Click slide to view full size and download

The Roman Road East from Oakwood Fort.

There was no acknowledged Roman presence in Selkirkshire until an examination of air-photographs in 1949 revealed a fort and adjacent temporary camp on the farm of Oakwood, three and a half miles SW. of Selkirk. This was excavated in 1951/52 and published in P.S.A.S. 1951-52, pages 81- 105 and in the Royal Commission’s Inventory of Selkirkshire, 1957.

The discovery of this isolated fort which was occupied for around the twenty years from 80 A.D. to 100 A.D. called for some re-appraisal of the Roman presence in the central Borders. It is situated in a blocking position where the Ettrick and Yarrow valley converge but The Inventory has no evidence of native forts in either valley and it was assumed that the area was sparsely populated. In fact there a number of quite large settlements in both but these are enclosed by wooden post palisades which leave very little ground trace and are only seen if time, place and conditions are right and even then only by lucky chance. Were the native Selgovae settled herdsmen rather than hunters? Probably.

The P.S.A.S report by Drs Steer and Feachem gives a good account of the excavation and some of the possibilities to be explored. As I spent most of my early life on Oakwood farm, I viewed the fort and its environments as my personal property. One day in the mid-1940s, the shepherd who lived next door came in with part of a flower-pot and two black discs which he had found while burying a dead sheep beside the small wood on the lower hill ( later revealed as a Samian base and two denarii.) This was the start of my Roman interest.

In the mid-1990s, a notional retirement left enough time to walk a wide area in the search of roads in and out of the fort. I established two (to my own satisfaction at least); one went south over the hills until it joined the Craik to Tweed-mouth road near the ‘cairn’ on Smasha Hill (NT 453 171); the other leaves Oakwood fort making eastwards to join Dere Street near St Boswells.

I agree with the excavators report that ‘on general grounds, it is reasonable to assume that Oakwood would be linked by a service-road with the great strategic centre of Newstead, 10 miles to the NE., and the absence of any definite evidence for such a road is of little account in view of the fact that most of the intervening ground had been long under cultivation. (P.S.A.S 1951-52 page 98.)

That summation holds good except for the section between NT44305 and NT 43695 25559 which is on ground that has been only partly ploughed in 18th century rig systems. There are two distinct roads on this general line. My presumed Roman road lies on the south side of the boundary dyke between the farms of Oakwood Mill and Middlestead. This is most convincing from the middle section eastwards where the road mound is about 5m across; in places it stands 1m to 1.2m above the ground level and 1.7m from the bottom of the ditch. The quarry trench is on the north side.

Why do I think that this is a Roman road? Principally because it can be seen on the ground coming out of the south gate of the fort to turn eastward and is visible until the Hartwoodmyres road. It can be traced with divining rods across three fields until it crosses the Hartwoodmyres burn at NT 44305 14730 in a series of paired posts (divined) at 3m intervals*. These indicate a wooden bridge or raised walkway to allow the troops to cross a burn or bog and are quite commonly found on minor Roman roads throughout the Borders.

*’Decempeda’ a standard Roman measure = 2.96m

The second road veers away from the first by as much as 30m. The road mound is not prominent and is about 3.5m across with a wasted turf dyke on the north side. The 18th century rigs do not cross the road and I concluded that this is the road shown, not very accurately, on the Ainslie Map of 1773.

The two roads almost converge at the burn crossing where the stone-built piers of the 1773 road are still visible.

On the Oakwood Mill side of the boundary dyke, a settlement or settlements have been noted although nobody is quite sure what they are.

Craig Brown in his 1886 History of Selkirkshire Vol I p 52/3 notes ‘a square camp with sides 180 feet – measuring along the top of the outer rampart. From this to the top of the inner rampart is a stretch of 24 feet over a ditch 7 feet deep’. He thought it might be a Roman but admits that it is ’a question is open to dispute’.

This might be the ‘earthwork which is best preserved at the SE corner which lies between the two dykes; here it stands 3ft 9in above the ground level and 4ft above the bottom of the ditch ( Inventory of Selkirkshire, No 140). In fact the outer bank of this earthwork is mistakenly referred to as ‘a turf dyke’ although at 5m across, it must be the widest turf dyke in the Borders.

North of the Roman road/dyke on the Oakwood Mill fields, I found the Inventory earthwork quite easily. This is marked as A in the above plan.

West from this at NT 43998 25682 is a double-ditched Roman fortlet c 18m square (marked B) and another (marked C) at NT 43952 25648, c 23m by 18m, both with simple bathhouses and latrines further down the slope. These are similar in size and shape to those I have found along Dere Street and The Devil’s Causeway Through Berwickshire.

NB. These await confirmation by Air Photograph. I am looking for a very dry year.

P.S.A.S 1951-52, page 99.

It must be concluded, therefore, that that Oakwood was not an intermediate station on a lateral road, but the terminal point on a branch-road from Newstead, and this lack of communication no doubt explains why, in the Antonine period, when the power of the Selgovae had evidently been broken, the site was not even occupied by a road-post. Yet the fact that the Antonine reorganisation involved not only the abandonment of Oakwood but also the replacement of the former forts at Milton and Raeburnfoot by small patrol-posts, would seem to demand a more permanent link across Ettrick Forest than was necessary during the Flavian period.

If Milton and Raeburnfoot forts were replaced by small patrol posts, it is quite feasible that Oakwood fort was the same. The two fortlets on Oakwood Mill are better placed to view up the two valleys.

Oakwood fort was not the terminal point of a branch road but rather part of a loop. From the south side of the fort, a road goes across the hills to meet up with the Craik to Tweed-mouth road. This is over uncultivated ground and much easier to trace. Work is currently being done on this. Here’s a snippet.

Construction Camp on Huntly Hill at NT 42288 23465

About 40m from the Roman road, a rectangular shape c 56m by 38m is one of the short-term construction camps. Single ditched and one gate on the NW side. Possibly in use two or three days, it has a latrine 7m by 4m on the SE but no bathhouse.

More to follow in due course.

Walter Elliot. 23 March 2016.

To navigate to the site you can download a gpx file below, for instructions see the Visiting Sites page.

Associated pages

External sites