The North of England in the late second and third century.
Dividing the province
It is one of the acknowledged ‘facts’ of Roman history, that Roman senators, who had successfully mounted a coup, attempted to ensure that the same methods could not be used against them by any would-be successors. 193 saw three Roman governors vie for the throne, all of whom used as their power bases the armies of their three legion provinces. It is thus hardly surprising to find that the victors, the Severans, are usually assumed to have been responsible for breaking up these commands into smaller, less dangerous units.
The earliest source for this division is Herodian, a historian of the late second and early third centuries, who has the advantage of being a near contemporary of the events described. Contemporaneity does not necessarily equate to being well informed, however, and Herodian can also be somewhat partial at times. He stated that one of the outcomes of the Battle of Lugdunum (197) was that Severus ‘set the affairs in Britain in order and divided the administration in the province into two commands’. There is certainly evidence that there were two British provinces later in the third century. The first was Britannia Superior (which probably refers to its greater proximity to Rome, and certainly does not imply any regional superiority) in the south, which contained the legio XX Valeria Victrix (at Chester) and legio II Adiutrix (at Caerleon), whilst Britannia Inferior, further north, held legio VI Victrix at York. Inferior was governed by a senator of praetorian rank, whilst the southern, more powerful, province remained the preserve of men who had held the Consulate. Nevertheless the early date given by Herodian for the division is contradicted by Cassius Dio, who places it in the reign of Caracalla. There are also inscriptions naming three governors of Severan Britain who all commanded a unified province, as well as one mentioning the legate of one of the southern legions operating north of Hadrian’s Wall under Severus. The latter could be explained by sending troops on detached duty to the neighbouring province, but no such convenient explanation exists for the three provincial governors.
Birley (2005, 333–336), following Dio and the epigraphic evidence, makes a good case for the division only occurring under Caracalla, after 213. It may seem surprising in view of the warfare on the northern frontier at this time that close support to Hadrian’s Wall was restricted to one legion. Indeed one might have expected Inferior to be the senior, two legion province. It is worth pointing out that the North contained the bulk of the auxiliary forces of Britain, which means that a division which put two legions and the fleet on one side, and a single legion, but with most of the auxiliaries on the other, may actually reflect a fairly even distribution of power in the island. Moreover the fact that two legions now had their bases in separate provinces does not seem to have diminished the presence of some of their members on Hadrian’s Wall. The legionary compound at Corbridge continued into the third century, while the fort at Carlisle appears to have been occupied by detachments from legiones II and XX.
Severan military activity in Britain
There is a surviving fragment of Cassius Dio (75(76), 5.4), which belongs in the immediate aftermath of Severus’ Civil War against his final rival Albinus (c. 197/198). As so often with such fragments, it lacks context, but it reads as follows:
Because the Caledonians did not keep to their promises and had prepared to aid the Maeatae, and because Severus was then devoting himself to the Parthian (or neighbouring) war, Lupus was compelled to buy peace from the Maeatae for a large sum, receiving a few captives.
The events described must have happened a substantial distance north of Hadrian’s Wall, as the Maeatae are traditionally associated with the area around Stirling, beyond the (now abandoned) Antonine Wall. This suggests that even at that point, Rome still took considerable interest in affairs in Scotland, a fact underlined by the substantial hoards of Roman coinage of the late second and early third century that have been found there (Hunter 2007, 23–32). The phrasing does, though, suggest that warfare was averted for the moment, although the presence of prisoners suggests that some sort of confrontation with Rome, and perhaps an incursion into Roman territory, had occurred earlier – albeit there is no indication of when this might have happened, or whether anything but the outpost forts north of Hadrian’s Wall would have been involved.
Virius Lupus was the first Severan governor of Britain and appears to have governed the whole province. In addition to his attested activities north of Hadrian’s Wall, he is also known from three building inscriptions, dateable to 197, from Corbridge (RIB 1163, building unspecified), Ilkley (RIB 637 mentioning the rebuilding of an unknown structure) and Bowes (RIB 730). The latter commemorates the rebuilding of a bath building, which had been destroyed by fire, and there has been a persistent tendency in archaeological literature to suggest that this fire, the rebuilding mentioned in the other inscriptions, and the burning of a barrack at Ravenglass, may be linked to the Maeatae trouble mentioned by Dio. Yet it is hard to see why a tribe from the lower reaches of the Forth, on the east coast, would be responsible for burning a fort on the west coast, and for widely distributed damage between the Stainmore Pass and the still more southerly fort of Ilkley. A bathhouse fire in the central Pennines is surely more easily explained as an accident, rather than the result of enemy action, that even with modern transport lived 4½ hours drive away, and was separated from the forts by Hadrian’s Wall.
