Home >












Roman Roads - The Known Network

Dere Street

Course of Dere Street. Based on a map of Scotland produced by Eric Gaba and made available on Wikimedia under a Creative Commons licence and Commons: GNU_Free_Documentation_License. With thanks. See original on Wikimedia.

Dere Street was one of the two main routes taken by the Romans into Scotland. Initially it ran from York to Camelon near present day Falkirk where there was a staging area used prior to the campaigns in the north. Later it may have had a branch to the eastern end of the Antonine Wall - this had a road, the Military Way, running along its length.

It is Flavian in date (AD 77-86/90), but was re-used in the Antonine re-occupation of the 2nd.century (AD 139-165) and also during the Severan campaign (AD 208-211). At a later date it was used by the Northumbrians as indicated by its name (Deira was the name of the area around York) and in places may have deviated from the Roman line.

The use of the name here is just for convenience and refers to the Roman road as far as Camelon, and Northumbrian Dere Street as far as Dalkeith. The latter may have had a different line to the Roman road in one or two places: it is not clear how the Northumbrians referred to it north of the Dalkeith area which is where the last known reference to "Dere Street" occurs.

Its course from the border to Newstead is well established though one or two researchers have suggested that in parts it incorporated existing native tracks - see for example Harry Inglis, The Roads that led to Edinburgh, PSAS, Vol. 50 (1915-16), p.32 ff. Beyond Newstead, its course through Lauderdale has not been established beyond doubt, one theory favouring a route just west of the Leader running close to some camps and another favouring a route on higher ground a couple of kilometres further west. This latter route is mentioned in several charters as Malcolm's Road and may have been made or improved by either Malcolm II (r. 1005-1034) or Malcolm III (Malcolm Canmore, r.1058-1093) to consolidate the territory gained from Northumbria after the Battle of Carham in 1018. Alternatively it could have been used by Malcolm IV (r.1153 - 1165) to reach Jedburgh, a favourite place of residence. Whether it was the Roman Dere Street or a new course taken by the Northumbrians (leaving the old Dere Street in the valley below) or an unrelated road made or improved by one of the Malcolms is difficult to say.

North of Lauder its course is more certain as charters refer to Dere Street, taking us past Oxton and Channelkirk to Soutra. Its course has been identified on the ground between these last two places. North of Soutra, both charters and archaeological evidence point to it making to Elginhaugh, a fort near Dalkeith. From this point onwards there are few clues as to its course other than a sighting at Liberton, marching camps at Gogar, a milestone at Ingliston and one or two possible antiquarian references although it undoubtedly made for Camelon from where a road leading northwards has been traced. In the Antonine period there were stations at Inveresk and Cramond that must have been linked to the Antonine Wall, probably at the fort at Carriden but practically the only clues we have to these are antiquarian references. A couple of these are quite intriguing but others are vague and do not inspire confidence.

From Hadrian's Wall to Newstead

The course from the border to Trimontium. Based on quarter-inch OS map, 1922. With thanks.

The road came from York to Corbridge (15 miles west of Newcastle) from where Dere Street continued north through Redesdale on the line of the A68 to the fort at High Rochester called Bremenium - there are several forts and marching camps on this stretch. Another road ran NE from Corbridge to Berwick via a fort at Learchild, near Alnwick, which was joined by a road from Bremenium.

From High Rochester, Dere Street runs into the Cheviots, reaching the present border at Chew Green where there is a fortlet of Antonine age and marching camps which probably date from both the Flavian (AD 77-86/90) and Antonine periods (AD 139-165). On this stretch Dere Street is called Gammel's Path. There is a small fortlet at Brownhart Law (NT 790096) just north of Chew Green which was used for observation and signalling.

A few miles further on is the native fort of Woden Law which guarded an important route over the Cheviots. Once Woden Law was overcome the Romans incorporated the route into Dere Street. It is not certain if nearby earthworks were native defences or practice siege works - the Pennymuir camps could have served as training bases.

Dere Street and Woden Law
Looking from Pennymuir over to Towford and Woden Law. Dere Street ascends the col to the left of Woden Law. Click for larger image.

Dere Street crosses a small stream at Towford and runs past the Pennymuir camps and Whitton Edge to a small fort at Cappuck that presumably guarded a crossing of the Oxnam Water. From here it is remarkably straight, crossing the Teviot (and possibly the Jed) near Monteviot, and running directly towards the Eildon Hills where there was a major fort at Newstead, occupied in the Flavian and Antonine periods and perhaps later.

Between the border and Newstead many parts of the road are well preserved and its line is more or less evident all along its course.

Charter Evidence
Both Curle (A Roman Frontier Post and Its People: The Fort of Newstead in the Parish of Melrose, James Curle - see section on the Roman Road on pages 7-15) and Hardie (The Roads of Medieval Lauderdale, R.P.Hardie, Oliver & Boyd, Edinburgh, 1942) give good coverage of charter references to Dere Street on this stretch of the road. From south to north these references are:

1. a reference to Deorestrete (south of the border) in the History of St Cuthbert (c.1104-1108) which was written by Simeon of Durham.

2. Melrose charters (Vol.1; p.122, no.131 and Vol.1, p.247, no.280 - see References below for links to relevant monastic cartularies) for Hounam and Rashaw refer to Dere Strete as it runs north from the vicinity of Woden Law by the lands of Hounam and Rashaw. The charters are somewhat difficult to interpret but this does not matter as the course of the road is well known in this area. It forms the boundary between Hounam and Oxnam parishes for several miles.

3. A Melrose charter for lands in Maxton parish (Vol.1, p.77, no.90) near to Lilliardsedge which were bounded by Dere Street.

