Ancient History & Civilisation

Margary’s (1973) network of Roman roads in Britain.

Chapter 1

The Prehistory of Roman Roads


What did the Roman find when he arrived in the first century AD? He found a trackway already 2,000 years old. It was not engineered, and would have abounded in hollows, ruts and obstructions of all kinds. At intervals along the route there were the banks and ditches of the Early Iron Age period, demarcating the territorial boundaries. Here was a route which he could use. From the several more or less parallel tracks he chose that which was most direct and most suitable, and straightened it where necessary. The Viatores1

Compare and contrast that view with this:

It is sometimes said that most main Roman roads in Roman Britain are based on pre-existing British tracks. While it is, indeed, certain that there were such tracks, and that they had clearly developed widely before the Roman conquest, even a general account would be uselessly fragmentary, since virtually nothing is known of them in detail. The well-attested examples, however, do not coincide in general with the Roman roads, which were unquestionably designed as instruments of conquest, as in other provinces. Collingwood and Richmond2

There is a popular misconception that the Romans brought the idea of roads to Britain (Plate 1), but nothing could be further from the truth. Every few years, scholars rediscover the idea of pre-Roman roads in British archaeology and then seemingly forget about it again. Dramatic headlines greeted the recent discovery of what appeared to be Iron Age road surfaces at Sharpstone Hill (Staffordshire); likewise, at the time of writing, a brand new book makes dramatic claims about the prehistoric origins of ‘Roman’ roads. However, earlier generations of historians and archaeologists had little trouble in dealing with the concept of roads before the Romans. Indeed, this very notion led Alfred Watkins to propose his Old Straight Track theory which, for all its flaws, recognized the importance of roads in prehistory. One of the most interesting approaches to Roman road studies in recent years has been a diachronic study of Akeman Street from the prehistoric into the Roman periods. What Sharpstone Hill provided was convincing archaeological evidence brought to the attention of a wider public.3

However, it is not necessary to cite large numbers of excavated examples of prehistoric roads in order to show that roads were not a Roman invention. To demonstrate this, we need only consider what might be termed Plautius’ Dilemma, which (hypothetically expressed in the modern form of a multiple choice question) is this:

You are the commander of the Roman invasion force of about 40,000 troops. Arriving at the coast of Britain in AD 43. Do you

a) begin building all-weather roads to move your troops towards their ultimate goal of Colchester,

b) start marching them (and their baggage train) across country towards that destination, or

c) make use of existing roads to achieve the same ends?

It does not take a genius to work out that a) will take too long (see below page 19), b) will take almost as long and be completely impractical for wheeled transport (historically, armies have never marched across country when it can be avoided), and that c) implies and requires the existence of pre-Roman roads of some sort. Assuming – for the sake of argument, as it is by no means universally agreed upon by scholars – that Plautius landed on the south-east coast of Kent at Richborough, was there an existing route available to him?4

The so-called Pilgrim’s Way was a ridgeway along the North Downs. Originating at Winchester, it passed through Guildford, Farnham, Snodland, Charing, to Canterbury, and originally continued as far as Dover (Figure 1). Despite its name, this was probablynot the route used by Chaucer’s pilgrims in the Canterbury Tales (who will more likely have followed the metalled Roman road – Margary 1). Four of the five places mentioned by Chaucer (Deptford, Greenwich, Sittingbourne, and Boughton) are indeed situated on the Roman road (as is the point of origin, Southwark) and another (assuming Bob-Up-And-Down to be Harbledown) probably was. However, the ridgeway does provide the most natural route from the south-east coast to the lowest Thames crossing and its use by the Roman invading force has long been accepted as a possibility by modern writers.5

Figure 1: A ridgeway? The Pilgrim’s Way (and its Roman successors).

So how did the Roman conquest of Britain progress from that point? It is not unreasonable to suggest that it continued in the same vein – using existing routes. Therefore, before we consider the Roman network, we must examine this prehistoric system and how it relates to its Roman successor.

Types of prehistoric roads and tracks

The existence of prehistoric trackways (Figure 2) is certainly accepted and referred to by modern writers – although the antiquity of the so-called ridgeways has been called into question – and in recent times some have even been excavated, most prominently the wooden trackways of the Somerset Levels. There are indeed writers on the subject of Roman roads who acknowledge the debt the Roman system owes to its native predecessor, including Hillaire Belloc. Nevertheless, the idea that roads that could be used by wheeled transport were already in existence at the time of the Roman conquest is seldom voiced, despite the fact that such early vehicles are indeed archaeologically attested: the earliest wooden wheel from Scotland (and, it so happens, Britain), for instance, dates toc. 1255–815 BC. However, one scholar (who had better remain nameless) remarkably even went so far as to suggest that wheeled Bronze Age vehicles were purely for the purposes of display, since no roads existed, so it seems clear that the role of the road in prehistoric Britain has not been overemphasized.6

Figure 2: The prehistoric trackways of central southern England according to Hippesley Cox (1944).

