Most of the Pompeian human skeletal remains have been stored in two buildings, which date to the original occupation of the site. The tradition of removing skeletons after excavation and storing the co-mingled remains in a building on site, designated as a ‘bone house’, can be traced back, at least, to the second half of the nineteenth century.1 Nonetheless, some skeletons and a number of casts are still in situ, especially in houses excavated from the time of the directorship of Maiuri and beyond.
The environment in which the bones are stored is as romantic as the novels that have served to popularize the site. The majority of the human skeletons have been stored in an ancient bath building, the Terme del Sarno (VII, ii, 17). This structure is situated to the south of the Forum. The Sarno Bath complex was first used as a repository for ancient bones and casts in the early 1930s when modern stone walls were incorporated into the structure and iron bars were inserted to deny access through doors and windows. It was also used to house fragments of marble statues, terracottas, carbonized rope, baskets and fishing nets as well as ferrous implements. The objects deposited in this building are largely unprovenanced but were apparently brought to the baths from all over Pompeii.2
When I commenced work on this project, the road to the Sarno Baths was not accessible to tourists and was overgrown with wild fennel. Entering the Sarno Baths was like being immersed in a classic B grade movie. The modern iron gates at the entrance to the baths had rusted shut and had to be forced open by guards. The entrance was completely obscured by brambles that had to be hacked away with a machete. Access was obtained via a dimly lit barrel-vaulted sloping passageway.
The first level down contained a side passage with two service rooms. One of these had been reused as a store for unprovenanced fragments of marble sculptures, table tops and broken sundials. The other room housed horse skeletons that had been rearticulated for long-since dismantled displays. The first room was so dark that one’s eyes could not adapt and it could only be viewed by torchlight. Shining a torch into the second room made the shadows of the horse skeletons leap up the wall. When the rooms were illuminated, bats would fly out into the faces of unsuspecting visitors who stood too near the doorway.
The next level down was reached by a crumbling set of stairs. This level contained what has been interpreted as the women’s baths.3 On the left was the pool for the frigidarium, which was adorned with paintings of pygmies in Nilotic scenes. Above this was a lunette with a painted stucco of the personification of the River Sarno. To the right were the tepidarium, caldarium and praefurnium. The ceilings are covered with stucco and vibrant paintings. Mounds of bones and the remains of casts that could no longer be displayed due to dismemberment of limbs littered the floors of these rooms, along with the remains of portions of marble statues, such as the odd disembodied foot. The photograph of the bones in the Sarno Baths (Figure 5.1) was taken some months after cleaning commenced, when the bone piles had begun to diminish in size.
The human bones stored in this building had been indiscriminately piled along with the bones of other animals, like horses, sheep, goats and dogs. Over time, the skeletons have become disarticulated and it is almost impossible to reassign bones to specific individuals, except in the case of certain pathological changes (Chapter 8).
In the late 1970s, a small collection of human skeletal material was removed from the Sarno Baths for study.4 This collection is stored by bone type, including femur, tibia, humerus and skull. It is housed in another ancient bath house, the Terme Femminile del Foro(VII, v, 2/8/24) This building also contains boxes of mixed human and non-human bones that were reburied after excavation in the nineteenth century and subsequently reexcavated in the second half of the twentieth century.5 Most of this material is unlabelled.
The Sarno and Forum Bath buildings house their own ecosystems, often directly associated with the presence of skeletal remains. The main animal life in the Terme Femminile del Foro consisted of rats and mice. Due to the comparative isolation of the Terme del Sarno and its increased exposure to the elements, the range of observed fauna was far more extensive. There were various rodents, cats, bats, snakes, spiders and various insects, such as carpenter bees and beetles. Birds had been nesting in the bones for many years; the inside of a cranium apparently formed an excellent basis for a nest. In a number of cases it was necessary to remove over a centimetre of bird lime from bones before they could be examined. Occasionally, a large green lizard would unexpectedly drop off the ceiling onto my workbooks.
