7

DETERMINATION OF AGE-AT-DEATH

Estimation of age-at-death is more dif ficult than the attribution of sex from skeletal material as there are only two options for sex, whilst ageing is a continuous process. This means that it is virtually impossible to age individuals, especially adults, with a great deal of precision. A further problem for the estimation of age-at-death is that an individual’s biological age may not reflect their chronological or actual age. This is because the relationship between the degree of skeletal development or degeneration and the actual age of an individual is not linear.¹

Juvenile skeletons generally produce the most reliable results. Criteria for age determination of immature individuals are relatively straightforward as they are based on growth and development. While there is some variation between individuals and populations in timing, these factors tend to be relatively consistent and predictable. Juvenile age-at-death is generally determined by extrapolation from standards that have been derived from data obtained from children of known age from modern populations. A number of variables may influence this, such as illness and nutrition. Ideally, it is preferable to avoid the use of ageing criteria that are likely to be affected by such variables. An example of this can be seen in the size of bones, which tends to be a good indicator of foetal age. Apparently, poor maternal nutrition is less likely to affect foetal bone length than malnutrition after birth. Bone length of a growing child is subject to too many external influences to be a really useful indicator of age. The incomplete nature of most archaeological remains, however, makes it impossible to discard any evidence, even if it is problematic.

Teeth develop from the crown to the roots, with root formation continuing to completion after the tooth has erupted. Dental development tends to be complete by the beginning of the third decade of life, though the last tooth to erupt, the third molar, or wisdom tooth, is the most variable. While there is some variation between individuals, teeth tend to be reliable indicators of the age of sub-adults, as their development appears to be less influenced by environmental factors. This would also suggest that the modern standards for tooth formation and eruption are applicable to ancient populations. This theory was tested, using 63 named and well-documented skeletons of children from the Spitalfields crypt. Though they only date back as far as the eighteenth and nineteenth centuries, it is notable that there was a high correlation between documented age and the results obtained from a number of standard dental ageing techniques. The ages obtained from the dental standards minimally, but consistently, under-aged the Spitalfields children. It has been suggested that this is a reflection of the effects of poor nutrition on dental development.²

After teeth, skeletal development provides the best indicator of juvenile age-at-death. Growing long bones are made up of three parts: the shaft or diaphysis and the ends, which articulate with other bones, which are known as the epiphyses. The epiphyses are separated from the shaft by growth cartilage, which is where growth occurs. When the growth period ends, the cartilage ossifies and the epiphyses are fused with the shaft. The majority of other bones also have epiphyses. Epiphyseal fusion occurs in an orderly fashion in the period between adolescence and early adulthood. The actual age at which epiphyseal fusion occurs for different bones can vary between individuals, sexes and populations. Epiphyses tend to fuse earlier in the bones of females, whose period of growth is generally shorter than that of males. The last epiphysis to fuse is that of the medial clavicle or collarbone. The age of fusion for this epiphysis can vary between 21 and 30 years of age, though generally all bones have fused by about 28 years of age in modern populations.

The determination of adult age-at-death is fraught with problems. After the completion of development, the only changes that occur are essentially degenerative and individuals do not degenerate at the same rate. This is readily apparent on living people. Some people’s hair, for example, goes grey when they are in their early twenties, whilst others can naturally retain their colour into old age. Differential degeneration is a biological fact that cannot be accounted for by any ageing technique. The sequence of changes after maturity is attained is variable and tends to be influenced by environmental factors; for example, the degree of tooth wear or attrition observed on an individual is determined by diet and lifestyle. Even with entire skeletons, it is difficult to establish the age-at-death of an adult. The addition of further complicating factors, such as a disarticulated unknown population, exacerbates the existing problems.³

Choice of age-at-death indicators for the Pompeian skeletal sample

As with the determination of sex, age-at-death can be more con fidently assessed with complete skeletons. The constraints of the Pompeian skeletal sample limited the use of certain ageing techniques; for example, it was virtually impossible to employ a standard multiple trait assessment based on the examination of the entire skeleton.⁴ Due to limitations of time and budget, emphasis was placed on the techniques that were deemed most useful at the time. The choice of the pelvis, skull and teeth as the indicators of age-at-death in the Pompeian sample was based on their well-documented potential to provide age information from birth to relative old age.⁵ Criteria that were used to give an indication of adult age included changes to the surface of the pubic symphysis, ectocranial suture closure and tooth wear. Assessment based on dental attrition was of limited value for this sample, as most of the skulls can no longer be articulated with mandibles due to the manner in which they were stored. This meant it was not always clear whether wear related to occlusal problems or dietary behaviour. Consideration was also given to a number of cranial features, which could be used to separate adults from juveniles, such as fusion of the basi-sphenoid and development of the frontal sinuses. Though less reliable, features like endocranial suture closure were also recorded, especially when only limited material representing an individual was available.

Age-related pathology, such as hyperostosis frontalis interna (Chapter 8 and see below), was employed to give an indication of the relative longevity of the Pompeians. The range of bony indicators generally associated with advancing years that could be used for this purpose was determined by the disarticulated nature of the sample. For example, it was not possible to do more than note most cases of osteophytic change, as age-related arthropathy cannot necessarily be distinguished from trauma-related changes when examination is based on a single bone. Some scholars have argued the possibility that certain disorders associated with old age in a modern Western population occurred at comparatively earlier ages in an ancient population. It was therefore necessary for their association with elderly people to be corroborated by other skeletal age indicators.

There was relatively little advantage to be gained from an examination of all the samples of specific bones in the disarticulated Pompeian collection to establish age-at-death. Since the times of epiphyseal closure vary between bones in an individual, a study of the degree of epiphyseal union in all these samples would do little more than separate adults from juveniles.⁶ For this reason only one post-cranial bone, the pelvis, was chosen to represent the entire sample. The degree of epiphyseal union was routinely recorded for juvenile bones that were included in non-metric trait scoring for long bones.

One of the constraints of this project was that it was not possible at the time of examination to obtain permission to perform destructive tests on Pompeian skeletal material. This precluded the use of various established methods, including bone cortex remodelling and root dentine transparency in teeth.⁷ It is notable that research by the Victorian Institute of Forensic Pathology at Monash University has produced results that question the reliability of the former method.⁸ Access to radiographic techniques, such as those suggested by Iscan and Loth,⁹ was also not possible due to financial constraints. This meant that the determination of age-at-death in this study was limited to macroscopic observations.

Assignment of specific ages

It is misleading to ascribe exact ages to archaeological skeletal material from an unknown population for two reasons. First, the actual ages for epiphyseal fusion, dental eruption and subsequent degenerative bony changes associated with ageing are variable. Variation for all these changes can occur within and between individuals and populations. In addition, it can be correlated with sex.¹⁰ As a result of these variations tolerances in age estimation can vary considerably. For example, those produced from the Suchey–Brooks technique of ageing from the pubic symphysis have tolerances (95 per cent) of between ± 5 years for phase 1 to well over ± 20 years for the later phases (see Table 7.1).¹¹

Second, age estimates that have been established for skeletal material have been determined from modern Western samples. There is a standard sevenpoint scoring scheme to estimate relative age.¹² Two additional scores were included in this work to deal with some of the vagaries encountered with the establishment of adult age-at-death (Table 7.2). This system roughly classifies age in ten-year increments, as the order of accuracy of most available techniques is very poor. It must be remembered that these age ranges are artificial as age-related changes are continuous.¹³ It is also important to understand that the classification of the last phase as relating to the sixth decade or older is purely a reflection of the upper limit of the techniques. It in no way is meant to indicate a shorter lifespan.

