Chapter 5

Rediscovery

But though antient ways are soon sunk in grounds formerly woody and soft, and now much improved and inclosed; yet such ways as were laid through this county would probably be well paved; which may so far make it more probable, that Stane-street has been antient and Roman. John Horsley¹

Roman roads are still an important constituent element in the modern landscape of Britain, not least through the role they continue to play in the infrastructure. There has always been a kind of tacit understanding of their importance, but it is only when antiquarians like Horsley begin to take a closer look at them that the mechanisms for their survival, and hence the means to study them, would become apparent. Tracing a Roman road requires the accumulation of a series of clues. Each of these, in and of itself, will be no proof of a route, but taken together, they serve to provide a tell-tale body of evidence. Understanding these clues inevitably requires some degree of comprehension of the role they have played in shaping modern Britain.

Traces in the landscape

It is quite unusual for a Roman road to have vanished completely. More often than not it will have left a variety of indications to its existence and students of Roman roads have used these to identify or verify possible routes. Margary outlined his methodology and these principles have been repeated and expanded upon at the same time as remote sensing techniques for archaeology have improved. Indeed, alternative methodologies have developed that are driven by serendipitous observation through aerial photography, geophysical survey, or even the excavation of unexpected roads.²

One of the most obvious ways that Roman roads have been identified in the landscape is where they have remained in use and a modern road exhibits the characteristics of its predecessor. Since, for the most part, the Romans laid out their roads in straight stretches with obtuse-angled turns where necessary, they differ markedly from normal sinuous British country roads, or the sweeping, gently curvaceous routes preferred by more recent roads and motorways that have to take account of the higher speeds of modern travel. However, as we have seen (above, page 104), modern roads do not always sit directly on top of a Roman predecessor, since they can undergo lateral ‘creep’.³

Occasionally roadworks afford the opportunity (albeit on a very small scale) to confirm the existence of an underlying Roman road and such opportunities are one of the few ways in which modern developer-funded archaeology can make a positive contribution to road studies. Even more rarely, major road engineering works will permit some form of larger-scale examination, as has happened in recent years on the Fosse Way in Nottinghamshire, Roman Ridge to the north of Castleford, and Ermin Street through Gloucestershire and Wiltshire. These are very rare, however, but nonetheless valuable for the detailed information they can provide on structural aspects of a particular short stretch of road.⁴

What such developer-funded archaeological work cannot do, however, is provide a synthetic overview of the road system of the sort Margary, and before him Codrington, attempted. This has to rely upon work undertaken by volunteer (one hesitates to use the term ‘amateur’) fieldworkers who devote time to tracing and, in most instances, excavating trial trenches across Roman roads. Until recently, attempts to consolidate this knowledge have been few and far between, but the Cadw Roman Roads Project in Wales has borne fruit with one report already published for the south-east of the principality. Elsewhere, preliminary work is underway to update the whole of Margary’s system by the North East Hampshire Historical and Archaeological Society (see below page 127).⁵

The course of roads that have gone out of use can be preserved in field boundaries, often continuing the route of a modern road where it veers from the Roman route for some reason. This happens for the reasons discussed above in relation to the role of a road as a boundary – even if a road used as a delimiter went out of use, the notional line it marked would usually continue to exist. As such, this has proved to be another valuable tool for road hunters.⁶

There is a wide and much-exercised range of place-names that are indicative of the proximity of a Roman road and they are cited by most of the modern authorities. The most obvious is ‘street’, derived from the Anglo-Saxon interpretation of the late Roman word for a road, strata (see above page 41). This can occur in the name of a town (Chester-le-Street: Margary 80; Stretford: Margary 7; Streatham: Margary 150; Stretham: Margary 23), village (Stretton: Margary 1 and 18), farm (Streatham Farm: Margary 140), route, or other significant landmark. Other common place-names include ‘path’ or ‘peth’ in some part and the same is true of ‘gate’ and ‘causeway’ (or ‘causey’). Another word that it is thought can point to the proximity of a Roman road is ‘chester’, derived from the Anglo-Saxon ceaster, coming directly from the Latin castra for a military base of some kind (but in this case interpreted more widely to include native British settlements).⁷

One of the more puzzling associations of names with Roman roads is that of ‘coldharbour’ and ‘caldecot’. Although the meaning of these words (particularly coldharbour) has been much debated, with little sign of a plausible interpretation being reached, it has long seemed fairly clear that there is an association between the place-name and the presence of a Roman road nearby. This apparent relationship has been questioned using statistical methods and, whilst objections have also been raised as to the antiquity of the name ‘coldharbour’, the conviction that this place-name can be a reliable indicator of the proximity of a Roman road persists, not least as any statistical analysis can only ever be based on known roads. Thus the presence of such a name well away from a known Roman road could equally signify the proximity of an as-yet-unidentifed road and thus render the statistical analysis as good as meaningless.⁸

These, then, are the first two major landscape clues to the existence of a Roman road: the right kind of alignments and related place-names.

