Even the Lines of Communication by Means of Military Roads, in some parts of the Highlands, have been productive of Benefit to the Country, though the Motives that gave rise to their Formation, having no Relation to Objects of Commerce or Industry, the Advantages derived from them are very imperfect and the Want of further Roads has been the greatest Obstacle to the Introduction of useful Industry. Reports of Commissioners for Highland Roads and Bridges1
Phasing the network
Most of the time, it is not easy to date a road archaeologically. The dating of Dere Street at Corbridge to cAD 85, for instance, is based upon the assumption that its construction coincided with that of the first Roman fort on the site. An earlier fort, at Beaufront Red House, slightly to the west, may mark the course of the original prehistoric route used by the army during the initial push north in the first century AD, which was then probably abandoned when the new road arrived. That is, however, a deduction based on limited evidence. To actually date a road archaeologically requires an understanding of the archaeological process, especially stratigraphy, and how an artefact found on a road surface provides a terminus ante quem (TAQ) for the construction (or repair) of that surface, and aterminus post quem (TPQ) for any surfaces above it, but sadly no guarantees of its contemporaneity with either act (how old are the coins in your pocket?). Moreover, as any excavator of Roman roads will tell you, finds are not exactly liberally scattered on their surfaces. Nevertheless, finds do occur occasionally and can be of assistance, but it would be a mistake to think that they are common.2
Although the major Roman roads of Britain are known by certain names of some antiquity (Figure 11), none of them date back to the Roman period. In most cases, these names seem to derive from the Anglo-Saxons. Dere Street (York–?Inveresk), for instance, is first recorded (as Deorestrete) in the early twelfth century History of St Cuthbert, describing the lands around the church at Gainford in County Durham. The name also occurs in various charters from the twelfth century onwards, notably in Scotland. Watling Street (London–Chester) is mentioned even earlier (as Wæcelinga Stræt), in the treaty between Alfred and Guthrun that establishes the boundaries of the Danelaw. Although only surviving as a fourteenth century copy, the treaty nevertheless dates back to 886–90.3
Figure 11: The principal traditional names for Roman roads in Britain (see Appendix 1).
Whilst we do not know any of the Roman names for their British roads (the Via Julia between Bath and Bristol, Margary 54, was a very obvious antiquarian invention) – nor even if they actually had any – we can guess that any that did will have been named after the emperors associated with their initial construction (e.g. via Claudia, via Flavia). Thus in Arabia, the Via Nova Traiana and the Strata Diocletiana bear the names of their imperial progenitors. It is unlikely that roads of Britain will have been named after provincial governors as was the practice in Republican times, since this would be seen as offering dangerous competition to the ruling emperor: the British governor Sallustius Lucullus succeeded in offending Domitian simply by naming a new weapon after himself, so an eponymous road would seem to have been out of the question. Additionally, we may be justified in suspecting that, by the Late Roman period, most roads in Britain were known as strata (like the Strata Diocletiana), rather than via, if for no other reason than the word entered the Anglo-Saxon language as streat.4
Since most of the major roads in Britain probably had their origins in the military campaigns of the conquest period, it is not unreasonable to suggest that the various components of the network can be dated to the aftermath of the earliest military activity in each area (Figure 12). Moreover, it may be possible to attempt to deduce the various stages of the development of the road network in Britain. In this, we must assume that the surfaced Roman roads more-or-less followed the courses of their native predecessors but, as has been outlined above, we can take comfort from the fact that this is not an unreasonable assumption. To some extent, what we must be witnessing is the selection of key components of the existing network that best suited Roman military needs. Indeed, in one or two instances early forts are overlain by a later metalled road, as at Baylham House.5
First there would be routes derived from the main line of advance in a campaign. In this category we could include the roads from Richborough to London (Margary 1), London to Lincoln (Margary 2), London to Colchester (Margary 3), London to Wroxeter (Margary 1), and London to Exeter (Margary 4). Although these probably originated as pre-Roman routes, they would have been formalized as all-weather roads in the way outlined above. This first stage would have been military in origin.
Figure 12: A hypothetical phasing of the principal components of the known early Roman road network in Britain.
Next, cross-routes would have been provided, linking the lines of advance. Their function would have been strategic, facilitating the lateral movement of men and supplies, linking bases along the main roads. Roads like the Fosse Way (Margary 5), that from Chester to York (Margary 7, 712), or the Stanegate (Margary 85) fall into this category. Again, pre-Roman routes would have been exploited and formalized, and this stage would also be set out by the military.
It is at the third stage that we start to see the interaction of the civilian sphere with the military. Civilian use of roads will probably have followed rapidly after the initial military construction and exploitation, as trade and social interaction led to the development (or, more likely, continuation and expansion) of movement around the province by the indigenous inhabitants and the many non-military occupants of (or visitors to) Britannia. This is a similar phenomenon to that commented upon by the Highland Commissioners in the quotation at the head of this chapter, for they were noting that whilst the eighteenth century roads of Wade and Caulfeild were military in purpose, they also had a commercial effect (and one which could have been greater). With regard to the Roman network, however, it will probably have been at this stage that elements of the existing (pre-Roman) network will have been tied into the formal military structure, perhaps not even as a deliberate or planned act.