Given that excavation of them is so rare, proving that a route is prehistoric is by no means easy. It is not as if one can trace the outline of a footprint and identify it as pre-Roman. Some of the upland trackways and ridgeways in Britain have a fairly obvious antiquity if they can be identified over long distances but have not been made into modern roads. Moreover, in the case of both such ‘obvious’ candidates and their less-obvious brethren, close association with prehistoric monuments may be a good indicator of a direct and tangible relationship (although it does tend to beg the question ‘which came first: route or monument?’). Matters are made even more complicated by the tendency of all roads to ‘creep’ laterally across the landscape, and we shall be returning to that problem later in a Roman context (below, page 104). Nevertheless, it is probably true that we ought to be thinking more in terms of prehistoric ‘routes’, with often more than one track running parallel and their use perhaps dictated by seasonal conditions.7

Fortunately, some trackways are demonstrably prehistoric. As we have just seen, wooden footpaths across the Somerset Levels have been excavated and, by means of dendrochronology, these can be dated very accurately to the Bronze Age and even into the Neolithic. Although not used by wheeled traffic, they nevertheless demonstrate a familiarity with routes and the need to produce them where they did not already exist. It would surely be unreasonable to believe that these were the only such tracks in prehistoric Britain. Indeed, in Ireland, a 3.5–4 metre-wide wooden plank trackway (at Corlea in County Longford), dating to 146 BC, appears to have been specifically designed to take wheeled vehicles. It is difficult to avoid the suspicion that trackways preserved by exceptional environmental conditions – such as bogs – may be just the visible component of a largely invisible network of trackways and routes across the British Isles. Possible (but by no means certain) Bronze Age and Iron Age bridges have been identified on the Thames at Eton and Vauxhall.8

In fact, rural pre-Roman roads are known and, once again, some have even been excavated. A track has been identified crossing the Bronze Age landscape of Holne Moor in Devon. Further north, a type of Iron Age settlement, known as a ‘ladder settlement’, was in fact a linear landscape focused on a trackway of some kind (which happens to look like a ladder laid out on the ground when viewed from the air). Excavation of an example at Melton (E Yorks) failed to find any evidence of metalling but did locate roadside ditches and confirmed the complex nature of this Iron Age and Romano-British rural settlement. At Mount Pleasant, near Crambeck (N Yorks), a similar settlement showed such continuity from the Iron Age into the Roman period (Figure 3), laid out along a side road leading to the main Roman road from York to Malton in North Yorkshire (Margary 81a). Not only was the surface metalled (albeit using the living rock), but wheel ruts were evident in the pre-Roman surface. The fact that the road served as a link to the Roman-period road points to it having served the same purpose in the earlier period: clearly, one might conclude that not only the side road but the road from York to Malton itself pre-dated the Roman invasion. Minor roads like these were certainly a feature of the Romano-British landscape, but it is becoming clear that these are the legacy of an earlier period. It is perhaps worth noting that examples of such pre-Roman roads associated with rural settlements tend to be slightly sinuous and not straight. In this, they contrast with formally-constituted ‘Roman’ roads (although many evidently continued in use into the Roman period) and, crucially, with the proposed ley line routes favoured by Watkins as prehistoric roadways.9

It has already been mentioned that a number of long-distance trackways are known that are commonly presumed to be prehistoric in origin. These include famous routes like the Ridgeway (now formalized as a National Trail from Overton Hill in Wiltshire to Ivinghoe in Buckinghamshire) and the Icknield Way (from Ivinghoe to Knettishall in Norfolk). As Rackham has pointed out, a principal reason for identifying these as prehistoric used to be the rather misguided notion that early people only settled (and travelled) on high ground, since all the valleys would be full of woodland and impassable rivers. We now know this to be untrue (not least because of evidence like that from the excavations in the Somerset Levels), but as most of them do pass close to prehistoric monuments, the claim that they are prehistoric is probably true. At the same time, it might be argued that it would be very difficult indeed for a route to pass through much of Britain without passing close to at least a few prominent monuments (Plate 2).10

Figure 3: Trackway with a ladder settlement at Mount Pleasant, Crambeck (after Abramson et al. forthcoming).

