The current study was undertaken to test the validity and reproducibility of the Purkait triangle method and some alternative proposals for sex prediction from the proximal femur in the adult population of Spain. To that end, sexual dimorphism of the maximum femoral head diameter and the minimum femoral neck diameter were also evaluated. The study was conducted on 186 femora (109 males and 77 females) taken from the San José collection of identified individuals (Southern Spain). Discriminant function analyses (DFA) employing the jackknife procedure for cross-validations were considered. Overall, more than 94% of individuals of both sexes were correctly classified. The most dimorphic single variable from the triangle method was the intertrochanteric apex distance (BC) that reached 85.5% accuracy, falling below those obtained for the femoral head and femoral neck diameter, respectively, (89.8 and 91.9%). Combining BC with the neck diameter, the predictive ability increased to 92.5%; when femoral head diameter was added to the latter two, the classification success rate improved further up to 94.6% (94.1% after cross-validation). We conclude that the classification success rates of the Purkait's method remained considerably below any of those obtained with the models proposed in the present study which proved to be a much better and more reliable choice both as single predictors and in combination with other variables.
Keywords: Discriminant functions; Femur; Forensic science; Proximal end; Sex estimation.