The fourth and final stage will have been the natural and gradual evolution of the network, with some roads (both formal and informal) going out of use as needs changed, and new ones added where appropriate. Thus, it is a mistake to look at a map of the Roman roads of Britain and think of it as unchanging. All roads systems are dynamic, evolving to meet the developing needs of their users and any such map is inevitably a palimpsest.
Roads can to some extent be phased by their structural relationships, just as an archaeological site can be phased by its stratigraphy. What does this mean? Quite simply, a side road that joins a main road at an angle of ninety degrees almost certainly postdates the main road (otherwise there is no reason for it to exist). Elements of the system can therefore, to a limited extent, be identified as earlier or later than other components. In this way, the road from High Rochester to Learchild (Margary 88) provides a link between the Dere Street and the Devil’s Causeway, but it is a road that cannot exist without the other two having first come into being. It is, of course, possible that it is a fossilized part of a much longer pre-Roman route that has been incorporated into the Roman system to meet a very specific need, but that must remain speculative.
Similarly, some roads appear to run counter to the natural development of the system from the south-east northwards. Margary 820 links the two early conquest routes of Dere Street (Margary 8) and the Stainmore Pass (Margary 82) by forming a short cut from Bowes to a point south of Binchester that would obviously facilitate traffic moving between the two, since it removes the need to travel south-east to Scotch Corner and then effectively double back on the other road, shortening the journey between Bowes and Binchester by some 20 km. Once again, however, this could easily be an earlier route which has been fossilized in the middle and lost either end. Thus Margary 8 is for north to south travel, Margary 82 for south-east to north-west movement, and Margary 820 is provided to facilitate travel from the north-east to the western side of the Pennines, avoiding the need to go via the Stanegate. On the western side of the country, Margary 731 may well have provided the complementary link between Margary 82 and that other early conquest route, Margary 7, for those wishing to travel between the two without visiting their junction at Brougham (Figure 13).
Inevitably, tied in with the question of a road network in any province of the Roman Empire is the question of centuriation. This is the term used to refer to the regular (even rigid) division of land for settlers, frequently veterans in coloniae, marked out by a regular grid of roads in rural areas. Many attempts have been made to identify instances of centuriation within Britain, largely unsuccessfully. Examples have been suggested in the region of Colchester and Gloucester, coloniae that had been legionary bases. Margary suggested that a network of roads he had identified in Sussex might demonstrate it, although nowhere near an established colonia and, some would suggest, within a possible imperial estate that was devoted to industry rather than settlement. Genuine cases of centuriation are based upon a land unit of twenty actus in length (710 m) and cases are known in other provinces, such as Gaul, from both documentary evidence as cadastres, like those from Orange or Béziers, and from aerial photography. A cadastre is a document, usually surviving in the form of an inscription, recording the apportioning of land. No cadastral documents survive for Britain, but claims for evidence on the ground for centuriation continue to be made periodically. Amongst the most recent are assessments of the evidence from a number of suggested implementations in the landscape in the northern home counties of England (Hertfordshire, Middlesex, and Essex) and in the Weald. The existence (or lack of it) of centuriation in the British landscape is only really significant because, if it did occur, it has proved surprisingly difficult to trace, whereas continental examples are normally readily apparent. As such, it may be indicative of the complexity of landscape development in Britain, even in rural areas, rather than the genuine absence of centuriation.6
Figure 13: Main east- and west-coast routes with trans-Pennine links (Margary 8, 82, and 820 with routes 7 and 731).
Who built, maintained, and used the roads?
