Chapter 4

After the Romans

But the Romans long exercised a beneficent influence on the development of the country, as when they went away they left their roads behind them. J W Gregory¹

So what happened at the end of Roman rule in Britain? It is very far from the end of the story of Roman roads. Just as we saw with the question of the existence of pre-Roman roads, scholars disagree about what happened next, some seeing the Roman road network rapidly decaying to the point of uselessness. However, others stress that the evidence points to continuity in many parts of the system, whilst at the same time other elements were deliberately abandoned. Put simply, if a Roman road is still in use today, it has been in use since the end of Roman rule; those stretches that were not needed soon being abandoned and succumbing to nature. This is, of course, no more and no less than the Romans seem to have done with the existing prehistoric road network when they arrived. The process almost certainly went on during the three-and-a-half centuries of Roman rule, so not all roads that were active in the first century AD would still be in use in the early fifth century. Road systems adapt to meet current needs.²

The speed and efficacy of plant colonization, even on modern road surfaces, can nowadays be seen in rural areas where modern single-track roads have been given tarmac surfaces. Whilst the wheels of vehicles using the road keep two strips on either side clear, the central area between the wheels begins to decay fairly rapidly and, with the aid of weathering, soon turns into a grassy strip (Plate 15). If this happens even on modern tarmac surfaces, which are far more resilient than the rammed gravel used for most Roman roads, it does not take much imagination to appreciate how long it would be before an unused Roman road disintegrated to the point of being unusable. Indeed, one of Thomas Codrington’s published works concerned the upkeep of pre-tarmac roads. Rackham points out that the earliest colonizing species was likely to be blackthorn and this goes some way to explain why Roman roads can survive as hedgerows and other boundaries. Where they were abandoned to cultivation, however, they were (and are) vulnerable to ploughing, particularly once more heavily mechanised forms of agriculture were introduced. It is all too easy to look at a modern road preserving the line of its Roman predecessor and imagine them to be indestructible. They are not and that is why some sections of road appear to disappear completely.³

Moving medieval armies

The English historian G.M. Trevelyan clearly recognized the true value of the contribution of Roman roads to the subsequent history of the British Isles. In effect, they were to provide a framework for that history: the skeleton and arteries of a medieval state.⁴

The Romans may have gone, but armies still needed roads. In 1993, N.J. Higham published a map illustrating the association of Anglo-Saxon battlefield sites with Roman roads. Although it went almost uncommented in his text, the significance of his observation is obvious: early medieval armies used Roman roads to move around. Could this have remained true in later periods? Road building in medieval Britain was unusual, even rare, and it is difficult to find examples of it having been undertaken on anything other than a local scale, although one man was fined for having constructed one from Yarmouth to Winterton in Norfolk. In 1235, a road was built to link the new bridge at Corbridge with the existing Roman road. It has been suggested that Edward I may have undertaken some limited road construction (more likely clearance: see above page 16), but this was very unusual. He certainly ordered a thirty-mile road cleared from Chester to the River Conwy, via Flint and Rhuddlan, in 1277. Almost exclusively, the Roman road system continued in use, often heavily repaired, sometimes slightly diverted, but it remained the core of medieval British infrastructure. At the same time, the large number of -ford place-names in Britain may indicate that, whilst the roads remained in use, bridges were often allowed to fall into disrepair. Roman bridges that survived into the Middle Ages are rare anywhere in the Roman empire and almost non-existent in Britain. We now know that bridges were being built in the post-Roman period: an example at Cromwell Lock was assumed to be Roman and marking the line of a road (Margary 590), but dendrochronological study of a surviving timber has shown it to be eighth century in date. An early date might be true for a medieval bridge at Corbridge on the Tyne, too, despite the fact that general opinion seems to be that it was a ford until the thirteenth century and gained its name from the proximity of a Roman bridge. Chollerford, a few miles away on the North Tyne, has the -ford component in its name, despite being close to the Roman bridge at Chesters. Were there no early medieval bridge at Corbridge, we might expect that too to have been Corford, rather than Corbridge, but its name appears in the form Et Corabrige as early as AD 786.⁵

So why were so few roads built after the Romans? The medieval kings of Britain were every bit as organized as their Roman predecessors and their engineers just as accomplished. The answer may simply lie in the availability of the existing Roman network. For all its faults, it was there and worked after a fashion, whilst its slow decay seems to have led to piecemeal responses rather than a massive construction initiative. Hence provision of a new road in the medieval period became worthy of comment precisely because it was so unusual.

The Romans had used their road network to move their armies around and there is a substantial body of evidence that indicates they were far from the last to do so. Even today, the modern British army is garrisoned along the major Roman roads. In fact, the military history of Britain was often enacted along those same roads. There is an undeniable broad correlation between battle sites in Britain – especially England – and the Roman road network. This is true of major conflicts like the Scottish War of Independence, the Wars of the Roses, and the English Civil War, as well as countless other smaller military adventures right down to the 1745 rebellion under the Young Pretender. When William Wallace defeated the English at Stirling Bridge and Bannockburn, he did so on a Roman road; when Richard III allegedly offered his kingdom for a horse, it was beside a Roman road (Margary 57b, regardless of which battle site is favoured); and when the Royalist forces met the Parliamentarians at Marston Moor, they did so less than two miles south of a Roman road (Margary 8). Roman roads thus played both a strategic and even tactical role in later warfare. Examples are not hard to find from any age, but are perhaps most striking in the earlier periods. As Barrett (1896) noted, ‘Nearly all the battles on English soil have been fought either across one or other of the old Roman roads, or in close proximity thereto’.⁶

Early Medieval (AD 410–1065); see Appendix 2

Soon after the Roman period, the road network will have most closely resembled that of the latter stages of the Roman occupation of the island. Priorities had changed, however, and some major components fell out of use whilst others, such as the ridgeways, may have assumed a new importance. It seems highly unlikely, however, that any construction of new roads took place. There may well have been the beginnings of a process that has been termed ‘roads making themselves’ (see below, page 104), but it is unclear how widespread this was, or even if in fact the old, pre-Roman network continued to be used and thus became fossilized in the landscape. The battles changed from being between the native British (or ‘Welsh’, a word derived from the Anglo-Saxonwealas, meaning ‘foreigner’) to between rival Saxon kingdoms, then between the Saxons and the invading Danes, and finally between the Saxons and the invading Normans.⁷

It is hardly surprising that earlier battles are usually harder to locate than later ones. As such, their strategic, as opposed to tactical, use of Roman roads will be harder to detect. In 508, the battle of Natanleaga, identified with Netley Marsh in Hampshire, was fought between Saxon Cerdic (later the first king of Wessex), with his son Cynric, and a British force, and this is one of the earliest post-Roman battles where a location seems likely. It is situated just over 2 km south of Margary 422, which branches off from the Winchester/Bitterne road (Margary 42b) and heads in a south-westerly direction through what was to become the New Forest towards Ringwood and perhaps on to Poole Harbour. The Battle of Cerdicesford in 519 (possibly to be equated with Charford, south of Salisbury) is intriguing as the suggested site is near some street place-names that are not on an identified Roman road. After succeeding his father, Cynric is reported fighting the British at Searoburh (identified with Old Sarum) in 552 and Beranburh (possibly Barbury Castle, south of Swindon) in 556. The first of these lay at a strategically important junction in the Roman road network, where Margary 4, 44, and 45 met. Barbury, if the identification is correct, lies on the Ridgeway, just 5 km from its junction with the Cirencester/Winchester road, Margary 43.⁸

Amongst these late fifth or early sixth century battles, that of Mons Badonicus, often associated with Arthur/Artorius (but not by Gildas), attracts the attention. Its location is unknown but there has been much speculation, suggested sites including one of the hills around Bath, Liddington Castle near Swindon (and not far from the above-mentioned Barbury), and Badbury Rings in Dorset. All of these are close to Roman roads, a fact that adds little to the debate. Nevertheless, place-name evidence and Gildas’ use of the term obsessio, implying a siege and therefore a place that could withstand an assault – such as a hillfort – is intriguing. Most hillforts, as we have already seen, sit comfortably within the Roman and pre-Roman road network.⁹

Another battle connected with Arthur (supposedly his last), was Camlann, assigned to 537 but not attested in a source before the tenth century. Nevertheless, a number of sites have been identified with it, a recent favourite being Birdoswald on Hadrian’s Wall (assuming the identification of the site with Camboglanna is correct). The proximity of the Stanegate (Margary 85), the Military Way (Margary 86), and, slightly further away, the west coast road (Margary 7) and the Maiden Way (Margary 84) to the east would do nothing to hamper the identification.¹⁰

The Battle of Catraeth (c. 600), the subject of the poem Y Gododdin, is often assumed to have been fought at Catterick, and thus would have been located on the Great North Road, Margary 8b, if this association is indeed correct. Whilst not a major nodal point, it was the site of an important river crossing, over the Swale, at the Roman town of Cataractonium so it is easy to see why the identification is appealing. A battle that was fought at a nodal point was Cirencester in 628, between Penda, king of the Mercians, and Cynegils and Cwichelm of Wessex. Once again, this was a Roman town, Corinium, acting as a focus for activity.¹¹

Heavenfield (Heofenfelth) in 633/4 marked the triumph of Christian Northumbria over heathen Britons, but despite the chapel at the site associated with the battle, there is little beyond tradition to pin it accurately to the accepted location adjacent to Hadrian’s Wall, 5 km west of the point where Dere Street (Margary 8) crossed the Wall. The added complication of the involvement of the place-name Deniseburn, suggested as lying south of Hexham, makes it difficult to interpret the battle, unless there was an episode of pursuit down the Roman road.¹²

Although not actually a battle site, the Viking use of Repton as winter quarters during the years 873–7 not only provided them with sea-going access for their ships via the Trent, but also placed them less than 3 km from Margary 18. This was crucial in the subsequent division of the army in 875, half going north to Tyneside, the other heading south to Cambridge. The activities of the Vikings were responsible for the decision to remove St Cuthbert’s body from Holy Island and it was taken all the way south to Crayke (on Margary 80) – a property originally given to Cuthbert for when he was travelling to York – for four months before venturing back up the road to Chester-le-Street.¹³

