THE MAIN ROMAN ROAD TO EXETER IN DORSET

Eggardon Camp - Horn Hill, Beaminster - Waddon Hill fort, Stoke Knap - Axminster

THE THESIS - 1970s by Richard Whaley

The Roman Road out of Dorchester makes a lot of effort to keep to high ground on the edge of the downs. It seemed unlikely it would then plunge down into difficult country through Bridport and Charmouth to reach the known route from Axminster.

A route was charted keeping to high ground and the edge of Marshwood Vale, just south of Beaminster, and along modern B roads to Axminster. I explored this last B road route first. I found these B roads occupied a small part of a much larger terrace construction, generally c 40' wide. In a short stretch to Stoke Knap (ST444 015) not on modern roads, massive but very damaged engineering was found.

At Stoke Knap the evidence did not go straight on as expected, nor South, but unexpectedly clearly went North over Horn Hill to north of Beaminster. Once there it was clear where the route would go, to connect up with the earlier charted route - thus the second part of this 1970s investigation was to link up from Eggardon to Horn Hill.

This second part of the 1970s investigation started at Two Gates, near Eggardon Camp (SY552 938), where Margary's route leaves the downs and descends to Bridport which I have examined recently. This part was on or near the charted route, mainly on tarred bye roads, sometimes tracks or footpaths - occasionally these may take a detour and the line may pass through a few fields for short distances. This makes surveying much easier, for we only have to add engineering structures to OS maps on the routes they show, measuring from identifiable features.

There are some two dozen engineered structures on this route to Horn Hill, which also show 40' terrace constructions. This width can be traced all the way on Margary route 4 from London to Two Gates, and along our route until it crosses the Foss Way at Axminster. After this terraces are the more usual 20' - 30' (as they are on the route via Bridport). This suggests an original military purpose. But it is still followed by modern routes almost all the way, and is heavily damaged on sloping ground, and is a candidate for being the main Roman Road to Exeter.

Two excavations sites were identified, one on this second section. On the section to Stoke Knap, where the modern track leaves the Roman line, leaving a classically preserved agger. Someone had recently cut a drainage ditch through it, producing a perfect section. Possibly it could be cleaned up. This is near the known Roman Fort on Waddon Hill. I do not think the Road system went through the fort, but it stands on a corner, where the Road direction changes from easterly from Axminster to northerly to Horn Hill.

I will put my field records, which are very detailed, on the website www.hants.org.uk/rrexeter.htm in chunks sufficient for a days visit to the remains - starting at Two Gates.

THE 1970s RECORDS

The first part of the investigation on the B roads Beaminster - Axminster is entered in my note book "New Roman Roads" and dated May - August 1973, though contain additions through the 1970s. These are probably not the original field notes. The second part of the investigation was recorded on "backs of envelopes", clearly in the field - these were entered in the same note book in Dec 2009. There was also an assessment of what I found, and as someone then new to archaeology a decision not to try and tell anyone about the finds. However there was also a programme of work outlined to counter the assessment - much of which has not been completed, but could be undertaken in the present project. However having explored many Roman Roads in the following decades it is clear that the concentration of remains on this Dorset route is higher than most of those regarded as major and established, such as the Devil's Highway. As I am proposing to start the study from the last know remains from Dorchester at Two Gates, I am entering the records for the "second part" of my 1970s study first. [To go to the first part of my 1970s investigation, from Axminster to Stoke Knap and Horn Hill Click here]

Two Gates, Eggardon Hill - Horn Hill

Known Roman Road to "Two Gates" noted as similar to B3164 - 5 (Broadwindsor) - Axminster and Stoke Knap sections with the "Hollow way filled in".

Survey 2010 in Blue

Road NW from Two Gates may not be the Roman Road - 1st alignment windy - 2nd alignment from 546 944, leads back to Margary's route 549 929 where the parish boundary turns south - Margary's route up to this point massive - not so beyond - possibly main Roman Road goes across fields almost due north - no sign on air photo [ OS c 1970, 1973 techniques only]

OS Air Photo Film 69 053 No. 49. Photo copy Normal + 2 Darkness shows 2 parallel dark line from said 549 929 - either side of a barn - c80' apart, both clear to first field boundary. West dark line has weaker indications running into W side of 2nd alignment above. Enlarged photo copy Normal Darkness shows weak light line, quite broad, running out of barn merging with the modern lane of 2nd alignment c half way along c 30' wide, east side of this line tinged dark (not same as east dark line of darkest photo copy). Broad light line likely to be Road metalling, & 80' apart dark lines the side ditches.

