From trackway to road
Corbridge, Roecliffe, and the case for a proto-Dere Street
The notes are linked with both the text and the bibliography. Once a text note -  - has been selected, the reader should be brought to the appropriate note - 44. - in order to read that note. Whilst there, selecting a reference will guide the reader to the appropriate reference in the bibliography. Their browser back button will then take them back to the note again. Selecting the note number - 44. - again will take the reader back to the original reference - .
It requires no great insight to
state that the road
Roman Britain is the product of centuries of development, improvement,
and neglect. Its influence stretches far beyond the needs of those
military commanders for whom it was probably initiated, for it has
played an important role in the later history of Britain, for example
the conflicts between England and Scotland, where the movement of
was frequently dictated by the course of the existing (in other words
Roman) road system. Even today, many of our main arterial road links
based upon that same system and it is by a nice irony that an
in advance of development to one of these roads provided the genesis
for this paper.
In 1993, work began to upgrade the A1 dual carriageway in North Yorkshire to full motorway status. Survey work undertaken for the Highways Agency by Northern Archaeological Associates indicated the existence of a Roman site in the vicinity of the Devil's Arrows standing stones just to the west of the town of Boroughbridge.  Subsequent geophysical survey identified two rounded corners and three sides belonging to a Roman fortification, together with an associated defensive outwork system, a road passing the site (complete with a spur linking it to the fortification), and strong indications of both intra- and extramural activity. The geophysical plot was of sufficient quality not only to show the site to have been double ditched, but indeed it also provided strong indications that it possessed a box rampart constructed within twin palisade trenches. 
Although the castra of Roecliffe itself was not to be affected by the motorway, the road corridor did impinge upon the extramural area, part of which already lay underneath the embankment of the A1 dual carriageway. A strip 20 metres wide and 500 metres long was area-excavated at the edge of the new embankment. Although badly disturbed by medieval agricultural activity, it was clear that at least three phases of timber structures bordered the east-west Roman road where it crossed the corridor and in fact spread some way back from it. The dating evidence was consonant with an early Flavian date for the founding of the site and likewise suggested that occupation did not extend into the second century; in fact, there was every indication that the site was exclusively Flavian.
Whilst the geophysical survey indicated that the road led westwards towards the River Ure, perhaps to a crossing point, excavation prior to the construction of the north abutment of the new Arrows Bridge at Langthorpe - immediately to the north of the Roecliffe site - located the extremely fragmentary remains of contemporary timber structures. The implications of these structures are profound and arise from two possible interpretations: either the settlement spread over the river from the presumed east-west bridging point identified to the west of the castra, in which case the settlement on the north bank was extraordinarily extensive; alternatively, there was a second crossing of the Ure at Roecliffe, this time north-south and more or less approximating to the position of the modern Arrows bridge on the A1. In the latter case, the Langthorpe buildings could then be interpreted as a small bridge-head settlement, rather than part of a vast, sprawling north-bank community.
Roecliffe and Corbridge
What makes Roecliffe so interesting is its proximity to Aldborough (Map 1), the Roman town of Isurium Brigantum, only 2km away to the east and also on the south bank of the Ure.  Isurium has long been suspected of having a military base beneath the town, which was founded under Hadrian.  Some fragmentary structural remains were identified during the 1930s beneath the town walls  but it is the finds of late 1st or early 2nd century military equipment that speak most eloquently of its origins as a castra. 
Traditionally viewed as starting at York, the road passes through Aldborough and up the line of the A1 as far as Piercebridge, when its line is adopted by the A68 as it passes through Corbridge and then Newstead as it penetrates central Lowland Scotland (Map 2).
Fortunately, Corbridge provides us with a possible dating structure for the road, as the Beaufront Red House site slightly to the west clearly did not respect the line of, and was therefore presumably earlier than, Dere Street and its crossing point of the Tyne. Red House was demonstrated to be Agricolan by excavation  - coincidentally also undertaken in advance of modern road improvements in 1974 - and was succeeded by the main site at Corchester, just to the west of the Saxon town (and bridging point) of Corbridge.  It has been shown quite convincingly that Corbridge main site must post-date AD85  and it seems reasonable to link its foundation with the withdrawal from Scotland after the replacement of Agricola,  perhaps some time in the latter half of the 80s. The reason for the location of the castra at Corbridge main site was presumably the important junction of the Stanegate and Dere Street and its associated bridge, and there thus seems to be a direct connection between the foundation of Corbridge and the construction of the Dere Street. The inference that we can draw here is surely that Corbridge was built at the same time as the Dere Street.
This leaves us with Red House, served by no surviving road and pre-dating the Dere Street. Selkirk has demonstrated a possible link between a suggested early alignment of the Devil's Causeway, the site of the Red House castra, and a deviation in the course of Dere Street south of the Tyne. It begins to look as if pairs of sites, such as Roecliffe and Aldborough, or Corbridge and Red House, are indicative of changing objectives in strategic route alignments in Britain. There must originally have been a crossing point of the Tyne at Red House, although all trace will probably long since have vanished. The castra - which pottery evidence suggests could have been occupied until the construction of Corbridge main site - perches just above the northern flood plain of the Tyne and, to judge from the alignment of the known Roman bridge at Corbridge,  almost certainly at the very edge of the river course in Roman times, the bath-house being situated on the river bank below. Palaeochannels indicative of course changes are clearly visible on the northern flood plain, immediately to the south of Red House and in fact the Red House Burn follows the likely old course of the Tyne between the castra bath-house and the position of the later bridge. No trace of a crossing has been found at Red House or, indeed, looked for, but these later movements of the Tyne may conceivably have destroyed any such evidence.