Two (some scholars think only one) major battles were fought at Corbridge in Northumberland during the tenth century. This again involved Dere Street, but this time at its junction with the (east–west) Stanegate (Margary 85); its strategic significance (like its fellow site at Carlisle) did not end with the Romans. So it was that the Bernician noble Ealdred and the Scottish king Constantine intercepted (and were sent packing by) a marauding party of Danes (travelling from west to east, and thus in all likelihood along either the Stanegate or its medieval successor, the Carelgate) at Corbridge in (probably) AD 914. Dere Street (which was a via regia of the Scottish monarchy) was a favoured route for Scottish incursions into England – it was often the case that when the English (like Edward I or III) went up the east coast (see below, page 81), the Scots would travel down Dere Street.¹⁴

In 934, Athelstan, the king of Wessex, marched north and invaded Scotland, apparently reaching as far north as Dunnottar. On the way, he stopped at the sepulchre of St Cuthbert which, at the time, was in Chester-le-Street, betraying the fact that Athelstan was heading north on Margary 80. The fact he was accompanied by a fleet suggests he kept to a coastal route in his progress north. His best-known battle, Brunanburh in 937, where he defeated a combined Scottish and Danish army, is commemorated in a poem but, unfortunately, unlocated, although many candidates have been proposed for its site. One of the most favoured, Bromborough on the Wirral peninsula, would lie close to Margary 670, but another candidate, Burnswark in Annandale, is situated on Margary 7, whilst a further site, Brinsworth, is near the junction of Margary 18 (which actually crosses the village) and 710.¹⁵

As we have already seen, when Eric Bloodaxe (the Norse king of Northumbria) died in AD 954, he did so traditionally at Rey Cross, on the trans-Pennine Roman road (Margary 82) that is nowadays followed by the A66 (and where there also happened to be a Roman temporary camp surviving as an earthwork). He had no army with him, but nevertheless chose to flee on a Roman road. The likely site of the Battle of Maldon in 971, unusual in being both precisely located and commemorated in a poem, was more than 9 km from the London to Colchester road (Margary 3). However, the identification of a short length of road to the east at Bradwell (Margary 31) hints at the possibility that it may in fact have lain close to or even on an as-yet-unidentified stretch of that Roman road.¹⁶

Amongst our final Early Medieval battles, Carham was fought between Malcolm II of the Scots and Uhtred of the Northumbrians in 1018 on the putative Tweed Valley road (see below, Appendix 5), although only the location is cited and thus the significance of the road can only be inferred. Dunsinane Hill took place in 1054 between Macbeth and Malcolm Canmore, and is situated only 6 km from the main road (Margary 9) running up the eastern coastal plain of Scotland. They fought again in 1057, this time at Lumphanan, resulting in Macbeth’s death. This last is some way north of the nearest known paved Roman road but a number of camps stretch even further north than Lumphanan (which is 22 km south-west of the 110-acre camp at Kintore), suggesting the existence of tracks which both the Romans and their successors were using.¹⁷

Medieval (AD 1066–1539); see Appendix 3

AD 1066 is a useful, if arbitrary, point at which to begin our consideration of the role of Roman roads in the battles of medieval Britain. In terms of a road network, the point at which national, rather than regional, states came into being might have been a better point at which to draw a line, had there actually been any change in policy towards roads, but there was no such modification. Such legislation as there was for maintenance of the existing system tended to be haphazard and generalized, as with Edward I’s requirement in the Statute of Winchester of 1285 that roads be kept clear on either side to prevent ambush. The importance of the Roman road system in the medieval period has already been underlined by other writers, as well as the informal mechanisms by which roads might come into being (such as ‘roads that made themselves’), but an additional factor that should be stressed is the hidden, yet potent, role of pre-Roman roads and trackways in helping along those self-made roads. Settlements followed roads, rather than being created de novo in the middle of nowhere, so if there was no formalized Roman road there, it may be presumed that some sort of pre-Roman track or road would facilitate foundation and growth. Only then could additional ‘self-made’ links form. Conversely, failure or removal of a settlement could kill a road, as we have seen for the immediate post-Roman period for sites like Silchester or Mildenhall.¹⁸

So it was that the system was continuing to evolve, albeit without the agency of a direct government contribution. To the pre-Roman tracks and roads and the all-weather Roman roads were added streets in settlements and even the occasional local stretch of road outwith, as well as tracks between new hamlets and villages.

The period certainly opens with a famous example of the post-Roman use of the road system which serves to underline their continued importance. In 1066, the invading Norse army of Harald Hardrada was met in battle at Gate Fulford by a Saxon army commanded by Walcar and Waltheof, between Riccal (where the Norse army had encamped) and York. Margary has no numbered road running south from York, but the place-name, incorporating as it does a Gate element, together with a -ford component, is at least suggestive of a Roman road. This would have been a branch from Margary 2, which at first adopts a southerly course out of York, before turning to the east. In fact, within his section dealing with route 800 (which heads north-east from York) he refers to the discovery of a Roman road running from Heslington down to Pool Farm, crossing the Germany Beck on its way. The Saxon army, which must have marched down this road from York, was defeated and the remnants retreated back to the city, whilst Hardrada and his ally Tostig moved to Stamford Bridge at the junction of Margary 80 and 810. There, they were soon met in battle and defeated by the army of Harold Godwinson, who had marched from London in order to intercept them. Whilst in the north, Harold must have received word of the Norman invasion of Sussex on 28 September, and promptly marched his army back south. ¹⁹

The accomplishments of Harold Godwinson and his army in that fateful year would not have been possible without the Roman road system. In a comparatively short time, he had to march an army from the south of England to York, fight a battle at Stamford Bridge, then march south again to tackle the invading Normans at Senlac.

His march north involved his gathering levies as he went. Setting out from London on 20 September, he reached Tadcaster (315 km, using Margary 2 and 8) on 24 September, and so achieved a scarcely believable average speed of 79 km (53 Roman miles) per day. Harold’s march back to the south of England was slightly more leisurely, involving another 400 km (250 miles) in just twelve days (an average speed of 34 km – 21 miles – per day, closely matching that claimed for a Roman army on the march). There is an inevitable compromise between the speed of forced marches and an army’s combat-readiness at the end of such ordeals. In 1942, the Second Battalion of the 506th Parachute Infantry Regiment of the US army set a record by covering 190 km (118 miles) in 75 hours during a march from Camp Toccoa to Fort Benning. This, it has to be remembered – at an average speed of 63 km (39 miles) per day – was neither preceded nor followed by combat.²⁰

Although there were to be many complaints about the decaying condition of the Roman system in centuries to come, Harold’s extraordinary achievement demonstrates that a shoddy road system remained better than no system at all so far as an army was concerned. Hastings, fought 4 km from the Roman road (Margary 13) that would ultimately give William access to London, was not the final battle of 1066, however. When William marched from Canterbury to London in December to claim the capital and be crowned as king, he did so along Margary 2. Reluctantly admitted to the city, there was then unrest that resulted in the deaths of Londoners.²¹

The Battle of Stafford in 1069, when William took to the field again and defeated the troublesome Mercian earl, Eadric the Wild, is of particular note because it seemingly occurred nowhere near a Roman road. The close correlation which we have so far witnessed between early battlefield sites and the Roman road network as we understand it renders this more than a little surprising, but for the time being it need only be noted. At the end of the year, William was in the north laying waste the region around York after some local difficulties. This is of less interest in itself than the fact that it directly led to the all-too-mobile body of St Cuthbert being returned to Holy Island to avoid this ‘harrying of the north’. Simeon of Durham even preserves detail of its route (via Jarrow, Bedlington, and Tughall) which appears to demonstrate the use of Margary 80. William’s marches from the Tees to Hexham and then from York to Chester in 1070 are described by Orderic Vitalis in terms that suggest the Roman roads proved quite challenging. Malcolm III ‘Canmore’, the Scottish king, caused additional problems for the Normans in the north, but was finally killed at the Battle of Malcolm’s Cross, near Alnwick, in 1093. This was another battle supposedly far from a Roman road which, like the peripatetic St Cuthbert’s route, will be discussed later, since both seem to have utilized what is probably a Roman route between Newcastle and Tweedmouth (see below Appendix 5).²²

The first half of the twelfth century is dominated by the civil war between Stephen and Matilda and continued conflict with Scotland; the two are not unconnected. David I invaded in 1136 in support of Matilda, attacking sites on the east coast between Berwick and Newcastle. When a peace treaty was signed he retreated but returned again in 1138, arriving at Corbridge after besieging Wark Castle, so presumably travelling down Dere Street. Stephen followed him up to Roxburgh, then headed back south via Bamburgh. It was not long before David was heading south again, this time down the east coast, ending up at Newcastle. Then, sending his army on to ravage as far as Teesdale, he returned to lay siege to Norham Castle. Meanwhile, part of his army under his nephew William had marched west and met and defeated an English force at Clitheroe in Lancashire (on Margary 72). Once the Scottish army was reunited they marched south via Bamburgh and Mitford, crossed the Tyne and on into Yorkshire. There they met an English force near Northallerton in the so-called Battle of the Standard, on the road to York (Margary 80). This resulted in a Scottish defeat and David retreated to Carlisle, presumably using the Stainmore Pass (Margary 82).²³

In the second half of the century, a confrontation between forces loyal to Malcolm IV of Scotland and the would-be invader Somerled, the Lord of the Isles, at Renfrew saw the defeat and death of the latter. Fragments of a road south of the Clyde, running towards the Roman fort at Whitemoss and the fortlets at Lurg Moor and Outerwards, were known to Margary (route 780) but more has since come to light and it is clear that a road ran around the peninsula towards Largs. Renfrew is situated on this route, some 8.5 km from Whitemoss, and it may have been Somerled’s intention to use it. Malcolm’s son, William the Lion, having laid siege to Prudhoe Castle in Tynedale, had the misfortune to be captured at a second Battle of Alnwick in 1174, during the revolt against Henry II, once again betraying the use of the east coast route by the Scots. Henry himself took his siege train from Huntingdon towards Framlingham Castle, intending to besiege the rebel Hugh Bigod there, but before he could reach his goal, Hugh came to him and they met at Syleham. This is only 3.6 km from Margary 35, a length of road on a north-west to south-east heading which may have linked in with a route to Huntingdon that incorporated Margary 37, towards which it is heading. ²⁴