Septimius Severus (193/211) is one of the most visible Emperors in the epigraphy and archaeology of Roman Britain. During his 14 years in power in Britain, he ordered the rebuilding (or rather finishing) of all three active legionary fortresses in stone (e.g. Ottaway 2004, 75), and substantially rebuilt or refurbished forts all over the north of England. Indeed on Hadrian’s Wall his activity was such that early Wall scholars spent considerable time debating whether it was really Severus’, rather than Hadrian’s Wall.
Much of this building was probably occasioned by the simple fact that forty years after the withdrawal from Antonine Scotland, the fort structures were likely to need ‘updating’ or, in some cases, complete rebuilding. Such was the case at Risingham, where a gate had collapsed through old age (RIB 1234), and another building inscription from Caernarvon/Segontium (RIB 430 + add), which must predate 209, shows that such activity was not just limited to the Wall. The governor most frequently mentioned in these inscriptions is Alfenius Senecio, who certainly appears on eight, and has been restored on a further three, and thus almost personified the programme. One of these inscriptions (RIB 1337 + add) is not a building inscription, but a dedication to the Victoria of the two Emperors Septimius Severus and his older son Caracalla. But Birley (2005, 191) has pointed out that this may just as likely commemorate the anniversary of the Severan Parthian campaigns, or relate to a war elsewhere, as to a victory in Britain.
There is comparatively little evidence for what was happening north of the Wall during the Severan period. Newstead appears to have been abandoned around 180, but High Rochester, Birrens, Bewcastle and Risingham remained in operation, documenting Rome’s continued interest in the area, and it would thus seem much more reasonable to assume, if any trouble was caused by the Maeatae in the aftermath of the Civil War, that it would have occurred in the Scottish Lowlands, in areas under Roman protection through the outposts, but north of Hadrian’s Wall and thus more accessible.
In addition to the refurbishments, some forts and installations were clearly new designs. At Vindolanda, the earlier stone fort was demolished and its stone frontage has been found apparently deliberately collapsed into the ditch. Its replacement was built over the Antonine vicus and surrounded on three sides by substantial ditches. The east side, facing the old fort platform, was only defended by a wall, and faced an area of roundhouses arranged in long lines, apparently covering most of the former fort (Birley & Blake 2005, 27–30). This complex is so far unique and its purpose remains debatable. Hodgson’s suggestion (2009, 32) that it provided accommodation for levees of civilians from the South, who were involved in the reconstruction of Hadrian’s Wall, is one possibility, but it does not explain why the accommodation should take such an unusual form, or be deemed important enough to replace the existing stone fort. South Shields, on the Wall’s eastern flank, on the south side of the Tyne mouth, was rebuilt in this period, as a fort plan dominated by numerous granaries, turning it into one of the most striking examples of a supply base in the Roman Empire.
Further north two forts appear to have been built on the East coast: Cramond and Carpow. Carpow was a legionary vexillation fortress serving detachments of both legio VI Victrix and legio II Augusta. It was located at the apex of the Tay estuary, just below its confluence with the Earn, and close to its lowest crossing point. It is currently the only known Severan installation in the area, and its isolation is striking, but Cramond parallels its basic location, this time on the south side of the Forth estuary.
Cramond was first built as one of the coastal forts on the eastern flank of the Antonine Wall. The withdrawal from the Wall from 158 onwards does not seem to have involved the fort’s demolition and some reduced occupation appears to have continued on the site (Holmes et al. 2003, 153f.), but it was fully reactivated in the Severan period. The date of reoccupation used to be fixed at 208, with reference to the historical sources, but there has recently been a tendency to shift this date back due to the evidence of coins. For example, Holmes (et al. 2003, 155) suggests a possible start date ‘in the very early years of the third century’. Like South Shields, Cramond seems to have had a role as a support base, with a considerable provision of workshops and industrial complexes, both internally, in the praetentura, and outside the fort, which included evidence for pottery manufacture.