Dere Street approaches Corbridge from the south-east, passing along the Tyne valley through Riding Mill on a west-north-westerly heading, before changing course to the north-west just to the north of Prospect Hill crossing Dilston Haughs and aligned on Bishop Rigg, half way between the main site and Red House (Map 3). Selkirk's postulated road network, although providing a link between the Devil's Causeway and Red House, makes no provision for the southerly approach to the site.
A 'proto-Dere Street'?
Thus Roecliffe and Red House would seem to be indicating that the general route, if not the specific course, of Dere Street was in use even in the first ten to fifteen years of the Flavian period. Hence my suggestion of a proto-Dere Street:  a route that was, perhaps, followed for most of its course when the new road was constructed in the latter part of the 80s. Naturally, where its course was followed exactly by the Dere Street, it would now be next to impossible to detect; but where their chosen courses differ, the older route may still be evident. Deviations certainly occurred and Roecliffe and Red House seem to be symptomatic of this.
South of Roecliffe, it may be questioned whether the proto-Dere Street would anticipate its successor and also make for York, or if in fact it followed the course of the Rudgate past Newton Kyme, near Tadcaster, thence on to Castleford.  Castleford itself has produced evidence of very early Flavian military activity and its earliest phases appear to be contemporary with Roecliffe. However, the Rudgate is thought to be a later addition to the road network, and a more attractive solution might be suggested by the course deviation just to the north of Aberford, where the road from Castleford suddenly veers onto a north-easterly heading to bring it to Tadcaster. Had the proto-Dere Street continued northwards from Aberford, then this would have brought it through Bramham and Wetherby to Roecliffe, coincidentally more or less on the line of the later Great North Road. That this route is no longer evident need hardly surprise us, if it was indeed superseded in the Flavian period.
South of Castleford, the proto-Dere Street may well have been followed by the later road (Iter V of the Antonine Itinerary ) which swings round to join the main road north out of Lincoln and - if genuine - might be taken as indicative of a two-pronged Flavian advance into Brigantia, the eastern advance crossing the Humber, the western skirting it. York is obviously an irrelevance to our postulated proto-Dere Street and its replacement with Dere Street proper would presumably be a sign of the importance of that site and the need to tie it in with the road network.
This does not of course tell us who first dictated the course followed by the proto-Dere Street, and to consider this we need to examine both the dynamics of Roman road development and the likely precursors for any such Roman routes.
The ergonomics of invasion demand ease of movement and communication. The Roman army was justly famous for its military roads, but those with which everyone is so familiar are only one of several phases of existence of a route. All-weather surfaced roads were designed to permit the much-needed ease of movement and communication for an army in occupation, but took a great deal of time to construct and can only have come as one of the many products of invasion. In the first instance, an army needed to be able to move rapidly and not be hindered by waiting for its engineers to provide a road. Thus movement could be accomplished in one of two ways: across country or along already-established routes, what I shall term a Stage One route. Whilst the former allows a much greater element of surprise to the commander, the latter would make it easier to move rapidly along an already established line which, most importantly could later be used by supply convoys and reinforcements. The off-road capability of the Roman army was comparatively poor, for whilst the fighting troops could pass unhindered across open country (and Roman cavalry, being unshod, were indeed largely restricted to off-road use), this was not true of the baggage train. Moreover, broken ground, dense woodland, or other such hazards complicated movement away from established tracks. Given the nature of Celtic British society, with the use of wheeled transport established and apparently widespread, existing 'native trackways' would offer an obvious route for a Roman supply train.
In the case of the initial invasion of Britain, it would not seem unreasonable to expect most of the advance to have been accomplished on Stage One routes. John Peddie has pointed out the need for tactical roads to be built promptly in the wake of the army's progress. These unsurfaced highways required the rapid clearance of a broad corridor to facilitate the movement of supplies up to the vanguard of the invasion army. Such roads, my Stage Two routes, almost certainly progressed more slowly than the army, taking some 69 days to construct from Richborough to the Thames by Peddie's estimate, a distance of 80 Roman miles. As such, these were campaign roads and could only really have been of use during the summer campaigning season. In time these could be replaced by Stage Three routes, the familiar 'Roman roads' with their metalled all-weather surfaces. These roads, maintained and repaired throughout the Roman period, would then become Stage Four routes after the end of Roman rule, roads which are not maintained and therefore prone to degredation in its various forms, most notably course shifting, whereby the avoidance of a poor stretch of road or obstacle leads to the course actually moving, although still running parallel, a phenomenon still seen today on the Trans-Saharan Highway. However the processes of decay could equally have applied originally to Stage One and Stage Two routes. Underlying this argument is the tacit assumption, which seems seldom to have been voiced, that all major Roman invasion routes in Britain follow existing native trackways. This has to be true if we are not to believe that parties of - presumably rather nervous - road-builders were sent out in advance of battle groups to prepare the way. Testing for Stage One and Two routes is surely largely dependent on detecting Stage Three deviations from earlier alignments and these minor changes are crucial for the purposes of this paper. A good example of such a deviation in the course of Watling Street to the south and north of London that has prompted discussions on the possibility of an original Thames crossing at Westminster.  Such deviations, I would argue, can also be detected, or at least inferred, at Roecliffe and Corbridge.