The thirteenth century saw John venturing north and exercising his acquisitive tendencies and a nascent archaeological bent by digging at the Roman site of Corbridge, looking (unsuccessfully) for treasure. Undeterred by a lack of success in February 1201, he apparently repeated the exercise in August 1208 and June 1212. What is of interest to us is that he was travelling along the line of the Stanegate (Margary 85) or possibly along the line of its successor, the Carelgate. In 1201, he ventured up from Newcastle to Belford and Bamburgh, before cutting back to Hexham through Rothbury and then west to Carlisle, whilst in 1208 he travelled from Newcastle to Carlisle via Hexham. In 1212, his route was from Richmond, up the line of the Stainmore Pass (Margary 82), via Bowes and Appleby, before heading up Margary 7 to Plumpton with a brief detour to Kirkoswald. His next move, to Calder and then Wigton, has no known Roman road, but it may indicate that a route from Brougham to Old Carlisle has yet to be found. He then goes to Carlisle and across to Hexham, before looping back down to Durham, presumably on Dere Street (Margary 8), on his way south to York on Margary 80. In these adventures (Figure 24), it is of course his visits to Hexham that afforded him the opportunity to explore Corbridge.²⁵

John inevitably became involved in the affairs of Scotland, but a confrontation at Norham in August 1209 was avoided when William backed down once the English king sallied north from Newcastle, presumably using the east coast Roman road (see below, Appendix 5). His problems came nearer to home with the revolt by his barons in 1215–17. This saw the French King Louis coming over to Britain and installed in London. The barons, meanwhile, are recorded as having moved from Dunstable to Mountsorrel on the way to Lincoln, utilizing two of the major Roman routes, Watling Street (Margary 1) and the Fosse Way (Margary 5), rather than going straight up Ermine Street (Margary 2). The decision to go on to Lincoln apparently only occurred after the besieged Mountsorrel had been relieved but, when they reached there, their army was defeated and the survivors fled south to London, completing the triangle, presumably by using the most direct route: Margary 2.

The Second Barons’ War saw the Battle of Lewes of 1264, with Simon de Montfort surprising Henry III and his son, who were based there. Simon had marched by the London to Lewes road (Margary 14), staying overnight at Fletching, less than 3 km from the Roman road. The events that led to the downfall of Simon in the following year saw some frenzied movements after he left his home at Odiham Castle in Hampshire for the last time. Proceeding to Northampton, he then made for Coventry and then Gloucester. His next move, to Hereford, is unhelpful, as there are several routes he could have taken, but his march to Monmouth, in an attempt to break out of the blockade his enemies had instituted along the Severn, can only have been undertaken on Margary 6. From Monmouth, he took routes 612 and 62 down to Newport, but failed to cross the channel and had to retrace his steps to Hereford. He finally managed to cross the Severn at Kempsey, to the south of Worcester. Kempsey has produced a milestone, although that could come from Margary 180 which runs just to the east of the village. However, his route 63, which passes to the north of Hereford, is pointing in the right direction and a Roman road is reputed to run from east to west through Kempsey, the two crossing at Palmer’s Cross. If this road did exist it would presumably have led to Evesham and would thus complete his fated journey. This untidy end only serves to underline how inadequate our knowledge of Roman roads can be, as does the aftermath of the revolt, where Simon’s supporters retreated to the Isle of Axholme, an area devoid of Roman roads (although this particular gap in our knowledge did not stop the young prince Edward reaching it, even building wooden bridges in the process, and signing a treaty at Haxey).²⁶

Figure 24: The travels of King John (after Hindle 1998).

August 1274 saw the coronation of Edward I, one of the most interesting English kings from the perspective of Roman road studies. Not only are his movements known in some detail, but he was one of the few medieval monarchs who actually undertook some limited road construction of his own. As early as 1277, he was campaigning in north Wales against Llewellyn ap Gruffudd, de Montfort’s former ally, clearing a route from Chester to Flint (20 km), a task which only seems to have taken ten days. Having made Flint his base, the route was then extended to Rhuddlan, evidently taking from 26 July to 20 August to do so. There was no Roman road for some 17 km of his preferred coastal route from Flint to Rhuddlan (c. 27 km), or he would not need to have cleared this route (Margary 67 turns inland at Holywell and stays to the south of his preferred route). In fact his ‘road’ turns out to have been little more than clearance through wooded landscape, reminiscent of the Roman practice of sending out an advance guard for the same purpose (see above page 16). The final stage, to Deganwy on the Conwy (c. 25 km), took only nine more days. The fact he undertook this route clearance is noteworthy for occasioning mention in the sources, as if a measure of how unusual his campaigning style was thought to be. Between 1,500 and 1,800 woodmen were employed for clearance from Flint to Rhuddlan, and 700 to 1,000 from Rhuddlan to Deganwy. It was not, however, all-weather road construction. ²⁷

Edward’s Scottish Wars tell us much more about his use of the Roman road system and provide tantalising hints about those parts of the system we do not yet fully understand. Like John, he made use of an east coast route between Newcastle and Berwick that afforded him access to Alnwick and Bamburgh. March of 1296 saw him going north from Newcastle, via Brunton and Bamburgh, only to branch westwards along the Tweed to Wark, before doubling back to Coldstream, Hutton, and finally Berwick, where he laid siege to the town through most of April. Success then drew him further north to Dunbar, via Coldingham, where the castle had been besieged by the English, falling after Edward had reached it and a battle was fought. From Dunbar he headed west to Haddington, then south to Lauder, presumably having cut across to Dere Street (Margary 8) on his way down to Roxburgh. These moves make little sense with the road system as portrayed by Margary, who has Dere Street as the most easterly route in the Scottish Lowlands, but as we shall see later (below, Appendix 5), there is good reason to suspect that there is a Roman basis for these elements of Edward’s itinerary.²⁸

A curious little addition was then an expedition through Liddesdale to Castleton, but it signified greatly in Roman road circles, for it appeared to exploit an old road, known as the Wheel Causeway, which James MacDonald, supported in print by Francis Haverfield, pronounced to be of some antiquity, but not Roman. Its exclusion from Margary’s work indicates he concurred. Haverfield helpfully pointed out its use by Edward and sundry other belligerent expeditions but left aside the issue of its construction. Whether a native track or a Roman road of some form, its existence is beyond doubt.²⁹

Edward was not finished yet. It was still only May and he now made for Edinburgh in June, then on to Stirling, Perth, Aberdeen, and beyond in July and August. This is much further north than any known all-weather Roman road, but he was closely following the route indicated by known Roman temporary camps, which, as Edward was to do, reached as far as the Moray Firth during the first century AD northern campaigns of Agricola. We have witnessed the exploitation of native roads and trackways for campaigns before, following the initial Roman invasion (above page 9), and here (as Maxwell has observed) we appear to see Edward doing the same and mimicking Agricola’s progress north. Looping round, he returned south, reaching Berwick in the fourth week of August.³⁰

Edward was abroad in 1297 but the Earl of Surrey marched north from Berwick to relieve Dundee and was met at Stirling Bridge by the Scottish army under William Wallace and Andrew Murray. The site of the battle was dictated by the fact that this was the lowest bridging point of the Forth and there was a boggy area, Flanders Moss, immediately to the west. These same factors also dictated the course of the Roman road past Stirling (Margary 9). Inevitably, therefore, the Roman road was an integral part of the battle, if seldom acknowledged as such. After the English defeat, Surrey beat a hasty retreat to Berwick.³¹

Edward’s next expedition north in person was in June of 1298 but this time he cut across from Alnwick and travelled north on Dere Street, reaching no further than Stirling, before going south to Peebles on Margary 7 and across to Ayr (presumably on the imperfectly understood Margary 79). He returned to Carlisle in September and remained in the border region until December. This bare account masks Edward’s important victory on 22 July at Falkirk. Some discussion of the location of the battlefield has hinged around the proximity of the Edinburgh/Stirling road (Margary 9, although its course is uncertain between Cramond and Camelon) to the two main favoured sites, although it should not be forgotten that the Antonine Wall Military Way (Margary 90) may also have been available.³²

In 1300, Edward campaigned in the south-west of Scotland. He started out from Durham in June and proceeded via Evenwood, Bowes, Brough, Brougham, and Skelton to Carlisle, which would have involved Margary 8, 820, 82, and 7. From Carlisle he went to Ecclefechan, Applegarth, Tinwald, and Dumfries, a route that involved Margary 7 and 76. His final move was from Dumfries down to Caerlaverock Castle, but there is no road recorded by Margary for this stage. However, the presence of a fortlet (Ward Law) next to Caerlaverock has led to the suggestion of a road running from Dalswinton, through Dumfries, to the end of this peninsula. Returning to Dumfries he next turned west to Kirkcudbright, another area with no known Roman road, but a fortlet at Gatehouse of Fleet and a fort at Glenlochar hint at a road yet to be found which Edward was perhaps using. By 30 August he was at Dornock, east of Annan, and probably about to ford the Solway near Burgh-by-Sands (where later he was to die), as he is found at Drumburgh the next day. He returned to Caerlaverock once more before heading south from Carlisle in November and taking the Stainmore Pass back to Yorkshire.³³

In 1301, Edward once more ventured into the lowlands, sending his son north from Carlisle whilst himself moving west from Berwick, through Norham, crossing the Tweed at Coldstream and making for Peebles. Margary records no Roman road along the Tweed east of Peebles but it is not unreasonable to expect one at least as far as Newstead and possibly all the way to Tweedmouth (see below, Appendix 5). Fieldwork by members of the Trimontium Trust has been productive in suggesting a route as far as the central lowland fort. Edward headed up to Bothwell Castle and then, via Dunipace, to Linlithgow, where he spent three months before heading south in February 1302. That last move suggests the continued use of the Antonine Wall Military Way (Margary 90).³⁴

Edward did not return to Scotland until May 1303, when he followed the familiar east coast route as far as Alnwick, but then moved westwards via Chatton to Roxburgh. He then embarked on another expedition into northern regions, supported by a fleet moving up the east coast, reaching the Moray Firth again, recalling his achievement of 1296 and presumably using much the same routes.³⁵

