A Foot in Both Camps: The Civilian Suppliers of the Army in Roman Britain

The civilian attendants and suppliers of the Roman army—those who were attached economically and voluntarily to the Roman military—are an important and understudied group. This article explores their roles and experiences during the Principate in Britain. It is argued that civilians were deeply involved in the functioning of the peacetime army’s supply-network; that these roles offered a scope for significant prosperity; and that the service community became increasingly integrated with local areas and incorporated many local agents. Further, the article argues that, in self-conception and in practice, they inhabited both ‘civilian’ and ‘military’ communities, the distinction between these two being weaker than is sometimes assumed. Finally, by examining the material in the light of globalisation theory, this article suggests that civilian attendants and suppliers can be seen as both ‘local’ and ‘global’, while also arguing for a greater appreciation for variability of experience among those traditionally seen as representing ‘global’ interests.


Introduction
This article investigates the roles and experiences of the civilians who were economically and voluntarily attached to the Roman army in Britain. 1 They constitute an understudied group, normally treated by scholars in passing, or in piecemeal fashion. This group included camp-followers/sutlers, contractors, long-distance traders, skilled manufacturers, and local producers. These may seem like neat, too-modern divisions, and there is indeed a danger here of minimising the ambiguity of the evidence. In many cases, we will be speaking about evidence for activity, rather than evidence for people. For example, it seems likely that some of the mercantile activity discussed may have been undertaken by the relatives of soldiers; i.e. their economic connection to the military is additional to their family connection. Similarly, some merchants may have been veterans. However, my primary interest is in the contexts in which these actors exhibit their economic connections to the army. As such, while acknowledging difficulties, I believe that ' economically and voluntarily' attached to the Roman army is a valid descriptor for the object of my study.
Furthermore, the use of the word ' civilian' with regard to army suppliers, and specifically the inhabitants of vici and canabae, may be regarded as problematic. 2 I will defend its use here, with the disclaimer that I do not mean to imply any kind of strict separation between military/civilian in these communities, and indeed that issue is discussed at length in the final section. In this article, ' civilian' is used solely to denote individuals who were not military personnel, i.e. who were not part of a military unit. The term ' civilian' is widely used in discussions of the communities explored here, and taken narrowly, I regard its employment as accurate. This of course does not mean that someone described as civilian was free from military association.
The basic presence of this group is undisputed by historians and archaeologists, but the scope of their activities has been unclear as they have rarely been a primary focus of scholarly discussion. Broad studies of Romano-British society have devoted attention to them as part of the Roman military economy (e.g. Mattingly 2006: 170ff), while examinations of specific industries, for example relating to textiles (e.g. Wild 1979; and food supply (e.g. Gerrard 2008) have necessarily touched upon my subject. Where these civilian communities themselves have been specific research focuses, this has either been on site-specific bases, often in archaeological reports (e.g. Buxton and Howard-Davies 2000: 414-415;Greene 2013;Trezzi 2017), or with a view to illuminate other issues, such as the operation of the ancient economy (Evers 2011). Therefore, while valuable, previous scholarly examinations have either been ungeneralised (focused on single sites or industries), or not specifically interested in the civilian groups themselves (treating them tangentially). A short article by Casey (1982) provides one example of a generalised, sustained discussion of civilians linked to the army in Britain. However, his interest was in the vicani, thus including soldiers' family and slaves, and excluding economic actors who did not live in extramural settlements. In addition, the great deal of archaeological evidence uncovered over the past three decades, as well as the exceptional discovery of the Vindolanda tablets, allows us to suggest answers to questions which could previously only be posited. I therefore believe there is value in treating these people as a coherent group, in a thematically and geographically integrated fashion. This article is thus intended as an interpretative essay exploring this group and their experiences. My geographical focus is the northern frontier of Roman Britain, and my temporal focus is the Principate, from the first to the late third century ad. These boundaries are not immutable, and comparative evidence shall be used where expedient. While my research subject (civilians supporting the army in northern Britain) is broad and includes people living all over Britannia and, indeed, outside it, much of my discussion will focus on the inhabitants of extramural vici and canabae, particularly those on the northern frontier. This is where the clearest evidence comes from, and where the association between the archaeological record and the military supply industry is most certain. As well as archaeological evidence from these sites (e.g. Manchester, Brough-on-Noe, Vindolanda, Housesteads, Newstead, Bearsden), we will closely examine tablets from the Vindolanda archive which reveal evidence for the activity of the research subjects.
After a brief discussion of the Roman military economy, we will examine the functions that evidence suggests civilians performed for the army. Next, we shall investigate the likely status of this group, and its ethnic makeup. We shall then explain their relationship to the military, and some aspects of identity and community construction.
Finally, an introduction to the theoretical framework I have adopted for my analysis is necessary. The role of the Roman army in Britain has long been an important battleground for theoretical debates in Roman archaeology, but again, comparatively little attention has been paid to the civilian elements of the military community from this angle. Today, most scholars agree that Romanisation is an inadequate framework, failing to explain a number of important phenomena. For example, it has no answer to the significant differences in local reactions to Roman influence, which imply active roles on the part of the ' assimilated' communities (Pitts 2008: 494), nor to the equally varied ways Roman actors engaged with local groups, which show that they, too, had a more complex understanding of their role and influence than as purveyors of a monolithic cultural force. Finally, Romanisation's inherent 'Roman/native' dichotomy ignores the fact that individuals never appear to be totally 'native' or 'Roman', but can exhibit both of these identities-and more-without contradiction (e.g. Laurence 1999: 388; see also Gardner 2013: 3-6).
The framework that, in my view, has come closest to reconciling these problems is the application of globalisation/glocalisation theory to the Roman world (Pitts 2008;Pitts and Versluys 2015), a development based upon the concept of connectivity (Horden and Purcell 2000). Connectivity stressed both the long-term (and largely unguided) currents of cross-cultural contact, but also the significant eddies within them, diverted by regional variation. Globalisation and its related frameworks preserve local agency and the importance of varied social practices, without denying or ignoring the dominant role of a generally homogenising and integrative cultural force (here, Rome). Globalisation also challenges Romanisation's privileging of material culture, by emphasising 'all forms of connectivity, be they cultural, economic or otherwise' (Pitts 2008: 504). Such contributions are especially relevant here, when talking about individuals-from the wider empire, and from Britain-who became attached to the Roman army through economic connections. Morley (2015) has developed this idea further, stressing a practice-based understanding of cultural change based on the adoption/imposition of, and admission to, networks and standards. As a research framework explicitly informed by our experience of the modern world, globalisation theory risks anachronisation, as Hingley (2014) rightly points out. Gardner (2013: 8-9) warns that globalisation may be inadequate to examine the 'Roman past as a distinctive social world'. At the same time, this is true to one degree or another for any research framework. Properly acknowledged, the contemporary relevance of globalisation as a concept is in fact vital to the pertinence of the observations it helps to generate -which Hingley also acknowledged. In any case, it is not necessary to argue that the Roman world experienced globalisation in a similar manner to the modern world to insist of the applicability of the concept (Hodos 2015: 252).