The seduction of scholarship by popular culture was one of the key factors that contributed to the neglect of Pompeian skeletal material as an archaeological resource. As already noted, Pompeian skeletal remains were not initially seen to have any research potential. They served merely as props for the creation of visual or verbal vignettes for visiting dignitaries or literary
Figure 5.1 Human skeletal remains stored in the Terme del Sarno (VII, ii, 17) as they appeared in 1987
works, like those of Gautier and Bulwer-Lytton. The early period of excavation was marked by the poor storage and subsequent disarticulation of a large proportion of the skeletons. Ironically, these skeletons now present the ultimate vignette of a romantic archaeological site. Contemporary popular culture, as exemplified by films like Indiana Jones, would be hard pressed to better the image of the mounds of bones stored in the Sarno Baths.
The lack of documentation of the skeletal sample meant that knowledge of the provenance of skeletons was lost. In addition, the disarticulation of individual skeletons over time resulted in a significant reduction of information. This, in turn, discouraged scholars from working on the bones, even when the archaeological value of the Pompeian skeletons was finally recognized.6 It is worth noting that disarticulation did not pose a major problem for nineteenth century physical anthropologists as they tended to concentrate their efforts on the skull. Even the more recent work of D’Amore et al.7 was essentially craniocentric.
Of course, no archaeological data set is complete. They are always constrained by the accident of what survives over time and the fact that the archaeological record is usually biased towards more durable examples of material culture.8 The human skeletal remains from Pompeii differ from material from other sites as most of the loss is the result of post-excavation activities. This has apparently been a greater deterrent to scholars than if the loss had occurred for other reasons. Some scholars have further argued that the value of the Pompeian sample has been superseded by the discovery of so many well-preserved bodies from Herculaneum.9 It is important to emphasize that even though Pompeii and Herculaneum were destroyed by the same event, the two sites are different and the data sets are not interchangeable.
Despite these problems, there is no doubt that the available skeletal material can yield valuable information about the inhabitants of Pompeii who became victims of the eruption, but it requires a pragmatic approach to deal with the constraints of the sample. The parameters for such a study are imposed by the nature and documentation of the excavations over time and the subsequent storage of the skeletal material. In addition, on-site conditions constrain the research design.
As a result, the main problem in developing a research design for this project was how to tease information from the difficult data set. The Pompeian victims provide a large sample of individuals who all died of a known cause within a short period of time. Further, unlike most other sites from antiquity, these bones provide valuable information about a Roman population that is not based on a cemetery.
Storage and access
As mentioned above, the main repository of human skeletons excavated in Pompeii is the Terme del Sarno (VII, ii, 17) with a smaller collection housed in the Terme Femminile del Foro (VII, v, 2/8/24). While the Sarno Bath collection was completely unsorted, the larger part of the Forum Bath collection had been specifically gathered for examination.10 The Forum Bath material was stored by bone type, including femur, tibia, humerus and skull. The sample sizes for each type and side of long bone ranged from 100 to 160. There are 125 crania and about 20 loose mandibles.
Light levels in both these buildings are low. This suggests the truth of Seneca’s statement that his ancestors kept their bath buildings gloomy as they only felt warm in the dark.11 The Forum Baths received so little natural light that it was necessary to wire the building and install a 100 watt light globe. This was found to be insufficient for making observations and reading measurements. A hand-held bicycle lamp was therefore used to direct light. Though cumbersome, it proved to be an essential tool.
Each building has its own wildlife problems. The main animal life in the Terme Femminile del Foro consisted of rats and mice which seemed to derive some sustenance from biological matter, such as paper labels and the remains of seeds that were presumably used in the 1979 skeletal study for the determination of cranial capacity.12 The presence of these rodents influenced the way bones were labelled for this study (see below).