4. Another Melrose charter (Vol.1, p.219, no.244) refers to Dere Street in much the same area. Interestingly it talks of Dere Street intersecting a road that ran from Roxburgh to Annandale. This is thought to have been based on the old Craik Cross Roman road.

5. On the boundary of Lessudden (St Boswells), there is a mention in a Melrose charter (Vol.1, p.76, no.88) of Dere Street descending obliquely to a burn. There is also a Dryburgh charter (p.44, no.58) that gave land to the abbey "west of the church of St Mary of Lessedewyn as far as the great road that led towards Eildon" (The History and Antiquities of Roxburghshire and Adjacent Districts, Alexander Jeffrey, vol. iv, p. 173).

Dere Street north of Tweed - click for larger image Dere Street south of Tweed - click for larger image

Maps to go with Dr Lonie's paper. Click for larger images. With thanks to Trimontium Trust.

Although the charters are vague as to the position of Dere Street in this area, it is at this point that William Lonie gives a detailed account of the road between St Boswells Burn and the Clackmae Burn, a distance of 6 miles (Roman Dere Street over the River Tweed from St Boswells Burn to the Clackmae Burn, William Lonie, The Scottish Naturalist, Vol.116, pps 3-28, 2004 - copy on Internet Archive). At this particular point he has the road cross St Boswell's Burn with traces up to Newtown St Boswells and Eildon village. From there it ran to a bridge at Leaderfoot, very close to the old (1798) road bridge there. This bridge is mentioned in Milne's Description of the Parish of Melrose (1743) as still being visible and from which worked stones were being removed when the water was low.

From Newstead to Lauder

Courses are approximate, particularly Roy's Girthgate route - see OS for known stretches. The eastern branch of the road running from the Tweed opposite Newstead is the "Hie Street" of Hardie and Lonie's Roman Dere Street. The western branch is Lonie's continuation of Malcolmesrode/Northumbrian Dere Street. Based on quarter-inch OS map, 1922. With thanks.

There is an immediate problem with the course north of the Tweed where three courses have been proposed by different authorities.

1. Girthgate
Roy's line, which is that of the Girthgate, would not be accepted today - see Military Antiquities of the Romans in North Britain, p.103.

2. Malcolmesrode
Using charter evidence Hardie thinks that the Roman road (Dere Street) had the line of Malcolm's Road from a crossing of the Tweed near Newstead up to the well-defined track that runs on the high ground about 2.5 km west of the Leader and which is used in part as the Southern Upland Way.

Lonie also has the Roman road take much of this line with the exception that he has Malcolm's Road diverge from the Roman road at Housebyres Moss to run south to the Tweed near Gattonside. He suggests that this latter line is that of Northumbrian Dere Street. North of the junction near Housebyres Moss he would see Northumbrian and Roman Dere Street as having the same line but with later improvements which he explains in detail.

While its central section is easily identified, its northern section as it runs down to Lauder is not so certain.

3. Leader Valley
The RCAHMS following Chalmers have it run close to the Leader, passing several Roman camps on the way and suggest that the other track was mediaeval (see for example NT53NE 72). Margary in his Roman Roads of Britain also thought that Malcolmesrode was mediaeval though possibly on an earlier Roman road, having noted some quarry pits. Whatever the case, it has the characteristics of a ridgeway which is suggestive of a very early development.

The Malcolmesrode and Leader Valley lines are dealt with separately below.



The central section of "Malcolmesrode"On the north side of the Tweed, Lonie found indications that the Roman road (Roman Dere Street) crossed the Tweed at a likely bridge at Leaderfoot and ran in a north-westerly direction to Malcolm's Road (used as part of the Southern Uplands Way), joining it a few hundred metres north of Housebyres Moss. South of this junction, he considers Malcolm's Road to have the line of the Southern Upland Way to reach the Tweed west of Gattonside and that this might be equivalent here to Northumbrian Dere Street, i.e. that a diversion from the Roman road had developed here. His remarks on the roads north of the Tweed are interesting, including what he calls the enclosed road to Lauder, which had the line of Malcolm's Road in its central section and which he sees as having been improved at a much later date, possibly to give access to the railway at Lauder.

Charter Evidence for Malcolmesrode
The charters are as follows (further details can be found in Hardie and a summary account here):

1. Hardie refers to a charter of 1590 that mentions a "hie streit" just north of the Tweed which can hardly be anything but Roman Dere Street, although unlike Lonie he thinks it likely to be a continuation of Malcolm's Road. The charter is from Selections from the Records of the Regality of Melrose, Vol.3, 1917, page 386 (Scottish History Society) with the relevant section reading: "...and fra thence (Lurdane Hole, a ford on Packman's Burn) ascendand south merchand with the landis of Drygrange on the eist as the hie streit gais to the Drycleughhead and fra thence descendand the said cleugh to ane pairt callit the Hauch yet and fra thence descendand to the auld dyk steid standis to the foresaid fuird in Tweid callit the Turf Fuird." Hardie identifies the Turf Ford as the one directly north of Newstead.

This charter is important as other than Lonie's work and the Sorulesfield charter it is the only evidence we have for the route taken between the Tweed and the central section now followed by the Southern Upland Way.
Although the location of the "hie streit" is quite vague, it is clear enough that it lay between Gattonside and Drygrange on the east and can presumably be linked to Lonie's findings.

2. Sorulesfield (Melrose, Vol.1, p.87, no.101). This also is important as it places Malcolmesrode near to Sorrowlessfield which lies above Drygrange . It is referred to as "....the road going towards Louueder along the causeway which is called Malcholmisrode....".

3. Kedsley (Dryburgh, p.80, no.113). This was an agreement between Dryburgh and Melrose on the boundary between Colmslie (Melrose) and Kedslie (Dryburgh). One of the boundaries was "the road that leads to Lauder" which from the context must be Malcolmesrode.