During 1306, we find Edward in the north again and in August he appears to have been moving along the Stanegate, with visits to Corbridge, Hexham, Carlisle, and repeatedly to Newbrough. In September he included Bradley, Henshaw, and Melkridge in his itinerary, but always returning to Newbrough. Finally, in October, he settled at Lanercost, since by this time he was seriously ill. He stayed there until March 1307, when he moved to Carlisle, before leaving for Burgh-by-Sands in July, where he was to die of dysentery before he could lead his army across the border again. Whilst he was at Carlisle, Robert Bruce achieved a significant victory over the English at the Battle of Loudoun Hill in May.³⁶

Loudon Hill shows how Roman roads could be exploited tactically. Bruce deployed his army on the road (Margary 79) running past the eponymous Roman fort and limited the available frontage (both because the road ran across boggy ground and he had dug additional obstacles), with disastrous effect for the opposing English force. He repeated this stratagem at Bannockburn in 1314, where this time he occupied the flanks of the road (Margary 9) with pitfalls.³⁷

Whilst an itinerary for Edward II has also been reconstructed, it reveals little more about the use of the Roman road system than can be gleaned by studying the movements of his father. For a while, internecine strife amongst the Scots helped him, but soon Bruce had turned to guerilla warfare, much to Edward’s frustration. The campaign of 1314 that culminates in Bannockburn is more illuminating. Leaving Berwick on 17/18 June he marched north with his army, accompanied by enough waggons to cover twenty leagues if placed end to end (about 44 km if a Roman leuga was intended). Heading for Stirling to relieve the siege there, his army was met and defeated at Bannockburn on 24 June.³⁸

There was widespread Scottish raiding in the north of England, reaching as far south as Teesdale, and in 1319 we witness a Scottish raiding party avoiding Edward’s forces by returning up the Stainmore Pass (Margary 82) and thence to Gilsland, presumably using the most direct road, the Maiden Way (Margary 84). They returned by the same route the next year, burning Gilsland and ravaging Stainmore as far as Brough, before heading west into Westmorland for more of the same.³⁹

Edward’s troubles moved south, with the revolt of the barons under Thomas of Lancaster. At the ‘Battle’ of Burton Bridge, at the beginning of 1322, Lancaster’s forces held the bridge there across the Trent, immediately east of Margary 18. Edward, approaching from the south-west (so presumably along that very road) split his force and crossed the Trent only to find he could not get back using a ford (a classic example of why bridges are necessary to an all-weather strategic road network). Fleeing north, without actually engaging the king’s forces, Lancaster reached Boroughbridge. Had he continued north on Margary 18, he would have reached the Don at Templeborough; the road network as proposed by Margary and others would have required him to head north-east to Doncaster on Margary 710 in order to continue north-westwards on Margary 28 (50 km), although the most direct route would have been due north to Castleford (35 km), so this may indicate a gap in our knowledge of the Roman network.⁴⁰

Boroughbridge is a very interesting battle from the point of view of Roman roads. Dere Street passed through the Roman town of Aldborough, but in medieval times this had slipped from ascendency, possibly when the Roman bridge over the Ure failed, and a new settlement at Boroughbridge came into existence (although, unlike (Ald) borough, Milby, or Langthorpe, it is not mentioned in the Domesday Book, which may imply that this change had yet to happen in the eleventh century). This meant the Great North Road was now running to the west of Dere Street. Intriguingly, excavation has shown that the likely site was very close to an early Roman fort, at Roecliffe, which seems to have been guarding a road crossing which may have been prehistoric in origin (see above page 7). It may therefore be the case that, when the Roman line of Dere Street past Aldborough went out of use, the more direct prehistoric line was reinstated and became the Great North Road at this point. Whilst a timber bridge at Boroughbridge itself played an important part in the battle, there was a nearby ford that Sir Andrew Harcla took the precaution of guarding with a schiltron of pikemen on 16 March 1322. Many interpretations of the battle assume the ford was near Aldborough (Figure 25) but, despite changes to the Ure around Boroughbridge, local tradition persists that it was fordable at Langthorpe, which is just to the west of Boroughbridge and on the north bank opposite the Roman fort, to the west of the later canalization.⁴¹

Edward followed up with a less-than-successful invasion of Scotland in August, retreating south through Berwick and Newcastle to Rievaulx Abbey in Yorkshire, but this only led to Bruce crossing the Solway, presumably travelling down the Stainmore Pass, and then confronting Edward’s forces in the Battle of Old Byland in October of the same year, which resulted in another defeat for the hapless English king. Although there is no known Roman road in the vicinity, Margary 814 from Malton may have supplied the missing link, since Barbour’s poem (and other sources) makes reference to a ‘peth’.⁴²

Figure 25: Aldborough, Boroughbridge, and the River Ure crossings.

Edward III, crowned 1 February 1327, tended to be more successful in his Scottish adventures, although that same year saw a defeat at the hands of the raiding Moray and Douglas at Stanhope Park. The English, having marched in a single day from Barnard Castle to Haydon Bridge, then lost patience and headed back down to Weardale to intercept the Scots. The subtleties of this manoeuvre are lost on us now, since we only have the fragmentary Margary 821 striding over the moors to the south of Stanhope, which may have linked Corbridge with either Bowes or Greta Bridge, but its proximity to Stanhope Park suggests it played a role, and John of Fordun’s comment, that the English army sat at the end of the road waiting for the Scots, indicates that they saw no other way for them to move.⁴³

The Battle of Dupplin Moor in 1332 is notable for the prominent role played by the Gask Ridge (Margary 9). Seemingly named as the location of the camp of the Scots’ baggage train, the road was still in use in the medieval period, since it survives in large part today. A fortified Flavian road, noted for its chain of signal stations, it allowed access to Perth from the west by means of higher ground to the north of the River Earn. Scots raiding at the beginning of 1333 saw Gilsland being burnt again, but this time from the west, followed by an English incursion. Edward’s journey to his triumph at Halidon Hill in 1333, thereby relieving a siege of Berwick by the Scots, revisited previous monarchs’ journeys between Newcastle and Tweedmouth.⁴⁴

In 1346, the Battle of Neville’s Cross saw the Scots defeated and their king captured. It took place just to the west of Durham, in the angle between two Roman roads, Margary 80 and 83, the latter providing a link between the former and Dere Street (Margary 8). By the time of the battle, it is possible this cross route had fallen from use (it is not shown on Ogilby’s 1675 map) but there is no way to be certain of this. On balance, the position of the battle would seem to argue against that.⁴⁵

The Scots were far from finished. Bolstered by a detachment of French troops under Jean de Vienne in 1385, they headed towards Roxburgh from Edinburgh (which indicates the use of Dere Street, Margary 8). Deciding Roxburgh could not be taken, they then moved off down the Tweed towards Berwick, on the way taking Wark, Ford, and Cornhill Castles. The English under Richard II retaliated by raiding into Scotland, travelling up the east coast once more to Berwick, but then venturing along the Tweed to burn Melrose Abbey (near the Roman fort at Newstead). They then continued north to Edinburgh and appear to have been observed on their way, passing up Lauderdale (if Froissart is to be believed), the size of their force obliging the Scots and their allies to avoid the advancing English army and go raiding in the south-west around Carlisle instead. After burning Dunfermline, Perth, and Dundee, Richard intended pursuing the Scotto-French force to the south-west, but was instead persuaded to return the way he came.⁴⁶

Shortly afterwards, a major incursion was mounted into the north of England in 1388 which saw Newcastle besieged by Scottish forces under Douglas, which introduces a new and interesting element to the story of the road system on northern England. The Scots returned home from Newcastle via Ponteland and ended up fighting the English at Otterburn. Although adjacent to Dere Street (Margary 8), so technically yet another battle near a Roman road, the route adopted by the Scots and the pursuing English is not a Margary route and does not appear elsewhere in Anglo-Scottish conflicts, Dere Street and the western route through Carlisle being favoured by the Scots, the east coast by the English. The possibility of a road linking Newcastle directly with Dere Street has not been considered by scholars but will be examined further below (Appendix 5).⁴⁷

The Welsh revolt against English rule by Owain Glendŵr in 1400 saw the battle of Mynydd Hyddgen in 1401. Although the precise location is unknown, its general vicinity has been suggested. No Roman road is known close by, but an east-to-west route (Margary 64) is pointing towards it.⁴⁸

After that brief interlude, the Scots returned to the fray and, in 1402, a force marched from the Forth towards Newcastle but, upon its return, was intercepted by Hotspur near Wooler at Homildon Hill. This resulted in a major Scottish defeat, supposedly commemorated by the nearby Battle Stone (which is actually a prehistoric standing stone, recalling the association of ancient tracks and roads with such monuments).⁴⁹

Hotspur’s unease over his treatment after Homildon was a contributory factor to his rebellion against Henry IV which culminated in his death at the Battle of Shrewsbury in 1403. Hotspur raised an army in Cheshire and then marched south towards Shrewsbury, presumably down Margary 6. Henry was at Burton-on-Trent when he heard of the revolt and moved to intercept Hotspur. Shrewsbury is puzzling as there is no immediately obvious link to the Roman road system, Margary 6 being more than 5 km to the east. A ‘campaign’ road, towards the Roman vexillation fortress at Rhyn Park, up the valley of the Perry from Wroxeter, has been proposed but even this may have been too far south so there may well be an as-yet-unidentified Roman road in the area, possibly linking Margary 6 and 64, or more likely anticipating the route of Ogilby’s road from Shrewsbury to Wrexham.⁵⁰

The existing situation with the Welsh rebellion began to merge with another evolving problem – France – in the year 1405, when a French invasion of Wales occurred in support of Glendŵr. Their force landed at Milford Haven, a port that was to see repeated use in later years for military ventures, united with the Welsh rebels and captured local towns, including Carmarthen, before marching through South Wales, presumably using Margary 60, route 6 up to Hereford, then the incomplete (from our point of view) 63 to get to Worcester for an indecisive confrontation with the English forces under Henry IV.⁵¹