The Military Economy
Before arriving at our subject proper, it is worth explicitly discussing the Roman economy, and its operation in the 'military zone' of Roman Britain, to make the assumptions of our later analysis clearer. This article concerns economic actors, and so it is important to have a sense of the system they acted within. Moreover, particularly when dealing with concepts such as globalisation there is a danger of applying anachronistic economic ideas to the ancient world; this must be explicitly addressed.
While the extremely 'primitive', embedded economic model advanced by Finley (1973) is today largely discounted, there remains a lively debate over the degree to which market mechanisms operated in the ancient Roman economy. This question is especially important regarding the application of globalisation theory to the Roman world, as the modern experience of globalisation is intertwined with that of international free-market capitalism. Questions of status and social obligation clearly played major roles in Roman economic relations, and political factors and the ' command economy' were significant also, especially in heavily garrisoned areas with (initially) poorly-developed systems of exchange, such as northern Britain. Whittaker (1994: 106-107) is undoubtedly right to problematise an unthinking assumption that army supply was regulated by the free market. That said, there is no necessary conflict between an acknowledgement that the ancient economy-particularly in military zones of the Roman empire-was very different from the modern idea of a free market, and the belief that certain market mechanisms were in operation, or that there was a significant degree of market integration across the empire (e.g. Temin 2001;Pitts 2008).
The army was evidently the major economic force in the frontier zone, and a significant driver in the adoption of Roman currency and forms of exchange. Creighton (2014) demonstrated that new coinage entered Britain mainly through the military economy, as pay for soldiers, some of which would also have gone to our suppliers, either in contractual payments handed out by units or the procurator's office, or in less official, more day-to-day transactions between soldiers and merchants. The command economy would also have encompassed land owned by forts; supplies for the army raised directly by tax-in-kind; and long-distance imports subsidised or managed by the state. This last point had a particular effect, it has been argued, in that shippers were able to take advantage of provincial import tax waivers to sneak commercial products into Britain alongside official supply cargoes at a discount (Whittaker 1994: 111-112). However, the fact that this can be regarded as commercially advantageous suggests the existence of more-or-less free markets through which traders sold items. Looking at the evidence from Vindolanda, there are some items that could be part of this hypothesised parasitical cargo-spices ordered through London, for example (Tab. Vindol. 588)-but there is also reference to items such as apples and eggs for sale at market, which must be from the surrounding area, and suggest the vicus' role as an exchange centre for the locality (e.g. Tab. 302). We also see grain arriving at the fort and being distributed to soldiers (Tab. 649, 180), which could have been collected directly as tax, or supplied indirectly, such as the 5,000 modii handled by Octavius and discussed below (Tab. 343). In this case, the responsibility for collection seems to have been contracted to an entrepreneurial intermediary who then drew on commercial contacts within the province. In general, I see no problem in speaking of a market, or markets, operating in Roman Britain, so long as it is borne in mind that integration was often politically driven. Evers (2011: 44) suggests the helpful idea of a tripartite economic system: a political economy, a market economy, and an embedded economy, which all interlink and overlap. The political economy existed primarily for the provision of official supplies for the army, which was achieved if necessary through tax receipts and subsidised long-distance imports, but 'intersected with a market economy which provided goods that could be obtained more cost-efficiently from local sources'. Finally, the mode of individuals' economic participation and their purchasing power was governed by social considerations, e.g. rank, status, ethnicity, religion, and ' communal obligations', constituting an embedded economy. As such, although local markets can be seen as embedded to a certain extent, they are in fact connected and integrated, partly by the political economy. This is not dissimilar to Temin's (2001) model of connected, localised markets forming a Mediterranean-wide network. Furthermore, to the 'social considerations' listed above might be added 'practices', such as the Roman usage of coinage, credit, and financial documentation, which possess both economic and social significance. This was, again, at the same time embedded in the local (Romano-British) context, and yet also part of an empire-wide discourse which the political economy in general and the army specifically helped to import and modify. Therefore, without devising an anachronistic empire-wide capitalist free-market economy, we can assume and demonstrate both the operation of market mechanisms, and the role of global processes and connectivity.

Roles: Occupations and Functions
Before proceeding, we must first establish the degree of civilian involvement in peacetime army supply, and the roles voluntary suppliers performed. There is no question that merchants sold ancillary items to soldiers, and luxury supplies, especially to higher-ranking officers. It is in this capacity that camp-followers, lixae, usually appear in ancient literature. There is not space to present all of the evidence for mercantile activity in military contexts here, but see in particular Bowman's (1994) and Ever's (2011) discussions of the Vindolanda material, where the ink tablets demonstrate the operation of a significant cash-economy in which exotic foodstuffs and spices, local produce, and trinkets were traded. Civilians also catered for the military community's leisure wants, demonstrated by taverns located at Caersws (Britnell 1989: 2-3) and Housesteads (Allason-Jones 2013: 73), replete with gaming counters and dice; a further example was tentatively identified at Vindolanda (Birley 2013: 90). Prostitution is assumed to have occurred at military sites, and is occasionally referred to in the literature (e.g. Valerius Maximus, Facta et Dicta Memorabilia 2.7), but it is more difficult to establish archaeologically.
Pottery production is also undisputed, with imported wares at forts giving way to local produce over time as locals attuned themselves to new markets (Whittaker 1994: 103). Occasionally we can discern the trail of individual potters chasing military money, such as Sarrius at Bearsden, to whom we shall return. Several kilns in or near vici and canabae have been identified (e.g. Greta Bridge (Casey and Hoffman 1998: 179) and Vindolanda (Birley 1977: 76)). The military market also stimulated production in the wider province, rather than just the forts' localities, as the success Black-Burnished Ware (BB1) in the late first-to-second centuries ad shows: a pre-Roman production tradition that found new life, possibly as beneficiaries of supply-contracts awarded by procuratores (Allen and Fulford 1996). While most of this potting relates to ' domestic' wares for food preparation and consumption, several civilian stamps on tiles from military bath-houses and structures inside forts show private potters discharging military contracts (Brough-on-Noe: RIB 2489. 10, 2489.55;Lancaster: 2489.56 A;Caerleon: 2489.65;Maryport: 2474). At Tarbock (Merseyside), Viducius, a contractor, the relation of another man attested as a negotiator Britannicianus, 3 produced tiles for the legion at Chester in the second century ad. Tight quality control and stock auditing were suggested by date-stamps on the tiles, and by a number of rejected examples with abnormalities. The short-lived operation was part of a 'privatisation of legionary tile-production' during the second century (Swan and Philpott 2000: 59-63).
These examples start to show that civilians did not just cater to the ancillary needs of units, but to more general supply. Similarly, we can demonstrate the involvement of civilians-most, probably, living in vici and canabae-in producing and maintaining equipment. The evidence for industry in extramural settlements has often been downplayed (e.g. Casey 1982), and the fact that soldiers did much of this work has diverted scholarly attention from the extra-military contribution (e.g. Bishop 1985). 4 However, the record is quite firm that civilian involvement in equipment production and maintenance was commonplace.
