It is a truism that the study of Roman Britain is divided between those who study the civilian south and others who concentrate on the military north and west. Very few manage to bridge this divide and work in both areas. Nor does there seem to be much communication across the divide. In some ways, the post-colonial agenda has exacerbated this division as imperialism, colonialism, violence and slavery are often dismissed as ‘bad’ and therefore not worthy of study; in such a climate, it is unfortunate that the Roman army operated in all four areas.

The past, however, cannot be so easily characterised in that way. Keeley (1996) has shown that warfare was more endemic in prehistoric societies than previously appreciated, while closer to our own times it is impossible to study the Nazi era and ignore the violence of the state that they created. The Romans were militaristic and brutal; their commanders and soldiers did awful things—the recent uses of ‘distasteful’ and ‘unpleasant’ to describe their actions are too polite. Yet, Rome created frontiers which had the effect, intended or not, of allowing her civilians to live in peaceful conditions for decades if not centuries and the economy to flourish. How the army achieved and maintained these conditions is surely worthy of investigation, as is the impact its methods had on the local population and the melding of peoples and ideas in the frontier zone and the improved understanding of Roman foreign policy that comes from a study of the frontiers themselves.

Within the framework of Roman Britain the study of Roman frontiers has a further advantage: it is supported by a large archaeological data base (Collins and Symonds 2013: 9; Breeze 2014: 146–147). In particular, the study of Roman frontiers should help us to understand better the relationships between Rome and her northern neighbours. The pattern of military deployment along the frontier can reflect the distribution of the indigenous population, routes, for both trade and invasion, into and out of the empire, and the attitudes of Rome’s neighbours, as well as indicating how frontiers were intended to operate.

My intention in this paper is to persuade my readers that the study of Roman frontiers is an essential part of the investigation of cross-border relationships. It is not my intention to rehearse the history of Roman frontier studies—two important papers offer relevant critiques (James 2002; Gardner 2013)—nor engage with the debate on the relevance of their study for contemporary debates (Gardner 2013: 34). I start with considering why Roman frontiers are so unpopular in academic circles, that is, as opposed to a wider public perception which sees them as interesting (Gardner 2013: 19).

Why are Frontier Studies Unpopular?

The primary reason, it seems to me, that Roman frontier studies have acquired a poor standing among archaeologists is that in our post-colonial world the colonialist, imperialist and violent nature of the Roman Empire renders its study repugnant, an attitude extended to military history generally (Black 2010: 214). Certainly, to modern eyes, its violence is ‘distasteful’ (James 2002: 1, 36) and ‘unpleasant’ (Mattingly 1997: 10) but that is no reason for it not to be studied. The Roman state came close to being a military dictatorship, and its army was the single most important element of that state and the agency of its expansion (Mann 1974: 509). In order to understand Rome and its empire it is essential to study every aspect of the Roman army. Its opponents appreciated that; the success of Arminius in AD 9 was based on his understanding of the activities of the Roman army. Roman frontiers were the membrane through which the Romans related to other peoples and understanding how they operated is an essential element in our study of the nature of the relationship between Rome and her neighbours. Not to study Rome’s army and frontiers would result in a very unbalanced picture (James 2002: 2).

This distaste for violence is not new, and in the field of Roman military studies can be traced back to the reactions to warfare of those who participated in the major wars of the twentieth century. After two World Wars, the scholars of the day did not want to study battles (James 2002: 12–14). Combined with this was the experience of military intelligence acquired by the leading exponent of Roman army studies in the Second World War, Eric Birley. War service spent studying the German army had demonstrated to him the value of understanding the structure of the Roman army and through that gaining a glimpse into the actions and attitudes of its opponents. In his obituary, Milton Shulman wrote that Birley was one of the men who helped us win the war, and he achieved that through his detailed study of his country’s opponents (Evening Standard, 24 November 1995).

A second reason is that students of Hadrian’s Wall have given the impression that ‘we already possess most of what we need to know and that there is little left to achieve’ (Hingley 2008, 25). This is a correct characterisation of the general perception of Wall studies, but it is not true (Collins and Symonds 2013). The concerted campaign of excavations on Hadrian’s Wall from 1892 to 1935 determined the broad outline of its structural history, and as a result the researchers of the day rashly stated that they had solved all its problems (Breeze 2014: 142–143). Unfortunately, these statements were believed. The truth is that these researchers had only solved the problems they had defined. One result was that there was no over-arching strategy for research on Hadrian’s Wall into the 1980s and few major excavations until the introduction of commercial archaeology, though even today the number of research excavations remains low (Hodgson 2009: 50). In fact, of course, there are many problems on Roman frontiers awaiting examination and a large data base awaiting further interrogation.

Thirdly, there is a current consensus that we must give priority to wider social aspects. This is clearly stated by Mattingly (1997: 8, 10):

‘The first priority must be to locate indigenous people in the power networks and colonial discourse that bound them to Rome, and seek to understand the prelude, processes and results of their complex negotiations (societal and personal) with the imperial power’.