By and large, roads were built by the army, maintained by civil authorities, and used by everybody. As such, their financing was partly hidden, insofar as the army’s input – technical skills and manpower both for labour and the acquisition of the necessary raw materials – was its to dispose of as it saw fit. The Roman army were and are famed for using their troops in engineering tasks: this not only aided them in their preferred form of engineering warfare, but served to keep a vast standing army from becoming bored through idleness. Thus the ‘costs’ of road construction are largely irrelevant. A minor consideration might have been ownership rights over the land through which routes passed (Plate 1), but more recent parallels, such as the construction of railroads in America, suggest that, at one extreme, scant attention might be paid to the rights of indigenous inhabitants, whilst at the other they could be bought through more-or-less valuable deals.7
Maintenance required real money, however. Town councils in Roman Britain operated by making officials increasingly personally responsible for the debts of a council. Taxes would need to be raised, a portion of which would have to have been allotted to the upkeep of roads in the region for which they were responsible. In many of the frontier regions, where there were no local town authorities, maintenance must effectively have been the responsibility of the army.8
Whilst it was normal practice just to resurface roads that had become worn out, at Inveresk the road through the civil settlement was also given new foundations at the same time, resulting in three successive roads built on top of each other within approximately twenty-five years (c. AD 140–c. 165) when the nearby fort was occupied. Most of the surviving inscribed milestones from Britain record repairs to roads, rather than their actual construction, and the majority of them date to the third century AD or later (Figure 10).9
Although wear and tear from use will have taken a toll on road surfaces, the example of the military roads of the Highlands suggests that the climate will have had a much more significant effect. Ironically, therefore, lack of use may have been more detrimental than frequent use, as weathering and the encroachment of vegetation slowly tore a road to pieces. The sort of rutting and flooding reported from Roman roads still in use in the medieval and later periods was a result of long-term use; roads constructed at the same time but no longer used will have long since vanished.10
One of the functions of a state road network was to provide remount and accommodation facilities for officials or those travelling with their approval. This system was known as the cursus publicus and, amongst other things, made it possible for couriers to travel overland as rapidly as possible, given the contemporary constraints on speed. The remount system was based around stationes and mansiones, examples of the latter being identified, although not proven, from a number of sites. These would typically be large buildings with a number of rooms for accommodation, a bath suite, and stables and other outbuildings. Important travellers would almost certainly be put up as the guests of local officials or army unit commanders, so it would be the many minor bureaucrats who would have been found at mansiones. Examples of these structures in Britain have been suggested at a number of sites, such as Chelmsford, Vindolanda, Corbridge, and Catterick, although it has to be said that most are deduced as being mansionesand lack any intrinsic, indisputable evidence of their function. Many of these were courtyard structures with associated bath buildings (Figure 14). Often set towards the edge of settlements, they would usually be placed near the roads they are thought to have serviced.11
Although Roman roads were probably not built for anything other than military use, that did not mean they were not exploited by civilians for the purposes of travel and trade. Of course, long distance travel and heavy haulage were often much easier and cheaper by sea or, in some cases, by river craft, but there could be no escaping the fact that for short distance travel and trade, the roads could not be surpassed. Study has shown how the distribution of pottery over a region can be linked to the road network, whilst shipments of pottery over long distances (such as ‘army contracts’ for particular wares, such as black-burnished or, more obviously, samian, pottery) would depend upon sea transport. The comparison of costs of land and sea transport from the limited available evidence shows why this should have been so.12
Figure 14: Examples of mansiones. 1: Godmanchester; 2: Silchester; 3: Caerwent.
Evidence from better-documented provinces, such as the Palmyrene Tax Law from Syria, shows how taxes and tolls were imposed on the movement of goods into and out of the province and for these reasons customs posts were set up near borders, usually manned by troops outposted from units. There would also have been ‘police’ posts manned by beneficiarii, responsible for keeping a watchful eye on movement along the road system. It must be remembered that brigandage and petty pilfering would have been endemic within the empire, particularly on those parts of the network remote from major troop concentrations or civilising influences such as towns.13
In literate societies, road systems invariably end up being mapped. A large map of the road system in the Roman Empire, now known as the Tabula Peutingeriana (or Peutinger’s Table) records the major roads and waystations throughout the empire, complete with distances between places. Unfortunately, only a small part of the British road system survives, mainly that from south-east England and East Anglia. The document survived only as a medieval copy in the monastery at Worms in Germany and was given to Konrad Peutinger of Augsburg in 1508, hence its name.14
Although the Tabula Peutingeriana is the only surviving graphical representation of the road network in Roman Britain, another document is arguably of greater importance, and that is the so-called Antonine Itinerary (Itinerarium Provinciarium Antonini Augusti). Navigation by map was probably rare, given the cumbersome (and presumably expensive) nature of Roman cartography, but itineraries – simple lists of places with distances between them, equivalent to the sort of information modern motoring organisations (or even Google) provide for drivers – may have been more widespread. Although purely textual in form, the Antonine Itinerary also covers the whole empire and provides a series of itineraria (routes), fifteen of which relate to the road system in Britain. Recording place-names and distances between them, they appear to be a record of a series of journeys, rather than any sort of recommendation of approved routes to take, and in this they resemble the medieval royal itineraries (see below page 110). Iter II, for example, from Blatobulgium (Birrens) to Ritupiae (Richborough) crosses the country three times unnecessarily (Birrens across the Pennines to York, back across to Chester, down to Wroxeter, then across once more to London and thence Dover). It would have made much more sense to either cross to York and proceed southwards down the east coast, or head southwards on the west coast and only cross upon reaching Wroxeter. The suspicion that this is not just a list of routes is heightened by comparing Iter V with Iter VI and Iter VIII. All three go from London to Lincoln (V continuing on to Carlisle and VIII to York), but take very different and, in the case of V, indirect routes (Figure 15).15
Figure 15: The Antonine Itinerary.