The industrial functions of extramural settlements have long been posited by some, especially at Vindolanda, where Birley (1977: 76) noted evidence for metalworking, and the famous dedication by the vicani to Vulcan implied that this was an important part of the inhabitants' self-representation (RIB 1700). This compares well with a similar dedication from the Old Carlisle vicus (RIB 899). Strip-houses, the rectangular, multi-function structures with short sides facing the main road, ubiquitous in vici, have been interpreted as mixed industrial/commercial and residential structures (Sommer 1984). Recent excavations have provided additional evidence supporting this interpretation. At Vindolanda, excavations uncovered a significant workshop-district with a common drainage system near to the above-mentioned dedication (Birley and Blake 2005: 49-59). Over 200 kg of iron slag was recovered from the Ribchester vicus (Buxton and Howard-Davies 2000: 337), and Manchester yielded a large second century ad industrial suburb (Gregory 2007: 185-186). 5 Such finds are common, despite the small proportions of these settlements generally excavated.
That this production was not only ' domestic', or destined only for use in the extramural settlements themselves, is implied by the regular discovery of military items in such settlements, sometimes multiple items in the same location. 6 Occasionally items bear civilian makers-marks (e.g. RIB 2426.3, a shield-boss from London). Moreover, there is reference to trade in equipment such as spears in the Vindolanda corpus (Tab. 861). The connection is cemented beyond doubt by a discovery at York, where a tent-fragment was recovered from an industrial district of the civilian settlement associated with abundant evidence for metal and leatherworking. The scrap was inscribed with a reference to Sollius Julianus (RIB 2445.16), a centurion attested on a Hadrian's Wall building stone (RIB 3454) from the Legio VI, based at Eboracum in the second century ad. A sword and several military fittings were found around the same location (Ottaway 2004: 102). It seems that civilians were producing, and at the very least maintaining and repairing, military equipment here.
As illustrated by the find described above, civilians did not only manufacture and maintain metal equipment, although these objects are the most archaeologically evident. For example, Driel-Murray (1985) detected a progressive heterogenisation of second century ad military footwear, which she suggested was due to the increasing outsourcing of production. Indications of tanning in the Manchester and Vindolanda vici might also be related (Birley 1977: 76;Gregory 2007: 186). Several tablets from Vindolanda seem to be merchants' accounts containing cloaks (Tab. 207,296,861); these are probably local products, but the possibility of production from further afield is raised by Clodius Super's statement that 'Valentinus' has approved of clothing ' on his return from Gaul' (Tab. 255), reminiscent of Hunt's Pridianum's record of soldiers sent to Gaul from Pannonia to purchase cloaks (RMR 63). Allason-Jones (2002) has commented on the certainty of local clothes supply at Vindolanda, the quality of which has been described as ' excellent' (Wild 2002: 32), and a number of spindle-whorls have been recovered from vici. 7 While Wild (2002) favoured requisition as the supply-mechanism, the reference to clothing in the Vindolanda tablets, the level of workmanship, and the continuing importance of the British textile industry into the later empire (receiving prominent mentions in Diocletian's price edict) point to a specialised industry. Given that soldiers in the Principate nominally purchased their own equipment (Speidel 1992: 131-136), it is not surprising that there should be significant scope for private manufacturers in extramural settlements to establish themselves. It may well be that much textile work, especially repair and maintenance work, was done by soldiers' families living in extramural settlements. Family 'back home' could also certainly have contributed: as noted in the Vindolanda letter recording that socks, sandals, and underpants have been sent to a soldier (Tab. 346). That said, the connection is largely suppositional and does not in any case dent the impression of a specialised industry in operation.
Turning to staples, traditional explanations for the grain supply of the Roman army have been found in compulsory methods such as taxation and forced purchase, supported in the British case by Tacitus' famous passage where Britons are forced to sell grain to the state (Agricola, 19). Subsequently, more complex models of frontier supply have been advanced by scholars. Whittaker (2004) emphasised a mix of tax-in-kind and wholesale purchase of grain, initially involving staple shipment from the continent, and later importation from the south of Britain to the frontier zone. Similarly, Gerrard (2008: 123) explains the 'irrational' distribution of Black-Burnished 1 (BB1) ware through a mix of voluntary and involuntary supply mechanisms. Kolb (2002) discusses the Roman state's use of private enterprise for official supply and transport. However, although they are important in beginning to suggest the complex role that civilians played in military supply, these arguments tend to underestimate the importance of local production sources. Archaeologists have long emphasised the likelihood that northern forts had a local supply of grain (e.g. Higham 1982), supported by pollen studies which have questioned the strict lowland/highland, cultivated/pastoral dichotomies which underpinned traditional theories (Van der Veen 1992: 159). Characterisations of the frontier as difficult highland underestimate the complexity of the landscape; there are in fact 'lowlands' and fertile valleys in the frontier zone, and the barrenness of upland areas should not be exaggerated.
On the other hand, suggestions of local production sometimes lead back to arguments emphasising tax-inkind, leaving little room for civilian actors; some archaeologists have endorsed the idea of military-controlled land for grain production (e.g. Higham 1991;Huntley 2013). Against this, and working from the Vindolanda corpus, Evers (2011: 25-26, 42-49) posited a nuanced sketch of a complex logistical system, incorporating multiple modes of supply, including local production, staple importation, forced purchase/taxation, and voluntary supply methods. Indeed, a close reading of Octavius' letter from Vindolanda (Tab. 343) supports and advances Evers' reading, and demonstrates that, even early in the province's development, civilians were closely involved in wholesale supply.
The letter, found inside a probable centurion's-quarters of the early second century ad fort at Vindolanda, concerns the provision of several supplies. Most significant is the fact that Octavius has acquired 5,000 modii of grain, for which he now needs 500 denarii from Candidus, the recipient. There is no question that this must refer to the regular supplies of the unit, and not ' ancillary' supplies as Whittaker (2004: 106) suggests; this amount would feed 1,000 men for a month, or the 296 soldiers listed as present in Tab. 154 for over 100 days (based on consumption of a 1/6 modius per-day, per-soldier), as Evers (2011: 48) recognised. However, against Mattingly's (2006: 512) supposition that Octavius is likely a soldier, and Bowman and Thomas' (http://vindolanda.csad.ox.ac.uk/TVII-343) comments that there is 'no way of deciding' his status, I believe it is possible to suggest his status as an entrepreneurial contractor. Octavius is market-conscious, complaining that Frontinius is reselling his leather at high prices. Furthermore, the statement that if Candidus (who may well be a soldier) does not send money, Octavius will lose his deposit and he will be ' embarrassed'-erubescam-suggests he is an entrepreneurial civilian who stands to lose personally if he cannot pay, and will lose face with his commercial partners. He was not acting directly as the state, and failure falls upon him. Finally, it is telling that although he uses some military constructions (e.g. fratri suo salutem, 'greetings to his brother'), 8 this has its limits, when he writes that contubernalis Fronti amici hic fuerat, ' a mess-mate of [our] friend Frontius was here'. Soldiers have messmates, but civilians do not, so it is notable that neither Frontius, nor the unnamed other soldier, are Octavius' messmates. Rather, Frontius, a soldier, is simply an amicus, and his friend is a contubernalis. Octavius appears a confident entrepreneur who has managed to capture a lucrative contract; he identifies with the army, and may have military connections beyond his work, but nonetheless signals that he is aware of his distinct status.