Then gives 5 alignments over downs, all but 1 v straight, keeping to flat ground, little of aggers seen - but straight with wide verges - 60' with modern encroachments

  1. As above, indication of Terrace at bend 547 947 Visited 4 April 2010
  2. Straight alignment across narrow saddle, short
  3. Straight alighment, part of which modern road goes W of parish boundary and overgrown hollow way Visited 4 April 2010. Large build-up east of hollow way, Agger Terrace 40' wide, modern lane on its west side.
  4. Not very straight alignment on the ground. Visited 4 April 2010. Summit of downs, whole of Marshwood Vale round which this Road likely to go could be seen. Alignments sensible to keep to edge. Ahead, a valley had to be crossed.
  5. Straight alignment, brings RR as far as practical along this section of downs, and leads onto:
  6. Alignment - gentle way down hill side, at top 550 960 strong indications of Agger Terrace. But area built on, and modern road enlarged: it first runs below Agger, the cuts back through it. Two paths/drives to houses cut through it. First fp shows its construction, second drive shows 18' wide 3' high made of packed flint - modern road here encroaching on it. Little further on indications of causeway over depression .[Barrowland Farm to West] Visited 4 April 2010. These cuts through Agger now grown over and construction not visible. A lady of the dog training school, resident for 25 years, knew this was a Roman Road, but thought it was the tarred lane below the buid-up.

    Route continues - with several short alignments, most of which are clearly very straight, and makes a gentle climb from a low point of c 400'

  7. Narrow road, not very straight on the ground - may not be original as its making for railway bridge 547 975 - though appears straight on map. Possibly faint indications of a more direct route on air photos [1973 techniques] Visited 4 April 2010 with photo copies of OS 69 114 no. 80. These indications are of an old field boundary with trees still standing. At the bottom of valley by railway a stream runs in its hollow way, where the shoulders of a causeway 62 yds east of modern lane found. 8 yds wide at top, 2 yds by stream - not visible on N bank of stream as this is now railway embankment. Does not line up with above field boundary. But on the air photo, light copy (normal - 1), oblique into desk lamp, shows thin straight line running from near this causeway to first bend in lane N of rail bridge. With powerful magnifier, a similar light line with dark edges runs from near this causeway to join the modern lane 156 yds south of rail bridge, according to scale of air photo 1:7500. Likewise, width of air photo line 1/32" gives width c 20'. Looks as if causeway crossed stream hollow way at right angles, and this arrangement was to keep the Road level on the ground (which was very wet) - walking up lane N of railway, if Roman Road went E of this lane it would require a terrace - none seen. 30 May 2010: Examined Causeway, identified its centre, measured its width W & E from centre -> 9 yards both (18 yds total at edge of hollow way); centre measured by wheel to centre of lane to west - 62.5 yards. STILL NEED to scale air photo
    14 Sept 2010, measured from centre rail bridge southwards up lane to wood boundary on west side: 93yds 16". On Normal -1 air photo this dist is .51". The air photo lite line meets S side of stream hollow way .33" from centre of lane. Thus this distance is 93.444yds x .33/.51 = 61yds. THIS AIR PHOTO LITE LINE IS GOING TO THE CAUSEWAY. Likewise, the southern lite line bounded by dark lines meets modern lane .68" S of rail bridge centre (as crow flies): 93.444x.68/.51 = 124.59yds or 374'

  8. Takes the hill 'level' across marshy ground(?) which is a rational detour, slightly built up, W verge has wide dense hedge: boundaries of whole v straight and 40' wide. Visited 4 April 2010, as described in last item, but this alignment is subject to the air photo light line in last entry above. It is not shown accurately on OS 1" sheet, the bend just N of railway is not shown (where the light line joins the modern road). Air photo suggests new alignment where light line joins modern road leading:

  9. To top of steep damp ravine, which the RR skirts round - from a distance appears to do so in v short but straight lengths: air photo -> 2 straight lengths. Visited 30 May 2010, terrace in 2 alignments, cut-back to west built-up on edge of ravine 4', 9yds - 10yds wide - wider than modern lane needs.