A lesser, but still irritating, problem posed by rodents was consumption of the filling of the ‘bean bag’, which was used to keep skulls stable during measurement. As dried beans were difficult to procure at the local supermarket, I decided to substitute small pasta shells as a filling. This proved most satisfactory until the bag was discovered by a local mouse. Although the bag was wrapped in several plastic bags and locked in a drawer when not in use, the contents kept disappearing. After the consumption of two bags of pasta (and a considerable portion of the cloth bag), I discovered that soaking the pasta in acetone prior to filling provided a mouse-proof solution. This anecdote gives some indication of the impact of vermin on the daily field work routine.
The wildlife in the Sarno Baths posed a greater direct risk to the skeletal material. Some of the bones had been partially destroyed by beetles as evidenced by damage consistent with chewing and the discovery of beetle elytra inside crania. Carpenter bees posed a threat to the plaster casts of humans as they bored holes into the plaster for their nests. As mentioned above, the use of crania by birds as nests meant that the skeletal landmarks were obscured by bird lime. This had to be removed very slowly and with extreme care to avoid damage and loss of information.
A key problem which affected the quantity of data that could be collected was access. The buildings used to house the Pompeian bones also served as a store for other finds from the site, such as marble and bronze artefacts. To maintain security, objects were generally not allowed to be removed from these stores for study or other purposes. Special permission was required to gain access to these buildings and to examine the material housed in them. Because of the value of the stored material, only three custodians had security clearance to handle the keys to these particular deposits. Access to the material was only possible when one of the guards was present. If one of the three custodians were absent, entry was usually only permitted in the morning.
The opening hours of the site vary throughout the year, the longest occurring in the summer months. Unfortunately, this is the period when the guards tend to have their holidays. As a result, for much of each field season, access was limited to four or five hours per day. In addition, I was usually locked in for the duration of each working day. This captivity entailed certain physical discomfort.
Access was also limited by the climate. It was more practical to carry out the bulk of the field work during summer and autumn as winter work conditions were far from ideal. The temperature and light levels inside the ancient buildings drops considerably during winter and the humidity rises. The length of each season varied from five weeks to six months, over a period of five years, the average lasting about three months.13
The literature variously suggested that approximately 2,000 individuals either lost their lives or had already been excavated (see Chapter 4). Such a large sample from a documented destruction is virtually unknown from antiquity.
Fewer individuals were available for study than suggested by the literature. It is known that a considerable number of bones were removed to the Regia Università di Napoli in 1853 for study by Chiaie. It is not known whether the hundred skulls that formed the basis of Nicolucci’s examination came from this collection or were additionally removed from the site.14 There has been no documented major removal of skeletal material from the site since these early anthropological studies. As many of the bones that are currently stored in Pompeii were probably excavated after the middle of the nineteenth century, it is likely that they would form a representative sample of the victims.
The sample has also been depleted by souvenir hunters and novel forms of secondary usage. Edward Bulwer-Lytton, for example, chose to grace the desk of his Knebworth house in Hertfordshire with a skull he collected from the excavations. Its shape and dimensions convinced him that it must have belonged to a man of great intellect and talent, a man like the evil Arbaces of his Last Days of Pompeii.15 Hester Lynch Piozzi, who published an account of his visit to Pompeii in 1786, observed a French tourist pocketing a human bone as a memento.16 In 1776, François de Paule Latapie commented on people who felt the need to have a piece of a Pompeian victim in their private collections and admitted to having removed skeletal material from the site for this purpose.17 This is an historical problem that, to some extent, still exists in cases where bones have been left in situ. For example, the bones of two skeletons were left as they were found in the Casa del Fabbro (I, x, 7)18 and were made available for the present study. Over a period of three years the number of bones in this house dwindled markedly. Apparently, tourists have managed to gain entry to the house and have purloined bones, such as mandibles, for souvenirs.
Some time ago, a ‘cottage industry’ was set up in the Sarno baths. This involved the transformation of human femora excavated at Pompeii into hinges for the reconstruction of ancient furniture found on the site. The original hinges were made of turned horse metatarsals which often required replacement for restoration.19 This has contributed a novel source of sample bias to the femur collection (see Chapter 6).
