[Back] [Section Summary] [Section 9-12 Summary]] [Contents] [Home]

8.7 Overall Interpretations

8.7.1 Problems of stratigraphy, percentage of site excavated and category assignment

There are three important aspects of this project which are fundamental to its validity. The first, in the case of an old excavation like Vetera I, is the difficulty of identifying artefacts which indeed relate to activities that took place during the occupation of the final Neronian fortress, and were not from earlier occupations or part of any redeposition process; the second is the representativeness of the excavated area, and the third is the difficulty of correctly assigning functional categories to these artefacts.

While trench provenances are provided for most of the artefacts in Hanel's catalogue, very little stratigraphical information is available. For example, a trench that was dug to investigate the wall foundations of one of the Neronian buildings can include a combination of undifferentiated contexts from the modern surface through to, in some cases, Augustan levels. As discussed, the stratigraphical contexts have been indicated where possible, and any items from identifiable pre-Claudian contexts have not been included in the analyses. However, there are many other items that may also have been from pre-Claudian contexts but this cannot be verified, most notably those from the excavations to explore the foundations of Building A. Identifying whether material has been redeposited is also difficult, although the short life, rapid abandonment, and nature of the area excavated at Vetera I suggest that the likelihood of this is minimal. Only through the consistency of any distribution pattern can it be argued that these assemblages do not constitute residual material from earlier phases or redeposition.

A second potential problem relevant to this analysis is whether the same percentage of each building was excavated. Visual analysis of the overall plan of the trenches indicates a fairly comparable coverage of most buildings although, as already noted, some buildings were only partially excavated. This applies particularly to palace P, Building T, Building F, barracks O and N, and Buildings R and Q. The latter two do not really feature in this analysis and, despite the small percentages of palace P and Building F that have been excavated, these buildings each contained a considerable range of finds. Building T also demonstrates a tantalising assemblage in the small area excavated. Thus, when their artefact distribution patterns are examined alongside the completely excavated buildings, these few fall within observed patterns. The lack of excavation of barracks O and N and of about two-thirds of the fortress is more significant for a study interested in distribution of activities across the fortress. It, therefore, has little to say on the distribution of activities within most of the accommodation of ordinary soldiers and probably that of storerooms, workshops, granaries and stables. That said, however, substantially more of this site has been excavated than many Roman fortress sites.

The assignment of functional and gender categories to certain items in this study may also seem, at times, rather positivist. However, the aim is not to prove that these items had such categories. Rather it is to explore the kinds of activities which took place at Vetera and the spatial distributions of these activities if these categories are indeed correctly assigned.

To demonstrate that the data and distribution patterns in this study are relatively robust it needs first to be shown that these patterns conform to what might be expected at such a site.

8.7.2 Traditional building function and gender distribution

This study can assess the relationships between artefact distribution patterns and their related functional categories, and the functions ascribed to the buildings at Vetera I by Lehner, and Hanel, and indirectly by von Petrikovits (see Section 4.4). It can also investigate the spatial distribution of gendered activities in relation to the various components of the fortress.

In many ways the study supports expectations of the distribution of activities and people within a 1st century CE Roman military fortress. The central administrative buildings, particularly Buildings A and B, contained a wealth of items related to male activities, especially combat activities, but also to administrative activities such as writing. In contrast, personal activities, such as dress (excluding combat dress), toilet activities, and leisure activities (i.e. gaming), were comparatively less well represented in these buildings than they were in the officers' residences, and interestingly, in the buildings along the east end of the via principalis. Similarly, items which are likely to have been associated with women were only sparsely represented in these central buildings but relatively predominant in the officers' residences. These distribution patterns suggest that the functional and gender categories ascribed in this study are likely to have some validity.

In addition, there is no clearly distinguishable pattern between these items with good Neronian contexts and those with less secure contexts. Thus, the overall pattern seems to be robust and valid. The errors that are no doubt present from the incorrect assignment of functional and gender categories, or from the imprecise stratigraphy, have little impact.

If these distribution patterns can, therefore, be accepted as valid, then it is interesting to investigate what distribution patterns at Vetera do not conform to the traditional functional ascriptions of its buildings or to the expected positions of women and children within a 1st century CE legionary fortress such as this.

8.7.3 Re-asssessing the functions of buildings

8.7.3a The barracks

In the first instance the lack of personal items, such as toilet and dress items, in the buildings identified, on architectural grounds, as barracks is perhaps noteworthy. This applies to infantry barracks N and O, although one toilet item was found in barracks N and generally very little was recorded from these buildings. There is also a relative lack of personal items in barracks Y, V, and W. In contrast, barracks Y in particular, but also barracks V, had a wealth of tools and items related to commerce. These were most evident in the row of rooms along the front of each of the barracks. It is interesting to postulate that these buildings fronting the street, and identified by Lehner (1930, 39) as shops, were indeed workshops for the craftsmen, the immunes, who may have resided in the barracks behind. They were well positioned to deal with the needs of both the fortress and any settlements outside. No comparable finds were made in Building S, identified by von Petrikovits (1975, pl. 4.10) as an immunes barracks. The activities documented in Building S had more in common with the officers' houses, although notably no items potentially related to women or children were found here.

