Home>Miscellaneous>Roman Roads












Roman Roads

With so much written about Roman roads, this section will just provide brief notes on the known network and on missing sections of road within that network, along with relevant links and references.

It will also look at those roads mentioned in the antiquarian literature that have not been established as Roman or are in fact quite speculative. Most of these are unlikely to be Roman but one or two may be worth further research.

The Known Network
Dere Street
- see also Roman Roads in Ayrshire
Roman roads mentioned in antiquarian literature


The network of known and accepted Roman roads. Many other roads have been conjectured.
The above is 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.

The Romans invaded Scotland three times. The first invasion was by Agricola in 78 AD and the occupation lasted until the end of the century when there was a withdrawal to the Tyne - Solway line. This was followed by the Antonine invasion in 142 AD when the Antonine Wall was built. The occupation lasted sometime into the 160’s when there was a withdrawal to Hadrian’s Wall though a presence was maintained in the south of the country. There were two further campaigns by Severus between 208 and 211 AD.

The majority of the roads appear to date from the first invasion. A main road (Dere Street) ran from Corbridge across the Cheviots to Newstead and the Forth near Edinburgh and then probably over towards Falkirk and up towards Stirling. Another main road (Watling Street) ran from Carlisle north to Crawford, continuing on the east side of the Pentlands to the Edinburgh area. An east-west route crossed this near Biggar which may have run from Newstead to Barochan, with an extension to Loudoun Hill. A loop road ran from the westernmost road near present day Lockerbie over to Nithsdale and then northwards to rejoin the main road near Crawford. From the Falkirk area, a road has been traced someway beyond Perth connecting the forts sited at the mouths of the glens.

In the Antonine period it is thought a road was built to link the east and west roads through Raeburnfoot, as well as the military road running to the rear of the Antonine Wall.

A great deal of work has been done to determine what other roads there might have been and also to fill in obvious gaps between known stretches of road. For example, in the south-west, a road is posited to Glenlochar and Gatehouse of Fleet and perhaps beyond; as well as roads running up the Cree, Ken and Nith valleys into Ayrshire. The road to the Loudoun Hill fort may have continued to the Ayrshire coast and it is thought roads may have ran from Castledykes to Crawford and up past Castle Greg.

There are also many references in the antiquarian literature to traditions of Roman roads in a locality. While we will deal with these in detail it is interesting to note some of these. A road had been thought to run from Berwick up the east coast towards Inveresk; another from Jedburgh to Roxburgh; and one from Carlisle to Hawick and onwards to Newstead. In Lanarkshire there was supposed to be one from Carluke up to the Antonine Wall and one on the north side of the Pentlands to Castle Greg. North of the Antonine Wall, one was conjectured to run north from Braemar to Forres and another through Aberdeenshire to the counties of Banff and Elgin. Most of these are not accepted as Roman.

The Known Network

Dere Street
See here for some useful references. There are also many references to Roman roads in Discovery and Excavation in Scotland and on Canmore.
See also Roman Roads in Ayrshire.

Roman roads mentioned in antiquarian literature

This section gives some details of roads proposed by various antiquaries and cartographers. Some of these exist but have been misidentified as Roman while others are quite speculative, i.e. no road exists but a conjecture has been made that a Roman road must have followed a certain route. Despite their generally speculative nature, some may repay further examination.

East of Dere Street
There are a couple of references in antiquarian literature to roads in this area although these have not been supported by any evidence. This lack of roads as well as of attested forts supports the view that the Votadini, whose territory this was, were on friendly terms with the Romans and so were never occupied. Despite this, it is still conceivable that there were roads and forts or marching camps, even if not for military use, and it is best if the evidence for these (which is not particularly convincing) is considered on its own merits.
The one Roman road in the area that is accepted as such is the Devil's Causeway running through Northumberland to Berwick on Tweed, perhaps to service a harbour there.

Based on quarter-inch OS maps, 1922, 1923. With thanks to Ordnance Survey.

East Coast Road William Maitland in his History and Antiquities of Scotland (vol.1, p.202-3) refers to an east coast road from Berwick via Coldingham Moor to Old Cambus (for which he gives an unlikely derivation from "old camp") and then Dunbar where there was a tradition of a Roman camp. From there it went to Camp-hill, a mile and a half NE of Haddington, and then on to Musselburgh where there were faint appearances to the west of the town. Its course then went by Leith, presumably to Cramond.

Modern research has not confirmed any Roman remains near Old Cambus (see for example A history of Coldingham priory, Alexander Allan Carr, page 14 that mentions these), nor at Dunbar or Haddington (Canmore entry for Camp-hill) which make a road very unlikely. However, the reference to Musselburgh would fit a road starting at and heading west from Inveresk which is close to Musselburgh, Inveresk itself being reached from the south rather than a coastal route.

Road to St Abb's Head
William Roy in his Antiquities of the Romans in North Britain notes the Roman road called the Devil's Causeway running from Watling Street (later known as Dere Street) at Bewclay up to the Tweed near West Ord from where it pointed towards Mordington. However, he says that no traces of any continuation had been found in Berwickshire.