Exactly, and this is where Roman frontier studies have not been as successful as they should have been. As James has pointed out, Roman military studies in general have ‘failed to keep pace with wider changes’ (James 2002: 26; cf. Gardner 2017: 35). Nevertheless, Mattingly’s aim is difficult to achieve in northern Britain when the necessary information is sparse.

A fourth aspect is that those interested in Roman frontiers studies have given the impression of being unduly narrowly focussed and obsessively concerned with pointless minutiae. There is no doubt that this has long been believed (Elton 1996: viii; Wells 1996: 438; Mattingly 1997: 8; Esmonde Cleary 2000: 89; James 2005: 501; Gardner 2017: 34); Hodgson has commented on an ‘image-problem’ (Hodgson 2009: 50). More recently, Hingley (2017: 2) has stated that:

‘The discipline of RFS [Roman Frontier Studies] has tended to lack a … concern with cross-cultural and cross-temporal comparison … Since the 1930s RFS has developed through the medium of the Limes Congress and the dominant tradition exhibited by these Congresses has been to uncover (through survey and excavation) and to reconstruct the tangible surviving remains of the physical works of the Roman army in the frontier regions of the empire’,

though he does acknowledge that ‘from the 1980s to the present day considerable attention has been directed to the complexities of Roman-native interaction.’ Hingley only cites one such session, at Canterbury in 1989, but there have been similar themes at all but one Congress since Carnuntum in 1986. These are part of a wider pattern of change at the Congress. Here, as in so many aspects of academic research, news of change is slow to filter into wider consciousness. Since the first experiments at Stirling in 1979, the Congress has been on a slow trajectory of change, experimenting with different formulae, moving from geographical sessions and excavation reports to thematic sessions, fully achieved at Newcastle in 2009. The process is charted in Wells (1996: 438–439), James (2005) and Breeze (2017) and noted by Gardner (2017: 36).

In any case, minutiae are important as it is through them that we can seek to answer the wider research questions and create the bigger picture, so long as the rigour does not ossify into over preoccupation with detail (James 2002: 25). In Roman frontier studies the minutiae help us approach an understanding of the development of frontiers and through that how they operated. This might bring us close to an appreciation of their function which, presumably, to an extent related to what was happening outside the empire. The wider conclusions are based on the survey, excavation and study of Roman military installations over the past 200 years and more, and, it must be emphasised, upon our knowledge of their opponents. If we omit the details our judgements are based on partial evidence. No one would consider excluding reports on weapons from excavation reports on the basis that they are ‘distasteful’.

To compound the archaeological perceptions, Roman historians have long avoided commenting on the structural elements of Roman frontiers. Until recently only Whittaker’s Frontiers of the Roman Empire (1994) has sought to interpret the archaeology of the frontiers within their landscapes while embracing the relevant textual evidence. It may be for this reason that we can forgive ancient historians who state that archaeologists have been digging up Roman frontiers for generations and still don’t understand them. Mattern (1999: 112) has stated ‘recent scholarship has argued that the purpose of Roman frontiers is uncertain in all cases’ Beard in her latest book, SPQR (2015), 600 pages long including the index, dismisses Roman frontier studies in one paragraph, 22 lines long: ‘it is surprisingly hard to know exactly what it [Hadrian’s Wall] was for’, offering no explanation herself (484; cf. Hodgson 2009: 51). Both sit firmly in the ranks of the ancient historians who state that archaeologists are unable to explain the purpose of frontiers. It is an attitude that Mark Driessen has characterised as ‘limes denial’ (pers. comm.). Why does this attitude exist? Are ancient historians scared of information which they do not understand—or cannot be bothered to try to understand? Now they do have a book which offers a detailed review of the archaeological evidence for all the frontiers of the Roman empire with discussion of what they can tell us by way of their purpose, operation and development (Breeze 2011).

How Do We Determine the Operation and Function of Roman Frontiers?

To my knowledge at least 21 theories have been offered to explain the function of Roman frontiers. These are listed below together with sample references.