Age estimation based on the pelvis

Juvenile age
Juvenile and sub-adult pelves were scored in relation to the degree of fusion and the maximum width of the innominate bone.¹⁴

Table 7.1 Mean ages associated with the phases of the Suchey–Brooks ageing system from the pubic symphysis

Phase Female Female Female Male Male Male 95% mean standard 95% mean standard range deviation range deviation

I 19.4 2.6 15–24 18.5 2.1 15–23
II 25.0 4.9 19–40 23.4 3.6 19–34
III 30.7 8.1 21–53 28.7 6.5 21–46
IV 38.2 10.9 26–70 35.2 9.4 23–57
V 48.1 14.6 25–83 45.6 10.4 27–66
VI 60.0 12.4 42–87 61.2 12.2 34–86

Source: Adapted from Brooks and Suchey, 1990, 233. Table 7.2 Modified standard scoring scheme for the attribution of relative age-at-death Nine-point scoring scheme for the attribution of relative age-at-death

1. Foetal This term applies to any time prior to birth.
2. Infant The period from birth to three years of age. The choice of age 3 as a cut-off point was based on the tendency for the completion of eruption of the deciduous dentition by this age.
3. Juvenile Consistent in age with between about 3 and 12 years of age in a modern population.
4. Adolescent Consistent with ages between about 12 and 20 years of age in a modern western population.
5. Indeterminate Cases where not enough evidence remains to distinguish between sub-adult

6. Young adult 7. Adult

8. Mature adult

9. Older adult and adult.
Consistent with an age attribution in the third (20–35 years of age) decade in a modern western population.
Consistent with an age attribution in the fourth decade in a modern western population.
Consistent with an age attribution in the fifth decade in a modern western population.
Consistent with an age attribution in the sixth decade or older in a modern western population.

Source: Modified from standard scoring schemes, like that of Buikstra and Ubelaker, 1994, 9.

The juvenile innominate bone is composed of three separate bones, the ilium, the ischium and the pubis. In a modern Western population the rami of the pubis and the ischium generally fuse in about the seventh or eighth year of life, though fusion can occur any time between the ages of five and eight. In about the twelfth year of life the cartilaginous strip that has separated the three bones begins to ossify. It can take up until the eighteenth year for ossification to be completed at this point. Epiphyses appear at the iliac crest, the anterior inferior iliac spine, the pubis and the ischial tuberosity at about puberty and fusion is usually completed by the twenty-sixth year. It should be noted that fusion occurs at an earlier age in females. In the case of the ilium, ischium and pubis, fusion occurs between the ages of 11 to 15 in females and 14 to 17 in males in modern Western populations. Various factors can influence the time of fusion including health, diet and the state of the endocrine system. In addition, there may be differences between populations.¹⁵

Correlation has been observed between maximum iliac breadth and juvenile age. The data that have been collected from a North American Indian population have been tabulated as a series of means, standard deviations and ranges of deviations for different age ranges and presented for comparison with other populations. It is important to note that comparative studies of growth rates between different populations have demonstrated that there is interpopulation variation in the rate of bone growth, e.g. the rate of growth of Americans of European heritage has been found to be greater than that of American Indians, which in turn has been shown to be faster than that of Inuit. This suggests that though these data were most appropriately applied to other American Indian populations, they could be used with caution to obtain a general estimate of juvenile age-at-death for other populations. Even greater caution is required as some of these age-at-death estimates have been based on extremely small sample sizes, e.g. estimates for juveniles between 10.5 and 11.5 years of age were based on a sample of one bone.¹⁶

The use of North American Indian material for comparison with the Pompeian sample is far from ideal but no other data were available at the time of study. It did not enable exact ages to be assigned to individual bones but it did enable the juvenile pelvic remains to be seriated.

Out of 196 left innominate bones available for examination in the Pompeian collection, 6.1 per cent were completely unfused. Comparison of the maximum width of the eleven bones in this category that could be measured with the mean ages established for North American Indian populations from maximum iliac width, indicated that the younger immature individuals in the Pompeian sample fell within the juvenile category with ages consistent with a range in modern Western populations of between about three and twelve years. Bearing in mind the problems of extrapolation between populations, rough age estimates are presented here purely to give some indication of the range of development of the juvenile pelvic bones in the Pompeian sample. In summary, approximately 11 per cent of the sample presented as juvenile, reflecting an age range consistent with 2.5 to 15.5 years. A roughly even distribution of juveniles for each of the ages in this range was observed.

Since it was not possible to relate clavicles to pelves in the Pompeian collection, only epiphyseal fusion of the iliac crest was recorded. The value of these observations is that, in theory, they give some indication of the age of adolescents and individuals in the early years of adulthood. In the current study, fusion of the anterior iliac crest was only used to distinguish between adolescents and adults. In modern Western populations, the anterior and posterior epiphyses fuse to form a single cap for the crest and then commence fusion with the pelvis from 15–18 years in females and 17 to 20 years in males. Fusion tends to be complete by 23 years of age.¹⁷

There were 15 cases, or 7.65 per cent of the sample, which exhibited partial union of the epiphysis at the anterior iliac crest. It was possible to measure two of these cases. The widths of both bones were found to be significantly larger than the North American Indian comparative data for older adolescents.

Adult age

The pubic symphysis is the joint where the left and right pubic bones almost meet. They are separated by a fibro-cartilaginous disc. The underlying bone at this joint displays progressive degenerative changes during adulthood. In younger adults, the surface is distinguished by a series of ridges and furrows with no clear margins. Over time, the ridges become less defined and the surface is delimited by margins. The surface of the pubic symphysis is marked by increasing porosity and a pitted and uneven appearance in older age. Emphasis has been placed on the use of the pubic symphysis for the estimation of adult age from the pelvis since 1858, when age related changes were first observed at this site.¹⁸

In 1920, Todd introduced a set of ten developmental phases for the identification of age-at-death. Though he observed some differences, he did not really account for differences between populations or between the sexes as a result of pregnancy and parturition in females. His system was modified by Brooks in 1955 to correct its tendency to overage individuals.

McKern and Stewart attempted to deal with the problems of variability at the site of the pubic symphysis and introduced a method which involved the individual analysis of morphological components to estimate male age. Gilbert and McKern later attempted to develop a set of standards that could be applied to female pubic symphyses. Their method was criticized by Suchey for not adequately addressing the issues of interobserver error and changes to the region as a result of pregnancy and parturition. Meindl et al. conducted blind tests on all the available methods of age determination from the pubic symphysis. They reduced Todd’s ten phase system to five which they considered had the dual advantages of dealing with variability and of being simple to use.¹⁹

The innominate bones with complete pubic symphyses were separated by sex. The male and female groups were then seriated, based on comparison with a set of casts of the pubic region that were produced for the estimation of age-at-death using the method developed by Suchey and Brooks. Each bone was then assessed separately using the appropriate set of casts. The determination of age-at-death from the Suchey–Brooks method was made on a number of occasions without reference to earlier assessments and with a considerable period of time (up to two months) separating each examination. The degree of concordance between assessments was found to be close to 100 per cent.