At least two trackways in the Peak District have been identified, passing a number of prehistoric monuments, including henges and hillforts, such as Arbor Low. The connection between prehistoric (and Roman) routes with prehistoric monuments is prominent and probably more than coincidental, as we shall see.11

Deduced examples

In some instances, the existence of a prehistoric route is indicated by its partial re-use by a Roman road. At Roecliffe, near Boroughbridge in North Yorkshire, a Roman fort guards a probable river crossing for a north–south route running parallel to the Roman Dere Street (Margary 8b), which itself crossed the Ure a mile to the east, where a fort (and later a town) was built at Aldborough (Isurium Brigantum, Figure 4). Interestingly, the Roecliffe crossing appears to have been indicated by a group of four standing stones, the Devil’s Arrows; the coincidence of standing stones with Roman roads is not unknown in other locations (see below, page 14). To the north of Roecliffe, on what might be termed the Great North Route, there is an impressive series of prehistoric monuments (including several henges, notably those at Thornborough) and, once again, their proximity to the road is striking.12

A similar situation can be identified near Corbridge in Northumberland, where an earlier Roman fort (Beaufront Red House) guards a crossing to the west of that later used by Dere Street (Margary 8d) and subsequently guarded by another fort overlooking a stone Roman bridge over the Tyne.13

A further example of Roman roads using prehistoric routes may be provided by Badbury Rings in Dorset, where two Roman roads meet at a hillfort (Figure 5). Field (1992) has gone further and suggested that other hillforts in the area may have been linked by a network of Roman roads, but if this was indeed the case, it is possible that these routes also had a pre-Roman origin. Across the Channel in France, just such a network of roads between hillforts has been identified in Burgundy, so the likelihood that something similar existed in Britain must be considered. There was a time when the proximity of a Roman fort to a hillfort would be remarked upon as a likely reason for the existence of the Roman fort (despite the fact that many such sites could not be shown to be contemporary), but it might be more credible if we see them both in the context of the same thing – their respective road systems. It is conceivable that later, more sophisticated, prehistoric oppidum-type settlements, such as Colchester-Camulodunum, St Albans-Verulamium, and Silchester-Calleva, may have anticipated their Roman successors and have lain at the heart of polity-centred networks of trackways.14

Figure 4: The east–west road (and north–south trackway?) at Roecliffe (after Bishop 2005).

Yet more evidence of the Roman use of prehistoric tracks may be provided by the locations of temporary camps. Two series in particular are of interest here.

Figure 5: From nodal point to redundancy I: the Roman roads at Badbury Rings.

The first runs up the south bank of the River Tweed between Tweedmouth (Northumb) and Maxton, near St Boswells (Borders), and is difficult to explain if it is not utilizing an existing route. The second, and probably the best known, is the line of camps that runs up the east coast of Scotland, representing multiple campaigns. Scholars have attempted to classify and date these, but for our purposes what matters is the fact that a metalled Roman road has only been traced as far north as just beyond Cardean (Tayside), whilst the camps continue about 140 km further up to the shores of the Moray Firth. This last route was also probably exploited by Edward I, once he too had exceeded the limits of the Roman system.15

Mastiles Lane in North Yorkshire was a monastic route which used to communicate between the estates of Fountains Abbey on either side of the Pennines. As it crosses Malham Moor, it passes through the Roman temporary camp at Malham (Plate 3), suggesting that the route not only pre-dates the monastic period, but already existed when the Roman army was campaigning in the area (after AD 71). It also hints at the fact that the Romans were interested in east-to-west movement in this area during this time (perhaps in concert with the use of the A66 route (Margary 82) during the advance on Carlisle (Cumbria)). Mastiles Lane went on to be used as a drove road after the Dissolution and survives today as an enclosed ‘green lane’ (and is still used by walkers, horse-riders, and until recently off-road vehicle enthusiasts). It was never officially adopted into the road system of any period, but nevertheless remained important from the prehistoric period onwards.16


Other than those few wooden trackways that can be dated by dendrochronology, the dating of prehistoric routes must – with the exception of rare examples such as Shapstone Hill – usually derive from the monuments associated with them. Hillforts, tumuli, henges, and standing stones all fall within this group and it is not hard to produce examples of each. What is lacking, at the time of writing, is any sort of statistical analysis of this apparent relationship. It is possible that we are seeing some routes that date back to the earliest occupation of the British Isles by mankind, but it is equally possible that each age has added to this proto-network to meet its own particular needs, and the invention and spread of the wheel must have had a key role to play here.