Documentary evidence

It has been stressed just how important the later use of Roman roads is in affirming their continued existence. We know of such use, particularly military campaigns, through surviving historical accounts and other documentary evidence, such as official records, and these comprise the third set of clues.

Medieval histories frequently record details of military campaigns and, as has already been pointed out, armies prefer (and have always preferred) to use roads for movement wherever possible. The recurrent use of the same route over a long period is usually indicative of a Roman origin for a road. This does not mean that medieval armies always used a metalled Roman road and ignored other parts of the network, but rather that it was usual for them to take the most convenient route (which for an army is almost always a metalled road).

Similarly, royal itineraries reconstructing the day-to-day progress of kings around Britain can prove a valuable confirmatory source for the continued use of large sections of the Roman road network. Medieval kings travelled widely around their kingdoms, and detailed records of their movements and the associated accounts were maintained, listing the places visited in each year in order. The nature of such expeditions, invariably accompanied by a large entourage, meant that the best road available was desirable. As with John, superimposing a map of the various surviving and reconstructed itineraries upon one of the Roman road systems as we currently understand it can not only act as a window on some parts of that system which were still in use, but also allow a much more comprehensive composite picture of the whole to be formed. However, it has to be remembered that just because a Roman route was not used by an itinerary, it does not necessarily follow that it was no longer active. Conversely, just because an itinerary uses a route with no known Roman predecessor it does not mean there is not one to be found.⁹

Legal documents can also be useful. The most famous example is probably the treaty drawn up between Alfred the Great and Guthrun, where the Watling Street was used as part of the delimitation of the Danegeld. This also works on a smaller scale, however, where descriptions of Anglo-Saxon parish boundaries incorporated lengths of Roman road. The road up the Greet Valley, near Osmanthorpe in Nottinghamshire, is recorded in this way.¹⁰

Medieval maps, like those of Matthew Paris (c. 1250) and the so-called Gough map (c. 1360), are other potential sources of valuable information. Of the four versions of Paris’ (which is based on an itinerary from Dover to Newcastle), only one actually marks the course of a road. The Gough map contains far more information and provides a number of interesting details, such as an east–west road across south-eastern England – from Canterbury to Southampton via Hythe, Rye, Winchelsea, Battle, Boreham Street, Lewes, Shoreham, Arundel, Chichester, and Havant (Plate 16) – which is only partly attested from known portions of Roman road (Margary 130, 140, and 421). The modern place-name Boreham Street (although only ‘Borham’ on the map) is of course suggestive of the proximity of a Roman road. This route appears to be confirmed by other ‘street’ place-name evidence (Ham Street, Brook Street, Reading Street, and Gardiner Street in Herstmonceux), and seems to have hugged the contemporary coastline (now much altered by sea-level change and alluviation). It may also have been employed for at least one royal itinerary, perhaps even providing a highway for William the Conqueror’s invading army in 1066 (see above page 76). However, the Gough map is also surprising for some of its omissions. Whilst a road from Canterbury to Southampton is shown, the arguably more important route to London (Margary 2) is not. Similarly, Hadrian’s Wall is depicted, but no road accompanies it, despite King John’s evident use of the Stanegate. One estimate suggests that forty per cent of the approximately 3,000 miles of road shown on the Gough map were Roman in origin, but it clearly presents a partial (in more than one sense) picture.¹¹

At a more local level, tithe maps and their associated documents can sometimes prove invaluable in providing details of field names, some of which can produce place-name evidence for a road, or even reveal a long-lost field boundary, ploughed out during the inevitable enlargement of field systems in recent times.

Bertram and Stukeley

An interesting footnote to Roman roads studies, and one which was to have repercussions into the modern era for the mapping of the road system, concerns what is known as the Bertram Map. The antiquarian Stukeley ‘discovered’, and encouraged Charles Bertram (an Englishman who had been living in Copenhagen) to publish a rather unusual map he claimed to have found, supposedly compiled from material recorded by a medieval writer, Richard of Westminster (identified by Stukeley with one Richard of Cirencester, who had indeed lived at Westminster). The whole map was subsequently shown to be a collection of data garnered from various sources, but interestingly it still has a habit of being able to ‘predict’ the course of undiscovered roads. A rational explanation for this might be that the hoaxer, Bertram, used contemporary roads to fill in the gaps in his other sources, thus – given the nature of the contemporary road network – most of his major roads were bound to be Roman in origin. It is therefore more by luck than judgement that he may have included as-yet-unrecorded stretches of Roman road.¹²

Fieldwork

One of the attractions of the study of Roman roads has been that the tools are readily available to any who take an interest, even the lone fieldworker. Unlike major archaeological projects, which nowadays require vast amounts of money and a huge army of specialists, the pursuit of a road need not be expensive nor particularly labour-intensive. Such research can indeed be carried out by the determined individual, Ivan Margary’s preferred way of working, but ultimately, the back-up of a small team, such as many archaeological societies can provide, has proved invaluable, and this was what lay behind the group known as the Viatores (see page 121). Whilst research can piece together many of the clues outlined above to show the likely course of a Roman road (and examples of alignments, place-names, and accounts of medieval armies using a road are pretty compelling circumstantial evidence), final proof requires some sort of archaeological investigation. This process can best be summed up as research, exploration, and examination. Even the last of these can, however, lead to less than satisfactory conclusions. Roads can often look Roman when in fact they were not. A number of routes excavated by Raymond Selkirk and claimed as previously unknown components of the Roman network have been disputed by other researchers, but the most famous contentious ‘Roman’ road has to be the paved section at Blackstone Edge. This last is certainly on or near a genuine Roman road (Margary 720), but the point at issue is whether that particular surface is Roman or later. It is salutary to recall for how long Roman roads have been used and repaired.¹³