It is clear from the table for age estimates based on the pelvic sample (Table 7.3), that the majority of the sample (81.1 per cent) is made up of bones that have been interpreted as adult. There were 160 cases, or 81.6 per cent of the sample, where pelvic fusion was complete.

Only 84 pubic symphyses were complete enough to enable age estimates to be made based on the Suchey–Brooks male and female sets of casts. The results of the raw Suchey–Brooks scores are presented as a histogram (Figure 7.1; also see Table 7.4). They reflect the pooled results for both sexes. The age estimate of 17 individuals, or 8.7 per cent of the sample, was consistent with the third decade of life in a modern Western population. Sixteen of

Table 7.3 Age distribution of the Pompeian sample, based on the pelvis Age-at-death Number of individuals Percentage

1. Foetal
2. Infant
3. Juvenile
4. Adolescent
5. Indeterminate (adult)
6. Young adult
7. Adult
8. Mature adult
9. Older adult
0 0 0 0 12 6.12 25 12.75 79 40.31 17 8.67 45 22.96 9 4.59 9 4.59

Table 7.4 Age distribution based on the Suchey–Brooks technique (note male and female scores have been pooled)

Age-at-death Number of individuals Percentage

Phase 1 4 4.76
Phase 2 8 9.54
Phase 3 13 15.48
Phase 4 41 48.81
Phase 5 9 10.71
Phase 6 9 10.71

these individuals were identi fied as male. Fortyfive pelves, or 23 per cent of the sample, displayed pubic symphyseal faces that were consistent with an age attribution in the fourth decade of life. Of these, 26 were identified as male and 19 as female. The pelves of nine cases, or 4.6 per cent of the sample, were interpreted as having belonged to individuals in the fifth decade of life. Four of these were identified as males and five as females. Nine cases (4.6 per cent of the sample) were assigned to the sixth decade or older at the time of death. Five of these individuals exhibited male characteristics and four female.

Note that if the adult sample (i.e. from phase 2–phase 6) is separated by sex, the distribution of male and female age groups is similar (see Figure 7.1).

Skull and teeth

The state of preservation of some skulls posed major dif ficulties for the determination of the age at death. The only criteria that could be employed to give an indication of age were ectocranial, or outer table, and endocranial, or inner table, suture closure, the development of the frontal sinuses, fusion of the basilar bone, the development of the Pacchionian depressions, tooth eruption and subsequent attrition. Of these, only ectocranial suture fusion and attrition could give an indication of the relative ages of the adults represented in this sample. Features like the development of the frontal

Figure 7.1 Sex separated Suchey–Brooks scores for the Pompeian adult sample

sinuses and the fusion of the basilar bone, could only be used to discriminate between adults and juveniles. Like the pelvic age indicators, the cranial ageing criteria suggested a much higher proportion of adults than juveniles in the Pompeian sample.

Cranial suture closure

The foetal skull is composed of a number of bones. The bones of the cranium are separate to enable some movement so that the skull can pass through the birth canal without damaging the brain. The cranial bones are separated by sutures. Growth of the skull occurs along these margins, which then fuse after growth has ceased. Age determination based on the order and degree of cranial suture closure was popular in the late nineteenth and early twentieth centuries, but fell from favour when studies revealed that these were unpredictably variable.²⁰ Since then endocranial, or inner table, suture closure has only been employed as a last resort in the absence of other skeletal remains.

Cranial suture closure was reassessed in the 1980s by Meindl and Lovejoy. Instead of using endocranial suture closure, which had previously been considered more reliable, they examined the ectocranial sutures, which are those that can be seen on the external surface of the cranium. They argued that these would be more useful for the calculation of age for older individuals as the ectocranial sutures close after the endocranial sutures. The authors stressed, however, that this technique should be used in conjunction with other ageing methods to produce an age based on a number of factors as there is no one reliable diagnostic feature for adult age-at-death.²¹

The endocranial sutures were open in only 15.4 per cent of the sample of 123 skulls. Thirty-six cases or 29.3 per cent of the sample exhibited partially fused endocranial sutures and the remaining 68 cases or 55.3 per cent of the skulls had endocranial sutures that had substantially fused.

The ectocranial suture scores give some indication of the actual age of the adult sample. From the histogram (Figure 7.2) of the ectocranial lateralanterior suture closure scores (EctsutA), it is apparent that the majority of the sample (69.4 per cent) was aged between the ‘adult’ and ‘older adult’ age range. These scores are consistent with ages in a modern Western population of between the fourth and sixth decade or older. This technique does not provide information about individuals that have not yet reached the fourth decade of life. There were 34 skulls or 30.6 per cent of the sample that exhibited no sign of ectocranial suture closure and which, according to this system, could only be classified as being of indeterminate age.

The ectocranial vault suture closure (EctsutB) scores yield slightly different results, which is a reflection of the difference between the two scoring systems. Generally, more observations could be made using this method as it

Figure 7.2 Estimated adult age-at-death based on ectocranial lateral-anterior suture closure scores (EctsutA)

involved parts of the skull that tended to have a higher survival rate. This can be seen in the larger number of cases that could be scored for EctsutB.²² Similarly, inspection of the histogram (Figure 7.3) reveals that it was possible to make more observations of sutures as evidenced by the smaller number of indeterminate cases (18.2 per cent). Nonetheless, this method is not so useful for the determination of older ages as EctsutA is demonstrated by the highest frequency of age scores being in the ‘adult’ category and the lowest (just under 5 per cent) in the ‘older adult’ range.

Development of the frontal sinuses

The frontal sinuses first appear as extensions from the nasal cavity and increase in size with age. Development of the frontal sinus commences in foetal life but they do not begin to expand until about the middle of the fourth year of life. Their extension some distance into the supraorbital region of an individual is a good indication of adulthood. The sinus can continue to increase in size well into the fourth decade. The walls of the frontal sinuses become thinner in elderly people, which gives the impression of an increase in size.²³

The vast majority of the skulls (77.4 per cent) that were complete enough to be inspected displayed considerable extension of the frontal sinus into the

Figure 7.3 Estimated adult age-at-death based on ectocranial vault suture closure scores (EctsutB)

supraorbital region. Only one skull, or 1.6 per cent of the sample, exhibited no degree of extension into the supraorbital region.
Fusion of the basilar bone

The fusion of the basilar portion of the occipital bone with the sphenoid is possibly the only example of cranial suture closure that occurs fairly consistently and can be used as a rough guide to separate adults from juveniles. In modern populations it is generally fused by 17 years of age in females and 19 years in males.²⁴

The scores for basilar fusion displayed a similar pattern to that observed with regard to the frontal sinus. Only two individuals (2.1 per cent of the sample) displayed a complete lack of fusion between the basilar portion of the occipital bone and the sphenoid, while there was partial fusion in three skulls (3.2 per cent) and complete fusion in ninety skulls (94.7 per cent).