It is possible that the sample has been contaminated by a small number of tomb burials from the region outside the walls of Pompeii. A previous director of archaeology at Pompeii stated that the level of contamination in the bone deposits would be minimal20. His premise is based on a number of reasons, such as the fact that until the final decade of the twentieth century, excavations were mostly directed at the AD 79 level. In addition, a great deal of the preAD 79 excavated skeletal material was lost during World War II bombing raids on the site. Any contamination would be so small in relation to the number of skeletons in this sample that it would probably have very little effect on the general statistical trends.
Site recording was erratic, especially during the early excavations, which means that it is virtually impossible to determine the exact number of bodies that have been uncovered. Although there is a tradition that two thousand individuals were found, recent work indicates that we can only account for just over half this number (Chapter 4).
The Pompeian skeletal project
The problems associated with the Pompeian skeletal remains limit the amount of information that they can yield, but they do not diminish the value of the material as an archaeological resource. These remains re flect previously unexplored aspects of an important classical site. Unlike other skeletal samples from this era, this represents a mass disaster rather than the more commonly encountered cemetery populations. Cemetery populations often cover large time spans and may be defined or skewed by sex, age, cultural practices, such as segregation, or pathology, as in the case of plague burials.
Figure 5.2 Box of bone hinges stored in Pompeii
The Pompeian skeletal remains, along with those found from other sites destroyed by the eruption of Mt Vesuvius, form a sample with a number of features that are not often found in the archaeological record. They reflect a living population, which can yield valuable information about the people who inhabited Pompeii on the eve of its destruction. The major asset of these collections of bones is that they provide a relatively large sample of skeletal material from the ancient world where two variables, time and the cause of death, are already known.
The human remains from Pompeii, along with the skeletal material from Herculaneum21 and other Vesuvian sites, are also important because they provide a major source of information about Roman populations in Italy in the latter part of the first century AD. In this period, cremation was the most popular form of disposal of the dead and while it is possible to learn about the characteristics of individuals and populations from cremated bone, the success of analysis of burnt material is dependent on the amount of bone and the number of diagnostic features that survive in each case. Burnt bone tends to be warped or otherwise distorted and can shrink. This makes the evidence far more difficult to interpret than complete skeletal elements.
When I started to work on this material in 1986, it was seen to be of minimal value by most contemporary scholars. It was apparent that any study would be an improvement on leaving this neglected collection to further deteriorate. The research design for this project had to be developed with respect to the limitations of the sample and the site. It was also constrained by a very small budget, which limited the amount of time that could be spent in the field and the laboratory work that could be undertaken. As a result, the majority of the work was based on gross observations and measurements. Such data provide the most important basic information that can be obtained from any skeletal sample and are a prerequisite for any further analyses, such as DNA or stable isotope analysis.
The project was designed to construct a population pro file in order to address questions about the sample of Pompeian victims. One of the key aims was to test the commonly held assumption that the sample of victims would be biased towards the old, the infirm, the very young and women.22In the absence of complete skeletons available for study, I decided to concentrate on statistical studies based on large numbers of particular bones. This obviously influenced the nature of the research questions. The bones selected for study were chosen for their ability to provide specific information. For the determination of sex, the best skeletal indicators are the pelvis, femur, humerus, skull and teeth. Age-at-death was based on pelves, teeth and skulls. Another issue that was addressed was whether the sample of victims reflected a heterogeneous population, as suggested by ancient writers who described the inhabitants of Pompeii and the surrounding region.23 The bones chosen for population studies were skulls, pelves, femora, humeri and tibiae. Because the skeletons were disarticulated, the femur was chosen as the most reliable single long bone for the reconstruction of height.
The most useful skeletal indicators for each feature under study could then be employed as controls for the determination of population norms for the other bone types in the Pompeian sample. For example, because of its biological function, the pelvis is the most reliable indicator of sex. Other bones, such as the femur, humerus and skull, also reflect sexual dimorphism but the degree of difference between males and females can vary between populations due to cultural and genetic differences. By using the most useful sexual marker as a baseline, it was possible to establish the sex-related parameters for other bones in the Pompeian sample.