The letter has more to tell us about civilian involvement in supply. The fact that if Candidus does not send money, 'I shall lose (perdam) that which I gave as a deposit (arre dedi), about 300 denarii' -reveals the exchange referenced here is no tax, forced purchase, or simple collection from a central storehouse. He is buying the grain, wholesale, likely from a single estate. That a deposit has been laid down, and he is already in possession of the grain, probably indicates he has a history of relations with this estate, helping to explain his worry about ruining an economic partnership. Moreover, as there is no reference to waterborne transport (only roads), and the place-names given are local (e.g. Cataractonium, Catterick), it seems reasonable that this purchase is occurring in the relative vicinity of Vindolanda, somewhere in the frontier zone. Birley (1997: 276) suggested that wheat mentioned in the tablets may have been grown in nearby areas such as the Tyne Valley; support of this proposition comes from the fact that in sites along Hadrian's Wall, spelt wheat is often recovered along with the chaff, indicating relatively local production (Huntley 2013: 46). Moreover, grain assemblages in the third century ad South Shields fort were at least partially local (Van der Veen 1992: 159). This may give us a context with which to understand the intensification of agriculture in the frontier zone which scholars believe has occurred (Jobey 1982: 15;Woolliscroft 2005: 228;Evers 2011: 13). This is also suggested by an increase in the number of attested farmsteads in the second century ad frontier zone (Brindle 2016: 313-315), as well as an apparent reduction in agricultural production in the hinterland of Hadrian's Wall between the fourth and fifth centuries ad, possibly reflecting the withdrawal or reduction of Roman forces and the collapse of the supply economy (Dark and Dark 1997: 143;Dark 1999: 265-266). 9 This does not match Tacitus' account, but he refers to wartime when the frontier was being extended, rather than to settled peacetime when it would make sense for the army to cultivate more constructive links in its garrison zone. Moreover, Tactius' rhetorical purposes with Agricola need to be considered; his forcedpurchase passage (19) is not intended to give a realistic account of army supply, but to demonstrate the humanitas of his father-in-law. While the lack of 'Romanised' objects on local farmsteads is often marshalled by historians to suggest a lack of Roman-local exchange, this is partly due to Romanisation's privileging of material culture. The narrowness of such a view is suggested by Tab. 213, a letter to Cassius Saecularis, apparently a military official involved with supply. He is instructed to interface (interpreteris) with some likely-native group, 'so they may have barley as merchandise from you' (ut hordeum commercium habeant a te), demonstrating the ability of Roman officials to cultivate links with groups who were not extensively integrated into Roman commercial and material culture networks.
This all suggests that civilians not only catered for the ancillary needs of soldiers, but were often involved with the procurement, manufacture, and provision of regular supplies. Civilians were deeply integrated with the functioning of the peacetime army. Moreover, it seems that their contribution to the military economy in Britain increased in the second century ad, which accords with suggestions below that during this period more native British joined military communities. As such, the military in northern Britain acted as something of a bridgehead for connectivity, encouraging the integration of the region into globalised networks, and the diffusion of material culture and practices. Haynes (2013: 184) sees soldiers and merchants as partners in imperial expansion. A prime example is the second century ad development of a network of productive towns in the hinterland of the legionary fortress at Chester (Ward et al. 2012: 413-414). A wellfunded institution, stationed in a poorly-developed region, the army of northern Britain was well placed to stimulate connectivity, and the end result was that it was able to source supplies more economically by drawing on the immediate locality, rather than being forced to continually subsidise long-distance imports. Purchasing large amounts of grain from nearby estates would also have been advantageous compared to collecting tax-in-kind as supplies. It seems unlikely that tax rates (usually assumed to be in the region of 10-20%) would have varied considerably within provinces. Relying on tax-in-kind for army supply would, therefore, have required significant grain transport, as well as official manpower for collection and storage, since the army's garrison was not evenly distributed throughout the province. However, if army units and provincial commands were provided budgets in coin from empire-wide tax revenues, they would be able to purchase large amounts of grain (i.e. more than c. 20% of the yield) from nearby estates. Octavius' letter, as well as the evidence for agricultural expansion in the frontier zone during Roman occupation, and its potential retreat in the fifth century ad, suggests that this is exactly what was happening. We see, then, the dual advancement of local and global interests, of which our research subjects are both beneficiaries and agents.

Status(es)
Having demonstrated the breadth of roles undertaken, we will now examine how those roles may have been viewed in Roman society, and the scope for advancement they offered. Civilians attached economically to the army are rarely mentioned in ancient literature, appearing most frequently in accounts of armies on the move, as lixae, the camp-followers or 'sutlers' who trailed forces in the hope of monetary and material reward. Roman authors normally display unfavourable attitudes towards these groups. Working from Republican sources, Vishnia (2002) argued the earliest civilians attached to the Roman army were slavers. However, they also appear in imperial literature; lixae follow the armies of all principal actors in Tacitus' Historiae, joining the Flavian forces in the sack of Cremona (3.33). He compares the unruly lixae to slaves (2.87), and historical anecdotes often seek to position lixae as a destabilising and dangerous element, defined against the army's better order. Thus, Sallust (Bellum Jugurthium, writes that Metellus Numidicus, in the Jugurthine war, forbade lixae from selling bread and ' cooked food' (cibum coctum) to soldiers and banned them from the camp, and Valerius Maximus (Facta et Dicta Memorabilia, 2.7) recounts that Cornelius Scipio expelled a great number of lixae, whom he lumps in with prostitutes (scorta). He characterises these undesirables as turpi atque erubescenda sentina, ' disgusting and shameful bilge-water', and the army only regains its discipline with their removal.
Further, lixa often appears as a negative epithet, implying bottom-feeding and parasitical following. Livy (21.63.9) writes that C. Flaminius, in acrimony with the Senate, quit Rome 'in the manner of a lixa (lixae modo) without insignias, without lictors', as if going into exile. In Seneca's Phoenissae (595-599), Polynices, bemoaning his status, asks if he is to be ' a wife's mere chattel … and, like a humble (humilis) lixa, attend upon her domineering father?' Finally, the late antique poet Ausonius (Ordo Urbium Nobilium,9), addressing cities as characters, says that Aquileia's greatest deed was its defeat of Magnus Maximus, armigeri quondam sub nomine lixa -' an erstwhile lixa acting as a warrior'. Maximus the usurper was not a lixa, but a commander of some stock, being Theodosius the Elder's nephew. This is a satirical insult, lixa suggesting a leech-like, follower character, sub nomine of a leader, because ruling is not his right. These references are separated by genre and time, and demonstrate an elite disdain for camp-followers. It seems the image of lixae as freeloading low-lives, living out of soldiers' pockets, was a well-known trope.