Development of Pacchionian depressions

Pacchionian depressions can be observed on the inner table of the skull on the frontal bone and on the parietals, on either side of the sagittal suture. These depressions become more distinct, deeper and more frequent with age and when well developed, suggest an older adult. Similarly, the impressions of the middle meningeal artery become deeper with advancing years.²⁵

Observations of the incidence and degree of development of Pacchionian depressions produced results that differed significantly from those of all the other features in that they appeared to be almost normally distributed throughout the sample.

These skull data were further explored using principal components analysis. This analysis confirmed the trends shown in the simple statistical studies.²⁶
Maxillary teeth
Tooth eruption

Tooth eruption is a good age indicator for sub-adults. The development and closure of roots subsequent to eruption is also a useful age indicator. Unfortunately, this could only be observed in the case of loose teeth, as no x-ray facilities were available.²⁷

The results for tooth eruption con firmed the evidence for a majority of adults in the sample. Of the 71 cases that could be scored, only four contained dentition that was consistent with that of an adolescent and 53, or 74.6 per cent, exhibited complete eruption of all the maxillary dentition. It was not possible to determine whether the remaining 14 cases were adult or sub-adult, either because of poor preservation of the maxillary area or because there was no sign of the third molars.

Tooth attrition

Tooth wear or attrition is entirely the result of environmental factors. In ancient populations it is usually interpreted as the result of eating unprocessed food or food that has been processed with millstones that contribute a certain level of grit into the diet. The use of basalt millstones (Figure 7.6) has obviously been a contributing factor to tooth attrition in the Pompeian sample. The quantity of certain foods consumed by an individual would influence the degree of attrition. Tooth wear can also result from industrial use of the teeth, poor occlusion or even from habit as in the case of bruxism, or tooth grinding. The early stages of attrition only affect the enamel of the tooth but over time, continual wear can lead to exposure of the dentine and even the pulp cavity in very severe cases, though generally this is protected by the formation of secondary dentine. It is possible for the entire enamel crown to be worn away (Figures 7.4 and 7.5).²⁸

Several factors contributed to the diminution of the value of attrition as an age indicator for the Pompeian population. First, the disarticulation of the mandibles, which generally could not be reunited with the skulls, meant that the state of occlusion for each individual could not be assessed and this could not be factored into interpretation. The other major problem was the fact that few of the mandibles or maxillae contained a full complement of teeth. In cases of incomplete dentition, the most conservative score was used. Because of the minimal number of cases of total tooth retention, these techniques were more valuable as a means of seriating the data than as absolute age indicators.

Figure 7.4 Maxilla of an individual (NS 86: 1), who was excavated in 1986, displaying severe tooth attrition

Figure 7.5 Mandible of the same individual (TF NS 86: 1), showing severe tooth attrition

Figure 7.6 Basalt mill for grinding flour

The two methods of scoring for tooth attrition produced slightly different results. The Brothwell scheme (described in this work as Att A) placed 24 cases, or 82.8 per cent, of the sample in the ‘young adult’ range, with ages consistent with the third decade of life. Only two cases (6.9 per cent) were scored as ‘adult’ and 3 (10.3 per cent) as ‘mature adult’.

More cases were included in the data set based on the Lovejoy scheme (described in this work as Att B) as anterior teeth could be scored for a minimum age estimation. It is worth noting that, even though more individuals could be scored for Att B, the sample size was only 33. Because the age range covered by this technique is wider, it was possible to identify four individuals (12.1 per cent) in the adolescent range. This method produced fewer ‘young adult’ scores. Four cases that been included in this category in Att A were identified as one ‘adult’ and three ‘mature adults’ by the Lovejoy system. Mandibular teeth compared with maxillary teeth

The evidence provided by the teeth of all the available mandibles and maxillae from the Forum and Sarno Bath collections confirmed the results of the pelvic and skull samples. There was no significant difference between the mandibles and maxillae for tooth eruption (Table 7.5). It is notable that neither series had any cases that could be classified as foetal or infants. Just under 24 per cent of maxillae and 22.2 per cent of mandibles were classified as being of indeterminate age. This was because they were either too incomplete to assess or required x-rays to determine whether the third molars had yet to erupt or had been lost ante mortem.

The majority of cases for both types of attrition assessment were identi fied as consistent with individuals in their third decade of life. As expected, the frequency of cases scored in this category was higher for Att A, as Att B covers a wider age range which enables the older cases to be distinguished (see Tables 7.6 and 7.7). It is evident from the tables that there was greater in situ preservation of mandibular than maxillary teeth. It is possible that differences in the way that crania and mandibles were stored could account for the higher post mortem retention of mandibular teeth (also see Chapter 8). It is important to note that many of the mandibles and maxillae did not have sufficient ante and post mortem dental survival to enable assessment, which is why the

Table 7.5 Age determination based on eruption of dentition Foetal/ Juvenile Adolescent Adult Indeterminate infant

Maxillary teeth 0% 3% (n = 3) 6.2% (n = 5) 67.7% (n = 65) 24% (n = 23) Mandibular 0% 6.2% (n = 5) 5.2% (n = 5) 65.4% (n = 53) 22.2% (n = 18) teeth

Table 7.6 Age determination by AttA Adolescent Adult Mature adult Old Indeterminate

Maxillary 2.1% (n = 1) teeth
Mandibular 7.1% (n = 5) teeth

8.3% (n = 4) 12.5% (n = 6) 0% 8.4% (n = 4)

18.6% (n = 13) 8.6% (n = 6) 0% 8.5% (n = 6)

Table 7.7 Age determination by AttB Adolescent Adult Mature adult Old

Maxillary 10.7% (n = 6) 8.3% (n = 4) 19.6% (n = 11) 1.3% teeth (n = 1) Mandibular 6.7% (n = 5) 22.7% 18.7% (n = 14) 0% teeth (n = 17)

Indeterminate 10.5% (n = 5)

9.9% (n = 7)

percentages do not add up to 100. This should be borne in mind when assessing the results of attrition as an indicator of age for the Pompeian sample.

Age-related pathology

Pathological changes that are generally associated with advancing years have been observed on a number of the bones of the Pompeian sample (see Chapter 8). Because of the problems of diagnosis of disarticulated material, only two disorders were identified with certainty. These were diffuse idiopathic skeletal hyperostosis (DISH), which is more commonly found in older males and presents as fusion, especially of the right side of the thoracic vertebrae, and hyperostosis frontalis interna (HFI), a syndrome associated with an endocrine disorder that is highly correlated with post-menopausal women.

Contrary to expectations based on the assumption that the elderly were more likely to have become victims of Mt Vesuvius, the results of the age assessment of the Pompeian skeletal remains suggest that the proportion of older individuals in the sample was relatively low. In view of the acknowledged tendency for macroscopic ageing techniques based on morphological examination to underestimate adult age-at-death, especially in the older range, it was considered that the presence of age-related pathology might prove a more useful indicator.