Some clue to the date of the Antonine Itinerary can be gleaned from the places mentioned, only two of which are not south of Hadrian’s Wall, whilst the title of the document itself is suggestive of a connection with one of the Antonine emperors (the first being Antoninus Pius, who came to power in AD 138). At least one scholar has argued for an association with Caracalla (or Marcus Aurelius Antoninus, to give him his formal name) and Septimius Severus, whilst others have pointed out that inconsistencies in the text may indicate that the routes have been accumulated over time.16
Even a cursory glance at a map of Roman Britain shows a scattering of forts, fortlets, and signal stations. The problem with an undifferentiated distribution lies in the fact that not all sites were occupied at once and that the sites demonstrate both a broad and a specific phasing. The two principal concentrations of post-Hadrianic sites lie in Wales and in the north of Britain. Earlier sites tend to lie in the Midlands and the south of England, for fairly obvious reasons (the Romans invaded the south and progressed northwards: the narrative of this progress may be disputed in fine detail but never in doubt in its overall scheme). The forts are linked by roads. Indeed they often occur at nodal points in the road network, which is only what would be expected if a logistical, rather than strictly strategical, interpretation were to be placed upon their distribution.
Perhaps the mistake scholars have made is always to think in terms of frontiers rather than networks. The distribution of the Roman army along certain elements of the road system in Britain was certainly a strategic placement in the sense that it was a logistical necessity – an army cannot be concentrated in one place for too long without straining its supplies – but by spreading it out in a network of strong points, it could also guarantee the security of its own supply chain. Thus it could be that these sites are as much inward- as outward-looking.
One writer at least noticed the illogicality of the frontier mode of thought when looking at fort networks in the East of the Empire. This was Edward N. Luttwak, a modern strategist and historian, who went on to formulate the theory of Defence in Depth when looking at that eastern Roman distribution of sites. In this, he suggested that the network of forts seen in provinces such as Syria was a response to the inadequacies of the peripheral frontier way of thinking and that by spreading troops into what was effectively a buffer zone, an invading enemy would be worn down, prone to attack from a number of sides, rather than free to roam in the hinterland once they had penetrated the hardened outer ‘casing’ of the frontier. His ideas attracted a considerable following at first (in what might be termed the ‘Invasion and Response’ phase of Romano-British studies), but fell into disfavour as a more critical eye was cast over his interpretation of the available evidence. Luttwak’s chief shortcoming (if such it be) was that he had viewed the ancient world with a modern eye. It must be understood that this was not necessarily a bad thing and that his rethinking of the situation was a refreshing breeze through the more cobwebby areas of Roman frontier studies. Nevertheless, it can be argued that he saw a genuine phenomenon but misinterpreted it from a modern perspective, influenced perhaps by the contemporary strategic situation pertaining in Western Europe.17
Frontier systems exhibit an intimate relationship with roads and not just because one usually accompanies the other. In order to understand the distribution of forts (and fortlets) along roads, it is illuminating first to examine a rather famous frontier. Romanists have long railed against interpreting Hadrian’s Wall as being defensive in nature, stating that it was never designed to be a fighting platform (despite the resolute silence of Roman authors on this fact; many seem quite happy with the concept of wall-top defence in other contexts). Luttwak could not help but take a strategic view of it and similar frontier works as ‘base lines for mobile striking forces, which operated against large-scale attacks in a tactically offensive manner’. Another modern writer, David Divine, former defence correspondent with The Sunday Times, went so far as to draw up a theoretical plan for how Roman forces could be concentrated in the field to the north of the mural barrier, so great was the Roman abhorrence of wall-top defence thought to be. This was the doctrine of open field conflict writ large: Hadrian’s Wall would never be defended by soldiers on top of it, but rather had to be managed by units in the field in front of it. The difficulty with such a model is the absence of the command and control element in the Roman world. Whilst radar made it possible for the Royal Air Force to defend Britain against the Luftwaffe by having advance intelligence of their approach, as well as an idea of their strength, the best historians could manage for Hadrian’s Wall was that the outpost forts (Birrens, Bewcastle, Netherby, Risingham, and High Rochester) might have been designed to gather intelligence and give similar advanced warning of enemy incursions. Units of scouts (exploratores) are attested from the last three of those sites. They of course lacked the facility to convey sufficiently detailed information quickly enough: signalling with smoke or fire could certainly communicate the fact that there was an incursion taking place, but little in terms of the necessary detail needed to decide how big a force should be deployed to deal with it. Nowadays those advanced forts are thought more likely to have served to protect friendly local populations.18
If Hadrian’s Wall was not a defence against invading armies, what was it? Over the past few years, archaeological work has shown how important the tactical component of the Wall was, with an increasing number of instances of berm obstacles being identified, mainly in the east so far. This, taken with the adjustments made to the system as it was built – milecastles adjusted from their hypothetical measured positions, additional towers included to cover blind spots, the careful use of terrain (notably the use of re-entrants and short stretches of ditch to cover gaps in the crags) to enhance the defensive qualities of the Wall, mean it is increasingly possible to view the Wall as a tactical, as well as, or even instead of, a strategic system. So what were all those forts doing along the Wall? There has been little serious questioning amongst writers that they provided the garrison for the eighty-one milecastles and upwards of 162 turrets it contained. Even if each milecastle contained only, say, twenty-four men on detachment (and assuming no permanent detachments within the turrets), a figure that fits within the estimates of between eight and thirty-two men for each milecastle, that would still have been a permanent drain of nearly 2,000 men from the theoretical maximum of 10,000 that those Wall forts could muster. How, then, a pre-emptive defensive system of the sort Divine envisaged was supposed to work is open to question. It is much less complicated to see the forts as the logistical hubs for the garrison of the Wall, facilitating the supply of men for its operation, and linked by a road (the Military Way) which ran from one end to the other.19