The straightness of the old Edinburgh to London road across Coldingham Moor is suggestive of a Roman road but confirmation of this has not been found.

The Devil's Causeway is also shown by Ainslie in his 1789 map of Scotland but crosses over the Tweed a little to the east of where Roy placed his crossing and continues through Ayton and Coldingham to a Roman camp at St Abb's Head.

The New Statistical Account refers to a Roman camp at St Abbs, on a hill west of the Head. From Chalmers' description of the various camps in the area the one in question was situated just north of Millar's Moss Reservoir and is now classed as a settlement (Canmore record).

Robert Stuart in Caledonia Romana, page 264 says that St Abb's Head was probably a naval station connected to Hadrians Wall by a road and notes some occurences of the placename Chesters in Ayton parish that may indicate camps along this road though no traces of these or a road can be seen.


Crachoctrestrete - click for larger image
Places mentioned in the charters. Parish boundary shown in green. Based on half-inch OS map, sheet 28, 1914. With thanks to Ordnance Survey.

This road is mentioned in mediaeval charters relating to the boundaries of Coldinghamshire, viz."from the boundary between Berewic and Lambertun, as far as Billie, thence to Drieforde, thence to Middlesdenehead, and from that by Mereburne to Crachoctrestreet, thence to Eiforde, and from that to the rivulet which flows into the sea at Auldchambuspethe." (Adam Thomson, Coldingham: Parish and Priory, 1908, p. 4). J Hewat Craw lists the various charters, one of which by Alexander II is worded "from midlesdeneheued by mereburnesheued to the west as far as Crachoctrestrete, thence by the same road as far as Eiford." A sasine from 1647 says "and while ye come to the Crawbockchaster-street, and thence be the same street to Ifuird..." These references indicate that the road went at least as far as a ford over the Eye. (Crachoctrestrete: A Forgotten Berwickshire Road, J Hewet Craw, Berwickshire Naturalists Club, Vol.27, Part 1, (1929), pps 93-95).

Although not on the parish boundary, this track continues the line of Crachoctrestrete as least as far as Ecclaw. Looking south from minor road 600m SW of Butterdean. Fawcett Wood on skyline.

Thomson suggests Drieforde may have been a causeway across Billie Mire though Craw thinks it more likely to be north of Billie as it is mentioned after Billie in the charters. Middlesdenehead would have been near Mayfield and Mere Burne would be Mire Burn (he may be referring to the Drakemire area north of the previous location though which of the two or three burns hereabouts is the Mere Burn is not clear). He notes two locations posited for Crachoctre: Prestoncleuch and near the parish boundary west of Warlawbank. Craw favours Warlawbank, suggesting Crachoctre is a corruption of Crawchester, noting the Auchencrow in the locality and the fort at Wardlawbank. He has a photograph of a holloway at the east end of Fawcett Wood and running along by the parish boundary towards the ridge of high ground where the fort is sited. Both Thomson and Craw have Eiforde as a ford across the Eye, on the old road to Butterdean - a ford is shown on the 6"map on the line of the parish boundary about 1 kilometre south of Butterdean. It is not clear if the boundary continued along the line of the old road that runs NNE to Ecclaw where the Heriot Water could be the rivulet running past Auldcambuspath or if it went more directly to the vicinity of Auld Cambus.

At its northern end, C A-Kelly (Canmore record NT76SE 35) examined the line from Billiesmire to Ecclaw which is clearly shown by the boundary between Bunkle & Preston and Coldingham parishes, field boundaries, and tracks. Traces of a ploughed-out mound, and flagging were found. It may have originally gone from Ecclaw to Oldhamstocks.

South of Billiesmire, the line is uncertain. One could speculate that if Roman it could have come from East or West Ord, or Berwick itself, i.e. an extension of the Devil's Causeway or from a putative Roman camp south of Chirnside but these would be very speculative. The Session Book of Bunkle and Preston (p.xxxvii) says that the Billie Burn was an extensive mire and was crossed by causeways - this was the origin of nearby Causewayend thought by some to be Roman (note: The OSA for Chirnside also mentions causeways crossing the Billie Burn but these are some distance from Causewayend).

Clearly, however, the road is very old, and possibly Roman. Four derivations of the name have been given. Barrow in Charters of King David I, page 74 suggests "Roman road by the oak tree frequented by crows (OE craca, crow)". Adam Thomson in Coldingham: Parish and Priory, 1908, page 4 tentatively suggests "the cross by the oak tree." Alan James has 'Upper-farm Crag' (Brittonic Language in the Old North database) and Craw, while not giving a meaning, refers to the link with a fort or camp through the word "chester". All these derivations indicate the age of the road.

As noted above, a Roman camp and possible road had been suggested for a site south of Chirnside. In the OSA for Chirnside (Vol.14, page 32), the writer describes what he thinks could be a Roman fort sited near Stuartslaw farm beside the Whiteadder.