  1. Defence against major invasions (Daniels 1979: 360; Bidwell 2005: 74, 2008: 142)
  2. Protection against raiding (Bidwell 2005: 74; Breeze 2011: 189–190; Hanson 2014, 7–8)
  3. To protect travellers in the frontier zone (Rushworth 1996: 303)
  4. To control civilian movement (Breeze and Dobson 2000: 40)
  5. In North Africa, to control transhumance (Daniels 1987: 244; Cherry 1998: 59–66)
  6. A customs barrier and passport control point (Cherry 1998: 59–66)
  7. An exercise to keep the troops busy (James 2013: 159)
  8. To keep the peace in a volatile area where the construction of the frontier had incited attacks (Isaac 1990: 214–215)
  9. To put the soldiers somewhere when towns were not available (Hodgson 1989: 177–189)
  10. To protect the soldiers (Cherry 1998: 59–66)
  11. A shield to allow the peaceful development of the province (Birley 1956)
  12. The creation of a stable frontier (Shotter 1996: 70)
  13. A statement of Hadrian’s commitment to ‘imperial containment’ (Everitt 2009: 225)
  14. A base for operations beyond the frontier (Bruce 1863: 26; Perowne 1960: 86–87)
  15. A reflection of Rome’s failure to conquer the rest of the world (Mann 1974: 508)
  16. A (psychological) edge to the empire for the Romans (Sommer 2015: 50–51)
  17. A piece of rhetoric (Mann 1990: 53)
  18. A symbol to intimidate the enemy (Robertson 1979: 34; Crow 1991: 59; Driessen 2005; Mattingly 2006: 158)
  19. In the case of Hadrian’s Wall, to build a monument to Hadrian (Breeze and Dobson 2000: 32)
  20. A symbolic act of building (DeLaine 2002: 220–221; Breeze and Ferris 2016: 27–29)
  21. To keep the soldiers fit (Forster 1915)

These models are not mutually exclusive. Soldiers based on the boundaries of the empire could have been involved in defence (1), countering raiding (2), protecting travellers (3), controlling access to the empire (4, 5), and collecting customs duties (6), as well as performing policing duties in the absence of a police force in the Roman world (Davies 1989: 175–185) and controlling the provincials. Frontier installations might have been built in a monumental fashion to intimidate the enemy (18) while the Antonine Wall distance slabs might have been erected as a symbolic act of building (20). Forts were required because soldiers needed protecting and on the European frontiers there were no existing towns in which they could be based (9, 10) and they might also be located to control farmland and the provision of supplies (Higham 1981: 108–111). Functions might change over time. Modern views might also change: it is probably no coincidence that Birley’s 1956 hypothesis (11) was promulgated at the time the American shield for Western Europe was at its strongest.

The theories may also be divided into different categories, practical and abstract. The latter include symbolism, intimidation of the enemy, rhetoric, failure to conquer the rest of the world or an edge to the empire (12–20). The protection afforded by the army allowed the economic development of the province (11), but was that the purpose of the frontier rather than an effect? These are all difficult to prove and I will place them to one side for the purpose of this paper and concentrate on the areas where literary and archaeological evidence might be able to support one theory or another.

Literary and Epigraphic Evidence

Some of these theoretical functions can be supported by the practical application of literary sources and inscriptions.

  • 1. We have statements about the defence of the empire. Aelius Aristides and Appian in the middle of the second century AD stated that the Romans surrounded the empire with armies and walls (Aelius Aristides, Roman Orations 26.80; Appian, History of Rome Preface 7). More specifically it was stated that after the invading Scythians had been repulsed more forts were placed along the river until it was not possible for the barbarians to cross the river again (Josephus, The Jewish War 7.4.3).
  • 2. Raiding was a problem on every frontier (Breeze 2011: 188–189). In the reign of Hadrian 60 raiders attacked a military post in the Eastern Desert of Egypt (Ostraca Krokodilo 87). Under the emperor Commodus forts and towers were constructed along part of the Danube ‘to prevent surprise crossings by brigands’ (CIL III, 3385). A fortlet was erected in Mauretania Caesariensis in North Africa to ‘close a route used by raiders’ (IRT 880).
  • 3. Fortlets were established on the main road through Mauretania Caesariensis for the safety of travellers (CIL VIII, 2495) while a tower was erected in Lower Moesia 45 km south of the Danube for the protection of both soldiers and civilians (CIL III, 12376). Towers constructed on the frontier in the late fourth century were stated to improve surveillance of the crossing points on the frontier (Theodosian Code 15.1.13).
  • 4. It is clear from our sources that the Romans sought to control access to their territory. Friendly states were allowed to trade within the empire, while others were kept at a distance (Tacitus, Germania 41; Cassius Dio, History of Rome 72.15–16; 72.19; 73.2; 73.3.1–2). Tacitus remarked that the people of one state in Germany complained that they were not allowed to trade with the nearby Roman town unless unarmed, under military escort and after paying a fee (Tacitus, Histories 4.64). The ostraca from the Eastern Desert of Egypt provide an insight into life in the frontier zone where civilians were only allowed to travel with passes (Maxfield 2005: 201–220).
  • 5. It is frequently stated that soldiers collected customs, but there is no clear evidence for this (Lewis and Reinhold 1966: 146, n. 137).
  • 6. Tacitus stated on two occasions that the army was put to work in order to keep the troops busy (Tacitus, Annals 1.35; 11.20), but there is no evidence that this was the reason for the building of Roman frontiers. Forster’s (1915) suggestion arose from his observation of the training of troops during World War 1.
  • 7. Writing about events in the AD 370s, Ammianus Marcellinus recorded that the Quadi protested at the construction of a fort in their territory; the Roman response was to murder their king which led to a serious invasion of Roman territory (Ammianus Marcellinus, History of the Roman Empire 29.2–14).
  • 8. & 9. On the Eastern frontier many regiments were based in towns; Dura is the best known example. In the northern and western provinces, some soldiers were placed in the hill-forts of south-western Britain but this only occurred in the early conquest years (Frere 1986). Soldiers were protected in their camps and forts, but these were often only lightly defended because, as Tacitus remarked after the double legionary base of Vetera was sacked by the Batavians, we never expected it to be attacked (Tacitus, Histories 4.23).