4. Elwinseley and Herdesley (Dryburgh, p.83, no.116). This charter relates to an outlier of Earlston parish on the west side of the Leader, where the westernmost boundary is given by Malcolmesrode. The old parish boundary used to run along the track now used as the Southern Upland Way, just to the west of Kedslie Hill.

5. Blainslie (Melrose, Vol.1, p.83, no.95). This charter of Richard de Morville allowed Melrose to cultivate land in Blainsley from "Windeslaue, and from there by the great road which descends from Windeslaue towards Lauder, as far as the marches of Lauder." (Monastic Annals of Teviotdale, J.Morton, p.263). Hardie uses a charter of 1547 to identify Windeslaue as the summit 200 metres west of Jeaniefield. The Milkeside charter below indicates that the "great road" ran as far as a road branching to Milkeside which was probably on the Southern Upland Way track a few hundred metres north of Fordswell. Beyond this point the course of the "great road" is not absolutely certain: early maps such as Armstrong and Blackadder show a line very close to the Southern Upland Way, although passing closer to Cold Shields; whereas Roy goes by Woodheads.

6. Milkeside (Melrose, Vol.1, p.96, no.108) As for the charter above this refers again to the "great causeway which descends from Windeslaue to Lauwder" and which then runs northwards to the road to Milckeside. As noted the line must be that of, or close to, that now taken by the Southern Upland Way.

Leader Valley
Chalmers and RCAHMS

Chalmers depended on the observations of his correspondent Kinghorn who surveyed the remains of Dere Street in 1803 (Chalmers, Caledonia, v.1, p.140ff; vol.3,p.90ff). While old antiquarian reports can sometimes be confusing, Kinghorn's account must be given credence if he says the remains of the road were quite distinct. The course he gives to Dere Street, although very close to the minor road running up from Drygrange through Kedslie and Blainslie, is not the same as this road. The minor road was in fact the turnpike built after the Turnpike Act of 1768 (see Border Highways, John James Mackay, 1998, chapter 9) and is shown on Armstrong (1775) and Taylor & Skinner (1775) but not on the Military Survey of c.1750.

Dere Street and Malcolmesrode
Course of Malcolmesrode, and of Dere Street after Chalmers. The easterly branch of the road running from the Tweed opposite Newstead is the "Hie Street" of Hardie and Lonie's Roman Dere Street. The westerly branch is Lonie's continuation of Malcolmesrode/Northumbrian Dere Street. Based on half-inch OS map, sheet 28, 1914. With thanks.

Chalmers, referring to Kinghorn, routes Dere Street from a ford opposite Melrose up the west of the Leader, close to the turnpike as far as a camp called Chesterlee (Kedslie-NT54SE 20) - the road in this area could easily be seen. It crossed the turnpike and a small stream that joined the Leader below Chapel which would place this about 400 metres above today's Kedslie.

It then ran up to a Roman site called Wass or Walls at New Blainslie (site NT54SE 18 is 200 metres from New Blainslie). Kinghorn says that the road was very evident for a mile and a half north of here when it crossed the turnpike again, and a nearby stream, half a mile ENE from Chieldshiels chapel. It then ran towards Lauder (The Rev. A Milne in his History of Melrose takes the road up through Chieldshiels chapel).

Details of the Roman camps (all temporary camps) can be found on the Canmore database (Drygrange-NT53NE 33; Kedslie-NT54SE 20; Blainslie-NT54SE 18; South Blainslie-NT54SW10; St Leonards Hill-NT54NW 14) - Canmore entries for Dere Street on this stretch refer mostly to Chalmers. It is interesting to see that the camps are all on or close to the presumed line of the road.While this is suggestive evidence it is not absolutely conclusive. By their nature, temporary camps indicate the passage of a military unit rather than a settled base served by roads, although of course they could use an already existing road. The St Leonards camp (currently the largest known camp in the Roman empire) has been dated to the Severan campaign between 208 and 211.

Charter Evidence
One interesting point is that charters of Richard de Morville and his son William refer to a road from Lauder that led towards Birkenside (Melrose, p.96, no.108, also no.109). In these it is called a magnum stratum which implies it was paved. It formed a parish/county boundary for about one kilometre which still exists and can be clearly seen on aerial photographs.

The feature runs through the St Leonards Hill camp (as shown on the plan in Roman Camps in Scotland, Rebecca H Jones) and touches on the north side of the Blainslie camp and possibly continues through it though it does not appear as a cropmark. As the Waas site of Kinghorn was at Blainslie it is very likely that this is the feature that he said was very evident north of here.

If it is a road, the question is whether it is Dere Street as Kinghorn suggests, or if it is a road to Birkinside and perhaps Roxburgh as Hardie suggests, though whether Roman or mediaeval is uncertain. As it stands the evidence is insufficient to determine which of these possibilities is correct.

Both the Malcolmesrode and the Leader valley routes have their merits but there is a degree of uncertainty that makes it difficult to decide between them though the presence of camps on the Leader valley line is quite compelling. Fortunately, a clearer picture is possible for the next section of the road, north of Lauder, which we deal with below.

From Lauder to Soutra

Roy's Girthgate route is approximate - see OS for known stretches. Based on quarter-inch OS map, 1922. With thanks.

The course immediately north from Lauder is not known exactly but there is a charter for land at a place called Newtoun (probably Lauder) between Derestrete and Thirlstan. Whether this meant the road ran through present day Lauder or to its west is not clear. A couple of miles to the north it is picked up (on charter evidence) at Midburn and probably ran up to Midburn close to the line of the old railway and the western edge of the Roman camp at Blackchesters (see Rebecca H Jones, Roman Camps in Scotland).

