Résumé
Traditionally, the separation of domestic pig remains from those of wild boar in zooarchaeological assemblages has been based on the comparison of simple size measurements with those from limited numbers of modern or archaeological reference specimens and then applying poorly defined cut-off values to make the identification calls. This study provides a new statistical framework for the identification of both domestic and wild Sus scrofa using standard molar tooth lengths and widths from a large modern comparative collection consisting of 407 West Palearctic wild boar and domestic pigs. Our study continues to rely upon so-called 'cut-off' values that correspond to the optimal separation between the two groups, but based upon a measure and visualisation of the error risk curves for erroneous identifications. On average, wild boar have larger teeth than domestic pigs and cut-off values were established for maximum tooth length and width, respectively as follows: 239 cm and 1.85 cm for second upper molar, 3.69 cm and 2.13 cm for third upper molar, 2.26 cm and 1.50 cm for second lower molar, 3.79 cm and 1.75 cm for third lower molar. Specimens below and above these cut-offs are most likely to be, respectively, domestic pig and wild boar and the risk of providing a wrong identification will depend on the distance to the cut-off value following a relative risk curve. Although likely containing high risk of inherent statistical error, nonetheless this basic metrical identification-tool (based only on recent specimens), is here shown to correctly re-identify 94% of the Neolithic pigs from Durrington Walls (England) as domestic pig. This tool could be employed not only to systematically re-evaluate previous identifications of wild or domestic Sus scrofa, but also to establish new identifications where more powerful and reliable approaches such as Geometric Morphometrics cannot be applied. (c) 2013 Elsevier Ltd. All rights reserved.