The situation in the south
Roman military installations south of the Pennines became rare during the course of the second century. Apart from the legions and a few large forts left in Wales (e.g. Segontium/Caernarvon), the principal sites were the Cripplegate fort in London (probably designed to house the Governor’s bodyguard and any troops passing through) and the fort at Dover. Because of its classis Britannica brick stamps, the latter is usually associated with the fleet, and this does seem more likely than the alternative: that it was only built by the navy, but was garrisoned by other troops, given that it has also produced inscriptions recording fleet commanders. It was originally constructed in the early second century and in its rebuilt form lasted to c. 210 (Philp 1981). The classis Britannica had been operating in the Channel and along the British coast from the late first century. Its largest base appears to have been on the French side at Boulogne, but in addition to building the fort at Dover, it was also active in Kent and Sussex, and was involved with iron smelting and brick making at sites such as Beaufort Park (built in the late second century and operational until the mid third). A detachment is also recorded on Hadrian’s Wall, where its construction skills, along with naval transport, may both have been useful. We know little about the detailed workings of the unit, but in analogy to other fleets, it is usually assumed that it would have served a logistical role (providing transport or at least escorts for supplies, officials and troops moving to and from Britain), as well as monitoring other maritime traffic.
By the late second and early third century, a need seems to have been felt for the military presence along the coast to be increased. The earliest new fort to be identified was at Reculver (Philp 2005, 216), where construction started in the late second century. After a hiatus of several years, it was eventually completed around 211, which matches comparable data from Caister-on-Sea in Norfolk, and our limited evidence from Brancaster. Philp would link these three with Carisbrooke Castle on the Isle of Wight (for which little dating evidence exists), and they may be the beginnings of the coastal defence system, which by the late fourth century had become the so-called ‘Saxon Shore’. In origin, though, in view of the dating evidence for Cramond and Carpow, it may be worth discussing whether the original forts were possibly designed as further elements in a supply chain reaching from the iron processing and supply harbours of Kent, to the Hadrian’s Wall complex and advanced forts in Scotland, and that they only mutated into a new coastal defence role later, once more costal forts were built over the course of the third century. Once again, more research is clearly needed.
Severus on campaign
Herodian tells us that Septimius Severus had three reasons for coming to campaign in Britain: a) to add a British victory to his other conquests in the East and North b) to get his sons (Caracalla and his younger brother Geta) away from the flesh pots of Rome and c) because the governor had requested help against the barbarians (Herodian III, 14, 1–3). A.R. Birley (2005, 191) points out that the latter is a topos, (a repeated theme that may be rooted in expectation rather than fact). To judge from similar descriptions in Herodian and elsewhere, the chain of logic would run something like this: in Roman eyes it was unjust to start a war without proper cause; this means that it was important to be able to claim that a war was fought in defence of the Empire or its allies. If the governor requested help, this showed that the Emperor was hardly likely to be just fabricating the evidence to justify military action, but was responding to a genuine emergency. In other words, this was a conscientious Emperor fighting a just war. In theory, this would balance the additional motives for waging war in Britain, even though the educational opportunity and self-aggrandizement would not usually be seen solely as acceptable reasons. Nor would a secondary reason given by Dio, that the legions were becoming restless and needed something to keep them occupied (76 (77), 11.1). Herodian, who counted Severus as one of the ‘good Emperors’, was thus able to provide suitable window dressing for a military adventure.
We have two main accounts for the campaign, as well as a number of other fragments. Cassius Dio, with his now customary disdain for geography and military detail, reports that in the first season Severus pursued the Caledonii and Maeatae, but that the enemy employed guerrilla tactics, luring the Roman army into ambushes and difficult terrain, while not allowing themselves to be engaged in set piece battles. Despite heavy losses (Dio mentions 50,000 dead, which must again be an exaggeration, as it would equate to nearly the entire garrison of Britain at the time), Severus reached the end of the island, where he took astronomical measurements on the length of day, before returning and concluding a treaty with the barbarians. In the second season, Severus responded to a revolt by ordering the soldiers to kill everybody involved. This was apparently carried out, but the Caledonii then joined the revolt of the Maeatae, and Septimius Severus died while preparing for a campaign against both. Again, the account offers little in the way of geographical detail (typical for Dio) and focused instead on the ethnographic peculiarities of the enemy, and on Caracalla’s hunger for power and supposed attempts to speed his father’s demise.
Herodian’s account agrees with Dio’s as to the major events, but says that the campaign was fought north of the fortifications that provided the defences of the Empire and adds that Caracalla commanded the second year campaigns, as his father’s health had already deteriorated. Beyond the fact of victory, most of the accounts left by the fourth and fifth century historians, Jerome, Orosius, Eutropius and Aurelius Victor, concentrate on Severus’ death in Eburacum/York, rather than on the achievements of the campaigns. Orosius (7.17) and Eutropius (8.19) also claim that ‘he defended the island with a wall’, the latter clearly a reference to the extensive work on Hadrian’s Wall conducted in the decade before the Imperial visit. These late sources thus add very little to our understanding of the military operations. Coins issued at the time show Severus riding forth, a standard issue for an Emperor on campaign, and the image of a ship (on one coin accompanied by the legend TRAIECTUM – crossing or transshipment point), which might be taken to suggest that naval operations formed part of the campaign, although it might also just reflect the fact that Britain is an island.