As a result of serendipity, one age-related disorder was discovered with a frequency that enabled comments to be made about age and longevity in the Pompeian sample. This pathology presents unequivocally on the inner table of the frontal bone and is known as hyperostosis frontalis interna. It is a syndrome of unknown aetiology related to an endocrine disturbance and is reported to occur almost exclusively in older, usually post menopausal, females. It has an incidence of, at least, 11.1 per cent in the Pompeian sample, which is equivalent to the upper end of the range of the frequency for this disorder in a modern Western population. Only a limited number of cases, both temporally and geographically separated from Pompeii, have been reported in the archaeological literature (Chapter 8). It has been suggested that the reason this disorder has not been found frequently in archaeological contexts is because the average lifespan was much lower in antiquity.²⁹ The incidence of HFI in the Pompeian sample does not support this assertion. Since the frequency in this sample is comparable to the expected incidence in a modern population, it could be argued that the Pompeian skeletal collections reflect a normally distributed sample with a comparable lifespan to that of a modern community. If this were the case then further doubt could be cast on the presumption that the Pompeian sample was skewed towards the elderly as a result of their inability to escape the _AD 79 eruption. The frequency of HFI in the Pompeian sample also suggests that a number of women were surviving to older ages. This is at variance with the assumption that women in ancient Roman society tended to have shorter life expectancies as a result of death during labour. It is perhaps worth noting that ancient authors suggested that women who survived childbirth tended to outlive males.³⁰

The only other age-related disorder that could be diagnosed with relative certainty was diffuse idiopathic skeletal hyperostosis (DISH). This is an abnormality that is most apparent in the thoracic vertebrae and is specifically associated with older, usually male, individuals (Chapter 8). Only two cases were observed in the Pompeian collections, which does not allow for much comment beyond the statement that, at least a few individuals were living long enough to exhibit these bony changes. It is possible that this pathology is underrepresented in the Pompeian collections because vertebrae, which generally have not been preserved, are required for its diagnosis. A higher incidence of DISH has been observed in other ancient populations.

From a survey of a sample of 134 adult Nubian skeletons from Semma South in northern Sudan, 18 cases of DISH were identified, which meant that it occurred in the sample with a frequency of 13.4 per cent. This sample covered the period from 350 BC to AD 350. The frequency of this pathology has been determined to be about 10 per cent in people over the age of 70 in a modern Finnish population and 25 per cent in males over the age of 65 in the Todd skeletal collection. This led the researchers to conclude that the ancient Nubians did not have the short lifespans generally attributed to ancient populations.³¹

Similarly, Molleson reported a high proportion of cases of Paget ’s disease from the medieval Cathedral Green excavations at Winchester. This pathology is also highly correlated with old age and is recognizable by gross thickening of the long bones and skulls of affected individuals. Molleson documented these cases even though she generally subscribed to the view that ancient lifespans, especially those of the Romano-British populations she studied, were much shorter than those of modern populations. She used this example to promulgate the possibility that ancient skeletal material may have been underaged.³²

This is a moot point. The demonstrated tendency for a number of ageing techniques to underage skeletal material can be used to confirm preconceived ideas about ancient populations. For example, palaeodemographic studies based on Roman skeletal material of the later Empire have presented the average age-at-death at about 24 to 25 years, with few individuals surviving into the fourth decade. Various reasons have been invoked to explain this phenomenon, such as poor sanitation and inadequate medical attention for the urban poor.³³

So strong is the presumption of premature death in antiquity that the discovery of pathology associated with old age is often explained away by the suggestion that such diseases occurred at earlier ages in ancient populations. Anderson,³⁴ for example, considered that the occurrence of a case of HFI in an Anglo-Saxon skeleton was possibly a reflection of the earlier onset of menopause. Because of the vagaries of the available ageing techniques, especially in the identification of older individuals, it can be difficult to corroborate old age from skeletal remains. This makes it impossible to refute the argument of earlier onset of certain disorders, which means that interpretation can become a self-fulfilling prophecy.

There is no basis for the assumption that there have been signi ficant changes in mortality over time and space. Similarly there is no compelling evidence to suggest that various stages in the life histories of earlier populations, like the onset of menopause, occurred at earlier ages than in current populations.³⁵

Historical sources for old age in the Roman World

The literary sources do not resolve the issue of longevity in the Roman world, though they are instructive.

Parkin has made a detailed study of the concept of old age in the Roman world.³⁶ He noted that there is no consensus amongst ancient authors about the time of onset of old age. It was not uncommon for writers in the classical world to use blanket terms for different age classes that were not bound to numerical age, such as infant, adolescent and old. Some ancient writers do, however, provide information about when old age was said to commence. For example, in the different Hippocratic texts, the seventh stage of life or old age is said to commence at the ages of 42, 56 and 63 years of age. Censorinus, writing in the third century _AD, suggested five stages, each being 15 years long, based on Varro’s description of the first century _BC. Other examples provided for the onset of old age include Galen, who gave an age of 49 and Isodorus who indicated an age of 70.³⁷

Suetonius usually included the age of death of Roman Emperors in his lives of the Caesars. Augustus died just before his 76th birthday,³⁸ Claudius expired in his 64th year³⁹ and Vespasian died at the age of 69.⁴⁰ Tiberius was about 77 years of age when his life ended. It is notable that despite his venerable age, there was some suggestion that he did not die of natural causes but was perhaps slowly poisoned.⁴¹ This suggests that such an age was not considered so remarkable, at least amongst those of the higher social strata. Similarly, Galba was murdered at the age of 73.⁴² One could perhaps argue that the upper classes would be expected to have longer lifespans as a result of better diet, standard of hygiene and access to medical assistance.

Information about the rest of the population is incomplete. Almost none of the census information for Roman Italy has survived and ancient medical sources are filled with generalizations. Evidence from tombstones has been employed to determine longevity, though it has been argued that the interpretation of such evidence has often been simplistic.⁴³

It would appear reasonable to assume that the census data that survive could provide information about the age range of ancient Romans. To highlight the problems associated with census information, Parkin cites the case of the _AD 73–74 census instigated by Vespasian and Titus. Both Pliny the Elder, who wrote his account a few years after this event and Phlegon, who wrote in the time of Hadrian, record results of this census in relation to the number of centenarians in the region between the Po and the Apennines. It is clear that Phlegon is not merely repeating Pliny’s report as his is much more detailed. Pliny reported ninety individuals who were one hundred years of age or older while Phlegon only mentions seventy. Some individuals are recorded as being up to 150 years of age. Apart from the inconsistency in the numbers presented by the two writers and the unlikely number of centenarians in that region, some of the ages that are reported just do not seem biologically possible. Parkin argues that rather than being an accurate reflection of actual age, these figures represent the high status associated with achieving a phenomenal age and that this example should serve as a warning against blind acceptance of other Roman census figures.⁴⁴

The evidence from tombstones also shows a tendency to exaggerate lifespan. There are vast numbers of tomb inscriptions, which include large numbers of adolescents and elderly individuals with relatively few cases of people in the fourth and fifth decades. Dyson⁴⁵presented an argument to explain this phenomenon. It is based on the likelihood that tombstone inscriptions explain more about Roman attitudes to premature death and what was considered a proper lifespan than the actual age composition of Roman communities. The inclusion of information about the length of life of individuals who survived into old age has been presented as a reflection of a fascination with the defiance of mortality in a society where extreme old age was perhaps desired but not common. This reasoning has been supported by the apparent interest of ancient writers, such as Pliny the Elder and Pliny the Younger, in examples of longevity in individuals or communities.⁴⁶ Further, it appears that some of these tombstone inscriptions record an unlikely frequency of extremely elderly individuals, such as the large number of centenarians on Roman African tombstones. This is probably more related to geographical variation in the method of commemoration of the dead than differential longevity in various regions of the ancient world.⁴⁷

These cases suggest that, at least on occasion, ages were exaggerated. Parkin⁴⁸ noted that there were no bureaucratic implications associated with the degree of accuracy of ages ascribed to individuals on tombstones. As a result there was no reason to prevent the use of guesstimates when age was not known or for desired, rather than real age, to be recorded.