It is also interesting to note that both these areas, barracks Y and barracks V and W, had items that were possibly associated with women and children. A small, silver decorated trumpet fibula, dated Claudian-Trajanic, was found in or near barracks W and two small bronze bells, possibly belonging to children, were found in the front parts of both barracks Y and barracks V. These are not particularly significant but the trumpet fibula is of interest in debates concerning the presence of women within the barracks.

8.7.3b Valetudinarium Z

It has been noted on several occasions in the preceding analyses that the finds in Building Z do nothing to substantiate its identification as a hospital. There was a lack of toilet items here, with the exception of glass bowls. Remains of three larger glass bowls or dishes were found in the rooms on the street front of the building and one just inside the central courtyard. Two more glass bowls, one small and one large, were found along the outer west side of the building. In contrast, thirteen possible dress items were found here, including at least one female item, a belt attachment. There were also a number of combat items in the rooms along the front. Although the recorded finds from within this building were minimal, the lack of toilet-related items compared to, for example, Building T would suggest that the function ascribed to Building Z is improbable, or else that military hospitals functioned without the usual equipment found in the toolkit of medical practitioners (see Jackson esp. 1994b; compare Baker 2001). However, it is not possible to ascribe any other function to this building through the distribution of the particular artefact types selected in this sample.

8.7.3c Buildings b, c and d

There was a relative dearth of finds in Buildings b, c and d. Finds from Building b, investigated in this study, consisted of a bronze pendant, a lead weight, two inkwells and a hinge. Finds in Building d consisted of a drop handle only and nothing was recorded in Building c. However, over 400 artefacts were recorded from Building b, being predominantly tableware, food-preparation and storage items, and some thirty artefacts from the small part of Building d which has been excavated, also predominantly tableware and food preparation items. While more investigation of other activity patterns is needed, this distribution pattern suggests that Building b was not used for any of the personal or gendered activities discussed here. That would tend to suggest that it was not a workshop and probably not a residence.

8.7.3d Building a

Finds in Building a included a number of dress-related items, one being the remains of a women's necklace. They also included a number of glass bowls and finds related to either toilet or eating and drinking. The finds from this building, but not included in this gender study, also consisted of a considerable amount of tableware and serving items, food-preparation items and coins (Allison n.d. 2). It is by no means conclusive but it is tempting to see this as an inn or guest house, like the one run by Belica at Vindonissa (see Section 2.3.2c).

8.7.3e Building F

Lehner (1930, 23) identified a pre-Neronian hospital beneath this building but Hanel (1995, 304) felt that this identification was unsubstantiated. In this study no certain toilet items were documented here but the remains of a number of glass unguentaria and bronze basins were recorded, all from Claudian-Neronian contexts. Other finds from this building, and from the street immediately in front of it, included a number of dress-related items, excluding three from a pre-Claudian context, and numerous tools and craft implements, notably six writing implements. Food preparation, tableware and serving items were also relatively prolific here (Allison n.d. 2). It is difficult to interpret this assemblage which seems to cover a wide range of activities. Given that Building F is only partially excavated and many of these artefacts were found along the street front, this assemblage may rather document a range of services and wares sold in shops along the via principalis. Three of these items may have been associated with women and children. With the possible exception of the comparative wealth of writing items, this assemblage would not seem to confirm Hanel's identification of Building F (1995, 100) as an officers' administrative building.

8.7.4 Re-assessing the place of women and children within the fortress

While it is noted above that a large proportion of items related to women and children were found in officers' residences, an equal, or perhaps even greater, proportion of such items was not. A scattering was found across the central administrative buildings, Buildings A, B and G, but it is difficult to assess the significance of these, not least because none of these particular items are definitively women's and children's items. Though, perhaps they warn us that these were not exclusively male zones.

8.7.4a The gateways, the street, the shops and the central market area

There is, however, a relative concentration of women's and children's material in the open area in front of Building A, particularly immediately above Buildings C and D. This material was mainly from Trench 288, pit 6, which also contained many complete utilitarian ceramics and glass artefacts and which Hanel (1995, 311-312 ) dated to 69/70 CE. Lehner suggested (1930, 40) that this open area was a market area. Pseudo-Hyginus (12) had referred to the area on the via principalis, in front of the praetorium, as the place where the crowd gathered. The high proportion of possibly female and children's items here, as well as in the via principalis and in the buildings opening on to it, and in the east gateway, may point to the equal, or greater, numbers of women and children frequenting these public and relatively commercial areas as the officers' private residences (see James 2002, 43).