Modern scholars have sometimes followed this characterisation, minimising the civilian attendants' supply functions, and emphasising the apparently ram-shackle, slum-like appearance of Romano-British vici and canabae, and the dirty activities that occurred within. 10 But without denying some reality to these statements, it should be noted that suggestions of prosperity in vici and canabae are frequent, and their material culture is regularly characterised as rich (e.g. Britnell 1989: 126;Millett 2005: 40;Allason-Jones 2013: 76). Several examples of large, impressive buildings suggest the scope for social and financial advancement afforded by serving the army was significant. 11 Good water supply and drainage has been oft-noted at Vindolanda, indicating relatively sanitary conditions (Birley 1977: 77;Birley and Blake 2005: 11, 25). Further, Raybould (1999: 159-160) has shown that those connected to the military unsurprisingly exhibited aboveaverage literacy. The evidence demonstrates that many of these people had decent standards of living.
That camp-followers could attain prosperity should not be shocking. After all, the rewards must have been enticing enough to endure regular uprooting. RMR 68 suggests that, in 81 ad, legionary soldiers were left with about 40 denarii annually as pocket money, after essentials had been provided, deductions taken, and deposits made-about 18% of the total stipendium. Legionary pay increased to 300 denarii under Domitian, with auxiliary pay 5/6 of this (Speidel 1992); assuming the proportion of 'pocket money' allowance was similar, auxiliary soldiers contemporaneous with the Vindolanda tablets would have had c. 45 denarii annually to spend as they wished. This amounts to 22,500 denarii for a 500-strong cohort, and 45,000 for a milliary cohort. A legion of 5,000 men in the same period could spend over a quarter-million denarii annually in its canabae. With the additional prospect of travellers, and military contracts, we can see how attractive extramural environments would be to traders. The Vindolanda tablets support the traders' prosperity: Tab. 181's author has earned 54 denarii, that of 182 earned 89, and Tab. 184 records over 75 denarii of business. If Tab. 596 is civilian-penned, an incredible amount of business has been done-well over 500 denarii-which recalls the sums Octavius was dealing with (Tab. 343). The situation became more lucrative during during the second century ad: RMR 70 shows that, in 192 ad, auxiliaries were now receiving 70-80 denarii per stipendium (240 a year) in cash, with compulsory deductions (and therefore probably official rations) having been drastically reduced. A milliary cohort could now spend over 200,000 denarii a year in its vicus. This, along with pay increases under Severus and Caracalla, perhaps gives us a context to understand the striking growth that vici underwent in this period (Bidwell and Hodgson 2009: 30-31).
This general picture is illustrated by two case-studies. The first is C. Nipius Ascanius, known from two inscriptions on lead pigs. The earliest is from Hampshire, stamped under Nero (RIB 2404.3); the lead is marked as the emperor's property, from the imperial estates, and Ascanius appears an imperial agent. The second pig, from Hampshire and likely of Flavian date, bears a private stamp simply recording it as Ascanius' product (2404.38). Ascanius is a rare cognomen, and the editor's note the only other known epigraphic example was a freedman from Verona (CIL V, 4008), raising the possibility that our Ascanius was also of servile origin. 12 This seems a fascinating glimpse into upward mobility-initially a slave, by 60 ad, Ascanius was an imperial agent overseeing a mining operation. Later, he has the wealth to manage his own venture, having moved northwards with the army's advance. 13 The other is a second century mortaria potter, Sarrius, known from 125 potters-stamps. In the 130-140s ad, he achieved a dominant position in the Wawrickshire potteries, and subsequently set up subsidiary workshops at Rossington Bridge, possibly manned by two lower-ranking potters on his behalf (Buckland et al. 2001); at another, unknown site in northern England; and at Bearsden fort on the Antonine Wall (Breeze 2016: 137-140). These workshops were active simultaneously, and the Bearsden workshop was apparently established contemporaneously with the fort itself. Both examples demonstrate the opportunities that military association brought to energetic entrepreneurs, and Sarrius' story suggests a degree of organisational coordination with military units. It must have been communicated to him, by some mechanism, that Bearsden was under construction and would require pottery supply, or at least, that an opportunity existed. Perhaps he had dealt with the garrisoning unit at its previous base. His workshop at Bearsden produced a large amount of pottery and seems to have been active throughout the fort's occupation.
It is clear that the literary tradition is misleading, through omission and misrepresentation. A fuller appreciation of our research subjects reveals that many enjoyed scope for significant success, afforded by military association. Their close alignment with the most visible arm of the Roman state placed them in a privileged position, and coffered a reasonably high level of physical and social mobility.

Origins
Can we reconstruct the ethnic origins of our research subjects? This is an extraordinarily difficult question, but again, the best way to attack it is to examine extramural settlements for clues. For Salway (1980), vici were cosmopolitan communities, while others have posited the exclusion of locals from the voluntary supply industry (e.g. Higham 1991). While specificity is impossible, we can sketch a general picture broadly supporting Salway's contention, while not ignoring the potential for exploitation and unequal power structures. The discussion of this point is important to the application of globalisation theory. We have already seen that civilians attached to the army were relatively privileged, and undertook important work; however, if we have here only an exclusive group of international traders and contractors, then the process is one of domination and exploitation by Roman interests, rather than anything more complex. If, on the other hand, local British agents were an important part of the voluntary supply industry, then the suggestion above of the progressive integration of local agents into global networks and practices, and the way our research subjects represented both local and global interests, can be demonstrated further.
It is most likely that in Roman Britain the earliest inhabitants of vici and canabae had followed the unit from its previous base. This is implied by the early constructions of military installations and extramural settlements (Sommer 1984: 7-9), the significant differences in architectural styles between northern vici and local farmsteads (Sommer 1999: 89), and the lack of native pottery in early periods (Webster 1990: 138-139). This would have meant large numbers of continental Europeans entering the province, and such an influx would not be confined to the first century ad, as new units were periodically transferred to Britain from other provinces throughout the period.
Likewise, pottery imports from Gaul and amphorae from Spain demonstrate that the rewards for supplying the Roman army in Britain were initially captured by those in and from older provinces (Fulford 1991: 36-38). The presence of continental immigrants is explicitly demonstrated by several pieces of epigraphical and documentary evidence. German individuals are attested in the frontier zone (e.g. Lurio, recorded as Germanus at Chesters (RIB 1483), and Maduhus at Carrawburgh (RIB 1526)); a Greek doctor demonstrates the arrival of foreign medical practices (RIB 461); and the trader Barathes had travelled all the way from Palmyra to South Shields (RIB 1065). 14 A trader at Vindolanda furthermore states that he is also not from Britain (Tab. 344). From the perspective of frontier locals, Roman arrival would therefore have felt very exclusive-imperial power was propagated by a large, foreign army, followed by an equally foreign conglomeration of traders, contractors, and family members who captured the lion's share of the early market. It is easy to imagine a two-tier economy in operation during the first century of Roman rule (e.g. Whittaker 2004), where there was little meaningful interaction between Roman and local groups.