We might presume that the first people to move across Britain will have followed animal tracks, and that some of these will have broadened with use, others fallen from favour. That the routes of some of these are still in use to this day seems at least a possibility. In the American West, animal trails were used by the native Americans pursuing prey, adopted as waggon trails by settlers, and later followed by both railways and modern paved roads. The bulk of the prehistoric road system may well have come about during the Neolithic period in association with an increasingly settled lifestyle and the concomitant need to engage in trade.17


This brings us to the question of how prehistoric roads were used. It is only too easy to assume that, from our modern perspective, they functioned in the same way in the past as they do now, but that would be naïve at best. Mankind’s fondness for ritual led other civilisations and cultures to exploit roadways in ceremonial and such a function has been suggested for the Avenue at Avebury and at Arbor Low. Archaeologists are only just beginning to understand how the people of the past interacted with the spaces around them. A road need not be just a road, but it could be a route for ceremonial processions. If this seems vaguely ridiculous to those unencumbered by the archaeologist’s obsession with ‘ritual’, we need only recall the Via Sacra in Rome, used in triumphal processions from the Republic right through the Imperial period. Roman Republican history provides hints at a darker side to the ceremonial use of roads, with the rebels of Spartacus crucified along the road from Capua to Rome. Roads have a place in ritual that is only dimly remembered now in the final ceremony of the cutting of a ribbon to open a new road (along which the celebrants will, of course, duly process).18

Roads facilitate travel. It is not impossible without them, and many bush peoples may have travelled vast distances all their lives without the benefit of any sort of road, but they do undeniably make travelling easier. As a direct result of their existence, trade will flourish. Some of the great prehistoric roads were trade routes – the Amber Route, the Silk Road – but what was true on an international scale was certainly even more so at a national or regional level. Travelling by ship or boat along coasts and up rivers was natural, but not all rivers are navigable, and do not always lead to where they are needed, so this could be supplemented and complemented by trade along roads.

Roads help armies. Just as it is true in later periods, it is important to remember that roads in the prehistoric period will have had a role in warfare. After all, the inhabitants of pre-Roman Iron Age Britain were given to using wheeled vehicles in war as well as other forms of wheeled transport. The Romans certainly borrowed their technical terminology for carts wholesale from the Celtic language, the clear implication being that the ‘Celtic’ peoples of Europe were pre-eminent in the construction of wheeled transport. In Britain, one of the earliest eye-witness accounts (that of Caesar) makes special note of the British use of chariots in warfare in the first century BC. They are mentioned again in the context of the Battle of Mons Graupius under Agricola, over 100 years later. Excavation has produced evidence of such vehicles, most famously from the chariot burials of East Yorkshire, amongst the Parisi, where their timber structure was preserved as ‘ghosts’ in the soil. The recent discovery of a chariot at Newbridge, near Edinburgh, demonstrates a further northern British example of their use and confirms Tacitus’ account. Four-wheeled waggons or carts are also mentioned in relation to the final battle of the Boudican revolt in AD 60/61 and what must be similar vehicles are shown in relief on the metopes of the Tropaeum Traiani at Adamclisi in Romania. It is a fundamental law of wheeled transport that it always moves faster on a road of some kind than if it has to move off-road or cross-country (little has changed in that respect and even the most powerful off-road vehicle, whether it be a wheeled Land-Rover or a track-laying tank, will invariably travel faster on a road). The wheeled vehicles of the Britons were no different.19

Prehistoric roads will also have been used for transhumance. Drove roads existed in the medieval period for taking animals to market (Dere Street providing an important route for Scottish cattle going to Stagshaw Fair near Corbridge) and may well be earlier in date in some instances. In areas of Britain with upland pastures, there will also have been transhumance routes between summer and winter pastures. Elsewhere in the empire, the Romans sought to control (but not stop) such movements of livestock and people and in Britain the Knag Burn gateway through Hadrian’s Wall – not itself on a Roman road – has been suggested as having been placed on just such a route.20

Early roads may also have served the dead as much as the living. The Roman use of roadside burial places is well-known (see below, page 34), but it is not beyond the bounds of possibility that this association between burial and roads may reach back beyond the Roman period in Britain. The linear distribution of tumuli in some regions has been commented upon before now and in some places the use of burial mounds continued into the Roman era (as at High Rochester and Great Chesters). Possible Bronze Age mounds called Coney Hills lie next to a Roman road near Gilling in North Yorkshire. In later periods, dedicated ‘corpse roads’ carried the dead to their final resting place; an example in the Lake District led from Mardale Green to a church at Shap (both Cumbria), some 11 km.21

From a practical point of view, whilst it is acceptable to manoeuvre war chariots over a field of battle, off-road use for long-distance travel was simply not practicable. Therefore, mustering a chariot army and deploying them to the scene of battle will have required some sort of road system over which they could be moved, always assuming that they were not in some way transported to battle (like modern tanks, which use specialized low-loader vehicles), perhaps being disassembled and carried on pack animals.

