Sex Estimation from the Pelvis: Phenice, Sciatic Notch and Sub-Pubic Angle

Thomas Phenice's 1969 paper in the American Journal of Physical Anthropology described a non-metric method based on three traits observed exclusively on the pubic bone. The pubis is the most anteriorly positioned of the three pelvic bones (the other two being the ilium and the ischium), and it contributes to the pubic symphysis at the front of the pelvis. In intact skeletons, it is typically among the better-preserved pelvic elements.

The ventral arc. The ventral arc is an elevated ridge of bone that runs across the ventral (anterior-facing) surface of the pubis in a curved line, sweeping from the pubic symphysis toward the ischiopubic ramus. In females, this arc is prominent and well-defined, forming a clear visual curve on the ventral surface. In males, the ventral arc is absent or indistinguishable from the surrounding bone surface. Scoring is binary: present (female) or absent (male). Intermediate cases exist but are uncommon in the reference sample.

The sub-pubic concavity. Viewed from below (the inferior aspect), the medial border of the inferior pubic ramus in females curves inward, forming a concavity that runs toward the pubic symphysis. This concavity is the result of the wider sub-pubic space in females, which is continuous with the wider sub-pubic angle. In males, the medial border of the ischiopubic ramus runs in a relatively straight line or convex curve; the concavity is absent. Again, scoring is binary: concave (female) or straight/convex (male).

The medial aspect of the ischiopubic ramus. The ischiopubic ramus, examined at its medial surface, shows a sharp and narrow ridge in females, reflecting the tension from the fascial attachments that differ between sexes. In males, the medial aspect is wider and broader, more shelf-like. This is the most subjective of the three traits, with the highest inter-observer error rate in published studies. Scoring distinguishes a narrow, sharp ridge (female) from a broad, flat surface (male).

Phenice's original accuracy on 275 Terry Collection specimens was 96 per cent using all three traits. The ventral arc alone performed at 83 per cent, the sub-pubic concavity alone at 80 per cent, and the medial ramus aspect alone at 72 per cent. Combining all three brought accuracy to 96 per cent. The method was designed to be applied non-metrically, without callipers, which made it accessible in field conditions and partial recovery scenarios.

Ashley Klales and colleagues published a revision of the Phenice method in 2012 in the American Journal of Physical Anthropology, using a sample of 939 individuals from the Hamann-Todd Collection (Cleveland) and the Bass Collection (Knoxville). Their key innovation was replacing the binary female/male scoring with a five-point ordinal scale for each trait (1 = definitely male, 5 = definitely female, with 2, 3, and 4 as intermediate scores), and fitting a logistic regression model that combined the three ordinal scores into a posterior probability of female sex.

The ordinal approach addressed two problems with Phenice's binary scoring. First, real skeletons often show intermediate morphology on one or more traits; binary scoring forced a decision where the observer was genuinely uncertain, adding error. Second, binary scoring discarded potentially useful information about the degree of expression of each trait. The ordinal scores, combined in the logistic regression, gave posterior probabilities that the osteologist could report directly: "probability of female 0.83" is more informative in a forensic report than "female."

The Klales 2012 method performed at 89 per cent correct classification in their validation sample, with a broader range of intermediate probabilities clustered in the 0.40 to 0.60 range that the binary method had previously forced into a categorical call. Critically, the method produced a posterior probability distribution that could be incorporated into a Bayesian likelihood ratio framework, which is increasingly the expected reporting standard in courts in the US (under Daubert) and the UK (under R v. Adams and subsequent forensic statistics guidance from the Forensic Science Regulator).

The Klales revision has been cross-validated on Australian, South African, and Eastern European samples with variable results, reinforcing the requirement for population-specific validation before application. It remains the current recommended approach in the American Board of Forensic Anthropology's (ABFA) methodological standards for professional practice.

The greater sciatic notch is the large, curved notch on the posterior border of the ilium, located between the posterior inferior iliac spine above and the ischial spine below. Its width is functionally related to the width of the posterior pelvis and the position of the sacrum, both of which are sexually dimorphic for the same obstetric reasons as the pubic morphology.

In males, the greater sciatic notch is characteristically narrow, with the notch angle typically less than 68 degrees. The two walls of the notch are close together, and the notch has a deep, V-shaped appearance when viewed from the lateral aspect. In females, the notch is wider and more open, with the two walls diverging to form a shallower, U-shaped or rounded appearance. The notch angle is typically above 68 degrees in females.

Budinoff and Tague (1990) and Walker (2005) systematised notch scoring using ordinal scales. The Walker 2005 five-point scale assigns:

• Score 1: very narrow notch (most male);
• Score 2: narrow, predominantly male;
• Score 3: intermediate;
• Score 4: wide, predominantly female;
• Score 5: very wide notch (most female).