The first priority of this study was to obtain the maximum amount of information from a large sample of different individuals. Initial long bone measurements and observations were carried out on left bones to ensure that each bone represented one individual. Where possible, this work was also carried out on samples of bones from the right side. Observations from both sides are particularly important for the determination of frequencies of postcranial non-metric traits and supposed stress markers, such as tibial flattening or platycnemia. Such traits appear to be side related in certain populations. Due to time and access restrictions, it was not possible to complete both sets of observations and measurements on all bones.
Preparation and recording of the material
Before the bones could be catalogued and measured, they had to be sorted, cleaned, coded, described and, where appropriate, photographed.
The first task was to organize the material into a form where it could be used. The disarticulated bones in the Sarno Baths were sorted into groups of skulls, mandibles, sacra and left and right long and pelvic bones. Bones that were not likely to provide useful information in terms of the aims of this study were placed in separate piles. Bones with evidence of pathological changes were segregated from the rest of the skeletal material. In some cases it was possible to reconstruct individual bones from fragments on the basis of pathological change, as in the case of the skulls of individuals with hyperostosis frontalis interna (Chapter 8). The unboxed bones in the Forum Baths had already been sorted, though it was necessary to separate left and right long bones. The conditions inside both the Sarno and Forum Baths meant that bones had to be recleaned each field season.24
Figure 5.3 Sorted piles of bones in the Terme del Sarno (VII, ii, 17)
Labelling the skeletal material under investigation involved some consideration. Evidence of the remains of labels from previous examinations of the skeletons suggested problems for long-term projects. A number of skulls display large numbers painted directly onto the frontal bone. These probably date from nineteenth-century investigations.25 This method of labelling is obviously durable but tends to obscure certain anatomical features, which makes scoring some of the non-metric traits problematic. Also, modern practice requires that any method of labelling is reversible, especially with regard to skeletal remains. Disfiguring skeletal material in this way would now be seen as a sign of disrespect (Chapter 11). There was a lack of clearly defined labels dating to Nicolucci’s study. This was deplored by D’Amore et al.26 as it meant that they could not remeasure the bones he used in his work. D’Amore et al.27 used adhesive paper labels. This suited their purpose of a one-season study. By the time I commenced my research, the adhesive had mostly failed and it was impossible to associate labels with specific bones. This limited comparisons with previous studies on the bones to general trends rather than specific cases.
I initially used tie-on tags for the long bones as this form of identi fication is less time consuming than painting numbers onto a treated area of the bone surface. After the discovery of a mouse with a penchant for paper inhabiting the femur pile, I decided that numbers applied with Indian ink on a cleaned bone surface coated with clear nail varnish would be more likely to survive from one season’s study to the next. Each bone was marked with small numbers in a place that would not be too apparent and would not obscure any diagnostic features. The nail varnish was used so that marking was reversible as it could be removed with acetone, leaving no trace of ink on the bone. This method of identification was used for all the bones under examination.28
As mentioned above, the available sample of skeletons was considerably smaller than the number of discoveries claimed in the archaeological literature. It was not possible to select a sample as a percentage of the total since the exact number of individuals excavated was not known. For this reason I decided to examine the largest representative sample for each type of bone.
Almost all the bones stored in the Forum and Sarno Baths were inspected during the course of cleaning, which permitted assessment of the scope of the remains. As it was not logistically possible to fully record all the available Pompeian skeletal material, I decided to concentrate on the Forum Bath sample for the majority of skull and long bone studies. There were several reasons for this choice. First, I thought that the bones in the Forum Baths would provide a good random sample that I had not chosen. Comparison with the cranial and long bone remains in the Sarno Baths suggested that the Forum Bath sample was representative of the Pompeian material that had been recovered. Second, the Forum Bath bones were specially chosen for their completeness. Many of the Sarno Bath bones were incomplete and a number of the skulls had been deformed by the pressure of the ash under which they had been buried. Third, it would be possible to make a general comparison with the results of the 1979 study of the Forum Bath collection. In addition, these bones required less preparation for examination as they had been cleaned previously29 and were stored in conditions which did not require the same amount of annual cleaning as the Sarno Bath bones. Finally, the Forum Baths provided a more desirable working environment since there was a table and some access to artificial light.
