An Imperial Possession, page 18
There was no attempt in this early stage to establish a blanket garrison pattern across the country and no linear frontier. Attempts in the modern literature to trace a frontier along the Fosse Way connecting Lincoln and Exeter are seriously flawed and there is no evidence to suggest that Rome ever intended to limit her expansion in Britain at this point. Conversely, the mineral wealth of Wales and the Pennines provided a significant spur for expansion beyond the Trent and Severn and, while the sources imply that Roman territory was often subject to unprovoked attacks by the Silures for example, the physical deployment of Roman battle groups suggests rather that offensive pressure was applied by the massing of troops at the head of lines of advance. Where larger numbers of smaller forts start to appear in the pre-Flavian period – as in Devon, for instance – these were perhaps areas where pacification was prolonged and strongly contested.
Only from the Flavian period do we see a more general policy of spreading the garrison thinly in occupied territory, though there are continuing debates about the speed with which such networks of forts were constructed behind the advance of Roman troops. In Wales and the Pennines, new tactical deployments were devised to achieve the pacification and control of highland zones. In the aftermath of Mons Graupius, it is clear that Agricola’s unknown successor initiated similar deployments in Scotland, but the scheme was abandoned within a few years. From the early second century there was an increasing concentration of troops in northern Britain, to be soon formalized as the Hadrian’s Wall frontier. While garrisons were thinned out in other regions, by and large areas that had a military garrison at the start of the second century maintained some military presence thereafter. With the exception of a few years under Septimius Severus, when a return into northern Scotland appears to have been seriously considered, the basic shape of Roman troop deployments in northern and western Britain was much the same in the third century as in the second. The main difference after this time was a shift in emphasis towards coastal defence, not only along the Irish and North Sea coasts, but also on the Channel coast. The so-called Saxon Shore forts were established over a considerable period of the third and fourth centuries, but eventually represented a separate military command in Britain.
Some elements of the deployment were clearly linked to the exploitation of natural resources, as exemplified by the establishment of forts in mining areas (Charterhouse, Pumpsaint, Brompton, Brough-on-Noe, Whitley Castle) or at notable salt springs or salt mines (Droitwich, Middlewich, Northwich, Whitchurch). More detailed investigation of forts and associated settlements in other areas of natural resources could establish further cases. One interesting instance is the Roman road driven across Wheeldale Moor to the north of Malton in the Vale of Pickering. The rationale for this road and the two forts that sit at the northern and southern edges of the moor at Lease Rigg and Cawthorn has never been clear, but it is worth noting that the road connects the two major iron sources in the North York Moors in Eskdale and Rose-dale (though to date no firm evidence of Roman exploitation has been found), and if extended north-east it would descend to Whitby, main source of British jet and one of the best natural harbours north of York. The Lake District is a similarly intriguing case of a substantial garrison being put into an area of relatively low population, but high mineral potential. More research at the fort sites there could determine whether mineral exploitation was a factor in the deployment.
Regional differences in garrison patterns perhaps reflect nuanced approaches on the part of Rome to her provincial subjects as well as varying levels of resistance. The location and longevity of garrisons within a particular region may also indicate both Rome’s tactical approach and the relative success of pacification. The following section explores the regional character of military deployments, relating them where possible to British peoples mentioned in the sources. Dating at many sites is imprecise and this account will provide a broad outline, rather than attempt to reconstruct successive phases in minute detail.
South-east and south-west Britain
Out of the eastern kingdom Rome recognized two main peoples: the Trinovantes centred on Essex and parts of Suffolk and Cambridgeshire and the Catuvellauni of Buckinghamshire, Bedfordshire and Hertfordshire. The Roman conquest appears to have affected the two parts of the territory differently (Fig. 6).
In the Trinovantian area, one of the key early dispositions was the establishment of a legionary fortress at Colchester within the heart of the oppidum, with a possibly contemporary auxiliary fort nearby. The impact of the legionary force will have been highly significant – in numerical terms, a legion represented the equivalent of up to ten auxiliary units. In addition, there were several other forts, a vexillation fortress and a probable naval supply base in the region. This was not a blanket occupation of the territory, but rather focused on establishing control
Figure 6. Military sites in southern Britain
at the key central site, supplemented by a few large bases around the periphery of the territory, where troops could influence affairs in neighbouring districts as well. The foundation of a colony in succession to the fortress at Colchester c. 49 permanently removed territory from native British control, further indication of a relatively harsh approach to the Trinovantes.
The Catuvellaunian territory appears to have been much less supervised, especially now that a long-argued-for military fort at St Albans is being reinterpreted as a civil feature relating to the early town. The Folly Lane burial raises the possibility that Catuvellaunian territory may have been briefly organized as a client kingdom. However, there are military sites on roads leading north and west that appear to have lain close to the probable borders of Catuvellaunian lands. These include the fortress at Alchester in Oxfordshire (see below), and the presence of a vexillation fortress at Longthorpe near Peterborough is also suggestive. Again, we should see these not only as springboards for advance beyond the eastern kingdom, but as points from which the territory could be kept under surveillance and, when necessary, reached within a day or two by rapid march. Longthorpe is the probable location from where the unfortunate vexillation of legio IX marched against Boudica in 60. The scale of the rebellion they encountered resulted in their annihilation, but in less serious circumstances they would have provided a rapid and effective response to local unrest in the ‘pacified’ territories.