Research

Researchers looking for Roman roads in Britain usually begin by examining maps. Even the modern 1:50,000 Ordnance Survey Landranger series can be of some use for this, but by far the best place to start is the First Edition Ordnance Survey six-inch (1:10,560) series, most of which date to just after the middle of the nineteenth century (although in Scotland William Roy’s maps of the 1749–55 Military Survey can be enlightening). The OS six-inch maps are of a large enough scale to enable details such as field boundaries and settlement names to be included, yet sufficiently small to allow something of an overview of a likely road course in relation to the topography and place it in the context of other monuments in a given area. Conveniently, all of these are now accessible online.¹⁴

One needs to be aware of the shortcomings of the Ordnance Survey’s recording of monuments in the past with particular respect to Roman roads. A marked tendency can be detected to ‘join the dots’ between possible sightings of a road on the First Edition six-inch series and Margary is occasionally dismissive of some of the Survey’s more idiosyncratic identifications (although not averse to using the same technique himself at times!). This inevitably leads to misrecording and misunderstanding of the course of a road, and many a vain attempt at exploration has resulted from a misplaced reliance on the clairvoyant capabilities of the Ordnance Survey.¹⁵

Copies of old Ordnance Survey record cards exist for most monuments recorded on their maps and these are usually available in the respective National Monuments Records (NMR) for England, Scotland, and Wales. These can be compared with the information on the OS maps, and the three NMRs integrate this data, together with other material, on their base maps for any given area and this is all now generally accessible online. The heritage environment records (HERs, formerly sites and monuments records or SMRs) for any county or regional council have their own records. These will often overlap with the appropriate NMR, but each will usually contain at least some material that the other does not have. Such national and regional databases include large numbers of aerial photographs from a variety of sources, and to these have to be added the coverage that used to be held by the Cambridge University Committee for Aerial Photography (CUCAP). An ever-increasing number of HERs and SMRs are also accessible online.¹⁶

Aerial photography is vital in understanding Roman roads. Coverage, both vertical and oblique, increased dramatically during the Second World War and, once its value had been recognized, has continued to expand in breadth and depth. The largest single campaign after the war was undertaken to mark the turn of the millennium in 2000 by Getmapping plc, producing what they called the Millennium Map. Starting with England, they have subsequently expanded their coverage to the whole of Britain. Aerial photographic surveys like this, combined with high-resolution satellite imagery, form the basis of virtual globe software such as Google Earth. Indeed, the advent of internet access to large-scale aerial and satellite coverage has revolutionized archaeological research and given rise to the tabloid use of the cliché ‘armchair archaeology’. Neogeographical tools (like Google Earth) provide the means to survey routes and compare more than one set of coverage, where available. Some areas of Britain are even covered by black and white photography from the Second World War in this way, all mapped onto the modern base provided by Google. To illustrate the potential of this visual data, a file containing Margary’s map of the road network plotted onto Google Earth is available on the accompanying website (see Chapter 7).¹⁷

Aerial photographs not only permit an overview of a section of road, in much the same way as maps do, but also provide evidence for previously-unknown roads. Crop or soil marks will sometimes reveal the line of a road in this way. A good example of this is a previously unknown minor road on the eastern coastal plain of Scotland, not far from the legionary fortress at Inchtuthil (Plate 17). Now under agricultural land, the road shows as a pale line flanked by its dark drainage ditches but also accompanied by what appear to be quarry pits on either side, beyond the ditches. Since the road is at this point crossing a river floodplain, the purpose of these was almost certainly to provide sand and gravel for the construction of the road. This road, for whatever reasons, is not indicated by any other form of evidence except for aerial photographs and is a clear demonstration of the value of aerial photography. Another road (Margary 64), near Trefeglwys (Powys), shows up well in an oblique aerial view (Plate 18), its course indicated in turn by a hedgerow, parch marks, and then a matching modern road alignment.