A charter for Samsonshiels allows it to be identified as having the course of a minor road from Midburn to Burnton and Oxton. At Oxton, another charter indicates it ran through the village. From there it ran directly to Channelkirk House and the camp one kilometre to the north-west - Canmore record NT45SE 52.00. From there its course is marked on recent OS maps as far as Soutra Aisle - see Canmore record NT45NE 21 which continues it closer to Soutra Farm.

Charter Evidence
A Dryburgh charter (p.145, no.201) refers to land at Newtoun that extended from the west side of Derestrete to the Marches of Thirlestan (....ab occidentali parte de Derestrete in longum et latum juxta metas et divisas de Thirlestan). Although others had placed it at Newton Don and Newtown St Boswells, Hardie is surely correct in placing it at Lauder which is close to Thirlestane.

This is a confirmation to Dryburgh (p.123, no.176) by Pope Celestine III in 1196 of lands in Samsonshiels (now Shielfield), "namely, a toft of one rood in front of, and a croft with land contiguous to, the same croft of three full acres, close to my house from the west, and also that land, arable as well as meadow, which lies on the west side between the aforesaid croft at the top, and the boundary of the burn which is between my land and Pilmuir, that is to say, beginning on the south side at a certain stone cross set up on the edge of the same stream, and extending as far as Derestrete in length northward. To this, likewise, an acre which belonged to William, Robert's son, with the land which lies between the same acre and ditch between Samsonshiels and Pilmuir in breadth, and from the aforesaid stone cross as far as the way which leads to Wenneshead in length, and so by the same road on the east side continually to the ditch at Pilmuir as far as Bradestrutherburn, and thence going on towards the north exactly as that stream formerly ran to the Leader." (transl. Allan, History of Channelkirk, Edinburgh, 1900, p.59)

Using the clue afforded by the diversion of the stream (formerly Bradestrutherburn, now the Mid Burn) just below the farm of Midburn, as well as the parish boundary following the old course of the stream, Hardie (page 87) is able to identify Dere Street as the minor road leading to Burnton and Oxton. Wenneshead is now Wanside on the north side of the Lammermuirs and is reached by a track running north from Carfraemill.
Canmore record NT55SW 40.


The line of Dere Street through Oxton parish. Based on half-inch OS map, sheet 28, 1914. With thanks.

The Oxton charter dating from 1206 was for lands gifted to Kelso (Kelso p.202, no.245). These lay "From the head of Holdene descending by the Holdene burn to Derestrete and then by Derestrete heading north to Fuleford and, by Samson's divisions, to the Leader, and so by the Leader ascending in a straight path to the east end of the said town of Hulfkeliston, and from the east end of Ulfkiliston taking a straight path by the south street and ascending to Derestrete...."

Both Hardie (Roads of Mediaeval Lauderdale - p.88) and Allan (History of Channelkirk, p.664) identify the Holdene burn as the Howden Burn near Overhowden (NT49 52) along which the boundary ran to reach Dere Street and then turn north. Dere Street here has the line of the minor road running from Midburn into Oxton, which ties in well to the last charter.

Fuleford may have been over one of the burns just north of here (see Armstrong map) or nearer to Carfraemill. In any case it followed the Leader up to Oxton where the old 6"map (Berwickshire, sheet XIII) clearly shows a road running from Nether Howden west into Oxton.

Hartside (Hertesheued)

This charter is mentioned by both Allan (History of Channelkirk, p.441) and Hardie, p.89 and is for lands at Hartside which mentions a road from Wedale to Derestre(te). There is not enough information to place Derestrete exactly although Hartside is only a kilometre from the camp at Channelkirk.


There is a charter of the hospital of Soutra (Registrum de Soltre p.24, no.28) that refers to a via regia leading towards Roxburgh, viz. "...and as far as the rivulet to the east in Lynnesden, and from that rivulet eastwards by the road which leads to the royal road leading towards Roxburgh."
".....et usque ad rivulum orientalem in Lynnesden et ab ipso rivulo orientali per viam que ducit ad regiam viam tendentem versus Roxburgh."

As Linn's Dean lies about one mile east of Soutra Aisle neither of the roads referred to in this locality can be Dere Street although the via regia may have been identical with Dere Street for some part of its length north or south of here. Hardie links the via regia to the Birkenside road south of Lauder.

From Soutra to Dalkeith

Course of road north of Soutra, and the lands of Gocelyn the Cook as determined by Hardie. Parish boundaries in green. Based on quarter-inch OS map, sheet 2, 1936. With thanks.

Archaeological and charter evidence indicate that the road ran much along the line of the A68 as far as Pathhead and then along the minor road to the outskirts of Dalkeith at Whitehill. From there it would have run to the fort at Elginhaugh.

It is thought that it ran from Soutra Aisle to the south-west of Soutra Farm then a couple of hundred metres south of the A68 to the west of Fala then crossing to the north side of the A68 to run towards Bleak Law. From there the line may be shown by a long stretch of parish boundary running to the outskirts of Pathhead from where it would take the line suggested by Hardie, viz. the minor road passing through Chesterhill.

Charter Evidence
A Newbattle charter (p.7,no.10) records a gift by Malcolm IV of the lands of Gocelyn the cook that lay between Newbattle to the west, Cousland to the north, and Cranston to the south. Derestrete is mentioned as one of the boundaries.

In his analysis, Hardie convincingly argues that the territory was that part of Newbattle parish that extends eastwards into Cranston parish. This would indicate that Derestrete was the minor road running from Pathhead into Dalkeith. He notes another charter (p.229, no.269) that mentions Derestrete in connection with Cowden Bog which is one of the boundaries of Gocelynton and another charter (p.9, no.12) that may contain a reference to Derestrete.

