The discovery of a road of Roman character in Broomhill wood aligned with an ancient bridge approach mound at Drygrange Bridge of the River Tweed, allows a Roman identity to be assigned to both features. Trace road features to the north and the south of the bridge site and Broomhill wood, of Roman road character and alignment but previously considered too fragmentary to publish, may now also be assigned Roman identity. These features enable the route of Roman Dere Street to be traced for some 10km between St.Boswells Burn and Clackmae Burn by way of the crossing of the River Tweed near Trimontium (Newstead) Roman fort.

Northumbrian Dere Street and its Roman precursor followed different lines in several lengths. The term Roman Dere Street (abbreviated as RDS) is used to distinguish the earlier road. The route of RDS is here presented first southwards from the River Tweed to the St.Boswells Burn and then northwards from the Tweed to the Clackmae Burn, those being the general directions of survey and the logic of discovery. Survey dates are given to facilitate access to field and personal journal notes. Some entries report surveys as early as 1983, when VVilliam Lonie was yet unfamiliar with the palimpsest of road systems in Greater Tweeddale.

Summary: Hill-slopes and the Course of Roman Dere Street

The map figures present the topography of the study area in hachure form. In previous Roman road work the hachure map has been found to highlight the concern of the Roman army for economy of effort in the use of its roads by the intended traffic, in this case infantry, cavalry, despatch riders and pack-horses. Hachure arrays indicate slopes. Closely and widely spaced hachures show steep and moderate slopes respectively.

It is not possible to quantify slopes, e.g. 20%, 50% etc, with hachures (as can be derived from contours), since the hachure marks obviously splay on curved hill faces. In very general terms no hachures are shown, even on undulating land, where slopes are less than 12% (c. 1 in 8). This is near the limit for the one-ton Scotch Cart drawn by a single horse, and for ploughing with a two-horse team. The 'going', firm or soft, and the make, good or poor, of the roads are obviously important factors.

From the author's experience of some hundreds of miles of Roman roads, mainly between but also beyond the Roman Walls, and some ten years of intensive riding, all paces, all terrains, he would suggest that the Roman roads north of Hadrian's Wall were designed for riding- and pack-horse use. Wheeled transport in the area was a rarity and, except for the remains of chariot wheels and fittings, has left little archaeological trace.

In the study area the hachures emphasise the confinement of northbound roads of all ages to the 'Eildon Gap' between the glacial tails of the Eildon Hills and the heads of the steep-sided glens of the burns flowing eastwards into the River Tweed. The Romans pioneered the Gap route south of the Tweed, but northwards took to the uplands on a splendid high line avoiding the marshy headwaters and deeper glens of the many burns on the western slopes of Lauderdale.

1: Roman Road and Tweed Bridge Approaches by Trimontium

NT57543462 to NT57573455: Survey WL, 15-6-99 (Lonie, 2003) confirmed the report by divers Mitchinson and Middlemass (see Acknowledgements). The mounded terrace, 10m wide and up to 2m high above the side-slope, rises south-eastwards on the right bank of the River Tweed immediately downstream of the south end of the 1780 Drygrange Bridge. The mound has a stony core with kerbs under the smaller stone metalling of a cambered road surface 4m wide. After 30m the mounded terrace bends to run south for 50m before disappearing under the approach embankment of the 1780 bridge. Near the river the mounded terrace is truncated by a track leading eastwards to the ferry (Roy, 1774/1793) that preceded the 1780 bridge, thus establishing a crude chronology.

The mounded terrace is interpreted as the southern approach to the putative Roman bridge of the River Tweed near Trimontium (Newstead) Roman fort. This approach work locates the Roman bridge directly under the 1780 bridge. Further traces of the Roman bridge on the river-bed are proving difficult to find. Any Roman bridge approach structures on the north bank have been obliterated by the 1780 and 1973 bridge approach works. This site is not incompatible with Roman bridge remains noted by Milne (1743): "...the Entrance to it, on the South side, is very evident......" Unfortunately it is highly likely that in the 1780 rebuild any "fine Stones (remaining to be) dug out of the Arches... " would be reworked and built into the new bridge. It is not impossible that Milne located another Roman bridge of Tweed, as may have Curle (1911), for the fort of Trimontium faced alternately west, then east, then west again in its several occupations (Clarke, 1996).

NT57563450 to NTS7483410: Survey DG, WL, 26-4-02 (Gordon and Lonie, 2003) found a linear mound some 4m wide and 0.3m high running for some 200m centrally in the northern part of the woodland strip north of Broomhill farm. The mound is absent in the southern part of the wood, having been slighted or otherwise lost there. The mound aligns to the north with the Roman bridge approach mound described above.

NT57563450 to NT57543437: For some 140m at the north end of the woodland strip the land surface has been cut and banked to form a slanting terrace, levelled 20m wide, with back and front scarps up to 3m deep, which greatly reduces the gradient down towards the river.

NT 57543442: Trial excavation (WL, DG, 24-7-02) found the topsoil over the mound to a depth of 15cm to be sandy silt containing many small stones of up to 5cm grade. Below this stony topsoil was a lens-shaped layer of larger stones some 5m wide and up to 0.4m deep. The stones of this layer were of 10 to 15cm grade and of irregular shape and deposition. To the edge of this stony layer, stones of up to 20cm in greatest dimension formed a rough kerb set on the subsoil. Over and beyond the kerbstones the small stone content of the topsoil was much reduced. The excavation site was towards the southern end of the terrace length, and the yellow clay sub-soil surface, marked by iron-pan nodules under the stony layer, followed the original slope of the land.

The carefully graded and levelled road-way terrace, the stony road-bed kerbed with larger stones, the small stone metalling of the cambered road surface, and the appropriate dimensions of these structural features combine to identify this structure as a Roman road. Apart from the woodland cover, the pristine condition of the road is remarkable. In its alignment with the probable Roman bridge approach mound at the River Tweed the road may be identified as a length of Roman Dere Street.

With the co-operation of the landowner, of the Trimontium Trust, and of Dr. J Dent, Environmental Officer with the Scottish Borders Council, more extensive excavation of the woodland road feature is intended when undergrowth conditions are more favourable.

2: Roman Dere Street south from the River Tweed

NT57563450 to NT57473408: The Roman road mound in Broomhill wood aligns to the south with the linear mound, crested with beech trees, which borders the lane running north to south by Broomhill farm. This linear mound, some 170m long, may now be accepted as overlying the agger, and marking the southwards course, of RDS.