It has also been noted that the figures on tombstones have been rounded up or down to make them multiples of five. One possible explanation for this is that multiples of five are shorter in Roman numerals and therefore more economical to use if one has to pay for an inscription to be engraved. Other possible explanations for this phenomenon are illiteracy and ignorance of the exact age of the deceased. Parkin suggested that knowledge of the precise age of an individual was not important to the daily life of most ancient Romans.⁴⁹ Examination of the various forms of evidence available led Parkin to conclude that there were no methods to reliably calculate or verify age in Roman Italy. Even though there were well-defined rules of age, for example, males came of legal age at the age of 25, these apparently relied on the statement of age provided by an individual rather than objective evidence.⁵⁰

No compensation can be made for the bias generated by creative documentation and attempts to determine average lifespan from this class of evidence are of questionable value. Other forms of literary evidence, such as mummy labels, legal texts and tax receipts, have similar shortcomings and it has been suggested that, like the skeletal evidence, ancient written sources alone are not very reliable for the reconstruction of age-at-death information for the ancient Roman world.⁵¹ Nonetheless, the evidence that does exist does not indicate that the potential lifespan of the ancient Romans was significantly different from that of a modern population.

Life tables and palaeodemography

It is appropriate to consider palaeodemography and the use of life tables at this point as they have been employed by scholars like Henneberg and Henneberg and Capasso,⁵² but eschewed by Bisel⁵³ and myself, to describe the ancient populations of Pompeii and Herculaneum.

Palaeodemography is concerned with the reconstruction of ancient populations from archaeological skeletal material. The value of palaeodemographic studies has been debated for some decades.⁵⁴ The key problem that had to be addressed was the description of populations using sexing and, more particularly, ageing techniques for skeletal identification, with their attendant problems. Various methods have been developed in an attempt to address the issues associated with demographic reconstructions.⁵⁵ A series of assumptions were developed and justified to enable palaeodemographic studies to be considered useful. One of the most important, which allowed the construction of life tables from records of mortality, was population stability. A stable population is not affected by immigration or emigration, meaning that they balance each other out or that the population is closed to migration. The sex and age distribution will therefore be a function of the population’s actual fertility and mortality. A stationary population is a stable population where birth and death rates are equal.⁵⁶

Life tables are used to demonstrate mortality and survivorship. They were initially devised in the seventeenth century by Edmund Halley for the purpose of computing annuities for life insurance. The original models were derived from known populations, which meant that the data were relatively accurate.⁵⁷ Life tables have been applied to archaeological material to provide overall population profiles. Modern models are used but the level of accuracy of such models decreases when extrapolated onto ancient populations due to the number of assumptions that must be drawn.

Bisel argued that the victims of the _AD 79 eruption represent a cross-section of a living population and that it would be meaningless to undertake a mortality study at Herculaneum as such studies are only valid for cemetery populations.⁵⁸ There certainly is a real difference between the make-up of a cemetery population and a sample of victims from a disaster. Survival of such an event would appear to be random and it is difficult to be certain that the victims reflect the actual _AD 79 population.

Possibly even more problematic is the likelihood that the Pompeian and Herculaneum populations were not stable in the last 17 years of occupation as the available evidence suggests that they were in a state of flux as a result of the AD 62 earthquake and subsequent seismic activity. It is also possible that the population was seasonal and the make-up of the population of victims would have been determined by the season in which the disaster occurred. Further, Pompeii as a port town might be expected to have a variable population (see Chapter 4 and below).

These factors indicate that the use of life tables is inappropriate for the Pompeian and Herculaneum material. The models that are used rely on certain assumptions to deal with missing data. Their application in this case is likely to produce highly speculative results. It is important to note that this does not mean that the data should not be explored to describe the victims of the eruption. I would argue that the use of this technique may result in misleading information with this particular material and that it is better studied without the use of demographic modelling.

Estimation of age-at-death in Pompeii and Herculaneum

The most obvious results of the determination of age-at-death for the Pompeian sample are the high proportion of adults to children and the lack of neonatal and infant bones. The low recovery rate of the bones of neonates and very young juveniles from archaeological sites in general has been documented (Chapter 5). These bones do not tend to survive as well as the more robust bones of adults for various reasons. The Pompeian skeletal remains that were available for this study were not excavated by people with anatomical knowledge. It is quite possible that workers on the site have not recognized the bones of neonatals and infants as those of humans. In addition, the recovery of human remains, with the exception of the casts, was not a high priority in Pompeii until the latter part of the twentieth century. Another major factor that could account for the bias towards the survival of adult rather than juvenile bones in the sample is the method of storage.

The suggestion that this problem is related to recovery and storage rather than a real absence of young juveniles amongst the Pompeian victims is supported by the comparative frequency of juvenile bones in collections of skeletons that have been left in situ for display purposes, such as those observed in the Casa del Menandro (I, x, 4) and the number of children represented in the cast collection. For example, in 1960–61 the forms of thirteen individuals were cast in what is now known as the Garden of the Fugitives (I, xxi, 2). Six of these were clearly children, the youngest of which have been very roughly aged at about four of five years on the basis of visual inspection. While age estimates based solely on visual inspection can hardly be considered reliable, it is clear that these individuals were very young. Another example is a cast of young child that was found in the Casa del Bracciale d’Oro in the Insula Occidentalis (VI, xvii, 42), Figure 10.1.⁵⁹

Other studies of age-at-death from the Pompeian skeletal sample

Nicolucci chose to study a sample of 100 skulls for his 1882 work. He considered that the entire range of ages was represented in his sample, with the majority being between the ages of 60 and 90. His age determinations were based on suture closure and examination of the teeth.⁶⁰ It should be noted that while the criteria he used to establish age-at-death were absolutely reasonable for nineteenth-century scholarship, they would no longer be considered reliable. Nicolucci unfortunately neglected to mention the criteria he used for his dental examination so it is difficult to assess the ages he established. It is probably reasonable to assume that he based his age determinations on attrition and tooth evulsion. It is quite possible that Nicolucci obtained such high ages for his sample as a result of extrapolation from his experience of contemporary Italian tooth wear and loss. Since his main interest was in the determination of ‘racial’ typology, he was not particularly concerned with establishing the actual proportions of age groups in the sample of victims.