In this way, the army's civilian attendants clearly represent global interests, especially in the first and early second century ad, and were some of the most effective agents of Roman imperialism. Following an army of conquest, they took with them an alien material and architectural culture, arriving with a foreign way of practicing business: coins, credit, contracts. They helped to extend the tendrils of long-distance trade over Britain into a 'global' trade of oil, wheat, wine, and more. Equally, the early benefits of this trade were reaped by those economic elites already plugged into these global networks of exchange, and who exhibited the standards that allowed them participation therein. Verboven's (2007) interpretation of the negotiator inscriptions along the Rhône-Rhine axis as signalling a ' class identity' among traders discharging military supply contracts suggests that this role as agents of the (Roman) global economy was understood, if implicitly, and contributed to their conception of their place in the world. In turn, this suggests a complication to Morley's (2015: 55) understanding that globalisation was an undirected process, and that economic impacts were purely side-effects of political and social developments. Our research subjects were not merely passive tools of an equally unthinking process, but played active roles in the extension of it.
However, there is reason to believe that as time went on, vici and canabae increasingly attracted locals, and cultivated links with frontier zone communities. We have already argued that British supply sources were progressively drawn upon. Archaeologists have reported on architectural evidence suggestive of 'native' integration into extramural settlements. At Ribchester, a likely military-run workshop, between fort and vicus proper, was constructed using local techniques, 'implying the use of native workers' (Buxton and Howard-Davies 2000: 415). In the Greta Bridge vicus, a large timber building exhibited walls of 'military construction', but the lack of corridors, and the use of a central kitchen as a focal-point incorporated 'traditional elements of indigenous social relationships', suggestive of locals inhabiting the settlement, negotiating with Roman building styles (Casey and Hoffman 1998: 129). In the Caerleon canabae, Evans noted buildings of 'vernacular' construction near the settlement's edge (Evans 2000: 470-472). Dearne's (1991) analysis of viciclosure in the Pennines found that the settlements must have had alternative economic links to the military ones, and were locally integrated to some extent.
A number of other phenomena also suggest cross-cultural contact, such as Roman-style dedications to local divinities and examples of religious syncretism at military sites (below), even if we cannot say that dedicators to Brigantian gods were necessarily Brigantian. Further, while we can speculate that much of the local produce mentioned in the Vindolanda tablets was grown by local farmers and brought to the vicus market (Higham 1982), we can be more certain that some trinkets, charms, and jewellery are examples of local production that were traded with Romans. Examples of such material culture include the 'Black Bangles' found on Roman sites in Scotland (Hunter 2014), or Celtic-style scabbards sometimes used by soldiers (Webster 1985). Allason-Jones (2002: 823) argues such finds indicate a 'shrewd group of people providing trinkets specially aimed at Roman tastes'. 15 Comparative evidence supports the picture: at Nijmegen, both Roman and Celtic coins were discovered in and outside the fortress, indicating economic cross-contact (Van der Vin 2002: 171).
A progressive integration of locals into the military economy might also help to explain the expansion that many second to third century ad vici underwent. This would have been paralleled and aided by local recruitment into auxilia, which has long been posited (Dobson and Mann 1973). Although a third century ad inscription at Vindolanda, dedicated by the ' citizens of Gaul' demonstrates that auxilia did retain traditional ethnic characteristics and continued to recruit from their places of origin (Tomlin 2012: 211;Haynes 2013: 127), the addendum to the inscription-' and of Britain'-indicates that local enrolment was also occurring (RIB 3332).
Finally, a late Roman phenomenon is instructive for indicating local integration of the army's supply community. Vici declined in size after the mid-third century ad, and were largely abandoned by the fourth century ad (Bidwell and Hodgson 2009: 33-34). We cannot discuss in detail here why this occurred, but it seems due to both garrison reductions, and changes in supply mechanisms to state-production of arms and tax-in-kind. Related might be indications of a reduction in the number of farmsteads datable to the third and fourth centuries ad in the frontier zone, although the sample size is too small to advance this connection with much surety (Brindle 2016: 313-315). However, Collins (2012: 93-101, 107-110), in examining the Roman frontier communities in Roman Britain of the fourth and fifth centuries ad, notes a number of trends observed in late Roman forts pointing to an increased localism of supply for those garrisoned there. For example, the partial destruction or conversion of granaries, indications of local cattle production at Birdoswald and Carlisle, and coin deposition patterns that suggest the existence of marketplaces actually within the walls of several fourth century ad forts. This was interpreted as evidence for the increased regionalism of the late Roman garrison, prefiguring their survival as 'war-bands' in the fifth century ad. However, it may also be seen that such developments in fact suggest some continuity with the image of the Principate sketched here. Even if reduced garrisons and changes in the supply system (to tax-in-kind) damaged the viability of the large second century ad vici, and the broader community of attendants we have been discussing, military communities would still have had need of economic interaction with local groups. In particular, the provision of marketplaces inside the fort walls may suggest that instead of selling wares in the now-disused extramural settlements, local producers came to the fort and traded with the unit in marketplaces inside the gates. Economic interaction with local communities thus appears a constant and progressive process in the history of Roman Britain's military garrisons.
Therefore it seems that especially in the late-second to third centuries, the army's service community was diverse and cosmopolitan, and became increasingly integrated with the garrisoned territory. Locals moved into vici and canabae, they sold produce to the army, they engaged in Roman commercial networks, and they expressed their religion in ways appropriate to the new context. They likely first became connected to these processes through non-British suppliers and Roman agents acting as intermediaries, such as Octavius the contractor who purchased grain for the army (Tab. 343), and Cassius Saecularis (Tab. 213) who was dealing with groups trading in kind. The difference in mechanism between these two examples neatly expresses how global processes were negotiated in varied ways depending on local realities. Thus, the Roman army's attendants and suppliers, as well as having ' a foot in both camps' of the military/civilian divide, also straddle the overlapping global/local divide. Globalisation theorists have argued that cities are both '100% local and 100% global' (Koolhaas et al. 2001;Laurence and Trifilò 2015); our research subjects, as individuals, were also capable of this fluid duality.

A Frontier Community
Traditionally, in both conceptual and practical terms, civilian and military communities in the Roman frontier have been treated as separate entities. This owes both to the lower levels of archaeological interest shown to vici and canabae than to 'their' forts (Birley 2013: 85), as well as to ideas about the Roman army's exclusive institutional nature (see Gardner 2007: 210-219). The importance of extramural settlements and other civilian groups were often implicitly peripheral. Recently, however, scholars have begun to emphasise the importance of analysing the intra-and extra-mural settlements together, and interpreting them as a larger whole, rather than distinct units (James 2001;Birley 2013). Indeed, clear definitions between the ' civilian' and 'military' communities, and between 'Romans' and 'natives' are simplistic and fail to accurately describe frontier society, and recent scholarship rightly emphasises the complexity of these communities (Haynes 2013: 19-20, 134). The people in whom we are interested were positioned at the malleable meeting-points of these definitions.