Such a view of Hadrian’s Wall helps explain why the Stanegate ‘system’ continued to operate in tandem with it and why the Stanegate forts could never have fulfilled what has been suggested as the original scheme for the Wall. Scholars have little doubt that the Wall as designed allowed only for the linear components (ditch and curtain wall), with milecastles every Roman mile and two equally spaced turrets between each of them. Troops would be supplied by the Stanegate forts, it is thought, but that the shortcomings of that proposal were soon evident, given the distance of some Stanegate forts from the line of the Wall and the additional time that would need to be taken to address an emergency such as an assault upon it. However, another interpretation might be that the Stanegate forts, designed to service the road and protect the movement of convoys along it, could not provide the additional capacity also to man the Wall. The Vindolanda Tablets provide penetrating insights into the operation of one of these pre-Hadrianic Stanegate forts, and a surviving pridianum, a document summarizing unit strength at a given point in time, makes it clear that the cohors I Tungrorum, whilst having a paper strength of 752 men, could only actually count on 265 fit for duty within its walls, since so many were either outposted or sick. The additional impact of having to provide more troops to man the new frontier can only be imagined. Hence the so-called ‘forts decision’ can be seen as a pragmatic recognition that road forts could not be expected to undertake additional tasks and that dedicated forts were needed for the Wall to act as resource centres for its operation.20
So, given the possible tactical nature of Hadrian’s Wall, keeping low-intensity threats to the northern inhabitants of Roman Britain at bay, why were forts needed to protect convoys to the south of it? The answer must lie in the fact that the two systems were dealing with different problems: one looking outwards, the other inwards. Brigandage was endemic in the Roman world but increased to epidemic proportions during the instabilities of the third century AD, with large numbers of individuals, including renegade soldiers, dropping out. It was clearly never envisaged that supply trains would be safe to move about without the protection offered by the army in its fortified strong points and also, presumably, by virtue of its patrolling the fortified roads. Naturally, if towns were included amongst the strong points, this would hold true of the whole province, not just the ‘military zones’ defined by the distribution of forts. Effectively, the whole of Britain was a military zone so the problem (the lack of security for supply convoys) was a universal one.21
So it is that Luttwak’s identification of the importance of fort-and-road networks should not be dismissed, but for different reasons from those proposed by him. He saw their distribution as strategic in the defensive sense, but it is equally feasible to view them in a logistical light. Moreover, these networks are not, as he thought, a feature of just the High or Later Empire, but can certainly be seen operating in Britain from the very beginning of the conquest. Early forts have now been identified at many sites in Britain to the south and east of the later military zones of the North and Wales. Some lie beneath later towns and have been identified from structural or, less certainly, from artefactual evidence. Others remain on what are still greenfield sites, rendered redundant by the advance of the army north and westwards. A series of sites along the Trent valley – Broxtowe, Osmanthorpe, Newton on Trent, and Marton – would seem to demonstrate this as a series but there are other isolated examples which must reflect changes in the perceived strategic requirements of the road network. Thus the overall network of forts that can be plotted from the map of Roman Britain (Figure 16) is inevitably a palimpsest of networks, stretching from the earliest Roman occupation right through to the latest forts, such as the Saxon Shore system. We can crudely phase it into pre-Flavian (Figure 17), pre-Hadrianic (Figure 18), and Hadrianic and Antonine (Figure 19) in order to see just how the networks change with time (it can, and has, been phased more finely many times). Once towns are factored in (Figure 20), it becomes clear that an evolving system can be made out, but it seems likely that this is only a crude approximation of the realities of the situation, since we may suspect a military presence at smaller sites like villas, wayside stations, and beneficiarius posts too. A case can therefore be made that Roman Britain was thoroughly militarized.22
One reason for this militarization, often overlooked, is the absence of a police force to provide civil security. Although we know the city of Rome had the vigiles, a watch force that combined policing and fire-fighting in its duties, and there is even anumerus vigilum attested from Britain, it has to be doubted whether these operated countrywide, rather than on just a per-settlement basis. Security in all its manifestations had to be provided by the army (and the beneficiarii were of course outposted soldiers). Modern armies tend to be uncomfortable with policing roles but there were no such scruples in the Roman world. The military were integrated into the justice system (military officials passed judgement in civilian spheres, military officials such as the regionary centurions had a role to play in provincial judicial administration) and thus their role in security was as much to deal with the enemy within as with the invading barbarian.23
Throughout it has been stressed that most major roads in Britain were used by the army, but it is worth asking how a campaigning army actually exploited a road. It is clear from contemporary accounts of army movements that no attempt was made to fit all the personnel and baggage onto the metalled surface. Roman cavalry were unshod so almost certainly did not use metalled roads. This meant that a marching army would have had its infantry and baggage train moving along the surfaced road, but its cavalry (and perhaps a screen of infantry scouts) operating to either side of it, so a proper military road was in fact not just the surfaced area, but also the cleared zone on either side of it. Accounts of Roman armies on the march confirm that this was often the case.24
Figure 16: All known and surmised forts in Roman Britain.