Walker validated this scale on a multi-ethnic US sample that included Black Americans, White Americans, and Hispanic Americans from the William Bass Donated Collection and the New Mexico Documented Collection. Accuracy for the sciatic notch alone was approximately 78 per cent for males and 83 per cent for females, lower than the complete Phenice triad but higher than any single Phenice trait. The notch is particularly valuable in fragmentary cases where the pubis is absent or damaged, as the posterior ilium is often among the more robust and better-preserved pelvic elements.

A combined scoring approach, using the Phenice triad where the pubis is preserved and the sciatic notch where only the ilium is recoverable, is standard forensic practice. When both are available, disagreement between the two methods (pubis suggesting female, notch suggesting male, or vice versa) should be reported explicitly with the posterior probability for each, rather than averaged or resolved by selecting the preferred result.

The sub-pubic angle is the angle formed at the inferior apex of the pubic arch, where the two inferior pubic rami meet at the pubic symphysis. In females, the wider pelvic inlet and broader sacrum push the inferior rami laterally, producing an obtuse sub-pubic angle typically above 90 degrees (often cited as 90 to 115 degrees in most female populations). In males, the narrower pelvis produces an acute angle, typically below 90 degrees (commonly 50 to 80 degrees).

Measurement is performed with a large bone protractor placed against the inferior rami at the level of the pubic symphysis, or by the "thumb method" (thumbs placed along the inferior rami; in males the thumbs fit comfortably below the arch, in females the arch is too wide for the thumbs). The latter method, while informal, is described in Krogman and Iscan's classic 1986 text "The Human Skeleton in Forensic Medicine" and remains in use in field contexts.

Metric measurement of the sub-pubic angle is sensitive to the condition of the pubic symphysis face: postmortem damage, erosion, or incomplete fusion can alter the effective angle measured. Published sex-specific cut-offs show population variation. In Walker's 2005 multi-ethnic US sample, a cut-off of 90 degrees gave approximately 80 per cent correct sex classification. In Mukherjee's 1955 study of Indian pelves from the medical school collection at Calcutta, the female mean sub-pubic angle was approximately 104 degrees (standard deviation 10.6) and the male mean was 67 degrees (standard deviation 8.8), with an overlap zone that produced approximately 85 per cent accuracy at a 90-degree cut-off. In a South African sample from the Pretoria collection (Steyn and Iscan 1997), accuracy was similar at approximately 83 per cent for females and 86 per cent for males.

The sub-pubic angle is best treated as a corroborating indicator alongside the Phenice triad and sciatic notch, rather than a standalone sex determination method. Its value is highest in fragmentary cases where only the inferior pubic arch is preserved and the pubic symphysis and ventral surface are damaged or absent.

The pre-auricular sulcus is a groove or depression located just anterior to the auricular surface of the ilium, at the site of the anterior sacroiliac ligament. It is more common in females than in males, and in females it has been associated with parturition because the forces of childbirth stress the sacroiliac joint and may leave a bone remodelling signature in this area.

The sulcus has been scored on various scales, most commonly a four-point scale: 0 (absent), 1 (slight groove), 2 (definite groove), 3 (pronounced pit or groove). In published US samples, the sulcus is present in approximately 58 to 70 per cent of females and in 5 to 10 per cent of males. In an Indian sample (Singh and Potturi 1978), prevalence in females was approximately 52 per cent. The presence of the sulcus strongly suggests female sex (specificity is high in most populations). Its absence, however, provides little information, since a large minority of females lack it entirely.

The pre-auricular sulcus is therefore asymmetric in its diagnostic value: presence is moderately strong evidence for female sex; absence is essentially uninformative. Iscan and Derrick (1984) critiqued the sulcus as a sex indicator precisely because its base rate of absence in females is too high to make it useful in the indeterminate zone. Most current forensic anthropology standards treat it as a supplementary indicator only, worth recording in the case file but not relied upon as a primary sex indicator. In jurisdictions following SWGANTH (Scientific Working Group for Forensic Anthropology, now incorporated into the OSAC Forensic Anthropology Subcommittee standards) or the UK Forensic Science Regulator's codes, reporting the sulcus result without pairing it with the primary indicators would be considered incomplete methodology.

The accuracy figures quoted for pelvic sex estimation methods are not universal constants; they are properties of the method applied to a specific reference population under specific scoring conditions. Every major pelvic sex estimation study has been conducted on skeletal collections assembled from specific geographic and demographic populations, and the reported accuracy reflects how well the method performs within that population's morphological variation.