  10. Next alignment v straight and carries parish boundary 544 981. Actual road meandering at one point, but wide E overgrown boundary appears straight. Half way along terrace starts (cut out of hillside on E, built-up on W). Modern road in sunken way 4' deep which runs near W edge of terrace leaving E part undisturbed. Very overgrown but level plain of Terrace can be clearly made out .. through undergrowth .. level with remains of agger on W side - built up 5' at end of alignment. If hollow way filled in -> 40' terrace. This is 1st major structure since Two Gates. Not now since causeway found between 7 & 8. Visited 30 May 2010, not so overgrown, stranded ledge visible for whole length, 3 - 4 yards wide from half way along. Terrace 9 yards wide to W edge - likely erosion here.

  11. Next short alignment continues as 10, v clear terrace for whole length. Direction change necessary to maintain height, and terrace necessary. Major structure, 3' to 6' raised at end.

  12. At end of (11) another road joins 543 985, which widens it & makes features difficult to see + deeply sunken 10'+ in places. But whole is a terrace cut out of hillside, hilltop being 20' above. remains of undisturbed terrace clear in places, level with remains of agger on W - built-up 6' in places above ground to W. If all filled in we'd have a terrace all along line of 40'. Parish boundary follows. (This does not look so straight on air photo.) Visited 30 May 2010

  13. Last alignment curves around top of Mount Pleasant, from where the previous alignments can be seen, looking like short straight lengths of thick woodland snaking strategically across the hillside. The next alignment straight down next hill and up the next. Road down is deeply sunken and v steep at one point - but at this point a terraceway can be made out into which the modern road is sunk on its E side. It levels out the gradient at the steepest part to something reasonable - v thickly overgrown. It is cut 3' down on W, level with ground on E. - which is made up at one point. Traced for 70 yds with hollow way - if filled in -> 40' terrace 536 990.

    Undergrowth thins to W, and verges continuing a wide dry level position raised a few feet above the marshy ground to W, making short "zig" to ease the gradient down the last part of the hill. Visited 30 May 2010: arrangement seems to be: Cut-back and stranded ledge on W, sunken lane on E; when ground become v steep stranded ledge degenerates in a hollow, but a broad stranded ledge appears on E of lane built-up on E - RR has shifted to E, one can postulate a Zig connecting the two, but could not now be seen. The build-up on E become huge towards bottom of valley - c 10' - and this line of build-up continues on other valley side Straight ahead almost on the same line (but for this slight displacement) a line of trees/hedgerows climbs the hill again, the modern road taking a detour to the W and rejoining it at the top of the hill 530 995.

    It appears to be the remains of an agger terrace built up a few feet for a while on the W, and then becoming 6' or more on E - one of the few sections not still in use. The possible remains of a causeway takes over from the modern road at the bottom - 4' - 5' higher over low ground crossing a small stream - but not very wide - but goes straight across the stream. A ditch has been cut down much of the agger terrace (its a field boundary), in places a band of packed small limestones can be seen near the bottom, well below the surface. Whole is v overgrown but impressive in places - agger terrace being built up to provide characteristic uniform gradients. Field to W is under cultivation - and no other dimension visible - but a wide roadway would have existed. At top a ruined building exists on the line - some form of cutting would probably have been necessary here (possibly indications exist). but this later work probably obscures it. This whole line is shown on OS 1" as fine dots (possibly indicating boundary of rough pasture which exists on E. Detour may have been caused by failure of bridge over stream, or its causing a dam. [Possible excavation site to find blocked culvert, stream makes detour at crossing, and if it had become a dam it would quite likely broken through elsewhere - and also investigate the structure of the agger terrace, and compare it with that on Stoke Knapp - 533 993]. Visited 30 May 2010. The stream has been dug out under RR with a deep trench with several pottery pipes, some collapsed, taking the stream under modern lane and part of RR, dating from Victorian times - so excavation here not worth while. The huge E build-up on either side of trench continues up N valley side. Part way up several building terraces have been cut (buildings now abandoned) - cleaning up the N cut-back of 1 Terrace may provide RR section - NEED TO MEASURE LIKELY DEPTH. Ditch seen in 1970s silted up, and no stones seen. 14 Sept 10: excavation would go equivalent of 80 - 100cms - should be sufficient. To go outside any 84' side ditch would need to clean up 24m of a housing terrace cutback. Possibly 12 people including drawing - 2 or 3 day weekend.