Lidar (Light Detection And Ranging) is a comparatively new archaeological technique that uses airborne laser survey of the ground surface to produce extremely accurate maps of the topography and has the potential to identify features that might otherwise go unnoticed, even in the field. It has the useful capability of being able to penetrate the canopy in wooded areas in order to reveal the underlying terrain. Most such mapping in Britain was initially done for the Environment Agency to assist in flood prediction and that can be archaeologically useful. Increasingly, however, bespoke surveys are being undertaken to examine archaeological landscapes.¹⁸

Exploration

A route suggested by the study of maps would next be checked on the ground. However good maps or aerial photographs might be, it is always deemed essential that a road be explored on the ground. Some roads may even first suggest themselves by being driven along. Nevertheless, there are plenty of ‘Roman’ roads that are in fact straightened modern sections of older roads, although these should always be apparent by comparison of old maps with newer editions. More importantly, the observation of an aggeron the suspected line can only truly be appreciated from the ground, although modern methods of lidar survey can pick out extremely subtle micro-earthworks. At Charthouse-on-Mendip, archaeological lidar survey has demonstrated the continuation of the Roman road (Margary 45) on the alignment described by Margary.¹⁹

There are many ways of exploring a Roman road within its landscape. A lucky few will be able to overfly a suspected road, but even this might not produce the same insights into the lie of the land or lines of sight provided by travelling along the surface of a road. This can be accomplished in a number of ways.

Whilst it might be thought that a car provides a useful means of transport for a survey of a fairly long stretch of road, in fact it has distinct disadvantages (besides the obvious environmental ones), not least that – should it prove desirable to inspect part in greater detail – it can be difficult to park safely (or even legally in the case of some roads). Walking is probably the most straightforward way of exploring a Roman road that has been employed, but is of course very slow and, where following a course along a modern trunk road in the countryside, can be positively dangerous. The ideal compromise might be to use a bicycle or (and this is admittedly rare) ride a horse. Slower than a car, but faster than walking, such methods permit an overview of fairly large sections of a road, while still at the same time allowing the researcher to stop virtually anywhere they wish. There is also the added advantage of unhindered access to routes that have been reduced to the status of bridleways (as often happens with Roman roads).

Examination

The only way to prove the existence of a road conclusively is by some sort of physical intervention or ‘ground truthing’, but there are other techniques that can provide qualitative data about the nature of a supposed road. All will require the cooperation of landowners and, in certain circumstances, some (probing, geophysical survey, and trenching) may even necessitate applications for scheduled monument consent.

Simple fieldwalking, with the permission of the landowner, has long been used to provide evidence of, and crude dimensions for, ploughed-out road surfaces. In most cases where this happens, however, such roads will be visible without even needing to enter a field, especially from the air.²⁰

Probing is quick and simple and supplies simple binary data about a road, since it either makes contact with a hard surface or it does not. Of course, frequent strikes on hard material need not necessarily be caused by the remains of a road, but if it is on the line of a road that has been suspected for other reasons, then it may well be another indication of the promising nature of the investigation. This will obviously only work on roads that have gone out of use, and probably in a rural area. Note that probing is illegal on a scheduled monument without having first obtained scheduled monument consent. The outlines of scheduled areas are accessible online using Defra’s MAGIC mapping facility (see Chapter 7).²¹

Some researchers have had success with archaeological dowsing, although there is as yet no objective assessment of the efficacy of the technique in relation to Roman roads, nor even any indication of how widespread its use might be. If there is uncertainty over how dowsing works, there seems little doubt that it can work, under the right circumstances (and the writer knows of at least one dowser who enjoys a measure of success in using the technique as part of his armoury of methods for tracing Roman roads). Nevertheless, it should be pointed out that any results supported only by dowsing are almost invariably going to be the object of academic criticism.²²

Geophysical survey (and related forms of remote sensing) provide more sophisticated data about extent, course, and features of a road, but there is still much that it cannot provide (except indirectly), such as dating evidence. In this respect, resistivity survey is perhaps the most useful in road prospection, since roads normally provide high-resistance readings, side ditches low ones. It is, however, labour intensive for large scale survey.²³

Far and away the best way to test the existence and nature of a Roman road is to excavate a section through it. Unfortunately, it is also the most costly and time-consuming method. A sondage will provide details of the materials and methods of construction, show the precise width and depth of the road at the chosen location, and how many (if any) resurfacings or repairs were made, as well as placing the road in the context of any flanking ditches or quarry pits, provided that trenches are made sufficiently long to encounter such items.²⁴

What others have done before

The study of the roads of Roman Britain has always been the province of amateur scholars, by and large, and none of the major scholars who might be named were tenured academics. Fortunately, this does not mean the quality of their scholarship is in any way inferior to that of their colleagues whose interests lie elsewhere, but is rather a sad reflection of the lack of attention most scholars pay to the road system in Britain.