Relationship to monuments

The apparent proximity of Iron Age hillforts to Roman roads has already been mentioned. In Burgundy, a network of pre-Roman roads has been identified linking hillforts; in this case it is distinct from the Roman road network, whereas in Britain, the frequent association of hillforts with Roman roads suggests a higher degree of integration of the pre-Roman and Roman networks. At Badbury Rings in Dorset, two Roman roads (Ackling Dyke from Old Sarum to Dorchester, and the Hamworthy to Bath road) crossed immediately to the north of the hillfort (Figure 5), and other examples of hillforts close to road junctions can be cited. Old Sarum lay at the junction of three roads, as did Silchester (Figure 6). The fact that both of these were major pre-Roman settlements which later developed into important Roman sites cannot negate the coincidence between their location and the Roman road network. That coincidence is obviously more prominent at sites that did not continue into the Roman period, but it does not make the observation any less significant because they did not have a later existence. Thus it is scarcely surprising that the junction of the Fosse Way (Margary 5b) with the road from Winchester to Charterhouse-on-Mendip (Margary 45b) is marked by the hillfort at Beacon Hill, a few miles to the east of Wells, nor that the east–west road is paralleled by a line of round barrows.22

Figure 6: From nodal point to redundancy II: the Roman roads at Silchester.

What is by no means clear is whether the course of pre-Roman roads determined the placing of monuments, or if they merely served to join existing settlements. Logically, if we are to accept that elements of the road network pre-date the Iron Age, then the location of hillforts must, to some extent, have been determined by the course of roads (although other factors will obviously have been equally or more important). From that assumption, we can project the same determining effect backwards to earlier times. When might such roads have been formalized? The possibility that some at least were in use in the Neolithic period (and, given the existence of Neolithic wooden trackways, this does not seem unreasonable), when settled farming first became a feature of the British landscape, must at least be considered. If they were that old, then doubtless many of them marked even older routes used by man when traversing the land.

Standing stones are another class of monument that can be associated with Roman roads (and, presumably, with their antecedents). Taking just southern Wales, Maen Madoc next to Margary 622 is one such example, Bwlch y Ddeufaen by Margary 67c another, and Gelligaer (Margary 621) another.23

Major prehistoric ceremonial sites also cluster around Roman roads – even Stonehenge is situated near the junction of the Mildenhall/Old Sarum road (Margary 44) and the Winchester/Charterhouse road (Margary 45). On a lesser scale, there is Arbor Low in Derbyshire and Y Pigwyn stone circles next to Roman roads (Margary 71 and 62 respectively). Avebury too is linked to the main road from London to Bath (Margary 53) by the stone-lined Avenue, the junction being marked by Silbury Hill (which the Roman road swerves to avoid) to the north, West Kennet long barrow and The Sanctuary to the south. It is now known that Silbury lies adjacent to a major Roman settlement. Monuments of various ages are thus seen to be respecting a route that is adopted by the Roman road.24

It is tempting to agree with Rackham when he suggests that the majority of minor roads in Britain may be prehistoric in origin. The similarity between the sinuous trackways of ladder settlements and the meandering British country lane is undeniable. However, for our purposes, the most important observation is that many of our major trunk roads are not just Roman in origin, but probably pre-Roman. The Romans did not invent roads, but they certainly knew how to improve those that already existed.25

Ghosts of former roads may even be preserved in anomalous Roman roads. The outpost fort of Bewcastle, some 10 km north of Hadrian’s Wall, has been suggested as guarding friendly local territory and it is nowadays thought unlikely that it and its fellow outpost forts (Birrens, Netherby, Risingham, and High Rochester) were placed on likely routes for incursion into Roman territory. However, three of these lay on Roman roads into Scotland (Margary 7f and 8e), Netherby on a short spur of road (Margary 868), and Bewcastle on a similar spur linked back to Birdoswald on the Wall itself (Figure 7). It seems fairly likely that all of these routes are therefore prehistoric in origin, that some were completely reused in the Roman period, others (like Margary 865 to Bewcastle) partially adopted and given an all-weather surface, but only so far as was necessary for logistical purposes. However, recent geophysical survey has shown a road running north out of the fort which may well reflect a continuation beyond Bewcastle, at least in part.26

Figure 7: Netherby and Bewcastle on dead-end spur roads north of Hadrian’s Wall.

If you find an error please notify us in the comments. Thank you!