NT57483394: The Roman road by Broomhill farm, identified as Roman Dere Street, is lost at its southern end under no less than two abandoned railway embankments, the A6091 Melrose bypass, and the Broomhill farm access road, but may be assumed to continue on line to the Red Rig Lane end at the NGR (National Grid reference) given, as the 4cm OS (Ordnance Survey) map of 1976, pre-bypass, shows.

Extensive survey on the projection of this line southwards over Peasehill Farm (demolished) land failed to discover even secondary evidence of a Roman road (traces compatible with a Roman road) though a marked footpath, hedgerows and a farm lane align to converge with the A68.

NT57343373 to NT57223358: Photographed when delineated by a light snow cover, a robust linear mound and upslope track hollow, 10m wide overall, traverses for 200m SSW on the south face of the Red Rig (Lonie, 1997). This linear feature projects back NNE to the south end of the length of RDS by Broomhill at the foot of the Red Rig lane.

More recently a reproduction of an early OS map (Trimontium Newsletter, No. 17, Figure p. 13) shows a field boundary and line of trees exactly on the line described. No track is shown. Later OS maps do not show this boundary.

NT57163351: The linear mound and track feature noted above projects SSW to meet the two well-established Roman roads (Jones et al., 1994) issuing from the 1st and 2nd century south gates of Trimontium (Newstead) Roman fort exactly at this point, where they converge before crossing the Bogle Burn (Lonie, 1998). This deliberate junction with established Roman roads suggests that the mound element of the Red Rig traverse feature is the agger of a Roman road continuing RDS in the same form of track and mound as established by Broomhill farm.

NT57183340: The combined roads from the fort are directed to a crossing of the Bogle Burn at this point (Lonie, 1998). In the banks of the burn are good partial sections of a Roman road agger of stiff clay and stones, its crest under some 1m of silt. There are no traces of culvert stonework at the burn in its present course. Culvert remains are probably buried in the flat ground nearby, close to the road junction point noted above, some 80m north of the burn.

NT57073337: A dark linear mark on air photograph Y36 of Cambridge University (taken in 1947), and track evidence on the ground, show the track element of the Red Rig traverse feature to continue SSW to a ford of the Bogle Burn at this NGR (Lonie, 1999), some 160m upstream of the Roman crossing point. The ford is now marked by cattle watering ramps on both banks and a line of heavy stones set across the burn bed. Southwards from this ford, a field wall uniquely aslant to the slope and an offset of field corners mark the line of a former track rising to Eildon Mains farm. This sequence of features led to the Red Rig traverse being initially identified as a track between farms, the Roman agger element unrecognised. South beyond Eildon Mains farm, the old A6091 (recently closed to vehicular traffic) leads down to the medieval ford of the Sprouston Burn.

NTS7183340 to NTS7253308: On the line of the Roman road over the flat, then rising, ground south of the Bogle Burn towards Eildon village there is no surface trace of any road. Air photograph JU43 of Cambridge University (taken on 13th July 1952), however, shows a light linear crop-mark running on-line to the entry to Eildon village by an old right-of-way lane. At this point the suggested Roman road line crosses a well-made stone block and slab culvert draining from SW to NE. The earth and stone over-burden of the culvert, some 0.5m deep, carries a field track in recent use for farm stock and vehicle movement E-W along the ridge.

NT57263290: As the lane runs on-line southwards for 260m into Eildon village it crosses another stone block and slab culvert here. By local report, this culvert has been rebuilt recently to maintain the right-of-way. A remote Roman ancestry for both culverts is probable.

NT57293284 to NTS7483216: The northern abutment of the putative Roman bridge of the Sprouston Burn is sited through and beyond Eildon village southwards exactly on-line at a distance downhill of just less than one kilometre. On this descent through Eildon village, Eildonbank farmstead and pastureland, no road feature of a Roman nature could be found (survey WL, 4-10-02).

NT57353255: At the drainage ditch here a girder and concrete slab farm bridge may be a fossil of the Roman road, or the anticipated remains of the Roman culvert may be hidden nearby in heavy over-growth.

A recent housing development at Eildonbank, immediately to the south of the main village, is off-line to the west. Intermittent watch found no traces of relevant interest. The 'Watling Street' of Roy (1774/1793) follows the medieval and recent road to the west of Eildon village. The village was misidentified by Roy as Trimontium fort: one wonders why.

NT57523197: RCAMS (1956: 469) notes "But on the N. bank of the Sprouston Burn some 30 yds. W. of the modern road, which is carried on a high embankment, there are remains of a road-mound which, though much dilapidated, shows massive stone-bottoming with cambered metalling above it where the bank has broken away. It is difficult to interpret these remains except as the Roman road [RDS] leading off the abutment of a bridge long washed away". Inspection on several occasions found nothing to contradict these observations.

NT57543197: Parallel to the Roman abutment on its east side is a stony ford with well-graded approaches aligned to the north with the tarmac road A6091 descending from Eildon Mains Farm and, to the south, aligned with a strip of common land, now allotments, and a covered passage through houses leading towards Newtown St. Boswells.

This ford may be identified as an element of Northumbrian and later medieval Dere Street leading northwards to fords of the River Tweed near Melrose and the early 12th century Cistercian monastery there. The allotment strip and the passageway are remarkable fossils of the early road systems. Mystery surrounds the fate of road systems and native settlements in the aftermath of the Roman departure, by stages, from the early 3rd century.

NT57543197 to NT57923152: In the 600m through Newtown St. Boswells between the Sprouston and Bowden Burns , town development culminating in the railway, with station and stockyards, of the 19th century, and the housing schemes and Regional Council HQ buildings of the 20th century, have obliterated all trace of RDS. The banks of the Bowden Burn immediately south of Newtown St. Boswells are revetted with stone walls and developed with industrial and domestic buildings. No obvious trace of a Roman crossing could be found. Attention was turned to road approaches to the Burn from the south.

The Bowden Burn Glen is some 15m (50ft) deep and very steep-sided for two kilometres from its mouth at the River Tweed. Newtown St. Boswells occupies the only break, some 300m wide, in the slope of the north side of the Glen. This break of slope must have attracted the Roman road surveyors. The south side of the Glen is an unbroken 1:2 (50%) slope which requires more recent roads to make long slanted approaches to the Burn on that side.

NT57923152: RCAMS (1956: 469) suggests Roman Dere Street from the south to have "descended much on the line of the modern road from Maxpoffle to a crossing of the Bowden Burn close to the existing low-level bridge". The road from Maxpoffle slants in from the southwest, well off-line from RDS. Of the bridges of the Bowden Burn at Newtown St. Boswells there are no fewer than three which may be described as ‘low-level'.

NT57893152: The old (pre-bypass) A68 and underlying medieval roads from the south enter Newtown St. Boswells on a long slant down from the east to a stone bridge, strengthened with concrete, at this NGR.