The rebellion of Henry Percy finally came to an end in 1408 at the Battle of Bramham Moor. This occurred at one of the major intersections of Roman roads in the north of England, where the north-to-south Margary 28 and 280 met the east-to-west routes 72 and 729. The confusion of roads here suggests a palimpsest of different phases in the development of the network.⁵²

The Battle of Piperdean took place on the much-travelled east coast road to the north of Berwick-upon-Tweed (1436; see Appendix 5). Hostilities between England and Scotland resumed in 1448, with the English destroying Dunbar, and the Earl of Salisbury’s army Dumfries. The Scottish response was to attack Warkworth and Alnwick, culminating in the Battle of Sark, which was fought near Gretna, just 2 km from Margary 7 and near a presumed route to the west.⁵³

The onset of the conflict later known as the Wars of the Roses in 1455 saw a small but significant battle fought at St Albans. Henry VI had marched there with a force of men and taken up position in the town whilst Richard, Duke of York, had marched from the north and set up his standard to the east of the town, the former having the misfortune of ending up captured by the latter. The significance of the Roman road network to this conflict goes without saying, since two major components (Margary 1 to the south-east and north-west of the town) were used to move the armies. The first major battle was Blore Heath in September 1459 which is, unlike most of the conflicts described here, more than 7 km away from the nearest Roman road (Margary 19, to the south-west). A Yorkist force was marching south-westwards, heading for Ludlow from Middleham, and was met by a Lancastrian army on Blore Heath. The Roman villa at Hales lies only 2 km to the south of the battlefield, however, so it might be anticipated that an access road ran from there to Margary 19. By October, the two sides met again at Ludford Bridge, just south of Ludlow Castle and the Roman road from Weston-under-Penyard (Margary 613) is aligned on this crossing of the Teme, although Margary only records it as far as Ashton, 10 km to the south of Ludlow. However, Richard of York retreated from Worcester to Ludlow and no direct Roman road is known for that route so this may well represent another gap in our knowledge of the system.⁵⁴

The next confrontation, in July of the following year, was the Battle of Northampton and it took place to the south of the River Nene at the abbey of St Mary de la Pré. This lay next to the river crossing, towards which Margary 17 seems to have been directed and, like the modern A428, it may have continued to Bedford. Henry VI had come from Coventry, so route 17 was his likely access to the town, whilst Warwick was approaching from London and his most direct route would have been Margary 1. The next battle, in December, occurred near Sandal Castle in what is now West Yorkshire. The Lancastrians, having marched west from Kingston upon Hull to Pontefract, encamped there and awaited the Yorkists. Sandal is some 10 km from Margary 28, but the short stretch of Margary 728 near Swillington points straight towards it.⁵⁵

The Battle of Mortimer’s Cross in February of 1461 involved a Roman road (Margary 6), although precisely how is unclear, given the uncertain nature of this battle. The movement of the Lancastrian army, from Wales into England, was intercepted by a Yorkist force. This may have been from south to north along the Roman road, in which case it was directly involved, or, as has been argued, from west to east along the London to Aberystwyth road recorded by Ogilby, when the presence of the Roman road will have been largely coincidental. A third possibility, an east-to-west interception of a south to north movement is also a possibility. The next engagement occurred at the second Battle of St Albans, in the same month, where a Yorkist force attempted to block the Roman road (Margary 1) north of St Albans to deny the Lancastrians access to London. The preparations for this action saw the attempted tactical use of a road once more, with obstacles being placed to either side of it by the Yorkist force. However, the advancing Lancastrians pre-empted this by changing course and approaching from Dunstable instead and encamping on Bernard’s Heath, which implies their use of Margary 21 to avoid the blockaded northern road. March saw the next encounter between the Yorkists and Lancastrians at Ferrybridge, 5 km east of Margary 28 as it passed through Castleford. The Lancastrians held the bridge at Ferrybridge and the Yorkists attacked both from the south and, using the Roman road to outflank them and cross the Aire further west, from the north. The Lancastrians retreated north to Towton, which lay 4.5 km east of Margary 28. From this, it seems clear that what was later to be known as the Great North Road (subsequently the A1), which left the Roman route at Barnsdale Bar, 11 km to the south of the Ferrybridge crossing, already found favour over the original Roman route. It is possible that the more modern road is actually a southern continuation of the Rudgate (Margary 280) or a prehistoric predecessor. Ferrybridge was the site of a henge and may have been a crossing point of some significance long before the Romans arrived.⁵⁶

Three years later, a small Yorkist force under the Marquess of Montagu was sent north from York to escort a Scottish delegation. An attempt to intercept Montagu was made by the Lancastrians at Newcastle and then on Hedgeley Moor, 10 km south-east of Wooler, where they were defeated after venturing out from Alnwick. This was a significant location as it appears to have been a fork in the Roman system between the original course of Devil’s Causeway (Margary 87) towards Tweedmouth and a later extension to a higher Tweed crossing at Cornhill (see below, Appendix 5). This location therefore afforded a choice of heading north-westwards to ultimately join Dere Street, whilst the more northerly Devil’s Causeway course led to the coastal route often favoured by English armies in the past. The Battle of Hexham was fought in the same year, some 3 km south-west of the point where Dere Street bridged the Tyne west of Corbridge and led to another significant Lancastrian defeat. Codrington noted the possibility of a Roman road running south-westwards from the crossing point towards Whitley Castle. Selkirk, citing additional earlier references, records some physical evidence for its existence; strangely, Margary made no comment on the possibility of the existence of this road, even if only to dismiss it. At least one aerial photograph taken in 2006 shows an ambiguous linear feature south of the Tyne that may be related to the road.⁵⁷

The Battle of Edgecote Moor in 1469 occurred within an area devoid of any Roman roads nearby, but with at least one road (Margary 166) pointing north-westwards towards it, once more serving to underline how incomplete our knowledge is for some areas. We are on more certain ground with the Battle of Losecote Field (Empingham) the following year, for here the Great North Road (Margary 2) was central to the deployment of the Lancastrian forces.⁵⁸

The accepted location for the Battle of Barnet in 1471 lay between two south-east to north-west oriented Roman roads (Margary 167 and 220), just 3 km west of route 220 and near the junction of two roads shown by Ogilby. The same year saw the decisive Battle of Tewkesbury, on the Gloucester to Worcester road (Margary 180). Involving a complex series of move and counter-move around the West Country, the Lancastrian forces, having headed north from Gloucester, were caught up with by the Yorkist army and battle followed. Exhausted, the Lancastrians had been obliged to abandon some of their artillery, highlighting the fact that whatever their inadequacies, Roman roads were of continued importance for the movement of the more cumbersome components of any medieval army.⁵⁹

The Battle of Bosworth Field in 1485 must be one of the most famous English conflicts and the debate over its exact location has been heated and long running. Archaeology has recently suggested that the location of the conflict actually lay adjacent to the Roman road from Leicester to Mancetter (Margary 57), whereas the traditional battlefield site, on Ambion Hill, had long been identified as lying some 1.3 km from the same road.

A prominent role was played by a Roman road in the Battle of Stoke Field in 1487, where the Lancastrian and Yorkist forces deployed between the Fosse Way (Margary 5) and the Trent.⁶⁰

The events that led up to the Battle of Blackheath (or Deptford Bridge) in June 1497, the culmination of the Cornish Rebellion, saw a rebel army march from Wells to Bristol, Salisbury, and then Winchester, before heading to Guildford, finally arriving at Blackheath. Movement from Bristol to Salisbury and then Winchester can comfortably be accomplished with the main network as Margary understood it, but beyond Winchester there is no route known into Surrey. However, medieval Winchester possessed a gate in the north-east (the Durngate) and the east (East Gate) and the Gough Map shows a road running from Winchester to Guildford via Alresford, Alton, and Farnham. Thus it would be surprising if there was no corresponding Roman gate and, indeed, road for at least one and probably both of them. The alignment of Alresford Road certainly looks suggestive of a Roman origin and recent work by the North East Hampshire Historical and Archaeological Society has provided further convincing evidence for the existence of a road between Winchester and Staines using documentary sources and some limited excavation. In September there was a second uprising in Cornwall, with Perkin Warbeck (who claimed to be Richard IV) landing at Whitsand Bay near Plymouth and then marching on Exeter and then Taunton. The inadequacies of our knowledge of the Roman road network in the Cornish peninsula have already been mentioned but this adventure may give some hint of its ultimate extent in the south. At the same time, the castle of the Prince Bishops of Durham at Norham was attacked by James IV, who marched south with an artillery train (which included the bombard, Mons Meg). Unusually, he chose not to use Dere Street or the east coast route, but rather took a road that does not normally feature in Anglo-Scots warfare, from Haddington, across the Lammermuirs, and into the Merse (see Appendix 5). The fact that the route was able to accommodate artillery may indicate a Roman origin for it. ⁶¹

The sixteenth century saw more border warfare in the north, with a minor battle at Milfield in 1512/13 occurring on a possible Roman route (see Appendix 5) before James IV returned to Norham in 1513, this time using the Dere Street route to bring up his artillery (perhaps indicating that the route over the Lammermuirs had proved less than satisfactory). This action was prior to his defeat at Flodden, again near a possible Roman road (see Appendix 5). His captured artillery was briefly stored in Etal Castle before being transported to Berwick.⁶²

So passes the great medieval period of Roman road exploitation. This was by no means the end of their role in British history, as will become apparent. The Roman road system, modified and decaying, patched and diverted as it was, was far from finished as an agency of war.

Post-Medieval (AD 1540–1900)

Two long-term wars must dominate any consideration of the post-medieval use of Roman roads: the continued conflict with Scotland, and the English civil wars. This is the era when complaints about the decaying road system became loud and frequent and the authorities had to take action. Ultimately, this resulted in a complete overhaul of the system and the foundation of the turnpike trusts, but this did not happen until long after the end of the civil wars, so the Roman network still provided the skeleton to the flesh of these bloody conflicts.