Comments on the Roman road network in Britain can be found in many antiquarians. Roger Gale’s Essay Towards the Recovery of the Courses of the Four Great Roman Ways was appended to the sixth volume of Leland’s Itinerary in 1764, so it is unsurprising that Horsley’s Britannia Romana of 1732 contains much on the subject of Roman roads, since Horsley originally undertook his journey in 1725 in the company of Gale. Equally, during his travels around Britain, Daniel Defoe noted the importance of Roman roads and even examined the structure of some examples. However, Thomas Codrington was arguably the first of the great amateur scholars to produce a detailed survey of the road network and Margary was quick to acknowledge his influence in his own work. An engineer by training, Codrington is otherwise best known for his volume The Maintenance of Macadamised Roads (perhaps less so for his Report on the Destruction of Town Refuse), so the precise nature of his background – and the likely reasons for his interest in Roman roads – is not hard to find. His work, Roman Roads in Britain, first published in 1903 when he was seventy-six years old and revised twice before his death in 1918, provided an overview of the road system of Roman Britain. In his review of the third edition of the work, Francis Haverfield, doyen of Romano-British studies, noted acerbically ‘as so often happens, the surveyor has excluded the geographer’ and ‘Mr Codrington cares little for what others have done before him’. Haverfield characteristically did not ‘get’ the cataloguing instinct betrayed by most students of roads and always sought the big picture. Codrington’s work remains worthy of examination, however, not least for those roads he saw fit to include but Margary chose to exclude (often without comment). A good example of this is Codrington’s route 17, north of Ilkley, one branch heading north towards Hayshaw (where two pigs of lead had been found in the eighteenth century) and another north-eastwards towards Aldborough. Margary includes the latter (as 720b) but ignores the former. ²⁵

Best known for his ‘ley line’ theories, subsequently transmogrified into something resembling a minor cult, it is often overlooked that Alfred Watkins was an ardent proponent of the existence of a prehistoric road network, even if his proposed routes were impossibly straight. What is significant is Watkins’ acknowledgement that pre-Roman peoples had the wit and ability to have and indeed use a road network, an attitude that has not always been shared by his more traditional successors.²⁶

Few prehistorians – and probably even fewer Roman scholars – have heard of Hippisley Cox. Whilst Alfred Watkins’ Old Straight Track has attracted much attention, and a fair measure of opprobrium, for his ‘ley line’ theories, a quieter voice was proposing an extensive system of prehistoric trackways running across southern England. Hippisley Cox noted the association between prehistoric monuments and his ‘green roads’, most of which were ridgeways, and for this he is important. Completely straight routes of the type proposed by Watkins may look good on paper and even work in very flat areas, but they are completely impractical in the British Isles: one only has to look at more recent examples to see how roads have always been moulded to the landscape. The same was true of railways and only modern air corridors come anywhere near approaching Watkins’ ideal.²⁷

One of the most-studied roads has always been Stane Street, running between Chichester and London (Margary 15). A monograph by Hillaire Belloc was followed by a ‘critical review’ by Capt W.A. Grant, which contained the memorable put-down ‘Mr Belloc has not feared to rush in where others have feared to tread’. These were followed by S.E. Winbolt’s masterful With a Spade on Stane Street, which – concentrating on the archaeology and, by and large, avoiding personal abuse – has set the tone for countless subsequent road reports.²⁸

These were the predecessors to Margary and – unlike Codrington, if Haverfield is to be believed – he was only too aware of their contribution to his labours. Ivan D. Margary began his interest in Roman roads purely locally, publishing Roman Ways in the Wealdin 1948, but soon expanded his compass to include all of the roads of Britain. He produced what was then the definitive account of the Roman network in The Roman Roads of Britain in 1955, and this went through two further editions before he died. The second was a major reorganization into one volume, but the third enjoyed just a few pages of addenda at the end. Despite much new work, this magnum opus, long out of print, has never been superseded and is still much sought after. Margary, like Codrington, was unusual in his willingness to tackle the subject on a nationwide basis, a much more difficult task than writing about a particular region known to a scholar, since the writer inevitably becomes more dependent on other researchers in a way that is less true of regional studies. The scale of his contribution to the study of Roman roads in Britain can be gauged by comparing his map of the system with that of Codrington, separated by only five decades of study (Figure 30).²⁹

Figure 30: Codrington’s network (1918) compared to that of Margary’s (1973).

There are, however, a number of problems with the road system as defined by Margary.

First, and most obviously, the hierarchy he employs (Main Routes, Principal Branches, and Minor Branches) is his definition and may not have been shared by the Romans (if, indeed, they even viewed their roads in such a hierarchical way). It is a simple matter (but no more valid) to take his overall network and re-assign priorities, whether it be based on nodal points or the routes themselves as they appear on a map, a manifestation which, we must remember, was denied the Romans in anything more than the most rudimentary form. An excellent example of this would be the road between the legionary fortresses of Chester and York, which one might expect to have been of some importance (it was certainly fortified in the first and second centuries AD). In Margary’s system, this consists of 7a (Main Route), 712 (Minor Branch), and 28b and 28c (Principal Branch).

