Internet Archaeol. 17. Allison. 3.2. Categorising artefacts at Roman military sites

3.2 Categorising artefacts at Roman military sites

As outlined in Section 2.3, with the exception of specifically military equipment, there have been few comprehensive studies of artefact assemblages and artefact function in Roman military studies, although the number has been growing in more recent years. Twenty years ago Nina Crummy (1983) published a catalogue of the small finds from the excavations at Colchester, organised by a series of functional categories. Lindsay Allason-Jones has studied artefact assemblages from turrets on Hadrian's Wall (Allason-Jones 1988) and Simon Clarke (1994) also stressed the need to study complete artefact assemblages from Roman military sites. Hilary Cool and Michael Baxter (2002) recently demonstrated how artefacts from a number of different Romano-British sites - urban, rural and military - can be divided into a range of functional categories, and then multivariate analyses carried out to identify different types of sites as well as the varying functions of different spaces within a site. Patricia Baker (2001) has argued for a need to investigate the different depositional patterns for various types of artefacts within different military sites. Andrew Gardner (2001; 2002; 2004) has also recently stressed the need for a more contextualised approach to artefacts and social practices at Roman sites, using a consumer approach to classifying these artefacts. He has used GIS software to isolate distribution patterns of artefacts in the vicinity of the west gate of the fort of Birdoswald on Hadrian's Wall (Wilmott 1997; Gardner 2004), as well as at other British military sites (Gardner 2002).

3.2.1 Ascribing functions and activities to artefacts

Allason-Jones argued (1999a) that there are many pitfalls in identifying artefacts as being either military or civilian equipment (see also Gardner 2002, 331). For example, she posited that a lead candelabra and Minerva head couch decoration could be considered military equipment as these were the kinds of personal luxuries which soldiers, and especially officers, could bring with them during their posting, which might last at least three years. She has therefore argued for a separation of purely military equipment (e.g. swords, helmets, shields etc.) from non-military equipment.

These distinctive dichotomies are not important for a study that is concerned with extracting social relevance from artefacts and artefact assemblages. Given that all the activities within Roman forts and fortresses fall within the military sphere, as opposed to the urban or agricultural sphere, this study aims to highlight, through the artefact distribution, that life within a military site also embraced the 'domestic' sphere and, as such, included many activities which did not pertain to combat and defence and which could be carried out by both military and non-military personnel. In other words, the current study is not concerned with separating out the professional, community and personal activities. Rather it investigates how these are integrated together and what we can learn about relationships between these activities and the people who performed them. But, we need to be clear about levels of functional categories. For example, 'military' or 'domestic' are spheres of activities rather than actual activities, whereas 'combat', 'defence', 'food-preparation', 'eating' and 'drinking' are activities and functions, in terms of the specific function for which a person uses an item.

Cool and Baxter (2002, 366) cautioned that the current methods used by archaeologists to summarise finds according to functional categories are misleading, particularly concerning small finds which are not ceramic or bone and which are usually found in limited quantities. They argued that the ascription of these functions rarely include any formal analysis of the artefacts and involve only narrow functional interpretations. To map activities that provide a more nuanced understanding of how the components of each individual fort functioned we need to consider the range of activities for which each artefact can be used and also the range of people who could potentially have used it.

3.2.2 Ascribing categories

3.2.2a Functional categories [Figure 10]

To counter 'narrow functional interpretations', the functional categories in this study include alternative (or queried) functions that are intended to provide more flexibility to these ascriptions. They also include sub-categories that refine the ascribed function, although these sub-categories are not presented in this article. For example, 'jewellery' is a sub-category of 'dress'. Pendants belong to the sub-category 'jewellery', but some of them could have been either for the adornment of horse harness, armour or worn as pendants round the neck. They are therefore classified as the queried category 'dress?/combat equipment?/horse equipment?'. This means that, for example, one can collect all the material that can clearly be identified as 'dress' and the material that could conceivably have been dress ('dress?'), both together and separately (see Section 8.2).

3.2.2b Age and gender categories [Figure 11]

Gender and status are more subjective categories. Most artefacts cannot be ascribed either a gender or a status category but in many cases it is possible to suggest these. For example, tools used for wood-, leather-, stone- and metal-working, and assault equipment, such as arrowheads, and remains of javelins, daggers and swords, as well as remains of armour, are more than likely to have been used by males. For items used for cloth-production this is less certain. Weaving could be carried out by men or women (Dixon 2001, 117-29) and sewing could be carried out by soldiers (Allason-Jones 1995, 28; see Section 8.3). Thus, many cloth-production artefacts have been classified as 'male?/female?'. However, those used for spinning have been classified as female as, to date, I have no evidence that men are involved in spinning in the Roman world (although cf. Nandris 1981, 251). Some items of dress can also be ascribed a gender category. For example, Böhme (esp. 1972) has argued that some fibulae (brooches) were most probably worn by women. On the basis of modern data for ring sizes and the knowledge that Romans wore rings on all fingers and all joints, Alex Furger (1990) has done a statistical study of the size-ranges of different types of rings from Augst and Kaiseraugst, in Switzerland and provided estimates for the range of ring sizes that would have been worn by men, women or children. Similarly, certain types of beads could have been worn by women, children or horses and so these are given queried gender and status categories (see Section 8.6).

3.2.2c Status categories

In the examples included in this article status does not really feature. For the project as a whole, many items are categorised as 'utilitarian' and some are either 'decent' or 'luxury'. These terms are based on those used for 19th century European middle classes where 'decency', or 'gentility', pertains to the items through which the middle classes aspire to imitate the upper classes (McKendrick et al. 1982, esp. 1-6; Young 2003). While this analogy is not quite appropriate for this study, 'decent' items are those which can be seen to be beyond 'utilitarian' because they are decorative (e.g. a harness pendant, or decorated belt plate) or have some level of decoration (e.g. a fibula with an incised pattern) or are a fine ceramic (e.g. thin-walled pottery or undecorated terra sigillata). 'Luxury' items include precious materials (e.g. a fibula with silver pearled wire) or fine workmanship (e.g. relief-decorated terra sigillata vessels).

3.2.3 Summary

This article involves a network of interrelated functional and gender categories which can perhaps be seen as a closer approximation of reality than more narrowly assigned categories. It by no means includes the complete range of artefacts and activities at Vetera I, or all the permutations within those selected. Rather it uses a sample of the analyses to demonstrate the approach. At this point it is also important to state that these functions are invariably ascribed to fragments rather than to whole artefacts. Individual fragments are likely to be of little significance but the consistent distribution patterns of a number of fragments are likely to be more meaningful.