After Flodden, many of the battles of the northern wars that became known as the Rough Wooing were more like minor skirmishes. The Battle of Haddon Rigg in August of 1542 took place just over 2 km to the south-west of Carham and close to the Tweed valley road (see Appendix 5). Just two months later, an English force made its way back along the Tweed from Berwick to attack Kelso and the surrounding region but had to return after running out of supplies. However it was not until November of that year that a decisive encounter occurred in the Battle of Solway Moss. James V led a force of 18,000 men south on the western side but was taken ill near Lochmaben, suggesting he was using Margary 7 for his attack (perhaps unsurprisingly). After crossing the Esk, the road passed along a low ridge before encountering the Lyne and it was between these two rivers that the small English force caught the much larger attacking Scottish army and the majority of the action took place close to the Roman road, according to Sir Thomas Wharton’s account (and thus some way to the south of the present extent of Solway Moss).⁶³

The Battle of Ancrum Moor in 1545 saw an English army, which had been raiding in the Borders, defeated by the Scots and once again a Roman road played a (quite literally) central role. Returning to Jedburgh from Melrose, so moving down Dere Street (Margary 8), the English army spotted a Scottish force moving westwards from Penniel Heugh (to the east of the road) and pursued them, the battle itself occurring to the west of the Roman road near Lilliardsedge.⁶⁴

Roman roads may have enjoyed similar significance in the Battle of Pinkie Cleugh in September 1547. The English army, under Somerset, had advanced up the east coast from Berwick with 900 carts and fifteen artillery pieces, accompanied by a fleet with supplies. Camping near Prestonpans, they met the Scots in battle to the east of the Roman fort at Inveresk, which probably lay at the junction of the northern end of Dere Street (Margary 8), the eastern coastal route (see Appendix 5), and Margary 7 (which is aligned on Inveresk for most of its course, even though Margary has it turn towards Cramond). Indeed, Dere Street even appears to be included in an eyewitness sketch of the battle by William Patten. The resultant Scottish rout went in three directions: towards Leith, to Holyrood, and south to Dalkeith, all of those probably along Roman roads in the form of the east coast route, Margary 7, and Margary 8 respectively.⁶⁵

The so-called Prayerbook Rebellion against Edward VI in 1549 saw a number of clashes between the rebels and loyalist forces, largely choreographed around the Roman roads of the region. The aftermath of a skirmish at Crediton (on Margary 493) saw the rebel force split, part going to besiege Exeter, the other part to the east of it at Clyst St Mary (Margary 49). The Battle of Fenny Bridges was then fought at the point where Margary 4 crossed the River Otter, the rebels under Arundell blocking Russell’s attempt to march from Honiton to Exeter. Russell’s second attempt to relieve Exeter avoided the main road and went south-westwards instead, with an encounter with the rebels at Woodbury Common suggesting a link road between Margary 4 at Honiton and Margary 49, nearer the south coast. Russell continued towards Exeter and the next battle was at Clyst St Mary. The execution of some 900 prisoners by the loyalists led to a renewed rebel attack on Clyst Heath, only for them to be defeated again and Exeter subsequently relieved. The whole sorry affair ended with a final battle where it had all started, at Sampford Courteney, which is situated on Margary 492, less than 3 km from the Roman fort at North Tawton.⁶⁶

The events surrounding Mary Queen of Scots and her ambitions towards the English throne led to the Battle of Carberry Hill in 1567 and her surrender after a day-long stand-off. She had spent the night at Fa’side Castle, which also featured in the Battle of Pinkie Cleugh, lobbing a few desultory rounds at the English forces. As has already been alluded to, the road from Carberry down to Musselburgh is probably the northern end of the Dere Street.⁶⁷

There was one last battle between England and Scotland and that occurred at Carter Bar in 1575 (sometimes known as the ‘Raid of the Reidswire’). Surprisingly, this did not take place close to Dere Street (Margary 8), but rather at the headwaters of the Rede, where the modern A68 crosses the border, so it is possible that this, together with the A6088, represents at least in part a link road running over towards Margary 89 in the vicinity of Hawick (a suggestion that may be reinforced by a number of -chesters place-names along the route).⁶⁸

Although the road system had somehow muddled through the medieval and the beginning of the post-medieval periods, the strain was beginning to show and the sixteenth century was the time when complaints about the condition of the roads became most strident. In 1555, the Highways Act placed the burden of upkeep of roads upon the parishes through which they ran. This was renewed and extended in 1563, 1575–6, and then made perpetual in 1587. However, these provisions were still concerned with repair and maintenance, not construction. New roads were still some way away, but something had changed: a number of the battles are indeed fought next to Roman roads, but roads which had been replaced by alternative routes. The Roman version may already have been in decay, either as hedgerows (and thus boundaries) or even ploughed out altogether. As we shall see, ironically, the alternative replacement routes may well have been a reversion to the older routes the Roman roads had supplanted. This was a new dynamic in action and the next stage in the evolution of our road network.⁶⁹

The series of conflicts often simplified as the English Civil War was played out on the evolved network, but the Roman system of roads remained at its core and still had a strategic role to play. That being said, the first battle to be considered here, at Newburn ford, was not on a known Roman road, although only a short distance south of Hadrian’s Wall and the accompanying Military Way which was, by all accounts, still a significant road in 1640. The ford, which was almost certainly in use in prehistoric times, to judge from finds of an Iron Age or Roman cart wheel from nearby, was defended by Charles’ English forces and attacked by Scottish Covenanters from Newburn, to the north, resulting in an English rout. It serves to underline how the pre-Roman network could still be influential in the post-Roman period.⁷⁰

Two years later, the Battle of Powick Bridge, generally regarded as the first action of the Civil War, was fought near Worcester, just 2 km to the west of Margary 180. Held to be of little account as a battle, it has a few intriguing aspects with regard to the Roman roads of the area. First, the overnight march of Nathaniel Fiennes’ Parliamentarians from Alcester to Worcester is indicative of a missing (and surely logical) link in the Roman network. Second, a series of alignments of lanes and footpaths in St John’s and Lower Wick, along which Fiennes advanced on the day of the battle, points towards the medieval bridging point and appears to line up with the change in course of the known Roman road from south-westerly to slightly east of south at the southern end of Rainbow Hill. This might indicate the existence of a spur road continuing in that same south-westerly direction, across the bridge, perhaps ultimately intended to link up with Margary 612 around Ross-on-Wye.⁷¹

The site of the Battle of Edgehill, which was fought a month after Powick Bridge, lies some 7 km from both Margary 5 and Margary 56. Whilst it is conceivable there is an unknown minor road nearby, this confrontation is more important from the point of view of the strategic moves made by the opposing armies as they moved towards it. The king marched his army back towards London from their base in Shrewsbury, naturally using Watling Street (Margary 1), in order to tempt a Parliamentarian attack. Most accounts then go on to say that Essex marched from Worcester and the two armies met at Edgehill, but this is difficult to understand in the context of what we know of the formal Roman network. This is perhaps the first hint at the revival of the use of the informal, pre-Roman network. The Battle of Turnham Green (and the skirmish the day before at Brentford) took place in November, near the junction of Margary 4 and 40.⁷²

The Battle of Braddock Down in January of 1643 was fought in Cornwall, more than 30 km beyond the westernmost extent of the Roman road system as recorded by Margary but well to the east of the Cornish milestones from the western extremity of the peninsula, so it is not beyond the bounds of possibility that the A390 has a Roman origin. Two months later, the action at Hopton Heath in Staffordshire, just north-west of Stafford, occurred when a Parliamentarian force with artillery returning from Lichfield was intercepted by the Royalists, and it may indicate the existence of a link road between Wall (which is near Lichfield) and Chesterton. Back in Cornwall, the Battle of Stratton in the May, like Braddock Down, was not fought on or near a known Roman road. However the place-name is suggestive (one of the standard variants derived from strata), as is a long alignment of the A39 between Kilkhampton and Helebridge, adhering very closely to a 205/25 degree line. This impression is only reinforced by the brace of isolated milestones further down the north coast, near Tintagel.⁷³

In June, the Battle of Adwalton Moor was fought immediately adjacent to Margary 712, although this was to some extent coincidental, with the Parliamentarian marching south-east out of Bradford in order to avoid a Royalist siege and the two forces running into each other at Adwalton.⁷⁴

The Battle of Lansdown Hill in July is a classic example of a conflict fought out over a landscape dictated by the Roman infrastructure. Margary 542 ran north-west out of Bath, over Lansdown Hill and across the battlefield. The A420 to the north is Ogilby’s London to Bristol road and may represent a link road between Margary 5 (the Fosse Way) and Margary 54 to Sea Mills and the Severn crossing. Just over a week later the Battle of Roundway Down occurred to the north of Devizes, on that same Bristol to London road recorded by Ogilby and south of Margary 53 which is now only marked by hedgerows along this stretch, so it seems likely that the Ogilby road had already replaced it by 1643. Continuing as the A4361, the alignment of the Roman road takes it to Avebury, so a prehistoric origin is likely: in other words, the prehistoric route was replaced by the Roman road which was in turn replaced by the reinstated prehistoric route!⁷⁵

The first Battle of Newbury in September of 1643 occurred when a Parliamentarian army under Essex, marching from Hungerford, was met on the western side of Newbury by the Royalists. This occurred just east of the point where Margary 53 joins route 41. Again, the Roman road is derelict for its whole length between the abandoned Roman town at Mildenhall and that junction with Ermin Street so it is likely it was already out of use by the time of the battle, replaced by the modern A4 following the line recorded by Ogilby.⁷⁶

In January of 1644, the Battle of Nantwich took place immediately next to the line of Margary 700. Once again this may be coincidental, since much of the route of the road is derelict and only short lengths of road still adhere to it. Moreover, the action occurred from north to south, not east to west.⁷⁷

The Battle of Cheriton occurred in March of 1644, near Alresford and to the east of Winchester, situated close to the likely line of a Roman road that was not included by Margary but has recently been investigated. Cropredy Bridge was fought three months later not far from Edgehill, and all the caveats that applied to the latter are associated with the former battle too. Lostwithiel, in June, was fought a little further to the west of Braddock Down and not far from the newly discovered Roman fortlet at Restormel, further pointing to Roman penetration of the peninsula. Marston Moor followed in July of the same year and was located between – and within easy reach of – two Roman roads that continue in use to this day: Dere Street (Margary 8) and the Rudgate (Margary 280).⁷⁸