The site of London, controlling the lowest crossing point on the Thames and a river port, was a critical one in the conquest phase and a Roman military presence there from an early date seems certain. This fits with the artefactual evidence from London that conforms more closely with early military bases than with civil centres, though unequivocal evidence of an early fort is lacking. The nature and scale of military activity thus remains imprecise.
The eastern kingdom’s influence extended south of the Thames into Kent, technically the heartlands of the Cantiaci. Early Roman military activity is known at Richborough, which remained for much of the Roman period a major port of entry. Other early forts have been claimed, with less certainty. Dover was developed from the second century as the main base of the British fleet and a sequence of forts is known there. Otherwise, the main fortifications in Kent are all late Roman. However, the military control of the terminal points of the route through Kent from the Channel ports to London could have given early Roman Kent a profoundly military character.
The continuance of an Icenian client kingdom in Norfolk down to 60 kept Roman military posts at arm’s length up to that point. But given the circumstances of the Boudican revolt, it is to be expected that Roman units were posted there in the aftermath. Although the Roman civitas centre of the Iceni was later established at Caistor-by-Norwich in eastern Norfolk, the most impressive archaeological evidence of the heartlands of the Iron Age polity lay in western Norfolk, focused on sites close to the Icknield Way, such as Thetford and Saham Toney. Several Roman forts have now been recognized in this area, including one at Saham Toney (Woodcock Hall). The Romans built the Fen Causeway – a road across the Fens linking the vexillation fortress at Longthorpe with Peddars Way, the north–south Roman road established in western Norfolk. There was at least one fort in the Fens themselves at Grandford and perhaps others still to seek along the continuation of the Fen Causeway into northern Norfolk. A Roman military presence at Caistor-by-Norwich itself in the years following the Boudican revolt is plausible but unproven.
The initial fate of the lands of the southern kingdom in Sussex, Hampshire, Surrey and Berkshire is uncertain. There are military-style buildings below and adjacent to the Palace at Fishbourne, suggesting an early supply base. More significantly still, there appears to have been a military presence at Chichester close by. There are also numerous early Roman military finds from Silchester. One explanation is that the kingdom of Togidubnus was set up c. 47, as a way of releasing troops initially deployed to that territory for expansionist campaigns elsewhere. Alternatively, some troops may have been provided in support of establishing Togidubnus as king. A third possibility is that Togidubnus modelled his own troops on Roman lines. What is apparent from the absence of later forts is that the kingdom was thereafter fairly successful in keeping Roman troops out of its borders and that the eventual incorporation of the kingdom into the province by the 80s was relatively smooth and did not require any long-term troop deployment.
The territory of the Durotriges in Dorset and Somerset was the scene of extensive Roman campaigning as a prime target of Vespasian. A vexillation fortress is known to have existed just north of Poole Harbour at Lake Farm and appears to have been occupied c. 45–60. This implies a sizeable garrison in the territory for a generation post-conquest and there are additional traces of military occupation within a number of late Iron Age hillforts, as at Hod Hill, Maiden Castle, South Cadbury Castle, Ham Hill and Waddon Hill. There were also more regular forts, for instance on the Fosse Way at Ilchester or at Charterhouse on Mendip. Inscribed lead ingots from the Mendip hills attest the involvement of legio II in mineral exploitation and indicate that production was under way by 49.
The picture of the military occupation of Dumnonian lands in Devon has been transformed over the last thirty years. The 1978 Map of Roman Britain featured only the legionary fortress at Exeter and two forts. The current total stands at a legionary fortress, a sequence of forts at North Tawton, and at least six other fort sites, as well as military occupation of an abandoned hillfort at Hembury. The bulk of these sites are located in eastern and northern Devon, indicating close supervision of the valleys of the Taw and Exe, and perhaps iron production on Exmoor. The earliest finds from Exeter suggest initial military occupation c. 50, though whether the site was immediately a full legionary fortress is unclear. Exeter certainly became the key military base of legio II in the south-west by 55 at the latest, and was in occupation until the mid-70s (though probably with a much reduced garrison by then). Forts at North Tawton and Okehampton were evidently situated on a route running west on the north side of Dartmoor. Another fort further west on this route at the crossing of the Tamar in the vicinity of Launceston remains a probability. The presence of two fortlets on the north coast and the route down the Taw valley strongly suggest a missing site at or close to Barnstaple, while numerous Roman finds from Plymouth harbour suggest a port and possibly fort site there. Overall, Devon appears to have been quite heavily garrisoned in the third quarter of the first century.