Archaeological Evidence

To consider the theories in greater depth we must turn to the archaeological evidence. The frontiers of the Roman Empire extended for over 7,500 km through a variety of terrains (mountains, deserts and swamps), along rivers and, where these did not exist, artificial or land frontiers were constructed. In order to test our hypotheses, three types of frontiers will be examined, the river frontiers provided by the Danube and the Rhine, the mountain frontier of Dacia and the land frontier known as Hadrian’s Wall (on the issue of comparing ‘natural’ and ‘artificial’ frontiers see Gardner 2017: 39).

Military deployment can tell us much as it responded to local circumstances. Along a stretch of the Danube Valley downstream from Passau, the river flows through a steep, densely wooded gorge; here there were no forts in the early years of frontier building, we may assume because of the terrain and the lack of people living there. But to the east, the flat and fertile Tullnerfeld offered easy access to the empire, as well as rich pickings, and here the Romans stationed five strong regiments (Alföldy 1974: 150–152).

Along the Rhine and the Danube, and on the eastern frontier, major units—legions —were located beside significant routes leading to or from the empire, and often supported by cavalry (Breeze 2011: 172). During the Julio-Claudian period (up to AD 68), many such concentrations were poised on the Rhine ready to advance into Germany, an order which never came. On the Middle Danube, however, the legions appear to have been placed to control routes into the province, notably Carnuntum at the southern end of the Amber Route (Figure 1). In the late first century, more legions were transferred to the Danube in order to protect the empire from invasion by the Dacians. In these examples, defence and the protection of people living in the frontier area were the major concerns.

Figure 1 

Map of the Middle Danube. The concentration of legionary forts was presumably required in the face of two strong enemies to the north, the Marcomanni and the Quadi (Source: Author).

Dacia was invaded and conquered in AD 101–106. The new province projected north of the Danube and was surrounded by potential enemies on three sides (Figure 2). Its vulnerable location and geography required different military deployment. The province was protected by the circle of the Carpathian Mountains, broken by passes. Rather than guard every pass, regiments were generally based further back to maintain surveillance over several routes, thereby being able to react to a particular point of penetration; the two legions of the province were placed in the centre of the great amphitheatre created by the Carpathians, capable of moving out in any direction. Extra defence could be provided by the blocking of some valleys by banks and ditches, erecting towers, presumably to provide advance warning of attack, and, where the mountain chain was lower, locating additional troops (Haynes and Hanson 2004: 25–26). This system of defence would appear to have been carefully planned and was a sensible reaction in a frontier zone where the attack could have come from any direction.

Figure 2 

Map of Dacia (Source: Author).

The study of Hadrian’s Wall is particularly important in helping us understand the operation of frontiers because most of its elements touch each other thereby allowing its structural history to be disentangled (Breeze and Dobson 2000: 13–87). Before the Wall was constructed the most northerly line of forts lay across the Tyne-Solway isthmus from Corbridge, then the lowest bridging point of the Tyne, to Kirkbride on the Solway (Figure 3). The first plan for the Wall was to place on its north side a linear barrier, fronted by a ditch where appropriate, with a fortlet (milecastle), containing a gateway through the Wall, at every mile and two towers (turrets) in between. At this stage, there were no forts on the Wall. While work was in progress, military strength on the isthmus was significantly increased by raising the number of forts from seven to, probably, 19. Most of the new forts, about 13 km apart, were placed on the Wall line and, uniquely within the whole empire, where possible they were placed astride the Wall with three of their main double portal gates opening to the north and two smaller additional gates provided south of the linear barrier supplementing the double portal rear entrance.

Figure 3 

Maps of the first (a) and second (b) plans for Hadrian’s Wall (Source: Author).

The existence of two plans points to two very separate functions in operation. The relatively small numbers of troops in the milecastles and turrets strung out along the Wall could not deal with a major invasion; that task lay with the regiments in the forts to its rear. The purpose of the former, we may surmise, was control of movement. A roughly similar division of responsibility may be seen on the German frontier where additional units were provided at various places so that in an ‘emergency the Roman military command could pull out quite a proportion of regular auxiliary units without interrupting the routine watch along the limes’ (Baatz 1997: 15–17). A significant result of the change in plan on Hadrian’s Wall was to increase the mobility of the army in the frontier area (Perowne 1960: 86–87). Crucially, the placing of the forts astride the Wall indicates an intention to operate in the field to the north of the linear barrier.