Figure 17: Pre-Flavian forts in Roman Britain. It is likely that more remain to be found.
Figure 18: Pre-Hadrianic forts in Roman Britain.
Figure 19: Hadrianic and Antonine forts in Roman Britain.
Figure 20: Towns (and later forts) in Roman Britain.
On a more mundane and everyday level, the army had to move supplies between its bases, and collect taxes delivered in kind (Tacitus mentions an abuse of the system whereby farmers might be made to transport goods long distances for this purpose), but probably also used roads for the shipment of bullion to fund their activities and coin to pay soldiers.25
Vegetius provides an important insight into the way the system operated, recording that troops had to be placed in fortified locations, either towns or forts, in order to safeguard the movement of supply convoys:
Amongst those things the commander will provide, whether based in camp or in a city, are grazing for the animals, transport of grain and other things, water collection, wood gathering, and foraging, made safe from attack by the enemy. Because otherwise it is not possible for our supply convoys to pass back and forth, if garrisons are not distributed at certain points, whether these should be cities or walled forts. For, if not already fortified, forts are quickly set up in suitable places, surrounded with large ditches (forts are named from the diminutive term for camp). Living in these, moderate numbers of outposted infantry and cavalry keep the route safe for convoys. For an enemy only dares to approach such places with difficulty, where his adversary provides support in front and to the rear to delay him.26
This is something of a chicken-and-egg situation. However, as such posts would themselves need supplies, so the very fact of establishing the network would have ensured its continuation. Inevitably, it is easier to supply a distributed army than a concentrated one and it can be argued that the ‘network’ of Roman military bases throughout Britain (and, indeed, other provinces) is as much a product of supply requirements as it was defensive or offensive. It is at this point that scholars start to think in terms of frontiers or frontier zones, and fortified roads such as the Stanegate come to be discussed in terms of a frontier system without any clear evidence that the Romans themselves viewed them in these terms.27
Moreover, that passage in Vegetius – usually at least partially dismissed as a late epitomist – bears closer scrutiny. Careful analysis of the text, combined with Vegetius’ own cursory list of his sources, allowed Dankfrid Schenk to suggest that it comes from a section of the work thought to be derived from the lost De Re Militari of Sex. Iulius Frontinus, formerly the governor of Britain (or, more accurately, as legatus Augusti pro praetore, the commander of the army in Britain) during the first century AD. Comparison of Frontinus’ text in his surviving Strategemata (originally an appendix to the De Re Militari) revealed telling stylistic similarities. Frontinus’ credentials as both a tactical and strategic writer are therefore of the highest quality and lend weight to Vegetius’ words.28
Not only does this passage tell us what forts were for – protecting convoys – but it also reveals that they must be seen in the context of garrisoned cities. Analysis of the finds from the towns of Roman Britain has revealed a consistent pattern of items of military equipment from the second and third centuries AD. Now, whilst there are several possible explanations for the presence of such artefacts at sites, the combination of the distribution of towns and forts in Britain from the second century onwards would appear to support Vegetius’ (or Frontinus’) claim (Figure 21). Earlier writers thought in terms of a ‘military zone’ in Britain, in simple terms seen as the North of Britain and Wales, but it may well be that this was a misunderstanding of reality: the whole of Roman Britain was ‘militarized’.29
By comparison, it is interesting to note how the Bozeman Trail in the American Mid-West, linking the Oregon Trail with the towns of Bozeman and Virginia City, was watched over by a series of forts – C.F. Smith, Reno, and Phil Kearney – all dismantled when the treaty of Fort Laramie was signed (Figure 22). The Myos Hormos road in Egypt provides a documented Roman example of a militarily protected route elsewhere in the Empire, with fortlets and towers protecting both army interests and those of civilian trade along it from tribal attacks.30
The picture we gain of Roman Britain, therefore, is one of a province that lives with an element of insecurity (perhaps crime would be a better term to use), the effect of which is ameliorated by the presence of the army at crucial points along key routes. The statement recorded by Vegetius makes it likely that these routes were the ones in which the army had a direct interest, since they comprised at least a part of their lines of supply (allowing for the fact that transport by sea must have been an important component in the overall resupply picture).
Figure 21: The combined distribution of towns and forts in Britain.
Figure 22: The Bozeman Trail and its forts (after Waldman 2009).