The Battle of Naseby, in June 1645, did not happen anywhere near an acknowledged Roman road, nor does there appear to be any evidence for any unrecognized examples in the vicinity. Students of the Civil Wars think of it as significant for the major defeat of the Royalist army, so students of Roman roads might be forgiven for thinking this conflict marks the end of the association of Roman roads with British battles, were it not for the motte and bailey castle which, when viewed in the context of the relationship of other such fortifications with Roman roads, may at least give pause to wonder whether the road from Naseby to Sibbertoft and on to Theddingworth may indeed have ancient origins.⁷⁹

The Battle of Langport was fought in Somerset in July 1645, on a ridge of high ground that protrudes into the Somerset Levels. No Roman road has been recognized in the vicinity but a number of important villas are known in the area as well as the important Saxon royal centre of Somerton adjacent to the north-to-south Margary 51, so it is perhaps unsurprising that the road running along the ridge, the B3153, is laid out more-or-less straight, with gentle angular turns, like a Roman road. Rowton Heath in September of the same year was to the west of Chester but some 2 km to the south of Margary 7, near Long Lane which runs for some 22 km down towards Nantwich with some straight alignments that may point to a Roman origin. In the spring of 1646, the Battle of Stow-on-the-Wold was fought on the Fosse Way (Margary 5), although that may be coincidental, as Astley’s troops were in fact marching from Worcestershire to Oxford; it may be that he was proceeding along a road of which the only known fragment is Margary 560. The Battle of Worcester in 1651, at the end of the Civil Wars, revisited the site of Powick Bridge, which of course was fought at the very beginning.⁸⁰

The turnpiking of roads in Britain began in earnest in 1663 with the first Turnpike Act. The system of levying tolls to pay for the upkeep of roads was already well established but was now formalized and extended and, as such, was a qualified success. By and large it is still only maintenance, not new constructions, but these do begin to happen.⁸¹

In 1685, the Battle of Sedgemoor was a part of the Monmouth Rebellion. The would-be usurper Monmouth led his army out of Bridgwater along the road to Bristol on a circuitous route that was to lead them to the royalist army. Although not recognized as such, there must be a suspicion that the road from Bridgwater to Bristol is Roman in origin. The road south-eastwards along the Polden Hills certainly was Roman and probably a pre-Roman ridgeway, but Monmouth seems to have kept to small lanes and employed a causeway in order to launch a surprise attack on his enemy. The royalists had likewise been approaching Bridgwater from the east, also using a causeway. These causeways were in existence by the fifteenth century and a Roman building nearby at Slape may hint at an even earlier origin for them.⁸²

The 1715 and 1745 uprisings saw the last major conflicts on mainland British soil that are relevant to this discussion, and Roman roads still exert an influence on proceedings. Prestonpans, to the east of Edinburgh and west of Haddington, had featured as the camp site of the English prior to the Battle of Pinkie Cleugh, and in 1745 the army of Charles Edward Stuart defeated government forces in the first major confrontation of the Young Pretender’s uprising. As before, it occurred next to the main east-coast road between Berwick and Edinburgh (see Appendix 5).⁸³

As the Scottish army marched south, an attempt was made to intercept it by the government garrison of Newcastle, under Marshall George Wade, as we have already seen.⁸⁴

As the rebel army marched south, a skirmish with government forces at Clifton Moor in Cumbria took place less than 2 km west of Margary 7 on the successor road running parallel to the Roman route and, as such, sums up the relationship between the Civil Wars battles and the Roman road system: it had evolved but was still influential in events.⁸⁵

Battles and roads

Intriguingly, the association between roads and battles can be a two-edged sword, as the reader will quickly have realized. If a majority of battles are fought on or near Roman roads, then unlocated battles stand a good chance of having also been fought in such circumstances. Hence the plausibility of the suggestion that the Battle of Catraeth (cAD 600) took place at Catterick, or that the Anglian battle with the Picts at Nechtansmere (AD 685) occurred at Dunnichen near the Roman camp at Lunanhead. Moreover, this would suggest that long-sought-after major battles associated with Arthur, like Mons Badonicus or Camlan, may very well also have occurred near a Roman road. The exploitation of the Roman road network by medieval castles can also prove instructive. As an example, the Norman castle at Stafford, and the battle fought nearby (above page 76) may indicate the proximity of a previously unidentified Roman road, in this case perhaps one linking Margary 191 with 793. The association can be even more productive when looking for confirmation of a newly-identified route for a Roman road, so that the battle of Malcolm’s Cross (1093) near Alnwick, where the Scottish king Malcolm Canmore was slain by an English army, serves as one of many indicators of a Roman road running from Newcastle upon Tyne to Tweedmouth, subsequently followed by the route of the Great North Road in Northumberland (see Appendix 5).⁸⁶

It is perhaps ironic that the conflicts that litter the British landscape have one thing in common: they are almost inevitably the result of civil strife. The few examples of successful invasions, like the Danes or the Normans, exploited the Roman road system and there can be little doubt that it would have had its role to play had Operation Sea Lion gone ahead in 1940 (particularly given the apparent German lack of appreciation of the minor road system).⁸⁷

Boundaries

Roads were not just routes along which post-Roman military campaigns were conducted, however. They were frequently used to delineate property boundaries, most famously with the Watling Street acting as part of the border between Wessex and the Danelaw (it is even named in the treaty of cAD 886 drawn up between Alfred and Guthrun). The Fosse Way and Icknield Way are both mentioned in early charters, being used as boundaries, and Dere Street as a medieval boundary has already been mentioned (above page 41). This suggests that names had become attached to the old Roman roads fairly early on in the post-Roman period and some of these names were evidently generic: Watling and Ermin(e) Street are certainly still applied to more than one road. Many roads are still coincident with parish boundaries, virtually all of which are Anglo-Saxon in origin (although some may even relate to Roman villa estate boundaries) so this has come to be one of the means of identifying routes once they have fallen out of use.⁸⁸

In a law of 1050, Edward the Confessor supposedly named the roads protected by the King’s Peace, these being ‘Watling Strete, Fosse, Hikenild Strete, and Erming Strete’. It is now impossible to be sure precisely which roads were intended, the only clue given being that two ran up and down (i.e. north to south) and two across England (east to west), but it is a not unreasonable assumption that these roads can be identified with those we know as Watling Street (Margary 1), the Fosse Way (Margary 5), Ryknild Street (Margary 18, rather than the Icknield Way, Margary 333), and Ermine Street (Margary 2, as opposed to Ermin Street, Margary 41). Similar legislation is enacted by later monarchs, reflecting their continuing importance.⁸⁹

It was not only the roads themselves that were used as boundaries. We have already seen how the Rey Cross, a stone set up on the boundary of the Diocese of Glasgow and said to mark the site of the death of Eric Bloodaxe, may in fact have been on the site of a Roman milestation. There is also a tradition that the London Stone, now situated on the north side of Cannon Street but originally on the south side, at the entrance to what is thought to be the governor’s palace, may in fact be a Roman milestone or even amilliarium, a central milestone from which measurements along radiating roads were taken. It was certainly acting as a landmark and boundary stone in the ninth century AD, when it is mentioned, and continued to do so into the eighteenth century. As a manuport (it is made of oolitic limestone from the Chilterns) it has certainly been brought to the city, but the truth of its origins may never be known.⁹⁰

Maintenance

Medieval monarchs were sometimes more concerned with the form of a road than they were with the state of its surface. Henry I required that two waggons should be able to pass each other on a highway, and that sixteen knights should be able to ride abreast. As with the Romans, a road was seen not just as the surfaced strip in the middle, but also the margins too. The Statute of Winchester of 1285 saw important provision for the road network of Britain. This required manorial landowners to keep the margins of a road clear to a distance of 200 ft to either side ‘so that there may be no ditch, underwood or bushes where one could hide with evil intent’. The upkeep of roads could be funded by pavage, a toll imposed on road users and supposedly intended for the maintenance of the road fabric. In 1353, Edward III decreed pavage for the ‘alta via’ westwards out of London, namely The Strand between Temple Bar and Westminster (Margary 40). Although a picture of poor quality roads is often painted for the medieval period, it has been pointed out that the fact that medieval kings could move around the kingdom throughout the year suggests this is perhaps a pessimistic view of the realities of medieval land transport.⁹¹

Deviations to avoid obstacles are most obvious in any long straight stretch of road which suddenly makes a loop to one side. Although the reason for the detour may no longer be apparent, its effect on traffic often remains with us until today, whilst the old line may often be preserved in boundaries of some kind. Not all such deviations may be recent, however, for part of Stane Street appears to include an original Roman one, where difficult terrain was avoided by a detour (Figure 26).⁹²

It is appropriate that Edward I, under whom the Statute of Winchester was enacted, should have travelled and campaigned widely on the Roman road network. Indeed, as will become apparent later (Chapter 5), his itineraries provide one of the aids to identifying previously unsuspected Roman roads.

In 1555, a highways act known as ‘the Statute of Phillip and Mary’ made parishes, rather than manors, responsible for the upkeep of roads and examples of individual bequests being made towards the upkeep of the road system are known. Individuals were required to give several days of ‘statute duty’ for the upkeep of the system, including the provision of waggons and draught animals. Still, however, this is maintenance and not construction. So what was being maintained? The existing system, of course, and that remained largely Roman.⁹³

Figure 26: A Roman course deviation on Stane Street.

Nodal points and medieval towns

Many Roman military and civil sites were subsequently occupied in the early medieval period, albeit not always in exactly the same location (the Saxon settlements at Corbridge and London being situated next to, rather than on, Roman remains). One of the major attractions of such sites would be their position at nodal points or hubs in the road network. Some sites did not continue beyond the Roman period, however, and one of the reasons for this could well have been roads falling out of use. Two examples of this are Silchester in Berkshire, now a greenfield site with no major roads passing through it, and Mildenhall in Wiltshire, both evidently key nodal points in the Roman and pre-Roman system. Examination of the fossilized remains of the road system in the surrounding modern landscape shows just how effectively they have been removed by comparison with other sites (Figure 27). Sometimes the node in the road system will remain in use but the Roman site be replaced by a nearby medieval settlement, as happened with Wroxeter and Shrewsbury or Corchester and Corbridge. At Mildenhall, the main north-to-south roads (Margary 43 and 44) were diverted to the new medieval town of Marlborough (which was placed slightly further west on the east-to-west Margary 53).⁹⁴

This continued use of large parts of the Roman road network serves to explain why so many prominent English towns and cities have a Roman origin, and equally why some do not. Just as Silchester or Mildenhall fell out of favour, so other settlements may have been located to take advantage of other routes, perhaps even some retained from the prehistoric period. This surely serves to emphasize the reliance upon trade moving along the road system in the medieval and – by implication – in the Roman period too. When populations were less mobile than they are today, it is primarily goods and services (and their purveyors) that would have travelled. Without roads, much of Britain would be deprived of trade.