There is a significant concentration of such items, mostly dress-related, in buildings in front of Building Z that open onto the street. The numbers are extremely small but, if the quantity of women's and children's items in the tribunes' houses documents their habitation there, then the quantity here might also document habitation.

8.7.4b Buildings T and U

Three definitively female items were found in the vicinity of the east gateway, including two bronze hairpins in Buildings T and U. Other items found in the relatively small excavated area of these two buildings, and included in this study, were: six 'male' and 'male or female' dress items, two probes, remains of eight glass bowls, two writing implements, one small lead weight, one knife, and one gaming counter. With the exception of one mattock, which seems to have been in the courtyard, no definitely male activities were recorded here, most notably no combat equipment and no combat dress. Among the other items recorded here but not included in this gender study were quantities of tableware, food-preparation and storage items, and some coins. Painted wall plaster, with vertical stripes and fields, was also recorded in these two buildings. It is difficult to interpret this pattern. The apsidal buildings may suggest that they had a religious purpose but the artefacts do not support this identification. The lack of combat equipment is interesting, as is the preponderance of dress- and toilet-related items. Is it possible that these buildings could have been part of a bath complex within the fortress, as was recorded at Vindonissa (see e.g. Unz and Deschler-Erb 1997, fig. 2)?

8.7.5 Statistical analyses

The data and distribution patterns in this study can also be tested for robustness by assessing whether the patterns portrayed by spatial analysis are also repeated using statistical analyses.

8.7.5a Multivariate Analyses

Correspondence analysis (CA), a multivariate statistical technique, was used to explore the pattern of variation in the gender data and especially to see if statistical analysis repeated some of the patterns identified visually on the GIS plots. CA is an ordination method, which summarises many numerical descriptive variables, in this case gender attributes, into a smaller number of axes of variation along which the statistical cases they describe (excavation trenches, buildings etc.) can be plotted. In this way, groups of buildings or excavation trenches, formed on the basis of these numerical variables, can be identified and displayed. CA has the added advantage of producing plots in which the position of the variables is also displayed, allowing the numerical relationship between all cases, all variables, and between cases and variables to be analysed and seen (Shennan 1988; Baxter 1994).

Bi-plots showing the first two axes of variation were prepared for two sets of data derived from the gender classification. Data from individual trenches were not used as they represented arbitrary divisions of space. Rather, gender data was grouped by building or area, with cases excluded in which the sum of gendered artefacts was less than 10. This was done to reduce the effect of cases with small sample sizes. It was sometimes difficult to map precisely where individual artefacts came from and some were excluded if they derived from trenches that crossed building boundaries. CANOCO (Ter Braak 1999), was used to perform correspondence analysis after standard reductions and transformations of the data were applied to minimise the influence of rare artefact occurrences. Variables occurring in less than 10% of the cases were removed and, because of positive skewing, all data were subject to a square-root transformation.

Correspondance analysis chart
Figure 28: Correspondence analysis between buildings, with more than 10 artefacts, and gender categories (A. S. Fairbairn)

The first data set was derived from all buildings and excavation areas in which there were more than ten gendered artefacts (Figure 28). This included 513 artefacts (of the 704 gendered artefacts) from trenches in Buildings, A, a, B, A-B (i.e. includes trenches that cut through both buildings), F, G, trenches from a-H and H combined, M, P and P-F combined, S, T and U combined, V, Y, Z, finds from the central market area, the gates and the via principalis.

Correspondance analysis chart
Figure 29: Correspondence analysis between groups of functionally related buildings and areas, and gender categories (A. S. Fairbairn)

A second analysis was carried out which grouped the buildings according to their overall functions (Figure 29). This was done to simplify the data further and had the added advantage of including artefacts from several important buildings that were excluded in the first, taking a sample size to 546. The remaining artefacts in the total sample for this study were either outside the Neronian fortress or unprovenanced. Buildings K, J, M, H, P, Q and R were grouped as officers' residences (OFFICR on plot), A,B and G as administrative buildings (ADMIN on plot), Buildings O, N, Y, V and W as the barracks (BARRAK on plot), the gates, central market area and via principalis were grouped as open spaces (STREET on plot). In addition, a, F, S, Z and a merged TU group were also plotted.