Second, the system is quite obviously incomplete. Fragments of roads float in the middle of nowhere (Margary 164, 731, or 821, for instance), whilst others head determinedly cross-country only to terminate abruptly (Margary 31, 151, or 713). Two stubs may point suggestively at each other (Margary 55 and 183) only to be given differing classifications (a Principal Branch and a Minor Branch in that instance). In some cases strategically logical links between nodal points – the pre-Hadrianic forts at Templeborough and Castleford, for example – are completely absent. Codrington at least considered two candidates for a northern road from Templeborough, even if he was uncertain about them; Margary ignored them. Other, more recent, fieldworkers have added considerably to our knowledge for a particular area, but there is no official coordinated attempt to do the same for the whole of Britain (nor does there appear to be any interest in one, but there is at least now an unofficial project to remedy this: see below page 127). Similarly, South-West Scotland is poorly served by known Roman roads, despite a number of Roman military sites and a number of what may be indicative medieval battlefield sites. Attempts to address this imbalance point up the inadequacy of the research undertaken on the region until Wilson’s comparatively recent studies.³⁰

Third (and this may be seen as nit-picking for a work of such a great sweep), and at the opposite end from Haverfield’s criticism of Codrington for his lack of overview, Margary could be less than precise in his descriptions of roads, misspelling of place-names, and the almost complete absence of Ordnance Survey grid references.

The incompleteness factor is one of the major obstacles to any attempt to draw scientifically acceptable conclusions about Margary’s network: as with the coldharbour study, statistics derived from an incomplete dataset are always going to be of questionable value. The other such obstacle is that Margary’s network is not necessarily that used by the Romans (which may have included native trackways to a greater or lesser degree). Furthermore, it might be argued that insufficient allowance was made for the fact that what we see now may well be a palimpsest, incapable of taking into account the possibility that roads came into and went out of use over time, so that there was (and still is) in effect an ever-evolving network.

Setting aside his idiosyncrasies (such as the frequent misspelling of place-names), a final problem lay in Margary’s own methodology. He was a southerner and the leading expert on the road network of Kent and Sussex. He worked closely with the Viatores, who covered the Midlands, but the further his subject matter moved from his own familiar territory, the more he became dependent upon other authorities. Not all of these were of the same standard as his own endeavours and one occasionally senses him questioning (if very politely) conclusions that have been drawn by fieldworkers elsewhere. He certainly attempted to visit many of the roads in his book, but there is little doubt that he did not know the whole system as well as the region with which he was familiar; nobody could.³¹

Therefore it is important to distinguish the entity that is Margary’s Roman Roads from what might be termed the Original Roman Road Network, the former being very closely defined for us, but the latter an ideal, the extent of which is currently unknowable beyond the bald statement that it was not the same as the former. It turns out that these are, however, only two of many possible networks.

After Margary, prominent researchers have mainly remained regional in their interests. Key amongst these was a loose association of enthusiasts who called themselves, rather romantically, ‘The Viatores’ (a Latin term meaning ‘road surveyors’). Working on the assumption that they would make more progress by cooperating than by pursuing their own individual courses (both real and metaphorical), they worked closely with Margary and concentrated on the south-east Midlands of England. The results of their labours can be seen by comparing their map with that portion of Margary’s from just nine years earlier that covers the same area (Figure 31). The Viatores employed a more detailed method of publication than Margary, however, including detailed descriptions of routes, Ordnance Survey grid references and, doubtless at great expense, two-colour map sections showing annotated routes overlain on the two-and-a-half-inch OS map base, together with drawn sections wherever they had been dug. This approach left little doubt about their suggested routes, even though some of them may be open to question in specifics. One of the most crucial observations that could be made about their work was the density of Roman (and Romanized) roads that could be found using diligent research followed up by fieldwork. Indeed, the comparison in Figure 31 appears to supply a salutary lesson in what may await discovery in the rest of Britain. At the same time, their methodology can be (and indeed has been) criticized, and it seems their findings need to be taken with a pinch of salt in many cases.³²

Figure 31: Margary’s network (1973) with The Viatores’ (1964) contributions.

Margary’s network without the Viatores’ contributions.

The Viatores were quite unusual in their approach and road chasing tends to be a solitary pursuit, by and large. Norman Field specialized in the roads of Dorset, particularly those he thought were connected with the early conquest period. In more recent years, whilst in North Wales, Edmund Waddelove has produced detailed accounts of his elucidation of the road network in that area. Martin Allan has undertaken similar work in the Lake District, tracing the roads from Brougham to Moresby and Hardknott to Ravenglass. Attempts are now being made to coordinate much of this ‘amateur’ research (with the ultimate aim of producing an update to Margary’s last edition of his catalogue of roads), whilst new examples are still occasionally found by excavation (as on my own sites at Roecliffe and Melton) or by aerial and geophysical survey.³³

One individual who combined many of these approaches, often controversially, was Raymond Selkirk. Often mischievously anti-establishment in his books, with provocatively teasing blurbs on their dust jackets, Selkirk was sometimes criticized for his more extreme suggestions, usually at the risk of overlooking his many genuine observations of value. Nevertheless, some of his proposed routes have been shown to be, at the very least, questionable. Whatever else he might have been, Selkirk was an acute and accomplished observer (something perhaps attributable to his maritime and aviation background), and this quality of observation tends to be evident in the work of all students of the Roman roads of Britain.³⁴