For the sake of analysis, let us suggest that there are essentially two interrelated axes of identity and agency at play here. On the one hand, there is a military/civilian axis, and on the other, there is a Roman/British, 'global'/'local' axis. It might be assumed that these two axes essentially overlap, and they do in many respects, at least to the extent that in northern Britain the closer any individual is to the 'military' end of the scale, they are also likely to be closer to the 'Roman-global' end. Indeed, as we argued above, a principal catalyst of global integration in the region was the army itself. However, I believe that the real interest of the group under discussion here is their liminal and ambiguous situation vis-à-vis these definitions, negotiating the relations between them. We therefore now turn to the evidence for the way our subjects constructed their community, to see where they saw themselves and were seen by others, and to demonstrate the lived experience of globalising processes in one corner of the Roman world. I suggest that what we can see is a 'shared' community that is both military and civilian, both global and local.
Architectural and structural considerations suggest a common community, inside and outside the fort wall. Sommer (1984: 15) argued that space was deliberately left for civilian settlements during fort construction, based on bath-house positioning. Moreover, he also discussed the defended enclosures often attached to British forts, usually called annexes. Often thought to be fully military in character, containing for example animal pens or industrial installations for the fort's needs, there is in fact scattered evidence for civilian occupation of at least some of these enclosures (Sommer 1984: 20-22). Despite this, there is generally a ' contradiction' between annexe and vici (Sommer 2006: 112). Strikingly, however, it seems that at many Scottish forts, entire vici were regularly contained within such enclosures, for example at Newstead, where the second century ad enclosures contained some 16ha (Sommer 2012), and on the Antonine Wall, where annexes are almost ubiquitous and seem to have, as a rule, contained vici (Sommer 2006: 123). It has been suggested that this reflects the instability of the area. The Antonine Wall forts were generally smaller than Hadrianic installations, and the vici contained in the enclosures would have been also. Therefore it seems that only particularly dedicated camp-followers went northwards with their patrons, and due to the region's instability, commanders undertook to protect those who did, mindful of their important work. This is not to say that all attendants would have been the traders and manufacturers in which we are interested, but given the evidence for ceramic and metallic industrial activity at Scottish forts (e.g. Sommer 1984: 20-21;Bishop 2002: 33;Rogers 2002;Breeze 2016: 143), it seems that at least some were. That the military sacrificed defensiveness to protect their attendants suggests they were viewed as important constituents of the community (Sommer 2006: 123).
Further, clear evidence of the same architects and masons working inside and outside forts can sometimes be discerned, for example the fine carved sandstone and statuary found in a religious complex in the late second century ad Vindolanda vicus, paralleling material used in building the principia from the same period (Birley and Blake 2007: 14). Finally, it is interesting that a Hadrianic milestone which reads 'from Lancaster, 4 miles', was in fact four and a half miles from the fort. The distance it measures is to the edge of the vicus (Sommer 1984: 25-26;RIB 2272). It should of course be remembered that this milestone will have been erected by the military (Sedgley 1975). This indicates that the apparent barrier signalled by the fort wall, between 'inside' and ' outside', was not as defined, nor as impenetrable, as is often imagined. Deposition patterns strengthen this: Birley has recently shown that there is no evidence for a 'great divide' in the material culture of vicus and fort at Vindolanda; military equipment is actually more commonly found in the vicus than the fort, and items such as hairpins and female bracelets are evident within the fort walls too, even if they are more common outside (Birley 2013). Further, glass beads at the site are ' evenly spread between the military and civilian areas', although there are some items, such as amber beads, which occur only in military deposits (Birley and Greene 2006: 10-13). This indicates shared ties, but that there were sub-groups within the broad community.
Moreover, scattered historical anecdotes suggest that when push came to shove, civilian attendants were viewed as part of the military community. Tacitus (Hist. 4.22) writes that at Xanten during the Batavian revolt, the canabae, which had 'like a town, grown up to the camp-walls (procul castris … extructa)', was destroyed, and its inhabitants joined the soldiers within the fortress. Livy (23.16) writes that Marcellus, at the battle of Nola, had the camp-followers add to the army's shouts, fooling the Carthaginians into believing the Roman force was larger than it was. Similarly, during the Ligurian war, slaves and lixae joined the battle cry and burst from the gates with the army (Livy,40.28). Even the elite authors who did not think highly of camp-followers sometimes reveal a sense of a comradery.
Religious evidence also hints at a process of community creation, and demonstrates how civilians linked to military communities represented themselves both as 'Roman' and 'local'. Vindolanda seems to have been the cult centre for a likely local deity, Veteris (or Vitris, Veteres, and Hutiris, the multifarious forms probably reflecting individual interpretations of a name from a local language into Latin), with many dedications recovered in the vicus stretching from the first century to the fourth century ad (RIB 1697(RIB , 1898. Evidently a Celtic god, Veteris was expressed in a new way-in Latin, on permanent, inscribed altars-a way that Romans understood. Similarly, we have a dedication to Manopus (2431.2), and an altar to Sattada, by 'the curia of the Textoverdi', a local group expressing both Roman religious and political forms to make themselves understood (1695). While the Celtic influences on these dedications are obvious, the small finds (figurines etc.) often appear more unproblematically Roman, leading Birley (1977: 76) to judge them the 'veneer of Romanisation'. This indicates that Roman and local identities co-existed (not necessarily without tension), and that there was effort exerted to make the two traditions 'talk' to each other. Military officials also engaged in this effort: a prefect of the fort at Vindolanda set up an official dedication to Cocidius, the Brigantian equivalent of Mars (RIB 1683), and similarly at Brough-on-Noe, two official dedications were recovered in the principia. One was to Mars, but the second was to Arnomecta, an unknown local deity assumed to be a personification of the town Buxton (Dearne 1993: 8;RIB 281).
In this vein, the aforementioned dedications by the vicani of Vindolanda and Carlisle to Vulcan are instructive. These are unquestionably 'Roman' dedications, made on behalf of the whole community by a council, seemingly expressing the importance of metalworking to communal identity. Similarly, from a number of vici, potsherds displaying representations of smith gods have been recovered, but these forms are much more 'Celtic', incorporating elements of Sucellus and Taranis (Leach 1962;Green 1978: 19-20). Both the official dedications and the potsherds speak to the same impulse of self-representation, the difference in execution demonstrating the importance of context; the vicani understood that when it came to a corporate, ' official' dedication, it was Vulcan whom they should patronise. Religious evidence shows how our communities were capable of being both 'local' and 'Roman' at the same time. Through it, we can see our research subjects negotiating the tension between the local and the global in a context that, while local and embedded, came about due to the influence of global forces applied by the army.