US and European reference samples. The Phenice method was validated on the Terry Collection (Washington DC, now at the Smithsonian National Museum of Natural History), assembled primarily in the 1930s and 1940s from individuals of Black and White American backgrounds with documented demographic data. Walker's 2005 sciatic notch and combined scoring used the Bass Donated Collection (University of Tennessee Knoxville), a more recent US sample with greater demographic diversity. European validation studies have used the Coimbra Collection (Portugal), the Athens Collection (Greece), and the CAL Milano Collection (Italy), all showing accuracy within ±5 per cent of the Terry Collection benchmarks for the Phenice triad. The ENFSI Forensic Anthropology Working Group's published best-practice guidelines (2019) recommend citing the specific reference population used and its demographic profile alongside the sex determination conclusion.

Indian reference samples. Mukherjee (1955) published measurements of 60 male and 60 female pelves from medical college collections in Calcutta (now Kolkata), providing population-specific means and ranges for the sub-pubic angle, sciatic notch angle, and other pelvic dimensions. Singh and Potturi (1978) added a study of sciatic notch width and pre-auricular sulcus prevalence on 200 Indian pelves from AIIMS (All India Institute of Medical Sciences, New Delhi). Both studies confirmed that the direction of sexual dimorphism in Indian pelves matches the pattern established in US and European samples, but that absolute measurements and trait prevalence differ enough to justify using Indian reference values rather than US-derived cut-offs when the remains are most likely from an Indian population. In practice, most Indian forensic anthropology casework applies a combination of the morphological (non-metric) Phenice and Walker criteria alongside the Indian-calibrated metric thresholds from Mukherjee and Singh-Potturi.

South African and East African reference samples. Steyn and Iscan (1997) used the Pretoria Reference Collection (a South African Black skeletal sample) and the Raymond Dart Collection at the University of the Witwatersrand to validate pelvic sex estimation. Their results for the sciatic notch and sub-pubic angle showed accuracy comparable to or slightly below US benchmarks. Patriquin et al. (2005) extended this to a larger Pretoria sample and found that while Phenice accuracy was maintained at approximately 90 per cent, the specific morphological expressions of some traits, particularly the ventral arc, showed greater within-sex variability in the South African sample than in the Terry Collection. The Witwatersrand collection remains one of the most important African skeletal reference resources for forensic calibration in southern Africa.

Interpreting mixed signals. When an osteologist applies multiple methods to a single pelvis and obtains mixed results, for instance a clearly female sciatic notch paired with an intermediate ventral arc, the correct approach is not to average the results or select the majority-rule call. Each method's posterior probability should be recorded and combined using a Bayesian approach, or the uncertainty should be explicitly reported as "most consistent with female but with intermediate morphology on one Phenice trait." This is the reporting standard under the ABFA, SWGANTH/OSAC, and ENFSI best-practice documents.

Forensic anthropology casework rarely presents an ideal, intact, clean pelvis for scoring. The practical workflow must adapt to preservation state, fragmentation, and the presence or absence of specific bones.

1. Inventory and condition assessment
   List all pelvic elements recovered: left and right os coxae (each comprising ilium, ischium, pubis), sacrum, coccyx. Note preservation state: intact, fragmentary, erosion, burning, rodent damage, soil staining. Document which specific trait zones are assessable (ventral pubis, inferior ramus, sciatic notch, auricular area, pubic symphysis). Only assessable zones contribute to the determination.
2. Phenice triad (if pubis intact)
   Score all three traits on both left and right pubis if available. Note score for ventral arc (present/absent or Klales 1-5 ordinal), sub-pubic concavity (concave/straight or ordinal), and medial ramus (narrow ridge/broad or ordinal). Record inter-side consistency. If Klales ordinal: input three scores into the logistic regression to obtain posterior probability of female.
3. Sciatic notch (Walker 2005, if ilium intact)
   Score on both left and right ilium. Apply Walker 1-5 scale. Compare with Phenice result: if consistent, report combined confidence is high. If inconsistent, report both scores and their population-specific accuracy independently.
4. Sub-pubic angle (if inferior arch preserved)
   Measure with bone protractor. Compare result against population-specific reference data (US Walker 2005 cut-off 90 degrees; Indian Mukherjee 1955 means 67 male, 104 female). Use as corroborating indicator only.
5. Pre-auricular sulcus (supplementary)
   Score presence/absence and grade if present. Report as supplementary finding: presence supports female, absence is uninformative.
6. Compile and report
   State which methods were applied and why (preservation state). Report accuracy of each method for the relevant reference population. Provide overall sex assessment with explicit statement of confidence level. If results conflict, report the conflict and explain the basis for the composite conclusion.