It is dif ficult to assess the exact number of individuals that are housed in the two bath buildings. The sample in the Sarno Baths appears to be strongly biased towards cranial remains. It would be very time-consuming to attempt to establish a minimum number of individuals from the skulls as many of them have been broken into small fragments which have been scattered about the building. At least 360 individuals could be identified from the combined crania in both the Forum and Sarno Bath collections. One possible explanation for the higher representation of skeletons by cranial rather than other remains is that the skull was the part of the skeleton most easily recognized by excavators that were not schooled in anatomy. Another is that the skull was considered the most important bone by nineteenthcentury anthropologists (see Chapters 3 and 9) and that other, or post-cranial, skeletal remains were not thought worthy of preservation. A possible problem that could be associated with this skewing was whether the skull sample chosen for analysis was comparable to the samples chosen for other bones. The sample sizes chosen for each type of bone, however, were large enough to be statistically significant. In addition, the analysis suggested that there was enough consistency between the results to generally dismiss sample bias between bones as a problem.
The sample also seems to be strongly skewed towards adult bones with a total absence of neonatal bones and few young juveniles. Under-representation of neonates and infants is a recognized problem in archaeology and is thought to primarily result from small and fragmentary bones either being ignored as they are not recognized by excavators or disintegrating in the ground prior to excavation.30
The method of storage also appears to be a contributory factor for the absence of neonatal skeletal remains and the relative scarcity of young juvenile bones. This notion is supported by the comparative frequency of juvenile bones in collections of skeletons that have been left in situ for display purposes and the number of juveniles represented in the collection of casts (Chapter 10)
I decided to concentrate on issues associated with adult bones to circumvent the problem of a sample biased against juvenile remains. Juvenile skulls, teeth and pelves, however, were recorded to give an indication of the proportions of different age groups represented in the available sample. These particular bones were chosen because they were useful for the determination of juvenile age-at-death. Also, sufficient numbers of each bone were available for examination to provide a representative sample of the age spectrum of Pompeian victims stored on the site.
All the pelves in the Sarno Baths collection were examined. All the available teeth and as many skulls as possible from both stores were studied. While juvenile skulls were not sexed or used for the metric study, they were examined for epigenetic traits (see Chapter 9).
Two groups of skeletons from the Insula of the Menander that were supposedly in situ were to be studied as a control sample of intact individuals (see Chapter 1). It soon became clear that the main group of about ten bodies had been tampered with for display purposes by people with limited anatomical knowledge and that the individuals had been, in effect, disarticulated. The other group of two bodies was not complete.
A range of basic measuring instruments was employed, such as vernier callipers, spreading callipers and an osteometric board.31 The choice of measurements was based on a survey of the then current literature. It should be noted that the Standards for Data Collection from Human Skeletal Remains32 was not published when these data were collected. Nonetheless, measurements and observations were recorded in sufficient detail to enable them to be applied to this sample. The main criterion for each measurement taken was that it would yield information about sex, height, age-at-death, pathological alteration of bone, or population affinity.
Reliability of the measurements
A problem that is frequently associated with the metric analysis of archaeological bones is that they tend to be incomplete or eroded as a result of post mortem damage. The areas around the landmarks on bones, which are used to define measurements, are often damaged and a certain degree of guesswork may be required to make a measurement. An assessment of reliability or confidence level in each measurement would therefore be prudent, although it is not often undertaken in skeletal studies. The scoring system that was employed for this study also acted as a reflection of the degree of preservation of the bones.
