Meindl and Lovejoy

In 1985, Meindl and Lovejoy re-examined the Hamann-Todd Collection. They chose 10 ectocranial landmarks, although it is unclear whether they scored for bilateral expression (Meindl and Lovejoy 1985:60). They employed the following scoring methodology: 0 - no observable closure; 1 - 1 to 50% closure; 2 - 51 - 99% closure; 3 - 100% closure. Small (1 cm) lengths of a suture or specific sites were selected for inspection, for which only the judgment of one observer was required (Meindl and Lovejoy 1985:58). Since they proposed that in the case of forensic anthropology, the regularity of closure during the early adult years was not critical, only ectocranial sutures were studied (ibid.). Meindl and Lovejoy (1985:58) thus chose 17 ectocranial points for 236 crania from the Hamann-Todd collection, based upon the reliability of stated age at death.

During the initial analysis, they found that some of these landmark sites were of limited value for consistent age determination (Meindl and Lovejoy 1985:58). These sites were as follows (Meindl and Lovejoy 1985:58-60): (closure of) 1) parieto-mastoid; 2) squamosal point; 3) occipito-mastoid; 4) zygomatic and malar; 5) frontolacrimal; and 6) frontoethmoid. The sites which were retained (Meindl and Lovejoy 1985:60) were the vault system (midlambdoid, lambda, obelion, anterior sagittal, bregma, midcoronoal and pterion) and lateral anterior system (midcoronoal, pterion, spheno-frontal, inferior sphenotemporal and superior sphenotemporal). Meindl and Lovejoy (1985:60) found a chronological age ranking at the pterion, sphenofrontal, midlambdoid and lambdoid, respectively. These were based upon the assumption of underlying continuity of ordinal closure scales (the Kendall coefficient) (ibid.).

The next phase of analysis, to determine combination of sites which could be employed in age estimation, was based upon the following assumptions (Meindl and Lovejoy 1985:61): 1) sutures should demonstrate a protracted sequence of closure; 2) sutures should correlate with age during the primary period of its closure activity; and 3) the information provided by each suture should be specific to that particular suture. From these criteria, and since the calotte is usually the most durable in archaeological populations, they determined that the 5 lateral anterior sites were the best overall predictor of age (Meindl and Lovejoy 1985:61). It was assumed that the sample crania would exhibit a commencement and termination sequence which correlated with long bone epiphyseal fusion sequence (Meindl and Lovejoy 1985:62).

Using composite scores for each sample specimen, they found that the lateral anterior sites closed in the following sequence: Commencement - pterion, midcoronal, sphenofrontal, inferior sphenotemporal, superior sphenotemporal; Termination - pterion, sphenofrontal, midcoronal, inferior sphenotemporal, superior sphenotemporal (Meindl and Lovejoy 1985:63). The vault sutures closed in the following sequence: Commencement - obelion, pterion, anterior sagittal, lambda, midlambdoid, midcoronal, bregma; Termination - obelion, pterion, anterior sagittal, lambda, bregma, midlambdoid, midcoronal (ibid.). The overall (linear) correlation with known age was 0.57 for lateral anterior sites and 0.50 for vault sutures (ibid.).

It was found that the average mean deviation, for lateral anterior scoring, was 7.5 years, and 14.2 years for vault sutures (Meindl and Lovejoy 1985:64). Meindl and Lovejoy (1985:64-65) also examined whether or not race or sex had any effect on suture closure, thus affecting the accuracy of age prediction. Using analysis of covariance (ANCOVA) on the residuals, they found no measurable influence of either race or sex (ibid.). However, Meindl and Lovejoy (1985:62) state that any age estimation should take into account postcranial indicators as a control for suture cranial variations.

Buikstra and Ubelaker

In Standards for Data Collection from Human Skeletal Remains, Buikstra and Ubelaker (1994) proposed combining several aging methods, based on different cranial sutures (Baker 1984; Mann et. al. 1987; Meindl and Lovejoy 1985; Todd and Lyon 1924, 1925a, 1925b, 1925c) to predict age. According to Hershkovitz et. al. (1997:394), this ‘refined’ method ignored the fact that correlation between the recommended areas for inspection, along the cranial sutures, and age are very low. They (Hershkovitz et. al. 1997:394) proposed that these new standards still turn out old results, which seems to be apparent in their own research of cranial suture closure.

Nawrocki

In 1995, Nawrocki examined all three categories of sutures (ectocranial, endocranial and palatine) using regression and analysis of variance techniques (Nawrocki 1998:276). He studied 100 individuals of known sex, age and race (black/white) from the Terry Collection. Two individuals were chosen from each race/sex category in every half decade, ranging in age from 21-85 years (Nawrocki 1998:277). Although, for whites, the lowest age was 27. No specimens were excluded for any reason, barring damage therefore there was no control when examining ‘abnormal’ skulls (ibid.). Following Meindl and Lovejoy (1985:60), 27 landmarks were scored on each skull: ectocranial (16), endocranial (7), and palate (4) (Nawrocki 1998:278). Both the left and right side of the skull were scored when bilaterally expressed (ibid.).

A second sample was obtained in order to test the results from the Terry sample. This consisted of the crania of 61 white individuals (27 males; 34 females), aged 58-102 years, obtained from medical dissection rooms in Syracuse and Indianapolis (Nawrocki 1998:279). However, while Nawrocki (1998:279) notes that the presence of soft tissues and saw cuts on the cranial vault, limited suture observations of the endo and ectocranial surfaces, he does not indicate how the age at death of the sample crania was verified.

In order to produce a score, all 27 landmarks were added together, resulting in a range of 0-81 (Nawrocki 1998:279). It was found that there was a moderately strong correlation between the predicted age and the age of the Terry samples (Nawrocki 1998:279). However, Nawrocki (1998:281) does note that summing or averaging the data will result in a loss of information. One area of possible concern, regarding this reliance upon linear regression models, is that it allows for a predicted range from 25.3-82.8 years, even though the sample crania were 58-102 years old (Nawrocki 1998:279). Additionally, when tested against the Terry samples, the predicted ages deviated, on average, between 9 to 21 years (Nawrocki 1998:286). It should be noted that Todd’s average age deviation was 6 years, which he deemed to be unacceptable.

Creating an equation to test Meindl and Lovejoy’s proposal that race or sex may affect suture closure rates, Nawrocki found that there was a correlation between suture closure and sex, but not race, although there was an interaction between race and sex (Nawrocki 1998:282). He suggests that this finding, which is contrary to that of Meindl and Lovejoy, is due to the use of more cranial landmarks (ibid.). Nawrocki (1998:282) created 8 different equations to test each case sample, although this required dropping the number of landmarks from 27 to 15.

In his conclusion, Nawrocki (1998:288) proposes two possible sources of error: sampling (random) and secular trends (non-random). Sampling error is said to occur when the two parent populations are similar overall, yet the test sample is drawn unevenly (ibid.). However, this seems to be a rather weak argument, considering the emphasis placed on obtaining ‘random’ samples for statistical analysis. Secular trends systematically change the nature of the second population, or portions of it, so that the regression derived on the first are no longer as accurate (Nawrocki 1998:288). In this instance, one is concerned that the sample is too randomly varied. In other words, researchers seem to prefer finding fault with the ‘sample’ rather than with their chosen methodology.