  14. Short alignment over hill top (Warren Hill), generous verges. On E side runs a broad low agger, clearly visible in short grass, c 1' high on downhill side. Gives flat surface ~ 40' between it and modern road. Clear in air photo and also in previous fields which had corn growing. This turns into a modern road on probable remains of a terrace - way down hillside, made up substantially on E. Visited 30 May 2010 - could not find agger - but only visited southern part - where excavation carried out 2000, finding nothing. NEED TO VIEW WHOLE and find terrace features above - entry implies visible in air photo in field viewed.Visited 14 September 2010 - need to go along modern lane until it starts to decend when the agger seen closing with the lane, then it becomes a terrace, then it appears to W of lane 40' wide - probable Zig Zag to ease gradient. The above air photo indicates the agger starts well into the field - excavation was too close to modern lane.

  15. The modern road diverges at the bottom, but the line can be traced as a terrace way down the side of a deep depression, and then a fine terrace way of 40' wide takes the Road round the side of the depression. This well preserved in woodland, and is cut out 20' above both the hilltop and depression floor. (527 000)

  16. The modern road then cuts back into it, and continues on the terrace (but sunken) in several straight lengths round depression and a natural saddle to next assent.( Original postulated route branched off to west to take the line S of Beaminster - no evidence found on this route.)

  17. Up the next assent is wider than the road, concealed to E by thick vegetation - build-up 4'. Then it turns and becomes a cutting ~ 5' deep on E - the remains of a small Zig Zag which the modern road cuts through at expense of a steeper gradient. A parish boundary follows E course of Zig Zag (524 004). As in the investigation at other end of road, its apparent it continues around the rim of the hills

  18. At top, same characteristics of remains of agger, large Cut-back to N, build-up to same level on ground on other side of road to S - to give a broad agger terrace. In places this N banks is 18 - 20' wide, 4' high & flat on top and its N boundary makes several straight alignments along side of hill - modern road meanders in its hollow way.

    At 517 007 a parish boundary joins possible line, going in 2 straight lengths - leaving a widening over grown gap between the modern road for half a mile - difficult to explore. Appears to be a wide level surface bounded by a bank with indications of a hollow way which becomes deep at the end. The road is placing itself to start to skirt round steepish valley. The bank builds up to 3' high, 15' wide, and with hollow way filled in would be remains of agger terrace 45' wide.

  19. A track (from modern road) joins the route again as it swings round from the NW on a general line which is held for over a mile. The first section of 1/4 mile is climbing to summit of downs encircling Beaminster, carrying parish boundary, and goes in 3 identifiable alignments and a curving section which appears to keep the road's surface nearly "flat" in the direction of motion. It is now a little used track/footpath overgrown with summer vegetation but passable; sunken nearly all the way, 40' verges at first nearly all taken up with sunken path, remnants of a bank on lower NE side, appears to be built-up 3' in places in sympathy with contours - and level with ground on SW - again suggesting remains of agger terrace 40' wide. It reaches summit of downs exactly on B 3163 509 013. (Dirty Gate)

    A made up road continues the general line with ?NE hedge removed, fairly wide verges. But on ground un-necessarily twisting - its objectives would be reached by a straight line (continuing alignment we have just left) running a little ?NE of present road in field on flat downland: possible of levelled platform suggests air photo could be useful. Air photo: the made up track continuing the alignment just left. Change of direction would then be required, and seems that farm has been built on this alignment, so moving present road slightly. Possible indications that modern track runs beside original, and rest of this general alignment continues satisfactorily. 2010 Note: Roman Roads coming down to us generally warped, especially when not anchored to engineering.

  20. Much the same is true for next alignment. Curving section is a wavy hedgerow, then slightly meandering cart track within 60' verges in places, then just a hedgerow now - which builds up to a flat bank where the ground begins to slope and it curves round to join the modern road. Needs to be a curving section or another alignment between 19 & 20 but it meanders. Neither can be identified as a straight section for the first part of 20. Air photo does not help, though probably meandering can be contained within 60'

  21. Main road nearly due west (Buckham Down), v straight (but several individual alignments, mostly wide verged. Not much evidence of agger - but although on edge of hills, little need for build-up. Joins up with section explored from other end [Axminster, Stoke Knapp, Horn Hill].