One of the major concerns in dealing with this sample was whether it was representative of the Pompeian victims or was biased towards the more robust bones, which are generally associated with males (see Chapter 6). If the sample were found to be representative, there was still the problem of whether the more gracile bones, generally assumed to be associated with females, were more likely to be incomplete. This would pose a problem for making valid conclusions from certain types of analysis, such as multivariate statistical analysis, which have a limited capability for dealing with incomplete data sets (see below). I worked on the quantification of bone preservation with a statistician using cranial measurements as an example of how this problem, which came to be known as the ‘crumble factor’,33 could be assessed.
All the bones that were used in the investigation were described by a series of measurements and observations. The accuracy of each measurement was dependent on the completeness of the bone and the ease of location of specific landmarks. To quantify this, a four-point scoring system of confidence, or V-score, was assigned, based on that employed by Howells for cranial measurements. The value of the use of V-scores can be demonstrated from an inspection of the three-dimensional graphs for the V-scores of each of the cranial measurements. These are standard measurements that are used to describe the dimensions and shape of the skull.34 The three-dimensional graphs display the frequency distribution of each measurement in the sample, along with the associated degree of confidence.
Certain measurements, such as glabello-occipital length (Figure 5.4),35 the frontal chord36 and parietal chord,37 were found to be highly reliable, whereas measurements like maximum cranial breadth38 and maximum frontal breadth39 (Figure 5.5) were demonstrably less accurate. Bizygomatic breadth (Figure 5.6)40 produced few results that were better than guesses.
It is also apparent that for the majority of measurements, for example, glabello-occipital length, nasio-occipital length41 and maximum cranial breadth, there is no appreciable bias towards robust bones in relation to the
Figure 5.4 V-scores for glabello-occipital length
survival of landmarks. Where there is a bias towards the destruction of the more gracile skeletal material, it is very slight, as can be seen in the measurements of biasterionic breadth,42 basion-nasion length,43 occipital chord44 and the frontal chord.45 Bizygomatic breadth, which reflects the greatest bone damage of the twelve measurements is, if anything, skewed towards bone loss in the more robust end of the sample. This is also true to a lesser extent for basionbregma height,46 maximum frontal breadth and biauricular breadth.47
Visualization of the con fidence in each measurement made it possible to establish which measurements could reasonably be used and which should be either discarded from analysis or interpreted with due caution. It was apparent that bizygomatic breadth was not very reliable and, where possible, was not included in statistical analysis of the Pompeian skull measurements. It is notable, however, that when it was used the results were comparable to those of other populations. It is possible that the bizygomatic breadth
Figure 5.5 V-scores for maximum frontal breadth
measurements of other population samples were based on the same type of reconstruction and consistently reflect the same artefact.
Although it might be expected that the ‘crumble factor’ would have greater impact on the more gracile bones,48 these graphs show that there was at least equivalent bone destruction in the more robust end of the sample. Assuming the sample is made up of both males and females, these results indicate that there is no significant differential bone survival for the adult material and that the use of multivariate statistical analysis should not produce results with a sex bias.
The main aim of the statistical analysis was to characterize the Pompeian sample and to answer specific archaeological questions about the material.
Figure 5.6 V-scores for bizygomatic breadth
The trends that emerged from this study could be tested against samples of more complete skeletons as they are excavated in the future.
Initially, emphasis was placed on the use of descriptive statistics to assess the potential value of different types of bones as descriptors in this disarticulated collection. Both univariate and multivariate methods were employed. The combination of the two types of data analysis fulfils several functions. Univariate statistics, which describe a single variable at a time, were used to ensure that all the data were presented and to establish whether any distribution trends, such as skewing, normality or bimodality, were immediately apparent. It is a valuable way of describing the data in a compact form by simple graphics, such as histograms, and statistics, such as range, mean and standard deviation. The results of these analyses can easily be compared with data from other sites. These analyses are essential for the appropriate application of the data to predictive models, such as the determination of stature from femur length from samples of unknown sex.