What is ironic is that the ‘establishment’ against which Selkirk and others rail is arguably profoundly disinterested in the study of the road network, finding it far less sexy than frontiers or Roman and native interaction. This may help to explain why the great road scholars in Britain – Codrington, Margary, Field and their ilk – were amateurs (in the sense that they were not paid to do what they did) and enthusiasts, and there were no professional archaeologists interested in doing what they did. This is not universally true of Roman road studies, with the likes of Chevallier writing on the roads of France or Graf and Kennedy on those of Arabia, and it appears to be a curiously British trait. Now, perhaps more than ever before, when archaeology is being increasingly ‘professionalized’ and some object that there is no room left in it for amateurs, it is a good time to remember that the steady increase in our knowledge of the Roman roads of Britain is almost exclusively due to the labours of amateur scholars. The one academic project in this field – the Birmingham Roman Roads Project – is run by the extramural department at the University of Birmingham; traditional university archaeology departments seem reluctant to engage in this field. Nevertheless, there may be hope in the rise of community archaeology, with one such project hoping to trace the route of the Stanegate (Margary 85) between Corbridge and the North Tyne.³⁵

One of the by-products of regionalism in interest in roads is that some areas invariably get left out of the system. Whilst some areas of Margary’s maps seem to lack much detail, others appear completely barren. Cornwall, for instance, has produced a number of milestones (Figure 9), but seemingly no roads west of Devon. It is extremely unlikely that these stones have all been transported a long way from a known Roman road, whilst place-name evidence, such as Stratton (a ‘strata’ place-name), provides hints that what appears to be a blank on the map may well have had its share of Roman roads that will, at some point, repay investigation.

Using documentary sources and a limited amount of field observation, it is still possible to find likely new roads that merit further investigation. The success of the Viatores in filling in the gaps in the northern Home Counties shows how this is possible and further examples are offered below in Appendix 5.

So it is that the factors governing any notional map of the Roman road network are more than just dependent upon geographical and historical factors, but must also allow for the presence or absence of research into a given area. So, as with any archaeological distribution map, a map of the Roman roads of Britain such as Margary’s is as much a record of archaeological endeavours as it is one of Roman strategic thinking or infrastructure planning. It will doubtless continue to change in the years to come.

Mapping Roman roads

The reader might expect the mapping of the Roman road system in Britain to have improved considerably in recent years. Unfortunately, this is not the case, at least so far as the published information is concerned.

The Ordnance Survey produced a succession of Roman Britain maps, arguably reaching an apogee with the fourth edition of 1978 at a scale of 1:625,000. Subsequently, the project lost its way somewhat, the nadir coming with the much-criticized fifth edition of 1989. For various reasons the Survey moved away from serious academic mapping of Roman Britain completely, the most recent edition (the sixth) being aimed solely at schoolchildren: in this it appears to be returning to the territory of old-fashioned educational wallcharts and, sadly, abandoning any academic aspirations (with some sensitive design, the two, of course, need not be mutually exclusive).³⁶

Their criteria for the inclusion of roads, or their categorization as known or suspected, have (inevitably) been open to criticism, and some lengths of road varied in status between editions, often seemingly at the whim of the cartographers. This all seems most curious, when it is recalled that Margary’s work was undertaken with the close cooperation of the Ordnance Survey, whose maps had, since their first editions of the nineteenth century, recorded the known and suspected courses of Roman roads.

Between the fourth and fifth editions of the Ordnance Survey Roman Britain maps, students of the road system were encouraged by the appearance in 1983 and 1987 of two British sheets of the Tabula Imperii Romani at a scale of 1:1,000,000, albeit on a different projection to that used by the Ordnance Survey. The TIR was (and apparently still is) a project to produce maps of a consistent quality for the entire Roman empire. The pace of new discoveries soon left the maps looking out of date, however, and there does not seem to be a policy of regularly updating the series. Ultimately, all paper-based maps and atlases are now at a disadvantage in comparison to digital mapping.³⁷

The problems inherent in studying the road system of Roman Britain can be summed up eloquently in one simple question: which road system? The differences between the various mapping schemes, their varying (and not always transparent) criteria, and the inevitable distance of the researcher from most of the primary evidence mean that an element of scepticism is always advisable before accepting a suggested course as a fact. This sentiment will be familiar to any professional archaeologist sent to a remote location on a watching brief because the Ordnance Survey records the likely course of a Roman road at that point, only to find their speculation overly optimistic.

By way of illustration, it may be worth reviewing one simple example, the Roman road from Bath to Sea Mills (Figure 32). The first edition of the six-inch (1:10,560) Ordnance Survey ‘county’ maps records its course (even naming it the Via Julia, after the spurious antiquarian tradition). By the time of the third edition of the Roman Britain map in 1956, its route is recorded as certain between Bath and just west of Bitton, and assumed until Sea Mills. The fourth edition is more specific, recording it as unknown immediately west of Bath, known to the east and west of Bitton, and uncertain as far as Sea Mills. This situation is repeated on the ‘revised’ fourth edition. However, the relevant TIR sheet chooses to ignore the road completely. This despite the fact that, in the first edition of his book, Margary gave a detailed account of its course (including references to at least one place where the road was tested by excavation), supposedly working in consultation with the Ordnance Survey. Admittedly, the OS convention for an uncertain course could look alarmingly similar to interrupted sections of certain road course, but the same excuse does not exist for the TIR omission. It is difficult to see how the various lengths of agger, excavated sections, and numerous alignments of this road can translate into a non-existent road on the TIR sheet.³⁸

Figure 32: The so-called Via Julia between Bath-Aquae Sulis and Sea Mills-Abonae.