The research done by D ’Amore et al. in the latter part of the twentieth century had similar aims to that of Nicolucci, which meant that they also did not attempt to use their study to understand age groupings. Instead, they concentrated on a sample of 123 skulls, which were mostly adult, though they did include a few older juveniles to determine their ‘racial’ affiliations. Their determination of age-at-death involved a four-part classifi- cation system based on the work of Vallois: juvenile, covering the ages from about 12 or 13 to 21 years of age; adult, which incorporated individuals ranging in age from 21 to 40 years; mature, covering people from 40 to 59; and senile, which included those of 60 or more years.⁶¹ They also applied this system to Nicolucci’s series for comparative purposes and tabulated the results.⁶² In their sample, 56 per cent of males were classified as mature and 58 per cent of the females as adult as compared to Nicolucci’s classification of the majority of both genders as senile, viz. 71 per cent of males and 41 per cent of the females.⁶³

D ’Amore et al. did not supply details of the actual criteria they used to place each individual into this classification system and one can only conclude that those of Vallois were employed. Juvenile age-at-death was determined by tooth eruption and adult ages were based on cranial suture closure. Vallois considered tooth attrition to be too dependent on general health and diet to be of value for the study of ancient, unknown populations.⁶⁴

Vallois ’ criteria for ‘juvenile’ age determination, namely from the end of the eruption of the second molar to the almost complete closure of the spheno-occipitalis synchondrosis and the first appearance of vault closures,⁶⁵ are quite reasonable, with the possible exception of the final criterion. Suture closure of the cranial vault, which was also the single component for Vallois’ ‘adult’, ‘mature’ and ‘senile’ categories, using the skull,⁶⁶ as mentioned above, has been found to be of questionable value in the determination of age-at-death because of its high variability.

It is remarkable that Vallois ’ method for adult age classification was apparently employed so uncritically, despite the fact that the validity of suture closure as an age indicator had been challenged both by authors like Krogman, whose work D’Amore et al. referred to for sex attribution and by the other participants in the conference at which Vallois contributed this paper.⁶⁷ Indeed, it is notable that after having devoted so much space to a discussion of the problems associated with sex identification, D’Amore et al. gave minimal consideration to issues related to the determination of age-atdeath. Their age estimates can be seen in Table 7.8. These results were compared with those obtained by Nicolucci (Table 7.9).

It is notable that D ’Amore et al. did not consider the possibility that the consistently older ages identified by Nicolucci may have resulted from a lack of experience with teeth from ancient populations, which were more likely to demonstrate a greater degree of attrition as a result of consuming stone

Table 7.8 Age-at-death determination from skulls examined by D’Amore et al. Age-at-death Number of individuals Percentage

Juvenile 2 1.62 Adult 49 39.84 Mature adult 62 50.4 Senile 10 8.3

Source: Adapted from D’Amore et al., 1979, 306.

Table 7.9 Age-at-death determination from the skull sample studied by Nicolucci Age-at-death Number of individuals Percentage

Juvenile 7 7.07
Adult 12 12.12
Mature adult 23 23.23
Senile 57 57.58

Source: Nicolucci, 1882, 10.

ground grain. D ’Amore et al. concluded that they had made an empirical comparison between the breakdown of age-at-death in their sample and that of Nicolucci. They suggested that the difference between their results and those of Nicolucci could be explained by different samples or perhaps by a different system of classification.⁶⁸

More recently, the Pompeian skeletal material has been re-examined by Henneberg and Henneberg. They based their estimates of age-at-death on essentially the same sample of material that was used for this publication. They used 364 skulls and 186 right-hip bones. The ageing criteria that were employed were the obliteration of the cranial sutures, the state of dentition and changes to the pubic symphysis and auricular surfaces (Table 7.10).⁶⁹ The results they obtained were not dissimilar from those obtained by the author.

The key difference between the two works can be seen in the interpretation of the available data, with possible variation resulting from differing expectations. According to the assessment by Henneberg and Henneberg, the majority of the sample comprised young adults, with an estimated age at death of between 20 and 40 years. They observed very few children and relatively few people in the very old age bracket. They explain the lack of young juveniles in the sample, with the same arguments mentioned above about survival and recognition of the bones of young individuals in the archaeological record. They argue that the age distribution of adults in the Pompeian sample does not vary significantly from that observed at ancient South Italian burial grounds, most notably those of Paestum, dating from the sixth to the fourth centuries _BC and Patanello, which dates from the sixth to the third centuries _BC. They also point out that the age distribution of the Pompeian adults is quite comparable with data from death records of preindustrial Central European populations. This led them to conclude that the age structure of the Pompeian skeletal sample could be considered normal for a living ancient population. They use the lack of variation from ancient

Table 7.10 Age at death estimates by Henneberg and Henneberg
Age-at-death Percentage

0 –5 2.9
5–10 1.4
10–15 2.8
15–20 5.4
20–30 27.1
30–40 25.6
40–50 12.8
50–60 8.7
60+ 13.7

Source: After Henneberg and Henneberg, 2002, 173. (Henneberg and Henneberg do not provide the sample size for this data.)

cemetery samples to argue that the Pompeian population was stationary and therefore appropriate to subject to demographic techniques. This can be questioned, as there is evidence to suggest that the population at Pompeii was probably not stationary during the last 17 years of occupation (Chapters 4 and 9).⁷⁰

The age pro files that have been produced from these two works also need some consideration. The incomplete nature of some of the sample meant that there were a number of adult individuals whose age could not be identified with certainty. These cases were assigned an indeterminate score. It is diffi- cult to assess this difference in result without access to all their raw data but it is possible that the use of the auricular surface by Henneberg and Henneberg provided them with fewer equivocal cases, though it is unlikely that the use of this technique would have completely resolved the problem. The net result is a greater sense of certainty in the results presented by Henneberg and Henneberg.⁷¹

While the overall results for age-at-death distribution are quite comparable, it is important to reiterate that the limitations of the available ageing techniques that were used for both studies means that it is likely that a number of the adults in the Pompeian sample have been underaged. The interpretation of age-at-death from the Pompeian skeletal sample should be tempered by other evidence, such as age-related pathology. The frequency with which HFI appears, indicates that there probably were more older people in the sample than the available skeletal ageing techniques could reasonably establish. This is at variance with the suggestion by Henneberg and Henneberg that the Pompeians had relatively short lives.⁷²

Estimations of age-at-death from the Herculaneum skeletal sample

It is notable that more techniques could be used for the establishment of age at death from the Herculaneum sample as the skeletons were articulated and, in general, better preserved than the Pompeian skeletal sample. Also, combinations of techniques could be used for individual skeletons, which means that the ages obtained for the Herculaneum sample are potentially more accurate than those obtained from the samples of individual bones in the Pompeian sample.

Bisel determined age-at-death of the Herculaneum skeletal sample from an examination of epiphyseal fusion, tooth eruption, changes in the faces of the pubic symphysis, skull suture closure and the general appearance of the bone, including age-related pathological change.⁷³ Like Henneberg and Henneberg, she produced an age distribution, with a five-year range for each group, which could only be described as optimistic as the margin of error for adult age-at-death based on macroscopic examination substantially exceeds that figure (Table 7.11).

Table 7.11 Age distribution of the Herculaneum skeletal sample studied by Bisel Age-at-death Number of individuals Per cent estimate in years

< 1 5 3.65
1–5 13 9.49
6–10 10 7.30
11–15 12 8.76
16–20 6 4.38
21–25 6 4.38
26–30 20 14.60
31–35 11 8.03
36–40 11 8.03
41–45 19 13.87
46–50 19 13.87
51–55 4 2.92
55+ 1 0.73

Source: Adapted from Bisel and Bisel, 2002, 474.

