The present study shows a significant difference in the scores between the sexes with a combined overall accuracy of 62.05%. When classification accuracy for each sample was calculated, it was found that there was a considerable difference between males and females, with a bias towards males in two of the four samples (Table
2). It is interesting to see that this bias seems to have no grounding in an individual's ancestry since male bias is seen in South African Black and Spitalfields samples and female bias in the South African Coloured and White samples. A difference was also found between the four samples tested. The most considerable difference in accuracy was seen in male samples between Spitalfields, UK, and the Coloured South African sample having a discrepancy of 40.5%. It is worth noting the difference in overall accuracy between the Spitalfields and South African White samples (75.35% and 69.3%, respectively), and the South African Black and Coloured samples (57.3% and 46.25%, respectively).
In terms of intra/inter-observer error, Eliopoulos [
27] achieved an intra-observer agreement of 81.4% and an inter-observer agreement of 50.8%, which mimic what was found in the present study. The low inter-observer agreement found in this study (
k = 0.529) causes issues when trying to implement new techniques; if they cannot be repeated accurately by other observers then the technique should not be used regardless of the accuracy it can produce.
Rogers and Saunders [
25] found that the obturator foramen was the second most effective sex indicator for their Canadian sample, with accuracies as high as 93.8% which contradict the results found here. There are several reasons why there is a discrepancy. Firstly, the individuals from Canada may exhibit extreme morphologies for males and females for the obturator foramen. Secondly, they use descriptions made by St. Hoyme [
26], where no visual diagrams can be found; therefore, trying to recreate their study regarding the obturator foramen is impossible. If the descriptions from Ferembach et al. [
24] were the same as St. Hoyme [
26], then caution should be taken as there are no visual diagrams for how to score this trait, and the only description for each score contains a maximum of three words. Fourth, because a new scoring method was created for this study, different criteria were used, which could account for some discrepancies between this study and Rogers and Saunders’ [
25]. However, there is nearly a 20% difference in accuracy when comparing Rogers and Saunders’ [
25] result with Spitalfields from this study. This study is more in agreement with the interpretation made by St. Hoyme [
26] that the foramen has little value when considering its use for sex estimation despite it showing signs of sexual dimorphism. Accuracies did differ between populations in this study but averaged an overall accuracy of 62.05%, similar to Arsuaga and Carretero [
41] (63.5% correct classification). Comparing the results to Beirry et al. [
29], significant differences in accuracy rates are present. However, this can be explained by using Fourier analysis and Discriminant Function Analysis to categorise the oval (male) and triangular (female) expressions rather than relying on the naked eye. Ridgeway et al. [
42] concluded after analysing 96 female pelves that the considerable amount of shape variation seen in the obturator foramen could be due to an individual’s height, noting that the greater the foramen area is, the taller the individual was. Ridgeway and colleagues [
42] found no significant difference between shape and a person’s ancestry. Because an individual’s height was not considered in the present study, as it would have been largely estimated, unlike the known height in Ridgeway’s study, their conclusions could not be tested. However, a large amount of variation was seen not only between samples but within samples as well (Table
2 and Fig.
4), which coincides with the findings of Ridgeway et al
. [
42] that there is no link between shape and an individual’s race.
Waldron [
43] and Stojanowski et al. [
44] observed the preservation rate of skeletal elements from different archaeological assemblages and found that when focussing on the pelvic bones, the sacroiliac joint was preserved at a much higher frequency than the
os pubis. The obturator foramen is not preserved when the break occurs along the inferior and superior ramus of the pubis. The low preservation of the
os pubis in archaeological and forensic cases means this method is not usually performed.
The obturator foramen seems to provide variable success across the literature, with more literature stating its limited use in sex determination [
26,
27,
41]. With this in mind, the results of the present study we can conclude that the obturator foramen should not be used to determine sex on its own as it is a poor discriminator. Arguments can be made to include it in a multifactorial assessment that includes other morphological traits of the pelvis. However, since most other traits outperform the obturator foramen, and the fact that it is prone to be non-assessable if the pelvis is fragmented, which happens frequently, we do not recommend its use for the assessment of sex.