At the same time as, or shortly after, forts were placed on the Wall line, an earthwork, known as the Vallum, was constructed along nearly the whole length of the frontier complex close behind the forts. The Vallum consisted of a deep ditch with a bank set back on each side (Breeze 2015). Causeways were only provided at forts, thereby reducing crossing points from an original 80 or so to 16. While many theories have been advanced for the construction of the Vallum (including that it was a second line of defence, or was designed to hinder raiding, that it was a temporary arrangement or served to protect the rear of the military zone: listed in Breeze 2015: 22–25) perhaps we should focus on its effect, the tightening of control of movement across the frontier (Breeze 2015: 66). When we couple this with the enhanced concentration of troops across the Tyne-Solway isthmus, we are forced to ask if the changes were the result of local opposition to the construction of the Wall? Sources state that at the time of Hadrian’s accession in AD 117 Britain could not be kept under Roman control, while two generations later the senator Cornelius Fronto wrote of major military losses in Britain under Hadrian (Historia Augusta, Life of Hadrian 5.2; Fronto, On the Parthian War 3; Haines 1920: 23). About this time a centurion was buried at Vindolanda having been killed in a war (RIB 3364). Archaeological excavation at Birdoswald has demonstrated a hiatus in the building programme (Wilmott 1997: 73–79). Was that caused by war in the frontier zone? We need better archaeological evidence to answer the question. It is worth noting that we have no evidence for destruction along the Wall at this time, simply indications of a pause in the progress of the work of building the frontier.

The bottom line is that we can recognise a strong build-up of troops along the line of Hadrian’s Wall and an enormous, and complicated, linear barrier, or rather barriers, with civilian movement across it severely restricted, suggesting that it had become a closed frontier, at least to civilian traffic (Bidwell pers. comm.). A possible conclusion is that the Romans faced a powerful enemy, perhaps enraged by the construction of the Wall and its impact on their lives. This is not a new discourse. One hundred and fifty years ago, John Collingwood Bruce (1863: 73–74) stated ‘that the Wall itself was not a mere fence, but a line of military operation, intended to overawe a foe, whose assaults were chiefly to be expected from the north’. Theodor Mommsen (1964 [1885]: 165) compared the slighter fence of the German frontier with the frontiers in Britain and concluded that the ‘Romans in upper Germany did not confront their neighbours as they confronted the Highlanders of Britain, in whose presence the province was always in a state of siege’ (Figure 4). There is plenty of literary evidence for warfare on the northern frontier, though of course this has to be treated with caution (e.g. Breeze 2006b: 13–20). After the wars of Hadrian’s reign, the frontier was shifted northwards at the beginning of the next reign. This move was temporary, with the Antonine Wall abandoned about AD 160 and Hadrian’s Wall reoccupied. There was trouble in the early AD 160s, warfare about 180, in 197 and 206, and the arrival of the emperor to take charge of the campaigning in AD 208 followed by a Roman victory two years later. There followed a century of peace—so far as we know; it might simply reflect the paucity of our resources—to be followed by a campaign against the Picts in AD 305, with further warfare in 360, 364, 367, 383 and about AD 400 when the troops in the island were strengthened by the addition of a small field army. If we did not have this literary evidence, how would we interpret the military deployment across northern England? Most certainly as reflecting the need for defence against a strong and persistent enemy and also perhaps local insecurity. Hadrian’s Wall provided that defence. Or rather the troops in the frontier zone provided the defence. We can still draw a distinction between the role of the soldiers in the forts of the frontier zone and the duties of those based in the milecastles and turrets on the barrier, as indeed we can with modern linear barriers—the Berlin Wall, the Morice Line, the Israeli barrier, and the fences being erected today in Europe—which are all to do with the control of individuals or small groups of people and not attacks by large, invading armies. The measures that have been regarded as being defensive—the height of the Wall, the ditch, the obstacles on the berm and the Vallum—are found on modern frontiers and could be interpreted as seeking to ensure tight frontier control, which we know was an issue for the Romans as it is for today’s states (Figure 5). None of these measures would help the soldiers defend the province, which could only be achieved through the army beating the enemy in the field. Collingwood’s (1921) assertion that in the event of an attack the Romans would open the fort gates and march out to face the enemy in the field still holds good (cf. Goldsworthy 2000: 157–159 for the aggressive nature of Roman warfare).

Figure 4 

A reconstructed section of the palisade in Germany with a replica tower behind; the design of the tower is based on sculptural reliefs on Trajan’s Column in Rome (Source: Author).

Figure 5 

MC 37 (Castle Nick). Each milecastle had both a north and a south gate, closed from the inside. One could envisage these as simply a protective measure for the occupants. This simplistic view of these crossings as merely entry points was challenged by Fraser Hunter who identified a more sinister purpose when describing milecastles in a recent television programme. He saw them as places where travellers could be frisked – and worse – by soldiers who we know could be corrupt and grasping (cf. Symonds 2015a: 66). We can only speculate on the impact of this level of control on the local people. Was it enough to spark rebellion in the frontier zone? (Source: Author).