So what was being moved in these convoys? During the civil war of AD 69, Tacitus relates how rebellious troops intercepted a convoy carrying money to pay some allied troops. This tale reflects upon what must have been a regular occurrence, the shipment and escorting of pay for the army, which most units would have had to have received by road. Wherever possible, they would probably have exploited water transport, but it was inevitable that the road system had to be used at least part of the time. An impression of the extent of its usage may come from the distribution of lead ‘pigs’ around Britain, since they are almost invariably found next to roads (Figure 23). These were the product of the silver- (and lead-) producing mines, which were of course directly controlled by the government. Mines under state control within Britain were producing a range of important raw materials, but few are more prominent in the archaeological record than the silver/lead mines (a state-controlled monopoly). Pigs of lead, often with their silver content already removed, were taken to coastal ports, possibly for shipment abroad. These cast (and often inscribed) pigs were so heavy that a Roman waggon could probably only take two at a time. A number of these ingots have been found, sometimes leading to suggestions of skulduggery on the part of some of those involved in the production or shipping of the lead.31
What cannot now be judged is the extent to which the precautions suggested by the fort network were actually needed: could the perception of threat have been greater than the reality? One possible insight into this problem can be gained by that same distribution of lead ingots, or pigs, around Britain (Figure 23). Surviving pigs carry official stamps moulded into their tops that clearly identify them for what they are. Nevertheless, a proportion of these were ‘lost’ during transport. Given the size and weight of each one, accidental loss is scarcely plausible in all instances as an interpretation for their presence in the archaeological record, particularly since the government will have had a direct interest in (and almost certainly a record of) their existence and movements. Some have come from sites next to water (as with two bars from Bitterne), perhaps as a result of sunken ships, whilst others come from mine sites (like Charterhouse), but a significant number have been found near roads, but far from either a likely source or a probable destination. It is at least possible, therefore, that the surviving items are the result of successful raids on convoys which were not subsequently retrieved by the villains concerned. As with coin hoards, what is unknowable with caches of pigs is how many were retrieved and thus do not survive for us to find; we can only suppose that it was a greater number than those actually recovered.32
Figure 23: The distribution of lead pigs (based on location data in RIB I and III).
Lead pigs also raise an interesting point about the manner of freight transport on roads. One lead pig weighed in the region of 60–70 kg. Whilst that is within the capabilities of a mule, it would be difficult to pack as a balanced load. Two pigs, however, is beyond the capacity of most mules, so balancing a load with two pigs would be out of the question. This might seem to suggest that wheeled transport was necessary to move lead pigs, but we cannot rule out the possibility that some form of secure pack harness was used to hold one pig in place in the centre of the animal’s back (although it would entail an alarmingly high centre of gravity for the load).33
The Vindolanda writing tablets provide further evidence on the operation of Roman military supply. One Octavius, writing to Candidus (at Vindolanda), bemoans the fact that a shipment of hides from Cataractonium (Catterick) has been delayed ‘because the roads are bad’ (dum viae malae sunt) and he does not want to injure his draught animals (iumenta). He evidently has a financial interest in this material, although the status of the two is unclear. Major supply shipments must also be seen in the context of the personalized trickle of items moving around the Empire, often accompanying travelling soldiers. Letters home asking for socks or pieces of equipment are known from Egypt and Britain, and this trade was evidently inter-provincial and was all presumably at the very least assisted by the cursus publicus, the official posting system.34
Just as Roman roads facilitated the movement of the army, both on a large and small scale, and helped it supply itself, they would also have had an important symbolic function (whether intentionally or not is open to debate). Comparison of an all-weather, metalled post-conquest road with its native predecessor must have been a powerful indicator of how much things had changed since the arrival of Rome. As for the native British attitude to the roads, there are certainly stories of Highland Scots refusing to use the eighteenth century military roads, even to the extent of avoiding setting foot on them and wading streams next to new bridges. Nevertheless, since much of the Roman system appears to have reused native predecessors, if similar obstinate token displays were undertaken, it must be wondered to what degree they were followed by any more than a small minority of the population, given the eventual longevity of the roads. Similarly, as was noted by the Commissioners for Highland Roads and Bridges (above), the unintentional effect of an enhanced road network on civilian trade may have been considerable too. There are, nevertheless, theoreticians who see Roman all-weather roads as a manifestation of Roman imperialist ambitions and decry pragmatic interpretations of the network, and there may even be a kernel of truth to that assertion, but it ignores the link with the pre-Roman system outlined above and inevitably underestimates and downplays the capabilities and intentions of the native Britons.35
The road system of Roman Britain was by no means perfect. As has been pointed out by a number of authors, some sections may have been impassable to draught animals due to impracticable gradients (see below, page 105). Moreover, even though Britain presented fewer challenges than many of the continental provinces, extreme weather conditions would have rendered many sections of road useless at times. North of the River Tees, Dere Street (Margary 8) enters the Pennine foothills and even today, its modern successor (the A68) is often blocked by snow when coastal roads (such as the A1) experience no such difficulties.
Such problems were vividly highlighted by the discovery of the dum viae malae sunt letter from Octavius about that cargo of hides from Cataractonium. Such a load would, of course, have been required to come up Dere Street (Margary 8) to Corbridge and then along the Stanegate (Margary 85a) to Vindolanda, a total distance of around sixty-six Roman miles (98 km). Interestingly, Octavius’ letter contradicts some writers who have seen all freight transport as having used waterways and confirms the views of those who see evidence for the use of a more balanced mix.36
Available transport and its limitations
The range of modes of transport available to users of the Roman road system in Britain was fairly limited. They could ride, walk, or be pulled and those options more or less reflect the range of speeds, from fastest to slowest.