South of the Bowden Burn Glen the old A68 is sinuous and runs some 500m to the east of any probable line of RDS over Whitehill and Merwick farm-lands. Southwards from its junction with the bypass, the A68 skirts west of old Lessudden and of St. Boswells village proper, and rejoins the well-recognised line of RDS at St. Boswells' Burn some 2km southeast of the Bowden Burn.

The road from Maxpoffle into Newtown St. Boswells from the southwest passes under the disused railway bridge on an even longer low gradient of quite un-Roman character to a second stone bridge 120m upstream of the 'old A68' bridge. A third stone bridge, on a short spur road north from the Maxpoffle road, crosses the Bowden Burn at NT57843143, immediately upstream of the railway bridge. This spur road and bridge terminates in a group of workshops and store sheds, and must be suspected to be 19th century railway compensation work.

The railway followed Gap route precedent but at a height over the Burn of 15m. The recent A68 bypass breaks the mould by flying the Bowden Glen on a modern bridge some 300m downstream.

NT58023101: Whitehill house and farm have an access road from the east from the old A68, and another from the northeast from the Maxpoffle road. The latter access road is linear and suggestively near the line of RDS established further to the southeast. Where this access road crosses a bridge of the abandoned railway line at NT57823118 it acquires a considerable offset of its two component lengths but fails to reveal any trace of previous roadworks. This farm access road again appears to be a railway improvement. No feasible direct crossing of the Bowden Burn Glen on the access road projection northeast could be found. As noted, the Maxpoffle road slants gradually down to the Burn in thoroughly un-Roman fashion: the Romans liked to get their gradients over with as soon as possible.

Across Whitehill plough-land in a wide swath southeast of the Bowden Glen absolutely no surface trace of a Roman road either in contour or stone content could be found. The field is very flat. It is not unusual for a Roman road structure to be obliterated by cultivation, as RDS itself is over much of its length and the Devil's Causeway Roman road in Northumberland is almost entirely.

Immediately to the west of Whitehill farm steading is a reservoir some 50m by 30m in extent, its purpose not immediately obvious. Northwards of the farm buildings no trace of a Roman road could be found. A 5m wide linear soil and vegetation disturbance running NE from the farm to the field edge at NT57843112 raised hopes but led to an engine watering point of the steam age, usefully placed just before the south-bound junction of the Waverley line to Carlisle via Hawick and the line to Berwick upon Tweed via Jedburgh (1 inch OS map, 1897/1901). Steam locomotives expired in the 1960s, but provide explanation for the elevated pond and the field disturbance. Search continued after a pause for thought.

The line of RDS was secure to the Roman bridge of the Sprouston Burn from the north and, as will emerge, was secure to Whitehill farm from the south. These points are only 1100m apart, and the line between them could not stray far from RDS. With this in mind, the direct line from Whitehill farm towards the Roman bridge-works at the Sprouston Burn was followed across the featureless plough land already searched to the edge of the Glen and into the woodland and thicket of the glen side.

NT57853144: Surveys 23-2-93, 7-11-02, WL: a steep terrace angles NNE down the Glen side for about 50m about this NGR at a gradient of 1:5 (20%). The terrace is generally 7m wide, grass overgrown, much slumped and disturbed in places along its length. No stonework or other structural features are evident. At its lower end the terrace emerges from the wood on a line directed towards the old A68 bridge of the Bowden Burn, but some 50m short. The terrace faded at this point and could not be followed further. At its upper end the terrace turns southwards over the brow of the bank but is effaced before any trace reaches the present field fence.

The terrace, slumped to 7m wide, would have offered a 4m wide road surface in antiquity. The 20% gradient precludes regular use by wagons, certainly by the one-horse carts of the 19th century. Riding- or pack-horses and pedestrians would have had little difficulty with the short sharp gradient. The degree of slump and absence of superimposed cart tracks confirms the terrace to have had little or no use in the high medieval period or later. Taking the robustness, gradient, wear and slump into account, the terrace could well date to the Roman period and, from its situation directly between securely identified features and the direction of its termini, be an element of Roman Dere Street.

NT58073091: The approach to Whitehill farmyard from the SE is by a tarmac farm road starting at the culvert noted below. A low linear mound emerges from beneath the farm road on its west side for some 50m about this NGR before running under a midden wall at NT58043096. The fully developed mound is 4.5m wide, 0.5m high. The midden is sited on this road-like mound with its SW corner overhanging the western edge of the mound. As a result, the midden wall has subsided and cracked. The mound is suggested to be the agger of RDS. On line with the mound for some 80m north of the midden, farm buildings obliterate any ancient road structure.

NT58103087: The tarmac farm road overlying the probable Roman agger leads straight in 100m south-southeast to an arched stone culvert with embanked road approaches at this NGR. Issuing from the culvert is a dry ditch leading off eastwards into obviously wet pastureland. The 4cm OS shows ditch lines to the east and the west in line with the culvert and the wetland. Close to the northeast of the culvert entry on the road bank is a small brick-built structure of uncertain purpose and a loose stone slab incised '1874'. The six-inch OS map of 1897 marks 'Hydraulic Ram’ at this site.

The culvert build is obviously recent and probably contemporary with the incised slab. A culvert would have been required in Roman times. The hydraulic ram, the reservoir, and the watering station patently supplied the railway which ran only some 200m to the west of Whitehill farm. Mrs A. H. Gordon of Whitehill was unaware of the railway involvement, but offered that the farm buildings included a small watermill, not noted by the writer. As elsewhere, the arrival of the railway occasioned work on roads and drainage for landowners.

NT58133087 to NT58433060: Between these points a hedgerow runs 400m southeast from the Whitehill culvert to the vicinity of Greenside cottage. Mr. H. Gordon, late of Whitehill, considered RDS to lie close to the northeast of this hedgerow, from the excessive stone there which hampers ploughing. In a ribbon some 5m wide along the northeast side of the hedge-bank there are clearly more and heavier stones than elsewhere in the field. The profile of the hedgerow bank and adjacent field-edges shows a slight depression 14m wide, with the hedge-bank 4m wide, 0.3m high off-centre to the southwest side. The 4cm OS map of 1906/1953 shows an unfenced track. The hedge-bank remains are entirely compatible with a Roman agger and roadbed used along one side by later farm traffic, much as we have encountered elsewhere.

NT58453063: At the southern end of the stony road-bed from Whitehill farm the continued line of RDS is marked by a broad slumped terrace slanting eastwards for 50m down an earlier left bank of the West Burn, to be lost under the A699 tarmac road.