Multivariate methods, or those that deal with a number of variables at the same time, correspond more closely to the way humans obtain and analyze information. For example, we process a raft of variables to identify and classify archaeological material. The main value of employing multivariate techniques is that they enable all the available data for bones with complete data sets to be viewed at once, revealing underlying structures and trends which would not be apparent from single variables. These techniques are particularly useful for unknown and incomplete samples. However, one of the main problems encountered with incomplete material is missing values.
Statistics copes with missing values in two ways. The first is to either delete variables that contain any missing values, or exclude objects that do not have values for one or more variables. The second is to replace missing data with average values based on those of the other data. Both these methods have their disadvantages.
Deleting objects that have missing values may not only signi ficantly diminish the sample size but can also skew the sample. It could be hypothetically argued that, in the case of some skeletal material, the more robust bones may have a higher survival rate. This means that incomplete bones could represent females. Deleting variables with missing measurements or observations preserves the sample size but can result in the loss of a large number of the variables. Replacing missing information with average values maintains sample sizes but this method is based on the assumption that all cases are similar, which does not necessarily reflect the bones in the sample.
Averaged values were not used in this study. It was considered that even though deleting missing cases could affect the integrity of the sample, the use of average values would be far more misleading and a greater source of potential error. In some cases multivariate analysis was not possible because of the large number of missing values in certain data sets. In these cases bivariate analyses comparing two variables at a time were undertaken.
There are a large number of multivariate techniques that could be used to analyse skeletal data. All statistical techniques have assumptions that must be met for their appropriate use. This is especially true for multivariate techniques. For example, discriminant function analysis49 could not be applied to the Pompeian skeletal data as it is based on data sets derived from known populations. This standard technique is used in archaeological and forensic applications to place unknown individuals from a known population into their correct group for characteristics to establish sex or population affinities.
Facial reconstruction involves the combination of scienti fic and artistic techniques to re-create soft tissue facial features from a skull. Some scholars consider the term facial reconstruction to be misleading as it suggests a greater degree of certainty than the technique provides, and prefer to use ‘facial approximation’ as a descriptor.50 These techniques cannot produce an accurate reproduction of a face because the available evidence does not yield information about a number of features, such as the ears, hair, lips, nose, eye colour or the amount of fatty tissue.
There has been some research about the reliability of facial approximation in forensic applications. This has been possible because the results can be tested against a known individual when a match is made. In a serial murder investigation in the USA, a number of artists created approximations of facial features from the same remains with the same baseline information about sex, age and population affinity and produced significantly different results. The main differences were in the interpretation of soft tissue, which resulted in variation for eyes, noses and profiles. There was greater consensus for areas like the cheekbones and chin.51 A further problem that has been observed with regard to facial approximation is the tendency for artists to incorporate their own facial features into the reconstruction and produce visages that bear more resemblance to themselves than the subject.52
These problems are exacerbated when reconstruction techniques are applied to archaeological material as the results cannot be validated. Facial approximation is also not generally considered to be an essential osteological research tool and tends to be used only as a means of making individual skeletons more accessible to the broader public. One could reasonably argue that in the case of Pompeii, the casts already serve this purpose (see Chapter 10).
Despite the loss of considerable skeletal information, as a result of site management in the eighteenth and nineteenth centuries, it is still possible to obtain a substantial amount of data from the remaining bones. Given the paucity of available Roman skeletal material representing the first centuryAD due to the practice of cremation, the Pompeian skeletons provide an important source of population information for this period.
There are virtually no ideal archaeological sites or samples. As with the disarticulated human skeletal sample from Pompeii, it is important to acknowledge the constraints associated with the material under investigation so that a specific research design can be developed to deal with these problems. The Pompeian skeletal evidence provides tantalizing and sometimes ambiguous glimpses into the lives and deaths of the people who became victims at this site. This study demonstrates the potential of what at first appears to be an extremely difficult data set.