The results show a similar pattern of variation in both plots, with the female and child artefacts causing the main variance along the first axis (horizontal) and the male and uncertain male/female objects causing little. Male and female categories were distinguished more along the second axis of variation, with male categories plotting to the positive side of the axis and female to the negative side. In short, the officers' quarters and public spaces, with the possible exception of the via principalis area in front of Building F, showed a tendency for containing more non-male artefacts. This tendency was shared by Buildings H (a_H_H on plot), a, Z, F and a combination of Buildings T and U (TU on plot). The barracks and the administration blocks showed a tendency to contain fewer non-male artefacts, as did Building S.

This analysis broadly supports the assessment of the GIS artefact distribution plots, even though sample sizes from many of the buildings were small.

8.7.6 Occupancy of the fortress

Given that only a small proportion of the fortress has actually been excavated it is difficult to use this portion to assess the state of occupancy immediately before its abandonment. It has already been noted that no finds were made in Building c. Does this mean that this building had been empty before the final destruction of the fortress? Does this emptiness reflect its habitual use and state of cleanliness?

8.7.6a Palace H

Perhaps a more interesting observation is that palace H was largely bereft of domestic finds from the Claudian and Neronian periods, compared both with the tribunes' houses and also, notably, with the relatively limited excavated area of palace P. In this study were recorded: two writing implements at the front of the palace, facing the via principalis; a few fragments of glass bowls, at the front, on the east side, and in the courtyard; and a small scatter of other items. The main concentration of artefacts was found in the north-west corner and consisted of dress-related items, including one bronze hairpin, a lead slingshot and a lead weight. Given the area covered by this building, it was also relatively bereft of other finds (see Allison n.d. 2, esp. fig. 10).

The trenches across palace H indicate that it was as extensively excavated as any other excavated area within the central part of the fortress. If the pattern does not relate to some anomaly in the excavation process, this lack of finds suggests that this palace may well have been empty at the time the fortress was besieged. It is tempting to suggest that this might mean that only one legate had been in residence in 69/70 CE, and possibly only one legion.

8.7.7 The ban on soldier marriages in the 1st century CE

The double legionary fortress at Vetera I was destroyed in 69/70 CE, over a century before Septimius Severus had allowed ordinary soldiers to live in wedlock with their wives (Herodian, Histories: 3.8.4-5). Thus, the study examines the traces of women and children within the fortress at a time when only officers' families, and perhaps their domestic servants, were supposed to have lived within Roman military fortifications.

Unfortunately, this study throws little light on the issue of whether or not, despite the legal ban, soldiers lived with their wives and families within the fortress at Vetera. This is largely because very little of the soldiers' accommodation has been excavated. What the study does show, however, is that women and children were as present in the more public areas of the fortress - the central market place, the shops lining the street and other buildings immediately off it - as they were within the officers' residences. One can perhaps postulate their role in these places. For example, is not difficult to imagine that local women, or women from the canabae came into the central market area to sell merchandise or that they provided services in the shops along the street. If indeed Buildings T and U were part of a bath complex, however, does the presence of women's hairpins here document one or more of those services and is it necessary to assume that such women lived outside the fortress? Given the lack of finds in barracks W why was a trumpet fibula found here? Why is there such a concentration of women's and children's items in front of Building Z. These traces are admittedly extremely meagre but they can be seen to support the epigraphical material from Vindonissa (see Section 2.2.2) and the findings of Carol van Driel Murray (see 2.3.2b and 8.2.1c).

8.7.8 Concluding comments

The observations presented in this study demonstrate how such an approach, if applied to a series of military sites, can start to tell us more about the life within a Roman military base, about the roles of non-military personnel, particularly women and children, and about the non-military activities in this community.

With the exception of male dress and male activities, this article has concentrated on a number of activities that are not normally documented by large numbers of artefacts within an excavated military installation. Consequently the analyses produce maps of only a small number of items and are, therefore, suitable for a methodological article such as this. However, much more work is needed, particularly more inter-site analyses, before the suggestions made in this article can be given more weight.

Also, the analyses of other activities, such as cooking, eating and drinking, which are represented by larger amounts of material and which produce more complex maps, are as important to our understanding of fort life. A preliminary exploration of this material has been carried out for Vetera I and showed interesting patterns (Allison n.d. 2).

One of the main reasons why it is possible for me to carry out research of this type, using previously published excavation reports, is that the questions in which I am interested are different from those of the original investigators. Until recently, as outlined in Section 2.1, most investigators of Roman military sites have been more concerned with the fort as a fighting unit and as an element in the expansionist machine of the Roman Empire, or with the evidence the excavated artefacts provided for the dating of pottery typologies. The applications of new theoretical frameworks to answer questions concerning social behaviour and of new technologies to already published data from old excavation reports demonstrate that much useful information can still be gathered by revisiting past research.

[Back] [Section Summary] [Section 9-12 Summary] [Contents] [Home]

© Internet Archaeology URL: http://intarch.ac.uk/journal/issue17/4/8.7.html
Last updated: Mon Apr 4 2005