NT58543056: At this point immediately south of the A699, the West Burn bends abruptly to run ESE for 40m then resumes its general course ENE. On the left bank of the burn here, parallel to the diverted burn, a linear mound 9m wide and 1m high runs 40m to a nose overhanging the loop in the burn. The burn here has a stony bottom and a stony ramp on its SE side, noted as a ford. On line beyond the ford, a hedgerow on a robust bank runs off to the southeast.

By comparison with similar stream diversions elsewhere on Roman roads, the linear mound undoubtedly marks the course of RDS on an embankment over a culvert. On the abandonment of the road the culvert has blocked and the stream, dammed back, has washed out the weakest point on the embankment. The Roman culvert will now be blocked and buried under the remains of the embankment. Two work-sheds and scrubby woodland conceal the Roman embankment. The 4cm OS map notes this point as Clatterdean.

RCAMS 1956: 469) suggests RDS to have continued "on its previous bearing" (that of the A68 by Forest Lodge) "which would have carried it across the West Burn just W. of the head of its deep and precipitous ravine". This point is some 250m east of the 'Clatterdean' crossing. The discrepancy is discussed below.

NT58543056 to NT58883029: The hedgerow bottom from the West Burn to Merwick farm is 9m to 10m wide and 0.5m high, and is in exact alignment with RDS on Whitehill land to the northwest. The hedgerow without doubt represents the agger and the line of RDS to the southeast.

NT59362990: Survey WL, 9-3-83- The 940m length of RDS from the Whitehill culvert to Merrick farm projects south-eastwards to this point on the St. Boswells Burn. Again, the line of RDS from the southeast by Forest Lodge projects to exactly the same point on the Burn. A railway junction, road junctions and industrial works conceal any physical remains of the crossing of the Burn by RDS, but there can be no doubt as to its location.

RDS changes in direction: the course of RDS north-westward on Merrick and Whitehill land is 309 degrees Grid, some 3.5 degrees west of the course of 312.5 degrees Grid closely maintained by RDS from the crossing of the River Teviot six kilometres to the southeast. An even larger change in northwards course, of 25 degrees eastwards, had been noted at the Whitehill culvert. Close inspection found no trace of any Roman road works in the wetland to the east of the culvert: RDS did not cut across this 'corner'.

It would seem that the wetland area east of the culvert was, in Roman times, sufficient of a morass to occasion a realignment of a major Roman road. Moreover, measures to avoid the obstacle were taken at St. Boswells Burn, some 1200m to the southeast. As frequently noted, Roman surveyors and engineers were not obsessed with straight roads, but used them where practical. They must have worked many miles ahead of the road-making in selecting the route.

3. Roman Dere Street north of the Tweed from Drygrange Bridge

Since none of the fabric of the Roman bridge of Tweed has as yet been located, the exact positions of the bridge abutments are not known, nor is the angle, if any, at which the Roman bridge lay in relation to the 1780 bridge known. The Roman bridge approach terrace and mound run in to the south end of the 1780 bridge from the southeast so that the north end of the Roman bridge may have lain to the west of the present bridge end. There is no structural evidence of Roman work on the riverbank there.

NT57503476: Some 80m west of the north end of the 1780 Drygrange Bridge a dip in the B6360 road from Gattonside lies some 2m below the present level of the bridge entry. Some time in the 19th century the bridge approach was built up to level the road deck. Prior to 1780 the Gattonside road led eastwards to a road terrace, now overgrown, descending to a ferry point some 160m downstream of the 1780 bridge. As noted earlier, this ferry point and approach track downstream of the bridge is mirrored on the south side of the river. The north abutment of the 1780 bridge and the massive concrete abutment and piers of the 1973 bridge intervene in the 130m gap between the road dip and the terrace to the ferry so that the pre-1780 level at the bridge north end is difficult to assess.

Roy's map of 1793 (Plate XXI: Environs of the Eildon Hills; probably engraved in 1774) properly omits the 1780 bridge and shows the tracks to the Leaderfoot ferry point on both sides of the River Tweed. Drygrange Mill at Leaderfoot is also shown (but not on the Pont-Blaeu map of 1654) (Stone, 1991). A dam and lade on the Leader supplied power to this disused corn mill. The site of the mill suggests that corn from south of the Tweed might have been unloaded directly from the ferry-boat. The 'new' bridge could have weakened the commercial position of the mill.

NT57643484: Although any remains of a rising terrace have been entirely obliterated by later road- and bridge-works, the steep slope of the high bank of the river at the north end of the postulated Roman bridge dictates a terrace rising initially northeast to a point near this NGR. Maps of the horse-drawn era show a 'smithy' there. From this point Roman Dere Street appears, from the traces noted below, to have climbed northwest for 700m to the shoulder of Easter Hill at the moderate average gradient of 1:9 (11%).

NT57383512: Survey WL, 30-1-93 on Leaderfoot Farm land: for 50m about the NGR traces of a terrace 11 m wide ascend the open field at a gradient of 1: 8 (12%). The terrace is directed downhill towards the confluence of the Leader Water and the River Tweed, a line confirming that the initial riser from the Roman bridge angled northeast to break the slope of the high bank of the river there.

NT57323521: The double hedge lying E-W across the slope here shows a 10m gap exactly on the line of the terrace. The plough slip below the gap shows an excess of stones uniquely along the lower hedge bottom. Cattle trample may have brought stones to the surface.

NT57243531: A low natural ridge E-W across the slope here is flattened over some 20m on line with the terrace and the hedge gap in the field below. Shallow cuttings or clearways of this nature are not uncommon on Roman roads, their purpose to avoid sharp undulations for horsemen or to reduce gradients, here both.

Extended search found these several surface features to be unique on the wide south slope of the Easter Hill ridge.

NT57163541: On Easter Hill, The Parks wood strip: along the south edge of the wood here an excess of heavy stones shows for some 15m uniquely about the established road line. As field clearance of kerb and roadbed these stones, like the other surface traces described above, are not incompatible with the remains of a Roman road. Survey 25-10-02, WL noted a dilapidated ha-ha wall across the line of the Roman road some 3 m within the wood, slighting the road.

NT57133544: Surveys WL, 30-12-93, 25-10-02: a shallow depression 25m in diameter within a low bank 5.5m wide, with a 4m gap to the ESE, is probably a wall-stone quarry for the crumbled ha-ha builds edging the woodland strips in the vicinity, and slighting the Roman road mound northwards to the Howdler’s Burn.

NT57053554: For 240m about this NGR a heavy bank showing large quarry stones on its western side runs down slope northeast to the Howdler's Burn. The stony bank probably represents an old estate boundary on and slighting the pre-existing Roman road mound. Lonie (1996) reports a very similar re-use of an agger on the Craik Summit Roman road at Howcleuch.