The people of Cornwall are normally treated as part of the Dumnonii because Ptolemy’s Geography did not specify the existence of a separate group in this area. However, basic distinctions in the late Iron Age cultures of Devon and Cornwall strongly suggest that the people of the latter area had a well-developed and separate identity before the conquest. Since the name Cornwall appears to derive from an ancient form Cornovia, it is reasonable to surmise the existence of a people called the Cornovii here. A site called Durocornovium (‘stronghold of the Cornovii’) is also attested somewhere in the south-west in the Ravenna Cosmography. Only a single fort is at present known in Cornwall, at Nanstallon by Bodmin – evidently dating to the late 50s, but given its isolation (seventy kilometres from the nearest confirmed fort) it is probable that several others existed, perhaps corresponding with Tamara, Uxellodunum and Voliba in Ptolemy.
The take-over of the East Midlands is often considered to have been a relatively peaceful event. There are no explicit references to warfare with the major people of this region, the Corieltavi, or to their participation in revolts. Nonetheless, there are hints that incorporation was not entirely consensual. One explanation for a series of major hoards of Corieltavian coins from Yorkshire is that these were taken there by refugees from the Roman take-over of Corieltavian lands, presumed to extend between the east bank of the river Trent, the Humber and the Wash. The establishment of the Roman Great North Road, Ermine Street, drove an access route through the eastern part of this territory, with a sequence of forts along it. Along with the vexillation fortress at Longthorpe, these forts also facilitated the supervision of the Fens to the east and important iron deposits and farmland to the west. Lincoln lies at the point where the north road descended from the Jurassic ridge to cross the river Witham. An important Iron Age settlement existed at the crossing point by the Brayford Pool and an early Roman base probably lay just to the south of this (perhaps a vexillation fortress to match the one at Longthorpe). This early base was probably abandoned after the Boudican revolt and replaced with a legionary fortress on the north side of the river. This site remained a key military base until the late 70s, probably operating in concert with a series of vexillation bases on either side of the Trent (Newton on Trent, Osmanthorpe, Rossington). These latter sites appear related to intimidation of the Brigantes to the north or preparations for advance into their territory. Several smaller forts are also known on or close to the Trent. There were a further two possible forts near the Humber crossing. The overall impression is that the eastern lands of the Corieltavi came under concerted military supervision, centred on Lincoln, with territory for a colonia also being annexed in this area. A possible Roman military ditch beneath modern Leicester probably represents a small fortlet rather than something larger. There is a marked lack of other military sites in the Soar valley and the fact that the civitas capital was eventually located here may perhaps be seen as preferential treatment of the western sept of the Corieltavi.
Another key region that lay on a major line of Roman advance away from the south-east was the Cotswolds and related areas of Gloucestershire, Avon, Oxfordshire, Hereford and Worcester. This is normally equated with the people known as the Dobunni and, as with the Corieltavi, the general view has been that they were relatively favourably treated by Rome. The military deployment again modifies this view. The eastern limits of the territory of the Dobunni are uncertain, but are sometimes equated with the Cherwell in Oxfordshire.
The site of Alchester just north of Oxford has emerged in recent years as a key location in the early years of the conquest phase. It is now known that a major fortress underlies the later small town here and a timber post from the gate of its annexe has yielded a felling date of autumn 44 to spring 45. This is the earliest dendrochronological date from any Roman site in Britain and provides evidence of a base for a large battle group at the end of the second campaigning season. Alchester lay at the cross-roads of a Roman road leading east to west from Colchester via St Albans and on towards Cirencester and a south–north route linking Chichester, Winchester, Silchester, Dorchester, with potential continuations to the north and north-west towards Leicester and Wroxeter. There was also a second north–west route towards the Severn near Worcester. Alchester thus sat at the centre of a web of incipient communications and may have had a long life as a base in the conquest phase. In addition to the legionary base at Alchester, a fort at Dorchester-on-Thames secured the river crossing and controlled an oppidum there. A few kilometres south of the primary oppidum of the Dobunni at Bagendon a cavalry fort was established beneath the later town of Cirencester, again at an important road junction. This site was well situated to exercise close supervision of the primary native centre, but also served as a link between Alchester and a sequence of two major military bases at the Severn crossing at Gloucester. The legionary fortress at Gloucester was established only in the mid-60s, probably succeeding a mixed vexillation fortress at Kingsholm (tombstones of serving soldiers of the cohors VI Thracum and legio XX have been found nearby). The start date of the Kingsholm site could be the late 40s, with abandonment in the mid–late 60s. The implantation of a colony at Gloucester in the late first century will have detached territory from Dobunnic territory. The southern limits of the Dobunni are uncertain, though coin finds suggest that their influence extended as far as Bath (where an early military presence has been long suspected). Finds of military equipment also suggest a small naval base at Sea Mills with access to the Bristol Channel. All in all, there was a heavy garrison in Dobunnic lands from the 40s to at least the 70s.