In the past, archaeologists have tended to define the area of operations by means of a series of marching camps, which some publications would project all the way into Moray. More recent work has significantly changed our perception of these camps, however. After extensive excavations at the Kintore example, the 110 acre group, which formed the chain leading to Moray, is now mostly considered to be Flavian, leaving only the 63 acre and 130 acres series, along with the 165 acre camps in the Lowlands, as possible contenders for the Severan campaigns (NB: the groups were defined before metrication and have become a label rather than a strict indicator of size). The 130 acre sites form a close knit series stretching from Ardoch, through most of the Strathmore, and we know that they must at least slightly postdate the 63 acre camps, thanks to Hanson’s excavations at Ardoch, where the ditches of an example of each intersect. More recent excavations at Innerpeffray, however, might suggest that the camps were built before the Roman road through the area, and so could be earlier, perhaps Antonine, especially as evidence for a more modern road on this line is currently lacking (Woolliscroft and Hoffmann 2006). Roman temporary camps have always been notoriously difficult to date and, as has been pointed out by Hanson, there has long been a tendency to try to associate them with the two best known campaigns in Scotland: the Flavian and Severan. Ye t there were several other occasions in the second, later the third and the fourth centuries where campaigning took the Roman army north of Hadrian’s Wall and, as most of the camps still lack dating evidence, it is possible that future work will change our understanding of these structures.
Withdrawal from Scotland
Cassius Dio tells us that after the death of their father, Caracalla and Geta quickly concluded peace with the Caledonians, withdrew from enemy territory and returned to Rome to make sure that their succession as Emperors passed off smoothly. This has usually led to the assumption that a number of forts (including Carpow and Cramond) must have been abandoned at this time. But a fragmentary inscription from the East Gate of the Carpow legionary fortress presents problems. Wright has suggested that it could only have been cut after Geta’s death in Rome in 212, suggesting that the base was still under construction then and so remained in use after the end of the Severan campaigns. This has led to the belief that there may have been a continued presence in Scotland, possibly into the 220s or 230s (Bidwell and Speak 1994, 29). More recently, however, John Casey has pointed out that this same inscription could be restored to refer to Commodus (180/192), giving an earlier start date for Carpow, with interesting historical implications. Moreover, Nick Holmes (2003, 156 with further references) would argue that the coin evidence for both Cramond and Carpow does not support an extended stay beyond 211, and nor have they produced evidence of sufficient post-Severan pottery to suggest continued occupation.
Interestingly the Vindolanda roundhouse settlement was abandoned at about the same time, and Stone Fort II was built on the site (Birley and Blake 2005, 31). Nevertheless some features of the Severan building programme were retained, including the now fully renovated Hadrian’s Wall, and the supply base of South Shields, which appears to have continued to function as such into the late third century. But was the abandonment of the new territories necessarily a change of policy by sons keen to abandon their father’s goals in favour of a rapid return to Rome? Collingwood and Myres (1937, 160) suggested that the fact that Severus reoccupied very few of the earlier Flavian and Antonine fort sites, despite their lying on his line of advance, implies that he never intended a permanent occupation of Scotland, but only a punitive campaign: ‘visiting the wrath of Rome on enemies of Rome outside their grasp, but not outside their reach’. They would see this as part of the same strategy that brought about the rebuilding/refurbishing of Hadrian’s Wall, and thus provided an armoured baseline from which exactly such punitive campaigns could be launched, keeping the area to the north cowed under Rome’s political control and the province itself secure. This model, sometimes referred to by Scottish medieval historians as the ‘Caracallan’ or ‘Severan’ settlement (Fraser 2009), has been seen as the foundation on which early Medieval Scottish history developed, creating, in effect, an inner/southern zone of close contact with Rome, with a number of outer/northern and western zones beyond, which had much looser connections with the Empire. It is hard to verify this model by archaeological means. Certainly there are differences in what and how much Roman material reached different areas of Scotland, but the data sets are so small, and the historical records so sparse, that it is unlikely that we will ever be able to fully prove or disprove this model. For the moment, therefore, it might be best to leave it as an interesting hypothesis in need of further testing, noting in doing so, that the site of Birnie, in Moray, which should definitely lie in the ‘outer zone’, contained Severan coin hoards. This suggests at least that the division between inner/southern and outer/northern regions should not be seen as a simple geographical Lowland/Highland issue.