NT56963565: Survey WL, 30-12-93, at Howdler's Burn (a howdler is a limping man): a field track in recent use crosses the Burn at this point by a recent pipe culvert, heavily embanked. There is no trace of an ancient burn crossing or approach terraces in the open plough-land for several hundred metres about the culvert. The woodland strip some 40m wide, noted above, descends to the culvert site from the SE and rises from it on the same line to the NW. The wood strips and field boundaries angle across the valley to give a rather complex system of plough slip and plough-gather in the field corners near the culvert.

The boundaries of both wood strips are tumbled stone and turf banks. The boundary bank on the SW side of both strips is considerably more robust than that on the NE. This maintains the impression noted above that this SW boundary overlies an older linear mound base. This evidence for RDS at and about the Howdler's Burn is secondary, i.e. features on line compatible with a Roman road.

NT56913572: Survey WL, 2-4-94: on the rise NW from Howdler's Burn, along the SW edge of the wood strip on Cooper's Rigs, a considerable excess of cobble sized stones strongly suggests an earlier made road or roadbed exposed and broken by ploughing.

NT56763588: Survey WL, 30-12-93, on Cooper's Rigs: at the head of the woodland strip rising NW on the side of Cooper's Rigs the woodland strip bends N then NE to run down the crest of the Rigs. From under the linear stony bank forming the SW boundary of the wood, suggested as on the line of RDS, emerges a faint linear depression 13.5m wide with a low central mound some 4m wide. The feature was well delineated by light snow cover and runs for 50m on line about the NGR given. The feature is then truncated by a shallow ditch with a fence beyond. Probing the ditch bottom failed to discover any stony structure. Continuing NW beyond the fence are traces of the terrace on a slight side slope. Deep tractor wheel ruts crossing and tending to follow the line of the terrace again failed to reveal any stony structure. In view of the certainty of the line and the characteristic surface traces, the apparent absence of stone structure presents a problem. Deep peat underlies the scrubby grassland.

NT56013608: Surveys WL, 30-12-93, 1-1-94 on the NE slope of Cooper's Rigs: an animal path following the much disturbed road traces NW from the crest of the Rigs, runs more steeply down a broad nose protruding from a low scarp along the slope, then crosses a recent drainage ditch with a stony bottom. The visible stony spread is only 1.5m wide, but uniquely across the line of the animal path. This stony spread was found, by probing, to be almost 6m wide, and to contain quite considerable stones. Such stones are absent for several tens of metres both up- and down-stream of the 'ford'. The feature is set in deep peaty soil and is not inconsistent with the exposure of a Roman roadbed deeply buried in mature peat.

NT56483627: Surveys WL, 30-12-93, 1-1-94 at Packman's Burn: on line with the drainage ditch stone exposure, a linear mound some 8m wide between shallow ditches runs down NW to a nose of earth and stones 10m wide, 2m high protruding some 15m into a small oxbow of the Burn. Both approach mound and nose are much overgrown with grass. A tractor track and an animal path recognise the approach mound, as does a system of drystone walls joining and crossing the burn at this point.

The burn bed immediately downstream, NE of the nose, is noticeably stony. Northwest over the burn, opposite the nose, the burn has carved out a low-level wet terrace some 40m by 10m revealing two groups of large stones, one to each side of the line of the nose. Water stained with iron wells from an area 2m across between the stone groups. These remains are consistent with, indeed diagnostic of, a robust Roman road embankment and culvert broken through by the burn.

NT56333638: For some 100m about this NGR a grass terrace, some 10m wide, rises NW on the western side of a broad gully. To the SE over the more level ground towards Packman's Burn the terrace back shows on line for 50m. To the NW the terrace line crosses the old road from Gattonside to Lauder at NT56233648 without being recognised, e.g. by a gate, thereby establishing a certain antiquity for the terrace and burn crossing.

NT56133656: NW beyond the old Lauder road for some 250m about this NGR the Roman roadway is marked by a broad level depression in the grassland. This shallow cutting eases the gradient to the crest of the ridge of the West Strips woodland. The average gradient on the 640m long slope is 1:9 (11%). This sophisticated grading is typical of Roman roadwork; I believe to facilitate the passage of despatch riders downhill at the canter or gallop, comfortable paces without stirrups.

NT55873678: Surveys WL, 30-12-93, 1-1-94, 2-4-94: the approaches to and the crossing of the head of the Meikle Linn Burn in Hut Wood have been much disturbed by forestry work. In the two earlier surveys, only a short length of possible Roman road terrace, on line, was found to have survived on the SW approach. On the left, NW bank a platform 15m long, 10m deep some 2m above the burn bottom, marked the crossing on line. There was no trace of a culvert, and no bridge abutment or land could be found on the right bank.

On the survey of 2-4-94 it was found that the burn had been cleared by machine digger since the visit of January, revealing an extended section across the road line in the north side of the ditch. For some 50m downstream, and a lesser distance upstream, of the road line the section showed 0.4m of peaty loam over 0.4m of brown clay down to a thin layer of stones of irregular size lying on a hard, red, sandy sub-soil. At about 20m intervals groups of larger stones in the clay layer indicated old field-drains. Uniquely about the Roman road line over a distance of 16m was a considerable excess of scattered stones, lying high in the clay layer under the peat. To the south of the burn the bank of upcast from the ditch on line showed a similar local excess of stones.

Although there are puzzling features in the detail of this section, there can be little doubt that it represents the Roman road.

NT55703696: In Hut Wood: at the 2-4-94 survey date the slope northwest of the Meikle Linn Burn crossing had been recently clear-felled and the surface peat broken by the passage of logging traffic. At the NGR, on the brow of the slope, a considerable scatter of stones in a strip 20m wide across the line of previous road features was noted. Such a feature had been anticipated and it was gratifying to find it. It must be noted that the 4cm OS maps mark tracks in Hut Wood uncomfortably close to the probable Roman road line. These forest tracks appeared on site to be clearways rather than made roads.

NT55643704: Survey 2-4-94: clear west of Hut Wood, a drainage ditch, lm wide by 0.5m deep, crosses pasture land on the projected road line. The 50m length of ditch examined showed mainly undifferentiated brown clay and small stones under a peaty loam. A group of pebbles at one point off-line suggested an old field-drain. On the projected road line the ditch sides show the outlines of two ditches, c. lm wide by 0.4m deep, filled with light grey clay silt, at 17m centres under the peat cover.

In the absence of a stony roadbed in the clay layer between the side ditches, this feature provides little evidence of a Roman road. In times past, simple double ditches were cut over moss land to give dry-ways that have been claimed, perhaps correctly, as Roman roads (Selkirk, 1995). A similar feature on the valley floor at Ramseycleuch may supply a link in the elusive Ettrick Water Roman road.