From Dalkeith westward
There is quite a contrast between this stretch of Dere Street and the previous stretch. There are no charters, no obvious lengths of road; just one or two pieces of archaeological evidence and antiquarian tradition. Sometimes these traditions are intriguing and have the air of truth, other times they seem speculative or so vague as to be useless. Clearly there are dangers in taking some of these literally but other than the all important archaeological evidence (which includes a sighting at Nether Liberton, marching camps at Gogar, and a milestone at Ingliston) there is little else to go on. In any case, these traditions do have their own historical interest.

Another difficulty arises when we assume the forts at Inveresk and Cramond are from the Antonine period and from this infer a road system at that time. The reason is that some evidence has been interpreted to suggest that the sites may have been occupied in the earlier Flavian period which could mean that some routes thought to be Antonine may have existed in Flavian times - see for example Canmore record NT17NE 137 and DES 1995 (a), page 53 (doi:10.5284/1000284) for Cramond.

It is generally thought that Dere Street was built under Agricola early in his invasion. It is likely to have headed from Elginhaugh directly to Camelon which was the jumping off point for the invasion of the north and beyond which a road led for a considerable distance into Perthshire and beyond. Two routes are feasible: one directly to Camelon, perhaps passing the fort of Mumrills, and going through what is now Linlithgow; the other passing along a ridge about a mile north of Linlithgow and again heading to Camelon - there are antiquarian references for this road.

In the Antonine period the situation is more complicated. The main purpose of the advances at this time seems more to do with the re-occupation of territory that had already been conquered and the establishing of a frontier line rather than aggressive campaigning in the north - only three forts in Perthshire date from this time. Camelon with its road to the north would be used to reach these but much of the movement on the east side of Scotland must have been to reach Inveresk and Cramond and ultimately the Antonine Wall with its accompanying Military Way which once built could then be patrolled.

The antiquarian literature postulates that there was a link between Inveresk and Cramond by roads running through Leith and through central Edinburgh. Inveresk may have been connected to Dere Street by a road said to have been visible in the 1700's and making to Lugton or nearby Elginhaugh, or perhaps more directly by a branch from the south, perhaps from near Pathhead. In addition, Maxwell refers to parallel cropmarks seen on aerial photographs on the right bank of the Esk near Dalkeith that could be a branch from Dere Street (Two inscribed Roman stones and architectural fragments from Scotland, G S Maxwell, PSAS, 113 (1983), 382 (doi:10.5284/1000184).

It is unclear if the section of Dere Street that ran from Elginhaugh to the west of Edinburgh was used. A milestone at Ingliston suggests it was but this could also fit a road coming from Inveresk through central Edinburgh (intersecting the earlier course of Dere Street) or even from Cramond.

A further complication is the road up the east side of the Pentlands and how this relates to the remains of a road said to have been evident north from Lothianburn and which was thought to point towards Cramond. The existence of this road may at least be proved some day as Sir John Clark of Penicuik had the modern turnpike laid over it for a mile of its length - see here.

From Ingliston and Cramond westwards the course is somewhat uncertain. Dere Street itself may have had, as we saw, two possible routes in Flavian times and the road must have continued in use as shown by the milestone dated to AD 140. While it may have kept its original course to Camelon, it could also have been redirected or had a branch to Blackness or Carriden fort where the Antonine Wall and the Military Way would have been reached.

So far as Cramond is concerned there is a long-standing antiquarian tradition that it was connected to the fort at Carriden from where the Military Way that ran along the length of the Wall would be accessed. The route given is Barnbougle, Ecklin and Abercorn to Blackness and Carriden with a possible link from Blackness to the road running past Walltown and over Airngath Hill.

The marching camps used in the Severan campaigns from AD 208-211 show that Dere Street was followed at least as far as Pathhead. Ultimately the invasion force made for Dunipace, near Camelon, prior to its advance to the north.

Flavian period (AD 77-86/90)

Possible course of Dere Street in Flavian times. Click for larger image.
Possible course of Dere Street in Flavian times. Based on quarter-inch OS map, sheet 3, 1923. With thanks.

The Canmore entries (NT26NE 76 and NT27SE 444) have the road follow the line of the A7 (this has been re-numbered as the A772 though the Canmore entries retain the old A7 number) as far as Nether Liberton, less than two miles from the city centre although recent work (NT26NE 70) has found possible remains of a Roman road a kilometre or two east of this line at Edmonstone. The Edmonstone find may link to a remark of Harry Inglis (The Roads that Led to Edinburgh, PSAS, Vol. 50 (1915-16), p.36-38) that there may have been a road continuing onwards from Whitehill along the parish boundary towards Newton, Woolmet and Craigmillar Castle and that it might be the "road of the Standing Stone" mentioned in a charter of Dunfermline Abbey (p.265, charter 384). Margary (section 8g) had thought there might be traces of an agger near Whitehill at the point where the parish boundary carries straight on.

The course through Edinburgh is now lost but the milestone near Ingliston (Canmore entry - NT17SW 22) and marching camps at nearby Gogar Green are likely to mark its continuation. The milestone is dated to AD140 but this is thought to indicate that the road was re-used at this time rather than the road being made at this date. It mentions Trimontium, i.e Newstead. The former Gogar Loch (where the South Gyle Centre is) would have had to be avoided and it is of note that the boundary between Ratho and Kirkliston parishes runs along the main road here for just over a mile. On two of his campaigns (1298 and 1303) Edward I stayed at Kirkliston on his way to Linlithgow and probably passed through also when travelling between Edinburgh and Linlithgow. This is a strong indication that a route existed as far as Linlithgow at that time although strictly speaking we cannot assume that it originated in Roman times.

