the actual comparability of data obtained from osteological analysis on analogous versus virtual modality. The aim of this work is to address this topic and quantify the deviation between the scoring on dry bone and virtual models of sexually dimorphic features on the human pelvis and skull. Materials and methods: We applied seven widely used sex estimation methods to 200 archaeological pelves and 223 skulls. We scored and measured each method-specific trait on dry bone and computed tomographic (CT) models of the same individual. We added observations on 3D surface scans from sample subsets of pelves (N=39) and skulls (=50) for comparability. We compared the scores and measurements obtained from the application of each method to the three sets of observations using Cohen's κ tests and relative technical error of measurement (rTEM). Results: 1) metric traits are more repeatable and consistent than nonmetric traits; b) virtual modalities compared better to each other than visual-tactile modalities and c) precise trait descriptions are more important than the modality. Discussion: Our data suggest that traditional sex estimation methods developed on dry bone (visual-tactile sensations) can be applied interchangeably to virtual specimens (visual-only sensation) without obtaining substantially different information. More than other factors, however, precise trait definition is pivotal for the interchangeability of analogous and virtual modalities