The Housebyres Moss has been much 'improved' as pasture and ploughland, as shown by an extensive pattern of old and recent enclosures and cultivation rigs. Old and originally substantial turf walls have decayed to form very convincing road mounds for a few hundred metres at a time before revealing their true identity.

NT55513716: Survey WL, BWL, 2-3-03, sought to confirm the negative finding of survey 1-1-94 (see later). However, search along the dry drainage gully running west to east to discharge into the Craigsford Bum, found the gully sides broken at the NGR by cuttings, much eroded, which reduce the gradients to the gully bottom. In previous search of the gully the remains of an embanked culvert had been anticipated; the cuttings were unexpected as a Roman solution to a fairly trivial obstacle.

NT55543717: Survey WL, 2-4-94: the 4cm OS map of 1976 marks a well here, only some 30m east of the cuttings feature noted above. The outflow from this spring, from a plastic pipe, drains eastwards through a stone block and slab culvert under a cart way and a dry-stone wall aligned north-south. The cart way leads north to sheep pens and is contemporary with the culvert and wall, and probably of late 19th century date.

NT558368 to NT549378: Survey notes WL, 1-1-94 record that from the Meikle Linn Burn for 1500m north-westwards over the Moss there is absolutely no trace of any ancient roadway despite search for some 100m on either side of the projected line. While the surveys of 2-4-94 and 2-3-03 reported above found traces attributable to an old road, only the alignment of these traces with the Meikle Linn Burn section exposure to the southeast and with the hill road to Lauder, arguably Roman, to the northwest, justifies their being offered as evidence for the line of Roman Dere Street.

Roman roads in deep peat land: it must be suggested that in relatively well drained deep peat land the Romans may have constructed both roadbed and agger of peat blocks, the road width of 4 or 5 metres metalled with the clay and small stones dug from the side ditches. The near-total destruction of this fragile structure by agricultural processes is probable. A rather more substantial simple lens of clay and small stones set mid-depth in c. 0.5m of peat marked the course of the Roman road by Loch Thorn, in Renfrewshire, between Lurg Moor and Outerwards Roman fortlets (Newall and Lonie, 1990).

NT54803787: North of Housebyres Moss: from under the dry-stone wall bounding the southwest side of the woodland strip here emerges the edge of a linear mound that diverges slowly from the wall over a 450m length, the exposed width gradually increasing to 3m. At NT54373817 the southwest side of this mound is cut into by a deep cart-way, one of two running parallel at this point. The disturbed mound and the two cart-ways continue northwest for some 50m, then run under a cross-wall some 10m long at a field corner offset. A hunt gate is set centrally in the cross-wall, between the two cart-ways.

NT54353822: From under the hunt gate and cross-wall and the continuation of the cross-wall as the northwest boundary of the wood, a road terrace 10m wide, grass overgrown, emerges to continue northwest over grassland for 250m along the foot of The Heathery hill. This terrace then slowly converges with the drystone wall on the northeast side of a cobbled road ('Malcolm's Rode'; see later), with last trace at NT54123846. This point is some 100m short of the junction at NT54043853 of the cobbled road with a metalled track from the south (Roy's 'Road to Lauder' of c. 1750). At the junction the metalled road clearly truncates the cobbled road, though this truncation feature probably dates to the railway era. The combined roads become the enclosed road to Lauder.

These road terrace traces extending for over 900m about the hunt gate are not incompatible with a Roman road, but depend for their certain identification as such on their convergence with the enclosed road to Lauder (and this being accepted as Roman Dere Street), and their alignment southeast-wards with the fragmentary but secure Roman road features described earlier.

The enclosed road to Lauder: This road is well-made; cobbled and dry metalled, ditched on both sides with regular if sometimes ill-placed culverts, with robust cuttings and embankments to lessen gradients, and enclosed between good dry-stone walls set from 12 to 15m apart. This visible road structure is almost certainly of the 19th century Railway Age. The road leads in 10km north to Lauder, Royal Burgh and railhead, and in its enclosure was no doubt intended to lead sheep to the railhead stockyards established in the 1880s. From the excellent make of the road, some traffic of wagons and carriages seems to have been anticipated, though the upland farms along the way are few in number and at its southern end the road descends into the River Tweed valley and the railway system there established in the 1840s. Nevertheless the route of the enclosed road seems to be of such antiquity as not to preclude a Roman origin.

That the enclosed road to Lauder overlies Roman Dere Street is not evidenced by physical traces save those by the hunt gate. RCAMS (1956) reports only slight traces of earlier roads at two burn crossings, and studiously avoids any suggestion of a Roman origin. Inspection found these traces to be no more than the late rebuilding of the road might require as temporary bypasses. The Commission suggests quite another line for RDS over and north of Tweed (see Notes later).

NT54003861: The enclosed road crosses the wet ground at the head of the Clackmae Burn on a fine culvert with stone block sides and slab roof and floor, 40cm wide x 30cm deep. The structure resembles other culverts locally on roads associated with railway construction.

NT53664062: Survey WL, 6-1-03: from the Clackmae Burn the enclosed road rises, as shown on the OS maps, almost due north (352 degree Grid) dead straight for two kilometres to the east shoulder of Kedslie Hill. There, at the NGR given, the road bends slightly westwards (337 degree Grid) to descend, again dead straight, for one kilometre to NT53254155.

From the Kedslie Hill shoulder it is strikingly obvious that the 2km straight southwards is directed to the saddle between the Eildon Hills Mid and North, off centre to the west foot of Hill North. The summit of Hill North (404m) is the site of the famous Roman watchtower.

The aiming point of Roman Dere Street in its 18km length from the Cheviot Hills to the Bowden Burn, only 3km short of Trimontium fort, is again the Eildon Hills; the east slope of Hill North. In both cases the aim is noticeably off to the right of the watchtower. Intent or error, though of what nature escapes explanation, would nicely link the two road lengths.

The direction of the 1km straight northwards from Kedslie Hill is equally interesting, for it aims at Dun Law, 18.5km distant, exactly where Roman Dere Street crosses the ridge of the hills by Kings Inch and Soutra Isle into the Lothians.

While these straights and alignments of the enclosed road about Kedslie Hill cannot prove the Roman origin of the road, they strengthen the case.

