Particle Swarm Optimization (PSO)

The basic demographic data produced in this study include the age at death of every indi- vidual and the sex of the adults. Because measurements of the postcranial skeleton were taken, premortem stature was also calculated for adults whose sex could be estimated. Techniques used to estimate age, sex, and stature were drawn from multiple sources, recorded separately, and then combined to produce the most accurate assessment. Care was taken to record enough information for comparative analysis with other published Imperial cemetery populations.

4.2.1

Sex Estimation

Sex of adult skeletons was estimated using a variety of techniques reproduced in Buikstra and Ubelaker’s Standards for Data Collection from Human Skeletal Remains (1994), including

pelvic morphology, cranial morphology, and long bone measurements. Phenice’s (1969) tech- nique for sex determination was used above all to estimate sex from the subpubic region of the os coxae. In addition, individuals with intact pelves were assessed for differences in the greater sciatic notch and the preauricular sulcus (Buikstra and Ubelaker, 1994). Sexually dimorphic cranial features were scored as per Acs´adi and Nemesk´eri 1970. Finally, in some cases, long bone measurements were used in the discriminant function analysis of FORDISC 2.0 (Ousley and Jantz, 1996).

In the pooled population, I was able to estimate sex for 110 adults. Of these, 77 presented intact pelves with sex characteristics (70%), 83 presented sexually dimorphic cranial features (75%), and 109 (99%) presented bones that could be measured and used in a discriminant function analysis. In total, I estimated the sex of 91 out of 110 adults (83%) based on multiple methods. The pelvic methods were weighted more heavily than cranial morphology, which was more heavily weighted than long bone measurements, in estimating sex as indeterminate (I), female (F), probable female (PF), male (M), or probable male (PM). For the purposes of analysis of these small samples, the probable females/males are grouped with the other individuals of the same sex.

4.2.2

Age Estimation

Adult age was also arrived at using several different techniques.7 First, I visually examined the pubic symphysis using the criteria of both Suchey and Brooks (1990) and Todd (1921a,b). The auricular surface of the ilium was used for age estimation as per Lovejoy and colleagues

7_{In the field, Roman bioarchaeologists often rely on dental wear to estimate the age of skeletons based on the} charts reproduced from Lovejoy’s (1985) Libben study. Lovejoy (1985) and others note that dental (molar) wear can be a reliable indicator of age at death, but populations must first be seriated, and activity-related changes must be removed. This has not been done for Roman samples. In addition, age based on dental wear assumes a more or less standard diet within the population, an assumption that has not been tested for the Roman population. Based on comparisons between my age-at-death profile for Casal Bertone and the one presented by Nanni and Maffei (2004) for the same site, it is clear that using only dental wear significantly underages the population. Dental wear was, however, recorded for all individuals according to Standards, which uses Smith (1984) for anterior teeth and Scott (1979) for posterior teeth, in the hopes of using these data eventually to create a more reliable age-at-death rubric from Roman dental wear.

(1985). Finally, cranial suture closure was assessed based on the rubric designed by Meindl and Lovejoy (1985). As above, the pelvis was considered the most reliable indicator of age, with cranial suture closure used when a pelvis was not available. Of the 74 adults whose age was estimated, 66 (89%) presented a pubic symphysis, 56 (76%) an auricular surface, and 35 (47%) a cranium for age estimation. I estimated the age of 66 out of 74 adults (89%) based on multiple methods. Adults were placed into the following age categories: 16-20; 21-30; 31-40; 41-50; 51-60; 61-70.

Note that for the purpose of biological classification, the 16-20 age range is necessarily a 5-year one, as it represents the earliest post-pubescent category but includes individuals whose age can be narrowed down on the basis of dental development and epiphyseal closure. In Roman culture, however, individuals in this age range would generally have been considered adults. Roman males and females could marry at ages 14 and 12, respectively. Although marriage was the rite of passage for a female into adult life, that moment for males was at the end of the 17th year, when a male was able to enter public life and join the army (Wiedemann, 1989). As such, an individual in the 16-20 age category should be considered culturally an adult, even if the person was still developing biologically.8

Subadult age at death was estimated primarily using tooth formation and eruption and sec- ondarily by epiphyseal closure and long bone length of the postcranial skeleton. Tooth forma- tion for both deciduous and permanent teeth was recorded using the charts in Standards after Moorees and colleagues (1963a,b). The approximate age of the subadult was then estimated using White (2005, fig. 19.2) based on work by Gustafson and Koch (1974) and Anderson and colleagues (1976). Epiphyseal closure of all available elements of the postcranial skeleton was scored as unfused, partially fused, or fully fused and compared with tables by Baker and colleagues (2005, tables 10.1-5). In rare cases, subadult age was estimated using the length of long bones (Johnston, 1962). Subadults were placed into the following age categories: fetal;

8_{An argument can, of course, be made for females in the 11-15-year age range to be considered adults in the} Roman world. However, because puberty is the prime mover in creating sexual dimorphism in adults, it is very difficult to accurately estimate sex from remains of peri-pubescent individuals.

0-5; 6-10; 11-15. Whenever possible, more precise age estimates were made and noted per individual.

4.2.3

Stature Estimation

Adult stature was estimated as another metric with which to compare the present samples with other published Imperial skeletal populations. Long bone measurements were input into FORDISC 2.0, which uses formulae from Trotter and Gleser (1952) and Ousley (1995) to approximate living height from the postcranial skeleton. The formulae for whites from Trotter and Gleser (1952) were used in order to compare stature at the study sites with published data from other Imperial sites. For individuals with multiple long bones, leg bones were favored over arm bones where possible. The stature of every adult individual whose epiphyses were fully fused was estimated based on the available long bone whose regression formula presented the lowest standard error. The stature formulae can be thus ranked as follows, from those with the lowest to the highest error. For females: fibula, femur, radius, ulna, and humerus. For males: femur, fibula, humerus, radius, and ulna. Individuals over the age of 30 were subjected to a correction factor (Trotter, 1970) to account for age-related stature loss. The stature estimates presented below, therefore, approximate individuals’ standing height at death.