West of Kirkliston there are two feasible routes (given the ultimate destination of Camelon). One would very directly run through what is now the town of Linlithgow to Polmont and the fort of Mumrills, from where it is less than 4 miles to Camelon. This option would require the Avon to be crossed near Linlithgow Bridge. The other would run along a ridge about one mile north of Linlithgow Loch.

Alternative courses of Dere Street near Linlithgow and possible later roads. Based on half-inch OS map, sheet 27, 1913. With thanks.

There are a couple of antiquarian references for this second route. George Waldie in A History of the Town and Palace of Linlithgow,1879, says a Roman road ran along the summit that lay between Linlithgow and the eastern end of the Wall (which he places at Carriden) and passed the farm of Walton (page 19). This suggests he was talking about the east-west ridge of which Airngath Hill is a part. He also says that remains of a causeway coming from Blackness were removed 50 years before on the ridge above Walton (page 20). From his descriptions it is not quite clear if these were one and the same road, namely a road coming from Blackness and then crossing over Airngath Hill; or a spur coming from Blackness to join a different road (Dere Street) crossing Airngath Hill. Certainly it has long been thought that there might have been an outpost at Blackness. A reference in Discovery and Excavation in Scotland (1998, p.38, C McGill) identifies a road near to the Allanvale Hotel (NT0150 7907) which could be the Airngath road or even the Blackness road.

Another reference is from Borrowstounness and District, Thomas James Salmon , 1913 who says on page 10 "that the road which comes east from Upper Kinneil and Rousland and over the Erngath Hills by the golf course and Bonsyde was one of the old Roman roads though not the Military Way. It may not have run direct from Bonsyde to the Walton but traces of it have been found about the Boroughmuir and also at Grougfoot, near the Walton."

There are also vague references from Robert Sibbald, William Maitland and one or two others which we will deal with below under the Antonine period.

While not proof, it is interesting to note the occurence on this putative line of a Cauldcotts one km west of the House of the Binns which Codrington in his Roman Roads of Britain says is nearly always associated with Roman roads. Also of note are some lengths of parish boundary on, or near this line - these are shown on the map.

From Dalkeith westward - Antonine period (AD 139 - 165)
Inveresk and Cramond

Possible roads near Edinburgh in the Antonine period - click for larger image
Possible roads in the Antonine period. The earlier course of Dere Street is shown in light purple. Based on quarter-inch OS map, sheet 3, 1923. With thanks. Click for larger image.

In the Antonine period, Inveresk and Cramond were established which raises a couple of questions about the road system at that time. There were antiquarian reports that a causeway was evident between Inveresk and a camp at Lugton (Sibbald, noted in Chalmers Caledonia, v.1, p.143) or Sherrifhall (OSA - Inveresk - go to "Browse scanned pages). A camp has never been identified at Sherrifhall - see the Armstrong map (NLS) where it would have lain just north and east of the modern roundabout but there is one at nearby Lugton as well as the fort at Elginhaugh, so this could be a valid reference for a link from Inveresk to one or other of these stations or even Dere Street here. No course for the link road is given although a fairly straight line could be assumed. We have also noted Maxwell's reference to parallel cropmarks seen on aerial photographs on the right bank of the Esk near Dalkeith that could be a branch from Dere Street to Inveresk (Two inscribed Roman stones and architectural fragments from Scotland, G S Maxwell, PSAS, 113 (1983), 382 (doi:10.5284/1000184).

Possible locations through Edinburgh as noted by Wilson - North Loch, Trinity Church, North Back of the Canongate and Fishwives' Causeway - see NLS for early maps of Edinburgh. He dismisses the idea of a link road through Leith. The Lang Gait and Donaldson's Hospital are also shown (DES 1958). The north-south line appears to have been based on finds of Roman materials that do not necessarily entail a road. Based on half-inch OS map, sheet 27, 1913. With thanks.

It is possible that a road ran more directly to Inveresk from the Pathhead area and that Cramond was reached by a road that Chalmers notes (v.1,p.144) went by the Fishwive's Causeway in Duddingston and by Leith, rather than following the earlier course of Dere Street through Edinburgh (by Elginhaugh). William Maitland in his History and Antiquities of Scotland, 1757 Vol.1, page 203 also refers to Leith saying that "the military way westwards faintly appears on the west side of Musselburgh and then goes n-w towards Leith, crossing the Water of Leith at the end of the Weigh-House Wynd and seen about 50 years since under the foundation of the pier, then repairing." It continues to Cramond then crosses the Moor of Ecklin, and joins the Wall at Kinneil.

There are a few references in the Prehistoric Annals of Scotland, (Sir Daniel Wilson, v.2, p.55) that fit a continuation of the Fishwives' Causeway (referred to as a via regia in a Kelso charter of 1466, and following an earlier parish boundary - see here) through Edinburgh more than a route through Leith viz. remains of a Roman causeway four feet below the modern road, just north of the Castle, where the North Loch had been; under the old Trinity Church where Waverley station now stands; and a stretch discovered in 1822 that ran along the north back of the Canongate towards the Fishwives' Causeway. However, he confuses the issue in a paper entitled The So-called Roman Heads of the Nether Bow, PSAS XIX (1884-85), p.205 (doi:10.5284/1000184) where he talks about the road from Cramond coming by Canonmills, Broughton, St Ninian's Row, Leith Wynd, the Royal Mile, St Mary's Wynd and the Pleasance towards the south. This appears to have been determined by the presence of finds along this line including the causeway under the old Trinity Church. The evidence for his original east-west route is more compelling than this north-south route as finds of coins etc do not necessarily entail the existence of a road.