Cursory search for several hundred metres about the summit of Kedslie Hill (293m), NT53254070, in conditions of light snow cover and low sun, failed to find any trace of a Roman watchtower. Air photographs, examined on 31-1-03 at the RCAHMS Air-Photographs Library, were also negative for possible Roman watchtower bases. For quite other reasons the summits of Dun Law (393m) and Turf Law (382m) had been searched for possible Roman watchtower bases, with negative results. The Roman fortlets at Oxton, 2.5km ESE from Turf Law, and near Chapel Mains, 2.0km ENE from Kedslie Hill summit, at the more comfortable altitudes of 245 and 210m respectively, might well have supported watchtowers at higher altitudes. From the altitude and contour data of the 2cm OS map of the area, corrected for Earth curvature, the fortlets themselves do not appear to be intervisible, even with 10m towers. The calculated shortfall is 20m at Chapel Mains fortlet.

At NT533416 the enclosed road is joined by the 'Bluecairn Road (Track)' of the OS map which rises from the Allan Water valley and the 16th century towers there. From Bluecairn farm northwards the enclosed road is surfaced with tarmac and leads by articulated straights trending north-eastwards towards St. Leonards farm and Roman camp. The camp is one of the very large camps, 165acres, of the Severan campaign of 209AD.

'Malcolm’s Road' and the 'Road to Lauder'. Mention was made earlier of the cobbled road and the metalled road which combined to form the enclosed road to Lauder held to be Roman Dere Street. Roy, in 1747-55, mapped both roads. The more westerly road, that by the Blake Burn valley, is labelled 'Road to Lauder' and so seems established as the more important, even before the railway era. The more easterly road, that by Housebyres Moss, may have been made as early as the 12th century when the Cistercian Abbey of Melrose was founded. This road to the north acquired the cognomen 'Malcolm's Rode'. If the Malcolm concerned was Malcolm 11 (King of Scots, 1005-1034), he of the Battle of Carham, 1018AD, that sealed the fate of the kingdom of Northumbria, the road may well date to the 11th century or earlier, and may mark the line of Northumbrian Dere Street. Northumbrian Dere Street had a 7th to 10th century existence on a Roman base for most of the route from York to Edinburgh. The loss of the Roman bridge of Tweed at Trimontium, in the 3rd or 4th century at latest, would require a ford and new approaches to the River Tweed to be established in Northumbrian times. Malcolm's Rode offers approaches with lower gradients than the Leaderfoot Roman crossing at Trimontium. The route diversion thus established is unlikely to have been abandoned on the annexation of Bernicia and the Lothians by the Scots in the 11th century or in the resulting tussle with Danish and Norman England. The name ‘Malcolm’s Rode’ is slight enough evidence for all this. Whatever the historical facts, 'Malcolm's Rode' and the straights over Kedslie Hill are of great antiquity. The latter, from their style, cannot then be attributed to any era other than the Roman, and with this conclusion the whole road-line, from the Bowden Burn to the Clackmae Burn must be accepted as Roman Dere Street.

Notes

1. Roman Dere Street about the River Tweed: RCAMS (1956. 470, No. 10). The account follows Curle (1911) in taking Roman Dere Street across the River Tweed from Newstead village to Kittyfield lands where "the hanging valley behind Kittyfield, immediately opposite the Eddy Pool, appears to offer an attractive line of ascent towards the hilly ground on the N". Thereafter the Commission suggests Dere Street to proceed for 5.5km (3.4 miles) to "near Kedslie" by an unspecified route over the ridges between and across the glens of all of the Packman's, Meikle Linn, Craigsford, Heathery, Clackmae and Kedslie Burns. Surveys about this line have been discouragingly negative and strenuous. Kedslie is the site of a Roman camp.

   From Kedslie the route picks up the articulated straights, attractively Roman in style, of the 7km of what is now a farm service road into Lauder. En route, the road clips the north-western corner of the great 67ha (165 acres) Roman marching camp of the Severan campaign of 209AD, this certainly near a Roman road. The road enters Lauder by the Stonyford Bridge. The old broken bridge there, and the preceding ford, lend a certain antiquity to the road line. A Roman make cannot be deduced from the name Stonyford. A length of robust stony road uncovered by ploughing some years ago about 1km to the south of the ford was reasonably mooted as Roman Dere Street, but was found to correspond to a track leading to an old chapel shown on OS and other maps. Several such abandoned chapels with surprisingly well-made access roads might be cited. The rural gentry came to kirk in carriages, it seems.

2. Hunt Gates: An association of hunt gates and Roman roads is found elsewhere in the Scottish Borders, as on the Craik Cross Roman road by Kemp's Castle, and on the Ettrick Valley Roman road on Huntley Hill. It is to be feared that the successful efforts of the anti-hunt faction will result in the closure of many of these accesses to open land.
3. Dyke-stone Quarry: As noted earlier, twin cart-ways lead to a dyke-stone quarry on The Heatheries at NT54323832, some 100m north of a hunt gate. Both cart-ways disturb the Roman road terrace as they cross it to the quarry. A deep way and a shallow way suggest out and in respectively. A drilled shot hole in a detached stone suggests a 19th century date for the quarry.

4. Early Roads on Housebyres Moss:

At St. Helen's Haugh an extensive cobbled hardstand, now buried under c. 1 m of sandy silt, suggests an important traffic in cattle and carts. Roy (c. 1774/1793) shows a major ferry crossing. From the ferry a narrow road, dry-metalled on a rough stone bed, runs west then north to Gateside Bank. For two lengths of about 200m each the road is remade with a tarmac surface, the first as the B6360 on the left bank of the River Tweed, the second as part of the entry to Wester Housebyres farm by way of Davie's Brae. The practice of naming road lengths and bridges, as Trow Bridge and Fly Bridge, though now endearing, was probably to define roadmen's work stints.

From the junction at Gateside Bank northwards for 1.7km to the Easter Housebyres farm road-end at NT54073725, the road is made some 2.5m wide with stones in a range of sizes from fist to head, a single layer of the larger stones forming an inadequate road-bed for the roughly cobbled surface. Though not obviously rutted by wheels, the centre of the road is deeply and widely broken, presumably by draft-horses and then water and frost erosion. Crude repairs have been made by throwing wall stones from the bordering dry-stone walls into the worst holes. Scrub roots and recent cattle-trample have not improved the road. The same make and state of disrepair is evident on the branch road from the fork at NT539361 northeast 4.5km to Earlston over Scabbet Hill.

At the Easter Housebyres road-end the road is transformed by careful cobbling 4m wide on a road-bed some 6m wide well-graded by cutting, banking and terracing, with top-side ditches and culverts where appropriate. At NT542375 the road embankment forms a dam for a shallow but extensive, c. 1ha, reservoir or wildfowl pond. The cobbled road rises past Easter Housebyres from Sloethorn Bank and the Blake Burn bottom wherein lies Wester Housebyres farm. The road in the Blake Burn bottom and to Easter Housebyres is now made with tarmac. The splendid cobbled road continues northwards over Housebyres Moss for 1.7km to NT54053855, where it is joined and very definitely truncated, cut into, by a continuation of the Sloethorn Bank road over the east ridge of Easter Hill.

