Is sex estimation from handprints in prehistoric cave art reliable? A view from biological and forensic anthropology

Abstract

Estimation the sex of the creators of rock art scenes from handprints left in prehistoric caves has been of growing interest in archaeology in recent years. It has been suggested that both males and females were involved in symbolic activities, which has shaped the view of gender roles in prehistory. The experience from biological and forensic anthropology suggests, however, that using recent standards for the sex estimation of prehistoric handprints may be prone to errors. The aim of this study is to document the accuracy and reliability of sex estimation from handprints in a recent European sample and to assess the applicability of recent standards to the sex prediction of prehistoric artists. Our sample consists from 100 handprints of recent males and females from southern France. The sex of handprints is estimated by two discriminant functions using five direct measurements (DFdirect) and two indices (DFindex). The results showed that DFdirect correctly predicts sex in 92% of recent handprints but only about half the handprints can be classified with a certainty higher than 95%. The accuracy of DFindex is only 63% and cannot be successfully applied to sex estimation. We further suggest that the accuracy of both functions is overestimated due to the correct classification of handprints by chance and that especially DFindex is able to predict sex even in randomised datasets with no sexual differences. Finally, we demonstrate that both DFdirect and DFindex perform poorly when they are applied to population with hand size different from that used to derive them, i.e. that functions do not generalise across different populations and time periods. We argue that, given the lack of information about hand size in the population of prehistoric artists, recent attempts to estimate sex from handprints depicted in Palaeolithic cave art using morphometric data from recent populations is inevitably associated with unpredictable bias.