All this means that students of the Roman road system (and of Roman Britain in general) effectively lack a modern interpretation of the available evidence, despite the huge advances in modern technology. Steps are now being made to redress this state of affairs, with an attempt by the North East Hampshire Historical and Archaeological Society to coordinate work on the road system in their region since Margary, with a view to updating his corpus, and a useful listing of criteria for the identification of a new Roman road produced. These are detailed on their website as follows:

5* Major Roman Road where the course on the ground is exactly known for most of its route. Applies to routes like the Devil’s Highway [sic] or Foss Way, and not many more are likely to be found.

4* Margary’s criteria for establishing a Roman Road: cumulative evidence upon an alignment, with excavation in at least one place on most alignments. While the line across country is established, it does not mean that each piece of evidence must be shown to come from Roman origins, nor that the Road’s exact position is known on the ground except at excavation sites or possibly at terrace sites.

3* Accumulative evidence upon the alignments, but without the excavation evidence. Quite a lot of the Margary entries are of this class. Individual pieces of evidence may be coincidence, but the whole gains credibility the more alignment evidence is gathered, until it becomes quite probable that the alignments are Roman.

2* Projects based on sound documentary theory, but without the full standard of accumulated field evidence upon alignments. May be becoming probable. Some Margary entries of this class.

1* Sound documentary theory of long lines of parish, field & property boundaries, paths, place names, Saxon Charter and other historical documents, air photos, but without significant field evidence. A few Margary entries of this class.

0* Some documentary or field evidence, but not enough to assign any probability. May be worth expanding research from evidence available. Margary entries in this class exist.³⁹

Whilst it is possible to quibble over details (even excavation of a road surface may not provide absolute proof of a road’s Roman date), this star-rating system provides an interesting alternative way of viewing the road system as currently perceived. Margary chose to grade his network subjectively upon his view of the significance of a given length of road, but the NEHHAS scheme plots them by the quality of evidence, thereby avoiding the assumptions of importance implicit in Margary’s method. Neither, of course, is (or can be) complete, but both hint at the potential for future research on the road system in Britain. The work of the Society to date is accessible on their website, together with summaries of work that has been submitted to them.⁴⁰

Where to next?

So, what is the future of Roman road studies in Britain? With the most recent edition of Margary (1973) now badly in need of updating, no serious themed mapping being published by the Ordnance Survey, and apparently no plans to update the TIR for Britain, the situation does not look too promising. Moreover, with site-specific developer funding providing the lion’s share of archaeological finance, and no sign of a major campaign of research being mounted by any of the national archaeological bodies, the study of the road system remains firmly in the hands of amateur scholars. The question is, how can it progress?

The answer must lie in the sort of initiative mounted by The Viatores in the 1950s and 1960s, where local researchers cooperated to draw a larger picture. Attempts are already in hand to update Margary’s work, as we have just seen, but clearly what is really needed is an open source map using GIS techniques, contributed to and accessible by all researchers, but free of any debt of copyright to any body other than that allowed by a licence similar to the GPL, perhaps the Open Content License (OCL). The internet renders such a project feasible and, obviously, desirable, in the face of indifference in most quarters. Whether it will come to pass or not is a different matter. There may also be hope in any future change in the funding basis for British archaeology that might one day occur: as an example, RESCUE (the charitable trust for British archaeology) have proposed a general development tax that would to some extent alleviate the close ties of current developer-funded project research to very limited goals and perhaps even permit wider funding to broader projects like the study and mapping of Roman roads, a task with which bodies such as English Heritage and the Royal Commissions for Ancient and Historic Monuments of Scotland and Wales would be well-equipped to assist.⁴¹

Much remains to be done at even the most basic level in terms of mapping the road network – the work of The Viatores, Allan, Field, and Waddelove shows just how much can be achieved and there are still huge gaps in the system that urgently require filling, with roads that seemingly go nowhere: Brough-by-Bainbridge has two known roads running to it, but these are not connected to anything; Margary 720b from Ilkley to Aldborough obviously needs completing, as does the trio of roads (Margary 17, 170, and 222) to the east of Northampton which appear to be heading for a junction. Some major Romano-British towns have known or probable gates and no known roads running from them, Winchester (Venta Belgarum) being an example, on its east side. Similarly, Aldborough (Isurium Brigantum) is defined by only one known road (Margary 8) passing through it, yet has more than two gates. These lacunae are poignant reminders of the gaps in our knowledge. Indeed, wherever there are roads in Roman Britain, there are loose ends: at least 500 km (311 miles) of them, at a conservative estimate, in a system of approximately 11,938 km (8,067 miles) of roads identified by Margary. New roads are found all the time (for example at Melton), by no means all of them expected, and the basic groundwork (in terms of research and fieldwork) in tying together the known network remains to be done.⁴²