To the north of the North Loch was the Lang Gait, later called the Lang Dykes after it was enclosed. Mary Queen of Scots is said to have used it in 1560. It had much the line of the later Princes Street and would be well aligned with finds noted at nearby Donaldson's Hospital and Eglinton Crescent (see E A Cormack - DES 1958, p.22, who also reports on a road at the south end of Corstorphine Hill - DES 1957, p.22/23).

At Cramond itself, William Maitland in his History of Edinburgh, published in 1753 says that there were traces of roads south, east and west of the fort at Cramond - book IX, page 508. This seems to be borne out by recent work.

Various places said to lie on the course of the road between Cramond and Carriden. Based on quarter-inch OS map, sheet 3, 1923. With thanks.

From Cramond onwards various antiquarian authors have much the same route to the Antonine Wall, mostly to the fort at Carriden. Thus Robert Sibbald in Part 1 of his Historical Inquiry concerning the Roman Monuments and Antiquities in the North Part of Britain called Scotland, 1739, page 31 (thinking that the Wall ran much further to the east than commonly agreed) says that it went from Carriden to Walltown (the name refers to a well or wells rather than the Wall) then Blackness. From there it went south of the Manor of Meadup (Midhope) to cross "a ditch of a huge deepness" (Midhope Glen) then Abercorn. He thought it did not go much further than the Laird of Maners Windmill, where there appeared to have been a fort (there is a Windmill Knowe, 800 metres east of Hopetoun House). As said, Maitland in 1757 wrote that the road went by Leith to Cramond, then over the Moor of Ecklin to join the Wall at Kinneil.

An indication of the course can be seen on Ainslie's map of 1821. Having shown the road from Borthwick etc as far as Cramond, Ainslie's eastern sheet shows it continuing as Graham's Dike along the course given by various antiquaries viz. Barnbougle, Ecklin, Walltown, Airngath Hill though when he changes to the western sheet it becomes quite confused. His 1789 map is much sketchier and not quite showing the same route which is probably an indication that on both maps he was just showing a rough approximation to its course.

Echlin Moor. The road is said to have ran past Mons Hill/Barnbougle Hill and Queensferry, then over Echlin Moor to Abercorn, Blackness and Carriden at the end of the Antonine Wall. Click for larger image.

Writing in 1831 Penney in his Topographical and Historical Account of Linlithgowshire, page 32 gives the course from Cramond as by the Amon (Almond), Barnbougle Hill and Ecklin Moor where remains could be seen, to the end of the Wall.

John Philip Wood in The Antient and Modern State of the Parish of Cramond, 1794, page 9-10 says the road was visible at Bowbridge where the modern road had been laid on its line and that it ran from there by Ravelston to Cramond. From Cramond the road crossed the Almond then went over Mons Hill past Queensferry and Abercorn to Carriden. He notes that no vestiges could be seen between Cramond and Carriden.

In Borrowstounness and District, Thomas James Salmon, 1913, page 6 considered that the Military Way may have continued eastward by the ridge on which Carriden House stands to Cramond and Edinburgh and ultimately joined Watling Street. Certainly it has long been thought that there might have been an outpost at Blackness. Entries by E J Price in Discovery and Excavation in Scotland - 1974, page 67 and 1977, page 38 refer to such a road. Another entry, also by E J Price, on page 38 of the 1977 volume, notes an old road west of Abercorn that may be the continuation of this road from Blackness.

We have already noted Salmon and Waldie's mention of the road running in an east-west direction over Airngath Hill as well as the indication of a road near Allanvale Hotel. In Waldie's case (A History of the Town and Palace of Linlithgow, p.19/20), when he talks about a road coming from Blackness it is not clear if it continued as the road over Airngath Hill or if it joined this as a branch. If it is the first of these (and assuming any station at Blackness was Antonine), this would suggest the Airngath Hill road was Antonine and hence not Dere Street; if it was a branch then the Airngath Hill road would be earlier than the Antonine period, i.e. Flavian and so Dere Street.

Once the east end of the Antonine Wall was reached, the Military Way that ran along its whole length would have been used to reach any location along the Wall that was desired. At Inneravon there was a concentration of worked stones found on the bed of the River Avon on the line of the Antonine Wall indicating a crossing of some kind (DES 1973, p.58. L Keppie, (doi:10.5284/1000284)).

Roy (whose route for Dere Street from Borthwick to Cramond would not be accepted today) continues the route from Cramond by Barnbougle Hill and Abercorn or very close to it, and running to Carriden near the end of the Antonine Wall.

One question of interest is where the Almond would have been crossed when the road (assuming there was a road) left the fort at Cramond. It is not impossible that Cramond Bridge, which is known to have existed in the middle ages and is only one mile from the fort, is sited where there was an earlier Roman bridge.

The route that passed the milestone at Ingliston (dated to AD 140) must have been used because the inscription dates from this time but it is not clear if the link was from Cramond or from the east, either by the route through central Edinburgh or the earlier Dere Street coming from Elginhaugh and Liberton. Additionally as noted immediately below, there could have been a link from the road running up the east side of the Pentlands.

The East Pentlands Road (NMRS title: Border-Crawford-Inveresk(?))
A road ran from Crawford up the east side of the Pentlands through Biggar, West Linton and Carlops. The most northerly camps on the road are at Flotterstone (about 2 miles n. of Penicuik on A702) and Boghall about 1.5 miles north of Flotterstone. The road is thought to have headed to Elginhaugh in Flavian times (S. Frere in Britannia, London, 1967, page 107 suggested that it and Dere Street would have served to cordon off the hostile Selgovae tribe) but it may have changed in the Antonine period as suggested by Boghall being very close to the remains of a Roman road said to have run north from Lothianburn, just over a mile further on. Details are below but in essence more than a mile of road was said to be evident and to follow a parish boundary, often a sign of great age. If these remains did exist and were a continuation of the road running up the east side of the Pentlands it could have gone to Cramond in Antonine times or equally have intersected Dere Street and gone in a westerly direction by Ingliston with its milestone dating from this time.