The View from the North

The literary and archaeological evidence combine to offer a vivid picture of the Roman position, but what about the view from the other side? The acknowledgement that our evidence is slight and an appreciation that we rely too much on the Roman sources is not new (Warmington 1974). In his review, James (2007: 170) concluded that the ‘evidence of physical violence … is thin and ambiguous enough for us to impose almost any favourable interpretation on it’, though he goes on to say that there is a ‘strong cumulative case for the widespread physical violence, and the ideological valuing of martial skills and material culture.’

To turn to the evidence from northern Britain, we must start with that provided by the Romans. At the Battle of Mons Graupius, the Caledonians reputedly fielded an army of 30,000, losing a third of it in the fight; what tends to be forgotten is that they had given the Romans a bloody nose the previous year when they broke into a legionary camp at night (Tacitus, Agricola 26, 29, 37). Tacitus stated that the Caledonians fought with spears, long slashing swords and carried small shields (36), and that their strength lay in their infantry, the charioteers being nobles who brought their clients to battles with them (12); they were still using war chariots a hundred years later (Dio 76.12.3). One of the Vindolanda writing tablets (Tab. Vindol. II.164), however, after noting that the Britons are unprotected by armour, stated that they had very many cavalry, but the cavalrymen do not use swords, nor do the Britons mount in order to throw javelins; cavalry, we may note, had been earlier mentioned by Caesar (Gallic War 4.33) in describing his fighting in southern Britain, while cavalry regiments were later recruited from Britain (Cheesman 1914: 170–171).

The Roman perspective of the fighting spirit of the Caledonians might lead us to believe that we would find archaeological traces, not least in the form of weapons, in settlements, burials, hoards or even as chance finds. These, however, are rarely recovered. And while we have some swords and spears (Hunter 2005: 47–50), we do not have arrowheads and slingshots. Less than one hundred warrior burials are recorded in Britain, with only five north of Hadrian’s Wall (Hunter 2005: 64–66), while in the short list of signs of deliberate wounding to bodies in Iron Age burials there are no examples from Scotland (Dent 1983: 125–126). We do, however, have evidence for decapitation at Sculptor’s Cave in Moray, though whether this was ‘human sacrifice, execution of prisoners of war, or judicial punishment is uncertain’, though a connection with ritual is likely, while at Hornish Point in Uist the body of a boy had a ‘deep cut though his lower back suggestive of a stab wound’, again possibly evidence for a ritual death (Armit 2016: 108, 123). The problem of obtaining evidence of warfare through finds is not unique to this period. As Roger Mercer (2006: 119) has remarked in relation to the Bronze Age, ‘warfare is only demonstrated archaeologically with great difficulty.’ We do have one item associated with battle, the Deskford Carnyx, a Celtic war trumpet.

Hillforts, so often taken as indicators of warfare, do not help us. Over 90% of the hillforts in Scotland are clustered in the land between the Tyne and the Forth, but most of the defensive elements of these had been abandoned by the time of the Roman invasion and the function of those that remained, such as Traprain Law, is unclear (Armit 2016: 49–72). Settlement in south-east Scotland was dominated ‘by apparently lower-status, undefended farms’ while some remaining forts appear to have become ‘the strongholds of the warrior élite’ (Armit 2016: 72). Even the brochs and similar sites can be interpreted as being constructed for prestige rather than defence (Macinnes 1984).

There has been much discussion about the nature of warfare in Iron Age Britain, one current view being that ‘in the British Iron Age warfare was normally an elite pursuit. While war bands may have been formed, they were restricted to a small, socially exclusive group’ (Hunter 2005: 62). If this was the case, how could Rome’s enemies in the north raise sufficiently large forces to give serious challenge to the Roman army? Surely, Roman accounts of battles and warfare provide forceful challenge to that statement and to the suggestion that Iron Age society was peaceful. The armies Rome faced in northern Britain were led by nobles but formed of infantry and cavalry and Armit’s (2016: 76) estimate of the population at the time has led him to accept that as many as 30,000 warriors could have been assembled at Mons Graupius.

The Great Divide?

Since the surprise discovery of a gate through Hadrian’s Wall in 1848, archaeology has frequently challenged the earlier view of Hadrian’s Wall as an impermeable divide. The location and layout of the forts demonstrate an intention by the army to operate north of the Wall. On a local level, aqueducts led water from the north to supply forts on the Wall and perhaps as many as three extra-mural settlements spread across the Wall to the north, Wallsend, Birdoswald and possibly Chesters (Breeze 2006a: 84). In the eastern sector of Hadrian’s Wall, large-scale excavations are providing new evidence and a new dimension to the discussion. Excavations on the Northumberland Plain, just a short distance north of Hadrian’s Wall, have revealed a significant change in agricultural practices about the time that the Wall was built, with the abandonment of long-established rural settlements and the creation of stock enclosures (Proctor 2009: 101; Hodgson et al. 2012: 216–217). This may reflect a reaction to the prodigious requirements of the Roman army, not least for cattle (Proctor 2009: 83, 101), though other views have been advanced, including the deliberate clearance of the land to the north of the Wall (Hodgson et al. 2012: 217–219). Changes in settlement patterns are not restricted to the area immediately beside Hadrian’s Wall but spread as far north as the Forth. The hillfort at Broxmouth was abandoned—deserted, not destroyed—sometime in the second century, and this is the pattern generally between Tyne and Forth; very few sites can be demonstrated to have been occupied in the third century (Haselgrove 2009: 231; Armit and McKenzie 2013: 499). It is difficult to avoid the implication that the Romans were somehow involved in this fundamental change but the specifics elude us, and as Haselgrove (2009: 231) states, ‘the virtual absence of third and fourth century Roman finds is still difficult to explain given their relative abundance on Traprain Law.’