The enduring importance of communications in medieval Britain is also reflected in the siting of important castles. This is as true of motte-and-bailey castles of the eleventh or twelfth century as it is of stone castles of later times (Figure 28). Indeed, the coincidence of location between castles and Roman forts is less than accidental: not only do the same strategic and tactical considerations hold true, they oversee the same communications network. In some cases medieval castles actually sit within Roman forts, as at Tomen-y-Mur, Bowes, Brough, Brougham, Cardiff, Portchester, or Newcastle upon Tyne (the former Monkchester) but more often they content themselves with being near neighbours.⁹⁵

Figure 27: Shifting the population centre from Mildenhall-Cunetio to Marlborough. Did the roads move because of the settlement change or vice versa?

The military were not the only element of medieval society with an interest in good communications. The church made widespread use of Roman military sites, with early medieval ecclesiastical centres being established at locations such as Chester-le-Street (where St Cuthbert’s mobile remains enjoyed a brief sojourn), Inveresk, and of course in legionary fortresses at Caerleon, Chester, and York (all three of which also attracted a castle for good measure). So it was that castra became -chester (from the Old Englishceaster) just as surely as strata became -street (Old English straet). The great monastic centres required good lines of communications, particularly to facilitate the gathering of tax and the transportation of their produce and this was one reason why Mastiles Lane continued in use from prehistory, through the Roman period, into the medieval and modern eras. As in the Roman period, much will have gone by river and sea, but a great deal still needed roads.⁹⁶

Figure 28: Roman roads and the timber castles of England and Wales (using data from www.gatehouse-gazetteer.info).

Whilst part (arguably a major part) of the Roman all-weather road network continued in use into the medieval period – one estimate, using the Gough Map (see below page 111), suggests forty per cent of the 3,000 miles of road depicted – it seems fairly certain that the same must have been true of those parts of the prehistoric network that were never formally adopted by the Romans, but nevertheless continued in use. Many, perhaps even most, of the drove roads of the medieval period may owe their origins to existing tracks. ‘Portways’, ‘portgates’, and – most particularly – salt roads will have been adopted and adapted to suit the changing needs of medieval society in Britain, just as seems to have happened to the Roman network. It is therefore very likely that a majority of modern byways, country lanes, minor classified – and unclassified – roads probably have a prehistoric origin. Additionally, Hindle has written of ‘roads that made themselves’ between settlements, effectively the same phenomenon we saw described when animal tracks became hunting trails, and so on.⁹⁷

The evolution of the Roman road

The most important maintenance a road could receive was, perhaps a little surprisingly, continual use. Lack of use would allow plants to colonise rapidly, especially blackthorn. This would not only break up the metalling but also, once established, tend to render the road impassable fairly rapidly. However, frost and wet would attack those portions that remained in use and lead to potholing and it is for this reason that various legal measures were taken in the medieval period to maintain the system. The fact that such legislation is repeated over the ages suggests an unwillingness to participate in such work, and that may be supported by the common complaints about the state of roads in Britain. However, it has also been suggested that the volume of complaints almost certainly exaggerated the true state of the road network.⁹⁸

Course deviations are frequently visible on Roman roads that have continued in use, usually manifested as a short loop before the original straight line is resumed. In some cases, deviations could be more substantial, as when the A68 runs parallel to, and immediately next to, the Roman Dere Street (Margary 8f), just to the south of Newstead. The way in which this came about could be seen on old twentieth century trans-Sahara routes, where surfaces were, in parts, so degraded that vehicles ran next to, but not on, the original course. In the medieval period, we can also see this on the Stanegate near Hadrian’s Wall, where numerous later trackways run alongside the Roman road, or again on Dere Street near Chew Green, forming multiple (or ‘braided’) trackways.⁹⁹

The absence of Roman roads or inadequacy of existing routes in much of central and western Highland Scotland led to the construction of the military roads of the first half of the eighteenth century to help quell the rebellious Scots. George Wade had encountered considerable difficulties trying to march his army along the Tyne valley during the 1745 uprising, the remains of the Stanegate being largely disused and replaced by the Carelgate (possibly the course of its prehistoric antecedent), which was in a poor state of repair, and that (through Act of Parliament) led directly to the subsequent construction of the Military Road (the B6318/A69) which utilized Hadrian’s Wall for long stretches, both as a quarry and – in parts – as a foundation.¹⁰⁰

After centuries of service, the Roman road network finally began to be enhanced and augmented during the nineteenth century. The advent of turnpike roads, with their system of tolls, finally saw the arrival of something approaching a viable means of funding road construction and repair, although the system did have its problems and in some places did not last very long. Moreover, far from constructing a completely new road system, it largely just upgraded what was already existing: for the most part, Roman roads. Construction techniques even resembled those of their predecessors, but the development of tarmac in the early twentieth century (the adding of coal tar and later asphalt to the ‘Macadamized’ rammed-gravel surfaces used on the turnpikes) provided a durability the Romans were seldom (but not always) unable to achieve. Engineers gradually became more adventurous, producing cuttings and embankments to ease the gradients of the old system.¹⁰¹

One of the few nineteenth century engineers who actually set about building new roads, rather than improving what already existed, was Thomas Telford. His project for an alternative Great North Road sought to link Morpeth, to the north of Newcastle upon Tyne, with the Dere Street, thus providing a more direct link to Edinburgh than already existed. He changed the line taken by the turnpike in many places and, for a time, both his new road (which is now the A697) and the original course ran almost parallel with each other for some distance (Figure 29). It is ironic that, in doing this, he was almost certainly doing exactly what the Romans did when upgrading the existing British network – keeping the general course, and even long sections of the original road, but modifying specific stretches where he thought it necessary. It is even more ironic that the turnpike road he was replacing may have been Roman in origin.¹⁰²

Figure 29: Parallel roads: the old turnpike and Telford’s successor at Thirlestane, Scottish Borders.

The first truly new road system to be constructed in Britain after the Romans was in fact the railways, the so-called permanent way. Requiring major engineering (in terms of cuttings, embankments, bridges, and tunnels), because the technology was so sensitive to gradients, the system spread across Britain. Even so, it was still influenced by the Roman road network, for the towns and cities it linked were often located where they were because of that same Roman road system. Moreover, many of the topographical factors that determined the course of a road would have much the same effect on a railway, so that it is not unusual to find a Roman road, railway, modern road, and possibly even a canal forced through a geographical ‘pinch-point’ such as the Watford Gap.¹⁰³

The popularity of the motor car saw roads improved even further, although this was still only in the form of strengthening and widening (particularly with the conversion of trunk roads into dual carriageways). Finally, the motorway system saw the first serious campaign of genuinely new road construction, with roads designed specifically for the needs of the motor vehicle, using gentle curves and gradients (rather than the sudden turns and hills of earlier roads), often requiring major engineering to modify the landscape to suit the needs of the road. Paradoxically, we seem to have done to the Roman network what the Romans did to the prehistoric system they found: replaced or upgraded the major routes and continued using the rest of it with only the minimum of modification.¹⁰⁴

Subsequent developments to the road system have in many cases fossilized Roman roads beneath later surfaces. Occasionally, the chance to excavate these occurs, but the practicalities of working on a trunk road often render this difficult. Nevertheless, roads buried in this way are preserved in the condition immediately prior to their being resurfaced and they are protected from the sort of degradation suffered by roads that are no longer in use.

Paradoxically, then, once a road goes out of use, it seldom disappears completely. Nowadays it is rare to find a modern road actually grubbed up once it has ceased to serve its original function, and as trunk roads around Britain have been upgraded, and their awkward parts smoothed out, small sections on the original Roman line may survive, often serving as lay-bys or farmyard access roads, possibly preserving their original surfaces below layers of modern metalling.

1. The popular view of Roman roads arriving in Britain. (courtesy Rupert Besley)

2. The Ridgeway, looking north-east towards Uffington Castle hillfort. (photo: M.C. Bishop)

3. Mastiles Lane confined by eighteenth century enclosure, near Malham Camp. (photo: N. Sheridan)

4. The Vindolanda milestations. (photo: M.C. Bishop)

5. Wade’s Causeway on Wheeldale Moor.(photo: M.C. Bishop)

6. Road foundation in the vicus at Brough-on-Noe. (photo: M.C. Bishop)

7. Milestones from Crindledykes. (photo: M.C. Bishop)

8. Milestation XXI on the Via Nova Traiana in Jordan with multiple milestones. (photo: © M.C. Bishop)

9. Painted milestone inscription from Jordan. (photo: D. Graf)

10. Milestone (from the Military Way?) reused as a gatepost near Great Chesters. (photo: M.C. Bishop)

11. The agger of Margary 54 just visible as a linear mound on Durdham Down in Bristol. (photo: M.C. Bishop)

12. The truncated agger on Dere Street (Margary 8) north of Corbridge. (photo: M.C. Bishop)

13. The Stanegate (Margary 85) at Haltwhistle Burn. (photo: M.C. Bishop)

14. Pack and draught animals on Trajan’s Column. (photo: M.C. Bishop)

15. A tarmac single-track road surface decaying. (photo: M.C. Bishop)

16. Part of the Gough Map, showing the Canterbury to Southampton route – a missing Roman road?

17. A road near Inchtuthil (unknown to Margary) found from the air. (photo: © RCAHMS (Aerial Photography Collection); licensor www.rcahms.gov.uk)

18. A road (Margary 64) indicated by a hedgerow, parch marks, and modern road alignment near Trefeglwys (Powys). (photo: © Crown copyright: RCAHMW)