Riding on horseback would have been the preferred means for government couriers, supported by the remount system at way stations. There is anecdotal evidence for the speeds that might be attainable in this way, as in the case of the future emperor Tiberius riding from Mainz to Drusus’ death bed in twenty-four hours, covering a distance of about 200 miles at an average speed of eight miles per hour. However, it is difficult to see this as anything other than an extreme case which cannot be used for anything other than to show the limits of what was possible.37
Walking was of course the way the bulk of the army moved around, covering about twenty – in extreme circumstances twenty-five – miles a day. Undoubtedly, a large proportion of those who were not military who used the roads would have chosen this way to get around. Being carried in a litter was a possibility for a select few, but probably reserved mainly for use in town.38
Freight, and any official users of the cursus publicus who did not merit the use of the remount system, would probably have gone by draught or pack animal. Two- and four-wheeled carts might be pulled by horses, mules, or oxen, the last of these probably only achieving an average of two mph. This produces a figure of around sixteen miles per day travel, assuming eight hours of travel, which interestingly approximates with the spacing between many neighbouring military bases in Northern Britain. Octavius’ shipment of hides would have passed from Catterick to Vindolanda via Piercebridge, Binchester, Lanchester, Ebchester, and Corbridge, in other words a five-day journey by cart. The Antonine Itinerary, on the other hand, breaks the journey into three sections: Catterick to Binchester, Binchester to Ebchester, and Ebchester to Corbridge (the section Corbridge to Vindolanda is not included in any of the itinera), each of around twenty miles. Given that an acceptable speed for pack animals might be around 4 mph, then three days for the journey from Catterick to Vindolanda was easily achievable and even two days not impossible.39
Fort spacing from Iter I
Diocletian’s Edict on Prices suggests that the preferred capacity for waggons was 1,200 Rlbs (393 kg), but such ideals almost certainly did not reflect actual practise, merely what the state hoped would be done. Some writers have cast doubt upon the ability of draught animals to haul loads up some of the inclines on the road system of Roman Britain. Anderson, on the other hand, has shown how actual loads carried could differ widely from those legislated for under Diocletian’s Edict on Prices, by perhaps as much as a factor of three in magnitude. He suggests that it would only have taken 143 ox-carts of grain per year to supply a cohort-sized military base. Since Anderson’s work, des Noëttes’ seminal work on the limitations of Roman draught harness has been questioned further, suggesting loads of 1,000–1,500 kg would not have been impossible. Nevertheless, whilst the various claims and counter-claims in this debate remain unproven by any pertinent scientific data or experimental archaeological work, it is probably best to suspend judgement on just how steep a hill a Roman ox-cart could negotiate when laden. To get to Vindolanda, Octavius’ shipment of hides would have had to come by a route that required a total height gain of some 279m, with maximum slopes in the region of 9 per cent, but since he was intending to use pack animals (iumenta), this would have been less of an obstacle than it would to waggons. Pack animals are attested elsewhere in our sources and could have been used to carry a range of goods, providing they were not too bulky. They are also the obvious solution to objections to Roman roads being too steep in places for draught animals pulling wheeled vehicles, although evidence suggests that the Roman army would have used both pack and draught animals when moving around – they are certainly featured on Trajan’s Column (Plate 14).40
It has recently been suggested that the role of water transport has been seriously underestimated by scholars in the past. Whilst this is almost certainly true as a generalisation, some of the more extreme claims, such as the widespread use of damming to enhance navigability of natural water courses, have been thoughtfully challenged. Some Roman military sites (such as the bases at Whickham or Housesteads) defy any sane attempt to get a barge anywhere near them for purely topographical reasons. Exploitation of coastal transport, along with the navigation of major rivers, is beyond doubt and well-attested by the evidence, but the use of an intensive local canal system is sadly unproven. This means that short-haul freight transport will have to have used the available wheeled transport or pack animals.41
Which side of the road did the Romans use?
A perennial favourite, the question of which side of the road the Romans used says much about modern perceptions (and misconceptions) of the past and much ink has been spilled on this subject, notably in the letters pages of Current Archaeology. It has even been suggested that ‘proof ’ that the left-hand side was favoured has been excavated (although the idea that such a concept is capable of proof is, to say the least, open to question). Ruts were found to be deeper on one side of a road leading to a quarry than the other, from which it was inferred that laden carts travelled on the left, empty ones on the right.42
Although the British use the left-hand side of the road when driving, there is seldom a comparable rule for pavement areas or pedestrianized zones; the nearest to such a rule for non-vehicular traffic can be found in some rail and underground stations in London, where pedestrians are advised to keep to one side or the other when using the stairs or a corridor. Such suggestions are frequently disobeyed. Nevertheless, practical considerations tend to make those on foot use one side of a pavement going one way, one the other, but there will invariably be individuals who dodge around as they do not wish to move at the same speed as the mass.
In reality, the main user of the roads – the army – used the whole road when on campaign, a column being flanked by its cavalry on the softer unmetalled margins (Roman cavalry were unshod; see above, page 59): any other road-user would simply have had to get out of the way.