On the west coast, there have been attempts to demonstrate that Hadrian’s Wall was a great divide. Jones and Walker (1983) sought to use aerial survey to identify distinctions between the types of settlements on each side of the Wall. They noted that settlements north of the Solway tended to be more defensive than those south, while the density of settlement was also less, with fewer related field systems. Their conclusion was that the erection of the Wall allowed the development of a more prosperous and stable agricultural economy to its south (cf. function 10 above). This may have been the case, but practically all of the sites are undated and there have been few excavations since 1983 to help us confirm—or deny—the conclusion.

Homeland Security

The outer edge of the Roman frontier zone in Britain faced the enemy, but to the south there was a dense military deployment. What can we learn about attitudes within this military zone?

Several theories have been advanced in attempts to explain the distribution of the forts. Haverfield (1904: 142–144) advanced the hypothesis of a revolt of the largest northern tribe, the Brigantes, in the AD 150s; in spite of this being debunked (Breeze and Dobson 1976: 105–108), its shadow still lingers. Rivet (1969: 192) described them as ‘unsuccessful forts’ because the locals had not become Romanised thus allowing the soldiers to move on. Dobson (1970) suggested that the many units based in the forts in the immediate hinterland of Hadrian’s Wall in the later second century AD were the result of the need to place the regiments returning from Scotland and the Antonine Wall somewhere, not least bearing in mind that the southern part of the military zone had been handed over to civilian administrations. Hanson (1986) suggested that several forts were maintained in Wales long after the local threat had evaporated in order to supervise mining, and this proposal might be extended to northern England, where forts like Bough-on-Noe might have been retained in order to allow for military supervision of the lead mines. The supervision of routes might also be cited, which is the generally accepted interpretation of the military stations in the Eastern Desert of Egypt (Daniels 1987: 229).

Was it that restless natives required a strong military presence? The defence—or otherwise—of extra-mural settlements might point us towards an answer. Here lived the ‘regimental family’, the partners and families (and slaves?) of the soldiers (if some were not in the forts), merchants, craftsmen, publicans, prostitutes, all intimately connected with the soldiers. It might be expected that the army would want to protect its own if there was a hint of danger within the frontier zone. In 2009, Bidwell and Hodgson brought together the evidence for possible defences round such settlements and argued that they existed at as many as 20 forts across northern England (31–33). Unfortunately, as the authors acknowledge, the evidence is slight in most cases and it is the settlements outside cavalry units which have produced the best evidence for defences. Such units, it might be argued, however, would have been least likely to need them, so perhaps the motive for the ‘defences’ related to status. Further, Symonds (2015b: 92) has pointed out that the fortlet at Maiden Castle on a remote and exposed road across the Pennines appears to have outside it an undefended settlement suggesting a certain level of confidence on the part of the Roman army. In short, there is no evidence that the Brigantes were restless.

This conclusion has received some unexpected support through the discovery and examination of settlements in County Durham. At Faverdale north of Darlington, a habitation enclosure dating to the second century AD has been discovered (Proctor 2012). This was perceived as a trading post with extensive connections, adapting Roman cultural items to their own use. There is a further ‘small town’ at East Park, Sedgefield, while villas are now being found further north than their earlier known distribution (Mason 2010; Proctor 2012: 168–169).

These discoveries challenge previous theories. The paucity of artefacts on rural settlements had been interpreted in different ways (see the useful discussion in Kurchin 1995). It has been suggested that the few remains of Roman material culture in rural settlements combined with the lack of a settlement hierarchy—villas are found south of the military zone and brochs, duns and souterrains to the north but not within it—indicates that the Roman army creamed off surplus agricultural production and prevented the full assimilation of north Britain into the Roman world (Breeze 1989: 230; Higham 1989: 168). We can now see that the lack of artefacts on rural sites in northern England is part of a wider phenomenon. The striking contrast between the Roman material culture of urban centres (within which I would include the fort and its extra-mural settlement) and the surrounding countryside in the north is paralleled at Wroxeter and its surroundings; the people in the countryside ‘were much less interested in the material culture of Rome, especially away from the road network’, which is also reflected, to an extent, in east Yorkshire (Millett 2014: 19; White 2014: 11).