Work has been carried out on a possible road on the other side of the Pentlands that may have ran from Castledykes past Castle Greg to Craiglockhart- see Canmore references.

Addendum - Roy's route from Soutra to Cramond and Carriden
Note: The following section as far as Cramond is included for historical interest only. For his course beyond Cramond see above. The course to Cramond would not be accepted nowadays although he notes that remains of a road were to be seen near Bow Bridge and Lothianburn which if correct would be significant. His reference to traces north of Borthwick are too vague to be of much value.

Detail from Cary's 1794 New Map of England and Wales showing the route postulated by Roy, Chalmers and others. They have it come from Melrose (Newstead) by the Allan Water (i.e. the Girthgate) to Soutra. Map courtesy of David Rumsey Historical Map Collection. The image is copyright Cartography Associates but has been made available under a Creative Commons license for non-commercial use.

Roy thought that the Roman road went from Newstead up the Allan Water to Soutra, confusing it with the Girthgate - see Chalmers, Caledonia v.1, p.143 &164 and Roy, Military Antiquities of the Romans in North Britain, p.103. From there he thought it went to Currie near Borthwick where traces could be seen heading north.

He suggested that it crossed the South Esk near Dalhousie and the North Esk near Mavis-bank then headed for Straiton. He then takes it by the east end of the Pentlands at Bowbridge where traces could be seen, then by the Buckstone and past Bruce Hill to the east end of Barnton Park and the fort at Cramond. From Cramond it headed west by Barnbougle Hill and Abercorn to the fort of Veluniate at Carriden.

"The Cast"Where Roy says it crossed near Mavisbank, it is relevant to mention the nearby feature called The Cast that the NSA for Lasswade, page 133 (go to "browse scanned pages") says ran down to a ford over the North Esk and may be a Roman road with its name coming from via ad castra. It is very much a hollow-way, some 6 feet or so lower than the surrounding land.

Although it crossed at or near Roy's crossing, its alignment is not that good with Roy's putative line of road and the name could easily mean excavation or ditch (see DSL -3/7) though for what purpose it would have been excavated other than easing the gradient down to the said ford is difficult to say. The slope certainly lends itself to the formation of a natural hollow-way caused by local traffic but it is not necessarily Roman in origin. Some modern materials have been used to repair its surface.

The course for the road is that shown on the 1st series of OS 6"maps (sheet 6). Later series show it following the main road. Based on half-inch OS map, sheet 27, 1913. With thanks.

Roy says there were no traces after the crossing until Bowbridge, other than the possibly significant place-names of Loanhead and Straiton. The Ordnance Survey shows over a mile of Roman road on the first edition of the 6" map running from Lothianburn through Bowbridge and Fairmilehead to what antiquaries thought was a Roman town at Comiston (some thought it a camp). The Rev. T Whyte in his Account of the Parish of Liberton, p.308, (Archaeologia Scotia, Transactions of the Society of Antiquaries of Scotland, v.1, 1792) notes that "the road here for near a mile is exactly cut out in the very line of the old Roman military way; and this was done on purpose by the direction of the late Sir John Clerk of Pennycuick." The fact that the road follows the parish boundary could be an indication of its antiquity.

As said, Roy continues by Bruce Hill to the east end of Barnton Park and the fort at Cramond. From Cramond he takes the road west by Barnbougle Hill and Abercorn to the fort at Carriden.


Links to some online resources have been made in the text: the following resources are also of interest.

Monastic Charters - Melrose; Dryburgh; Kelso; Soutra; Newbattle; Dunfermline
Exploring Scotland's Heritage: Lothians and Borders, John R Baldwin, RCAHMS, Edinburgh 1985.
The Antonine Wall, David J Breeze, John Donald, Edinburgh 2006
The Northern Frontiers of Roman Britain, David J Breeze, Batsford, 1993
Canmore References - 26 entries; see also map for these locations
A Roman Frontier Fort in Scotland: Elginhaugh, William S Hanson, Tempus 2007.
History and Antiquities of Roxburghshire, Alexander Jeffrey, vol 1, p.213ff
Roman Camps in Scotland, Rebecca H Jones, Society of Antiquaries of Scotland, 2011
The Legacy of Rome: Scotland's Roman Remains, Lawrence Keppie, John Donald Publishers, Edinburgh 2004
Notes on the 'Roman Roads' of the One-Inch Ordnance Map of Scotland, The Roxburghshire Roads, James MacDonald, PSAS, Vol. 29, (1894-95), 317-28
Roman roads in Britain: north of the Foss Way - Bristol Channel (including Wales and Scotland), vol.2, Margary, I D (1957) London
The Anglo-Scottish Border Roads of The Cheviot Hills - An historical and archaeological background, Northumberland National Parks
The Roads of Mediaeval Lauderdale, R.P.Hardie, Oliver & Boyd, Edinburgh, 1942
RCAHMS (1956) The Royal Commission on the Ancient and Historical Monuments of Scotland. An inventory of the ancient and historical monuments of Roxburghshire: with the fourteenth report of the Commission, 2v Edinburgh. Page(s): Vol. 2, Appx. A. See online pages 252 to 265. This is a very detailed account of the physical remains of the road as far as the Dalkeith area.
The Midlothian and West Lothian volume, pps xxviii - xxxiv (1929) continues the account as far as the Antonine Wall - see online pages 35 to 45.
Scotland: The Roman Presence, ScARF (Scottish Archaeological Research Framework), June 2012. This report contains very useful and up to date information on the Romans in Scotland - see in particular section 3.
Atlas of Scottish History, Edinburgh 1996, The Scottish Medievalists.