It is signi ficant that Bisel⁷⁴ reported a lower than expected incidence of juveniles in the Herculaneum sample she examined. This sample was excavated in the 1980s under the guidance of Bisel and it is highly unlikely that any remains were missed. Out of 139 skeletons, five (3.6 per cent) were aged at less than one year, 23 (16.5 per cent) covered the span of one to ten years and 12 (8.6 per cent) were interpreted as between ten and 16. Bisel considered that the proportion of sub-adult bones should have been much higher to ensure that the population could be sustained. Initially, she dismissed the possibility that a disproportionate number of juveniles were able to escape or that they sought shelter in a chamber that has not yet been excavated, though in more recent work by Bisel and Bisel, consideration was given to sample bias.⁷⁵ One argument presented by Bisel for the comparatively small number of children she observed was that it was a reflection of decreased parity amongst the Herculanean women, as a result of the ingestion of lead or other causes. The issues associated with this suggestion are discussed in Chapter 8.

Capasso used a raft of macroscopic and microscopic methods to establish the age-at-death of the Herculaneum sample, including: tooth eruption and attrition, epiphyseal fusion, ectocranial and endocranial suture closure, changes to the surface of the pubic symphysis, the auricular surface of the ilium and the sternal extremity of the ribs, accumulation of osteons in cortical bone and radiological examination of bone to establish degree of thinning of bone cortex.⁷⁶ Only 143 of the 163 skeletons that were examined by Capasso were sufficiently preserved to enable age at death to be determined. He excluded two foetal skeletons from his palaedemographic study.⁷⁷ The ages that Capasso obtained can be viewed in Table 7.12.

Table 7.12 Age distribution of the Herculaneum skeletal sample studied by Capasso Age-at-death Number of individuals Percentage of sample estimate (years) (n=143)

0 –4.9 17 11.9
5.0–9.9 12 8.4
10.0–14.9 14 9.8
15.0–19.9 7 4.9
20.0–24.9 17 11.9
25.0–29.9 17 11.9
30.0–34.9 14 9.8
35.0–39.9 12 8.4
40.0–44.9 11 7.7
45.0–49.9 10 7.0
50.0–54.9 9 6.3
55.0–59.9 3 2.1
60.0+ 0 0

Sources: After Capasso and Capasso, 1999, 1826; Capasso, 2001, 959.

The 17 other skeletons were aged in groups based on a general age classification system suggested by Vallois. Capasso found that there were five that could be classified as infant or juvenile and 12 as adult. No adolescents or older adults were identified in this group.⁷⁸

Capasso argued that the sample was not biased and was probably a good reflection of the Herculaneum population in AD 79. Based on an assumed population of 5000, Capasso calculated the numbers of different age groups in Herculaneum at the time of the eruption. Capasso suggested that the comparative lack of sub-adults in their mid to late teenage years was a reflection of a lowered birth rate as a result of a birth rate crisis between AD 59 and AD 64. He argued that the main reason for the demographic anomaly he observed was the major earthquake in AD 62.⁷⁹

Petrone et al. published a preliminary study of 215 Herculaneum skeletons, including those studied by Bisel and Capasso, in 2002. The criteria they used to estimate age at death were: tooth eruption and attrition, epiphyseal fusion, ectocranial and endocranial suture closure, the degree of resorption of spongy bone in the proximal epiphyses of the humerus and femur, the changes to the surface of the pubic symphysis, the auricular surface of the ilium and the sternal extremity of the ribs.⁸⁰

Like Bisel and Capasso before them, they separated their age estimates into fi
ve-year intervals, with the percentage breakdown shown in Table 7.13.

Petrone et al. argue that the sample is representative of the population, which they estimate at 4000. They considered that the proportion of subadults to adults was insufficient for a stable population and like Capasso, suggest that the reason for the population imbalance was the impact of the _AD62 earthquake. They presented various scenarios that might account for

Table 7.13 Age distribution as calculated by Petrone et al. Age intervals Percentage

0 –5 11.1
6–10 8.7
11–15 11.1
16–20 5.8
20–25 6.3
26–30 14.9
31–35 11.5
36–40 9.1
41–45 9.6
45–50 9.1
>50 2.9

Source: Adapted from Petrone et al., 2002, 71.

the proportions of different age groups in the Herculaneum sample as a result of this disaster.⁸¹ The discrepancy between the population estimates for Herculaneum for Capasso and Petrone et al.’s work is a reflection of the lack of evidence for the number of inhabitants.

Conclusion

The skeletal evidence does not support the popular notion that the very young and the elderly inhabitants of Pompeii were more likely to have become victims of the eruption of Mt Vesuvius. The frequency of HFI suggests that the adult sample was not skewed, though the overall sample is obviously biased towards adult material. Further, the evidence of age-related pathology indicates that the Pompeians did not necessarily have a shorter lifespan than modern populations. Survival, therefore, was probably more related to personal decisions about whether and when to escape than on issues related to age or sex.

In the majority of cases it was possible to separate adults from juveniles and subadults. The factor which determined whether this was possible was the degree of completeness of the specimen. The available techniques that were used to establish adult age-at-death in this study were not very reliable. This is only partly related to the constraints of the Pompeian sample. Techniques, like tooth attrition, which are generally considered good for age estimation could only be used with limited confidence for the Pompeian sample because of the lack of complete dentition in virtually all of the jaws and the disarticulation of mandibles and maxillae in nearly all cases. Other techniques, like changes to the face of the pubic symphysis, which have been established on the basis of extensive research, have major limitations because biological and chronological age do not necessarily correspond. The margin of error associated with the Suchey–Brooks technique can be as high as 14.6 years for females and 12.2 years for males. Further, it is difficult to modify these techniques to account for the acknowledged interpopulation differences for pubic symphyseal age changes. This is a particular problem for archaeological samples. In addition, it has been demonstrated that there is a tendency to underage older individuals with this and other ageing methods.⁸²

Because of human variability for age changes it is unlikely that an accurate, objective test can be developed. The use of histological techniques, such as dentine root transparency and cemental annulations, appear to be the most promising for the future, though they are destructive and costly to perform. Jackes, in her review of current methods of age determination from skeletal remains, suggested that a complex system based on a number of techniques involving dentition may ultimately be able to produce results with a relatively high correlation to ‘real age’. She did, however, concede that it is impossible to extrapolate ages with any degree of certainty onto ancient unknown samples as it would be impossible to account for environmental variables.⁸³

Despite the increased potential for more reliable age estimates from the Herculaneum sample it should be noted that the tendency to apply ages within five-year intervals suggests greater accuracy than the methods that were employed in these studies can provide.

The disparity between the conclusions of this Pompeian study and that of Henneberg and Henneberg about the presence of a significant number of older individuals in the sample can be seen as a reflection of the inability of the available techniques for the determination of age-at-death to discriminate between adult ages from the skeletal record. In contrast with the conclusions of Henneberg and Henneberg about the demographic makeup of the Pompeian sample, the Herculaneum studies all suggest that, at least at Herculaneum, there was not a stable population. It could be argued that had the earthquake of_AD 62 had such a devastating impact on the population of Herculaneum, it would be likely to have had a similar effect on the Pompeian population. It is therefore possible that the sample bias against very young individuals in the Pompeian skeletal collection may not entirely be due to failure to recognize infant bones in excavation and poor storage.