A different theory has been advanced by Carol van Driel-Murray (2008) to explain the poor material culture of rural settlements in the Lower Rhineland, relating it to the absence of local men on service in the Roman army and the survival of their women within a subsistence economy with no money to buy Roman goods. In other words, the lack of Roman artefacts was not a conscious decision reflecting an antithesis to Rome but an enforced one. The Batavians of the Lower Rhine were unusual in supplying troops directly to Rome so it is not possible to transfer this interpretation to the northern frontier of Britain. But it should remind us to be cautious in our interpretations.

Some Conclusions

The archaeological databases for Roman frontiers, often coupled with the literary and epigraphic sources, allow us to test the theories that have been advanced for their function. There is considerable agreement between archaeologists on the purposes of the frontiers. Raiding rates high in the list of reasons, as does the very modern theme of the control of access to the empire (Birley 1961: 169–170; Dobson 1986: 24; Hanson 1989: 59–60; Bidwell 2005: 74; Hodgson 2009: 44–45; James 2013: 158; Hanson 2014: 8). On Hadrian’s Wall, the main distinction lies between those who would see a stronger defensive role for the frontier, with its function that of ‘holding up an attack so that reinforcements could be brought up’ with soldiers playing a part in its defence from a wall-walk (Bidwell 2005: 74; Hodgson 2009: 44–45), and others who reject the idea of the Wall as a continuous fighting platform (Collingwood 1921; Donaldson 1988: 132; Mann 1990: 53; Hanson 2014: 6–9), or would see the Wall as a springboard for actions to the north, downplaying the role of the Wall ‘in battle and siege-warfare’, noting that in relation to scaling ladders the Wall was ‘not exceptionally high or formidable’ (Dobson 1986: 7). For some, there is a clear separation between the role of the linear barrier in hindering raiding and controlling access to the Empire and that of the army in the forts along the frontier whose purpose was the protection of the province (Birley 1961: 270; Breeze and Dobson 2000: 62).

The primary issue here is whether north Britain was so volatile—the enemy so fierce as Mommsen (1964 [1885]) argued—that special measures were required, these being the presence of an especially strongly defended linear barrier and a particularly strong army. According to the Roman sources, the Caledonians were a formidable foe, and the strength of their fighting force, as evidenced at Mons Graupius and by the scale of Roman losses during the reign of Hadrian and the campaign of Severus, was clearly substantial. But in order to believe that the situation in Britain was special, one has to overthrow all accepted perceptions of the Roman army as a highly successful fighting force. We must look elsewhere for the reasons for the unusual nature of Hadrian’s Wall—the involvement of the emperor?—and for the unfinished business of failing to conquer the whole of the island (Bidwell 2005: 72; Breeze 1989, 2009). When the Romans chose to take action against the Caledonians they were successful, as at Mons Graupius in AD 83 and under Severus in AD 209, but for much of the time imperial politics were balanced against serious Roman interventions in the north. The success of Agricola in AD 83 was marred and ended by Roman losses on the Danube leading to the withdrawal of perhaps a quarter of the Roman army from Britain. Severus’s gains were thrown away after his death in AD 211 by his son who abandoned his father’s conquests and returned to Rome. In between was the expedition of Antoninus Pius, absorbing territory previously Roman and presumably under Roman surveillance for the previous 60 years. This was not an attempt to solve the problem of the British frontier, but was a limited operation designed to provide a new, unproven, emperor with military prestige. A glance round the empire, its geography, and its enemies, might lead one to suppose that if the Romans had set their mind to it, they could have vanquished the Caledonians; it was simply that there were few occasions when an emperor sat safely on his throne, with no other concerns, and a willingness to sort out an issue on the most north-westerly frontier of his empire, a combination which rarely happened. In the meantime, Hadrian’s Wall was built and maintained to control the incursions of Rome’s powerful enemies.

In seeking to determine which is the correct conclusion to be drawn from the evidence, we are all conscious that it is very one-sided discussion. It is not just the lack of non-Roman literary sources (except the Talmudic material for Judaea) but the paucity of artefactual evidence for arms and armour and defended settlements in Iron Age contexts in northern Britain. In the face of the lack of evidence from the non-Roman—and the difficulties of obtaining more—it makes sense to see what we can understand about how Roman frontiers worked and what they can tell us about the neighbours of Rome. And about its provincials. The existence or otherwise of defences round extra-mural settlements helps challenge the long-held view that the Brigantes were hostile to Rome, and actually indicates the reverse. The general lack of artefacts on rural farmsteads in northern Britain can now be seen as part of a more general pattern within the province, but still hinders deeper understanding of the relationship between Rome and the indigenous peoples of northern Britain.

The fact that we are still learning about the frontiers and still discussing how they operated is surely exciting in itself. Controversy does not lead to the abandonment of research into a specific topic but rather a greater intensity of research and a widening of discussions. The controversy demonstrates that the subject is alive and kicking. So it is with Roman frontiers and their impact on the populations which adjoined them.