The Vindolanda tablets support a blurring of definitions and a sense of a shared community. We have noted how Octavius (Tab. 343) seeks to identify himself with some military identity-markers, and the find spot of his letter, a probable centurions' quarters, suggests he worked closely with soldiers. In Tab. 218, Gavo, a trader who sold bedspreads and honey to the commander's household (Tab. 192), appears in connection with a personal request to Flavius Genialis, Cerialis' predecessor as prefect. The author tells Genialis he will communicate via ' our friend Paternus and … Gavo'. The sense is that Gavo plays some role in a patronage network alluded to, from which non-military status does not disqualify him. Similarly, in traders' accounts, such as Tab. 180, we have soldiers (and even beneficiarii) listed alongside apparent non-soldiers, such as 'the oxherds', and 'Lucco, in charge of the pigs'. In Tab. 181 we meet 'Alio the veterinary doctor' and 'Vitalis and bathman', and 182 lists 'Atrectus the brewer', indicating a community in which specialised occupations are integral aspects of identity. The emphasis is upon the role played in the community, and many individual's statuses are left quite ambiguous (and the fact that there was generally no need to identify individuals using exclusive status-markers is itself instructive); the implication is that interactions were occurring on these individual, role-specific bases.
Tab. 344, a draft-petition, found with Octavius' letter in the centurions' quarters at Vindolanda, is a good case study here, supporting both the idea of the 'shared community', but also unequal power structures within. The author-apparently having been beaten by a soldier-is certainly a civilian merchant. He speaks of his merchandise (merx), highlights his status as a foreigner (homo transmarinus), makes no claim to soldierly status for himself, and refers to the cohort in which his assailant serves as 'his unit' (numerus eius), i.e. the author was not also part of it. The merchant complains that he has been 'beaten with rods' (virgis castigatus), and his merchandise (likely grain) has been 'poured out'. He has complained to several centurions already, but was unable to petition the prefect, who was indisposed. He appeals to 'your majesty' (tuam maiestatem imploro), so must be addressing at least the governor. Birley (1997: 267-277) suggested it could be Hadrian, who was in Britain around this time, c. 120 ad. The offender was apparently a centurion, implied by the use of virga, and the fact that the author says he has complained to 'the other centurions of his unit' (centurionibus ceteris numeri eius-although it should be noted that ceteris is the editors' hypothesis). Evidently then, our supposed communal spirit was not perfect: even officers were not necessarily above abusing their suppliers. One could treat this petition as just another example of the brutality of Roman soldiers, and their privileged practical and legal position, an angle covered well by Peachin (1999).
However, this is only a part of the story. Firstly, it is notable that the author is careful to cast the transgression as individual: numerus eius, 'his unit'. He is not alleging institutional abuse. Further, he was confident enough in his relationships with other individual soldiers to complain to several centurions, and tried to reach the prefect (compare to Gavo above). The reason for this is clear: the author also wrote Tab. 181, an account of over 50 denarii of business done with soldiers, including several outstanding debts, suggestive of a degree of trust on behalf of both parties. The author did not therefore have an antagonistic relationship with the military, and indeed it was a relationship upon which his livelihood depended (the flip side being that he might therefore have been reticent to make a more general complaint about soldierly conduct). Again, that the petition was found in the fort suggests he approached a military official for help. Moreover, the statement that he is a man de cuius fide inquiras, ' about whose (good) faith you may inquire', implies that he expects officials at Vindolanda to support him if the governor or emperor took an interest in the case.
Despite this, we see the ugly side of the military's institutional nature: his trust in his relationships went unrewarded, and his complaints unheeded. The petition reveals an institution closing ranks following an individual's transgression; the frontier community is deconstructed as the threatened cohort looks inward, forcing the author to go outside the community for dispute resolution. This emphasises that the military was the senior patron, and had the recourse of asserting its exclusive identity when expedient. The author seeks to highlight and exploit this, by recording his travails in detailed, repetitive Latin, and by his selfrepresentation as a homo transmarinus, ' a man from overseas'. His motivation for this statement has not been fully explained, but I suggest this is intended to showcase his dependence on the army, following his narrative of neglect. He is away from his homeland, having followed the unit from its origin or previous base. This highlights the severity of the crime, as a soldier has beaten a trader for whom he has some responsibility, upon whom he depends for supplies, and with whom he shares a community. Therefore, where Haynes (2013: 19-20) sees evidence that a ' clear sense of insiders and outsiders was maintained', I think that the inside/outside dichotomy is much less clear. In the reader's mind, the author is attempting to rebuild the validity of the shared community that the unit is ignoring by failing to properly acknowledge his grievance. This document therefore demonstrates both the idea of a shared community on the frontier, as well as its limits, and the potency of the army's institutional character. Pitts and Versluys (2015: 14) remind us of a 'paradox' of globalising processes: they simultaneously create both unity and division, as 'social fault lines according to social and spatial mobility' form. In other words, access to the privileges and benefits of globalisation and connectivity are not equally distributed. But our investigation of the Roman army's civilian suppliers, who certainly counted as privileged among the inhabitants of Roman Britain, shows that inequality was not confined to distinctions between those with and without access to the 'global class'. Globalisation/glocalisation rightly emphasises the variability of local reaction to the dominating power; however, it also seems useful to argue for a greater emphasis on the variability of experience within the global class, between agents of the dominating power, which would have been dependent on a number of factors. Those within the global community were not simply the purveyors of Roman power-colonisers are not immune from colonisation themselves. We see this in the beaten trader's search for justice; certainly privileged compared to the poor inhabitants of the farmsteads in the frontier's hinterland, he was still the victim of status divisions within the locale's 'Roman' community. This community was not homogenous, despite the phenomenon of community-creation demonstrated above. We see this in the distribution of items in forts and their attendant settlements, the high levels of military items found in vici suggesting that this was a mixed space, while the proportionally much lower representation of civilian items within forts implying that civilians did not possess the same privilege to go where they pleased. Further, we noted above that at some Scottish forts, vici were defended by annexes, indicating that the inhabitants were valued members of the community. But this protection came at a price: at Newstead, which possessed several annexes, the particular enclosure dedicated to residential use changed in different periods likely reflecting in large part the fort's needs (Hoffmann 2009;Sommer 2012). It seems unlikely that the vicani had much say in whether, or when, they would have to uproot and relocate because their current space was required for something else.
The experiences of the civilian suppliers of the Roman army in Britain were therefore highly complex, and sometimes even appear contradictory. Even within interest groups and networks, there was immense variability of experience, and those who are privileged in one context may be disadvantaged in another. A closeup view of people's engagement with globalising, integrative forces reveals the constant negotiation that a more general view obscures, and gets us closer to the lived experience of globalisation in the Roman world. The multitude of interconnections and interdependencies that Roman imperialism made possible, and that our subjects actively furthered, created new opportunities to make money and win success, as we have seen. At the same time, however, these also generated opportunities to be divided and exploited.