He argues that if, as Tacitus says, the main invasion route was along the coast with support from the navy, this fort could have served to guard Agricola's flank. The argument is not as strong as it seems as this coastal advance is now thought to have taken place much further north, however, Agricola could still have advanced near the coast and may have had outposts guarding his flanks.

The Canmore entry (NT85NE 8) notes the tradition that this was Roman although one comment does not think the remains are Roman - hardly anything can now be seen at the site.

Fogo Parish
The New Statistical Account for Fogo says:
"Antiquities.—The only vestige of antiquity in the parish is an old Roman camp at Chesters, in the west end of the parish; the stones of which have been mostly removed to make way for the plough. The situation shows how careful the Romans were of the health of their armies,—the soil being light and dry, - and the situation being near water, and by its elevation, commanding an extensive view of the surrounding country. There was lately trenched up what appeared to have been a Roman causeway, passing through a marsh adjoining this parish on the south, and probably leading to this camp." (page 225).

The Canmore entry for this site (NT74NW 7) assesses it to be a mediaeval fort.

Between Dere Street and Watling Street

Jedburgh to Roxburgh
Stobie's map of Roxburghshire refers to vestiges of a road that ran from Roxburgh to Jedburgh and then to the Eildon Hills although the road was not shown on the map and the wording does not indicate its route. Jeffrey (The History and Antiquities of Roxburghshire and Adjacent Districts vol.1, p.253) suggests it crossed the Teviot at Bonjedburgh and in fact Roy notes the reference in Stobie and thinks it a possibility, given its situation, that Roxburgh could have been a Roman station. He shows it as a direct line on his map with pecked lines indicating it was not completed or perhaps conjectural.

Carlisle to Hawick (Tarras Water)
The line of this supposed road is shown very clearly on Ainslie's maps of 1789 and 1821. The road is mentioned by Roy on page 105 of Military Antiquities where he notes a road leaves the
Annandale road at Longtown and heads northwards towards Canonby. From Liddell Moat he suggests it led into the valley of the Tarras and from its direction no doubt headed for Hawick from where it could have gone either to Bonjedburgh or the Eildons. This would have given a link between Carlisle and the Lothians although he notes that it may not have been completed, perhaps because of shortage of time or because the Annandale road was adequate for this purpose. His map shows it running to Melrose and he mentions the author of the history of Melros (Milne) having said that there was a great military way running south from the Eildons. However Milne talks of Halidon Park and not Midlern Park as Roy does, so that the "road" is likely to be the earthwork raised by David I to mark the boundary between lands given to Melrose and Kelso abbeys (see The History and Antiquities of Roxburghshire and Adjacent Districts, Alexander Jeffrey, volume iv, page 51)

Continuation - Hawick to Newstead

Based on quarter-inch OS map, 1922. With thanks to Ordnance Survey.


Near Over Cassey
Near Over Cassey

OSA, vol.12, p.614 The OSA refers to a Roman road from Netherby to near Langholm and notes reports that it may have been discovered near Over-Causeway (now Over Cassey) at the head of the parish (Eskdalemuir).


The Ayrshire Road and Nithsdale to Ayr
Older OS maps show a Roman road running south from Ayr to just beyond Dalmellington. It then continues as a "Pack Road" towards Carsphairn and Dalry.

Chalmers, however, notes a tradition that south of Dalmellington the Roman road ran eastwards over to Nithsdale.

These notes (see link) look at the various courses that have been proposed for the road and assess the likelihood that there was a Roman road from Nithsdale up to Ayr.

Based on half-inch OS map, sheet 31, 1914. With thanks to Ordnance Survey.

Dumfriesshire outlier via Daer Water - Ainslie's maps
The road is supposed to have ran across this landscape (from the right or south-east) and then up the valley on the left towards the Daer Water


Based on quarter-inch OS map, 1922. With thanks to Ordnance Survey.

The road is shown on Ainslie's map of southern Scotland printed in 1821. The main Annandale road, while it has the general line of the actual road, is incorrect by a mile or so, and even more so beyond Moffat though it heads for the camp at Little Clyde as does the actual road.The more westerly road has an odd course, unsupported in the literature, that takes it up to the Daer Water. For some reason, both the roads terminate at a sheet edge but it is reasonable to assume they would both have continued to Crawford.

His earlier map of 1789 is more accurate in the sense that it runs across to the Nith near Thornhill then by the Wall Path to Crawford. Nevertheless the map is so imprecise and deviates so much from work that has been done to determine the route actually taken that we can be confident that it just sketches in a rough course for an actual road.


The Well Path - looking towards the Roman fortlet and road north of Durisdeer

Roy shows the road on his map which is too small a scale to be accurate. However, his description of it is more accurate taking it from Dryfesdale church, Wood Castle, Murder Loch, Lanegate and Duncow to Dalswinton where there was a fort. It then went past Thornhill and by the Well Path to Crawford.





Further roads will be added in due course