This Easter Hill road appears to have a shallow roadbed of ungraded stones through which the black peat of the surrounding moss is seeping, and is thinly surfaced with a scatter of hand-broken road metal of c.5cm grade. Though this road is clearly, from the road metal presence, the route finally taken by vehicular traffic, it is not badly broken and can only have carried a light traffic.

North of the junction the road resumes the well-made structure surfaced with both cobbles and road metal. Elsewhere locally this 'dry Macadam' make is associated with railway 'compensation' roads of the 19th century. For the 1.1 km length to the Mosshouses cross-roads, and for 0.9km beyond to the crest of Kedslie Hill, the road is in two perfectly straight lengths and set between dry-stone walls 12m apart. It is evidently, from slanted expansions at gates, intended for droving. Beyond Kedslie Hill crest the descent is by similar straight lengths joined by bends which take it by the hill-farms of Bluecairn, Jeaniefield, Fordswell and Upper Blainslie apparently to St. Leonards farm. At NT532417, near Bluecairn, the embankment of this cobbled road again acts as a dam for a pond, fed from the Covenanters' Well, emphasising the unity of that constructional phase. Hereabouts the road is patched with quarried whinstone chips as road-metal, and occasionally with tarmac. From Bluecairn northwards a tarmac surface prevails.

Despite the determined line down to St. Leonards, the Royal Burgh and railhead of Lauder, via the Stoneyford of the Lauder Burn, seems a more likely terminus. This uncertainty in destination may indicate that an early road crossed the Leader Water above the Boondreigh Water confluence to old Thirlstane with its Castle, Chapel and High Cross on the great road from Northumberland to the Lothians. To add to the uncertainties, a green track, now a Long Distance Path, holds to the ridge to Woodheads Hill before dropping down to Lauder, or over the Lauder Burn to Lauder Common and its ancient Fair. A trunk road serves many purposes and the approaches to the Tweed ferry at St. Helens Haugh are every bit as complex.

In terms of standards of construction, the hard stand of St. Helen's ferry seems to compare well with the graded and cobbled road over Housebyres Moss. The ferry-point is noted by Roy in 1750 and was presumably in use in the previous century. The several Turnpike Acts of the late 17th century did little to improve either trunk or rural roads, but from mid 18th century, in the wake of Wade and Caulfield, the turnpike trusts were very active (Dickinson and Donaldson, 1961: Vol. 3, page 358).

Acknowledgements

On 13-6-99, divers Mr. Rolfe Mitchinson and Mr. Robert Middlemass of the Northern Archaeology Group, acting for the Trimontium Trust in searching the bed of the River Tweed for traces of the probable Roman bridge near Newstead (Trimontium) fort, reported a linear mound on the south bank of the river similar to other Roman bridge approach works they had examined. Their report proved crucial to the study.

The search for RDS over Whitehill land was made at the invitation of Mrs Ann H. Gordon. Mr. Dunlop of Merrick farm made me welcome, as did Mr. S. Swinton, then of Eildon Mains, Mr. W. Allan of Leaderfoot, Mr. W. Hogg of Cairniemount, and other landowners and tenants along the way. Householders in Eildon Village were most helpful. The year of the Foot and Mouth disease, 2001, interrupted the study. It was a great pleasure to be welcomed back on the hills.

Mr. Frank Newall, principal co-author in several previous Roman road papers and prime mover in many upland search-surveys for the elusive roads, is now prevented by advancing years from rough walking and louping dikes. His vicarious pleasure on reading the draft of this account was a fine bonus to the pleasures of compiling it.

References

Clarke. S. (1996). Trimontium: a Roman Frontier Post and its Phases. Preliminary Report. Melrose: Trimontium Trust.
Curle, J. (1911). A Roman Frontier Post and its People. Glasgow: James Maclehose and Sons.
Dickinson,W.C. and Donaldson, G. (Eds.) (1961). A Source Book of Scottish History. (Revised edition, Volume 3), Edinburgh: Nelson.
Gordon, D. and Lonie, W. (2003). Broomhill wood and lane (Melrose parish). Roman road. Discovery and Excavation in Scotland, 3(NS) 2002: 102.
Jones, R., Clarke, S., Clerk, K., Dent, J., Lucy, D. and Cheetham, P. (1 994). The Newstead Research Project: 1993 Field Season: Preliminary Report. Department of Archaeological Sciences, University of Bradford.
Lonie. W. (1996). Howcleuch and Borthwickbrae (Roberton parish). Roman road traces. Discovery and Excavation in Scotland 1995: 1 0.
Lonie, W. (1997). Red Rig, Broomhill Farm (Melrose parish). Early road terrace. Discovery and Excavation in Scotland 1996: 90.
Lonie, W. (1998). Red Rig and Bogie Burn (Melrose parish). Roman road. Discovery and Excavation in Scotland 1997: 67.
Lonie, W. (1999). Red Rig, Broomhill Farm (Melrose parish). Early road and ford. Discovery and Excavation in Scotland 1998: 82.
Lonie, W. (2003). Drygrange Bridge (Melrose parish). Probable Roman bridge approach mound. Discovery and Excavation in Scotland, 3(NS) 2002: 102.
Milne, A. (1743). A Description of the Parish of Melrose. Quoted in Curle (1911) page 2, and R.C.A.M.S. (1956) page 469.
Newall, F. and Lonie, W. (1990). The Romans and Strathclyde: the road system. 2. The western flank of the Antonine frontier. Scottish Naturalist, 102: 27-49.
Royal Commission on the Ancient Monuments of Scotland. (1956). An Inventory of the Ancient and Historical Monuments of Roxburghshire. Vols. 1-2. Appendix A: Dere Street, pages 463-474. Edinburgh: H.M.S.O.
Roy, W. (1793). Military Antiquities of the Romans in North Britain. (Plate XXI: Environs of the Eildon Hills; probably engraved in 1774). London: Society of Antiquaries.
Selkirk, R. (1995). On the Trail of the Legions. Chapter 6: The lost Roman roads of Northumbria. Ipswich: Anglia Publishing.
Stone, J. (1991). Illustrated Maps of Scotland from Blaeu's Atlas Novus. Lauderdalia. [T. Pont's surveys, 1583 to 1586, published by I. Blaeu in 1654]. London: Studio Editions.

Dr. William Lonie,
11 Dean Place, Newstead, MELROSE, Roxburghshire TD6 9RL.