There is a very narrow bridge of high land connecting Horn Hill with Buckham Down, completing the circuit of the downs. (The A3066 tunnels underneath at its narrowest point.) The Roman Road crosses this bridge of land in 3 alignments from the summit of Horn Hill, which are clearly straight when viewed from outside their verges, which are for the most part 62' - 64' wide. These also form clear agger terraces, built-up 5' - 6' to the S in the first 2 alignments, and to the N in the last. The verges are v overgrown, but the 40' roadway is clear on investigation. The present roadway is narrow, fairly well made up track, which meanders a little within the 40' roadway, and probably outside it (to the S) in the last alignment. It is not generally significantly sunken all along here.

Just over the A3066 tunnel a section of the road to the N has been washed away (468 033), by what appears to be water draining off the present track just to the S. This has uncovered a carefully made dry stone culvert, covered over at the top with slabs of stone, and on the bottom. It is about 1/2 full of earth, but continues back as far as one can reach with a stick. It was found on the far extremity of the other side of the road, with water running into it. It was partly uncovered, or broken into, and appeared to end there. On a wet day it was found to be working, with water seeping through channels in the earth. Removal of earth indicated it was stone lined at the bottom, and its construction was as in the section below.

It was 1.5 - 2' below the ground, and where it is now visible is not a road but waste ground, so it is difficult to see it being a recent construction (although a few ft in front of its present opening a brick wall has been constructed to prevent further erosion. It looks very much like a Roman drainage culvert, partially still working (see above). This culvert was found again in 2010 by the Dorset Roman Occupation Group, but in a much more damaged state.

The 3rd alignment then merges with the modern road (branching off from A3066), continuing a series of straight alignments across Buckham Down (2010 note: Alignments 18 - 21 from Eggardon), and joins up with the route found in the second part of this investigation, leading back over the downs to Eggardon Camp (2010 note: This second part of the 1970s investigation is transcribed first in this reproduction of the 1970s field notes).


Sept 78 Stone blocked up drain, perfect channel, walls straight and true, to other side of road - half filled with mud

Axminster - Stoke Knapp - Horn Hill

The following points were observed (May 1973):

A
The modern B roads from A30 near Axminster, B3165 (to Marshwood), B3164 (to Broadwindsor) appeared to follow a carefully chosen course, to run in straight lengths, with parish and country boundaries following some of their lengths. Traversing this route confirms the straightness at the beginning, with quite wide verges, and one is impressed by the lack of hill climbing having regard to the nature of the country. There does not appear to be many unnecessary bends except as described later.

Another striking feature is that the road is necessarily terraced out of the hillside before Marshwood, and around Pillsdon Pen. But the terraceway is consistently much wider than the width or needs of the present road - well over twice as wide for much of the way. (This is only obviously visible where car parks now exist, elsewhere it is heavily overgrown.)

B3165 joins the A30 just before the first definite Roman Road remains on Margary's route though Bridport since Eggardon - a 40' terrace perfectly preserved.

Older terrace can be traced all the way round Lamberts Castle Hill SY370 990, just above the modern road -> 40'. Probably the road got sunken then widened by removing the resulting bank.

Below Pilsden Pen, ST419 017, B3164 is 15' wide, terrace continues to 33', an earlier terrace a few ft below takes it to 40', which is over 6' above the fields. Modern road veers off the terrace leaving 50yds of it overgrown. Where it is not terraced on the region of Pilsden Pen it can be windy, but in almost every case considerable evidence of encroachments - narrow fields or houses exist on either side which when taken into account form a 60' boundary with the road meandering inside it.

B
The recently found Roman settlement at Stoke Abbot is on the course originally plotted. Exeter in Roman Times, E Fox (University of Exeter) shows this a fort on Waddon Hill ST446 015. Therefore decided to explore our proposed route where it leaves B3164 to climb Burstock Down ST424 017. A parish boundary has been following the B3164 - or just slightly to the S for the last stage, and continues up our proposed route which is a track.

There was nothing very special about this track, it is quite narrow but straight & slightly sunken, a bit steeper than the road previously - but not unreasonable. Noticed near top of climb that a narrow field to S follows the lane, and closes with it at the top.

At the top the character changes. Almost a total blaze of bluebells. The route followed exactly the edge of a ridge which was very narrow, and appeared to be a pathway (10 - 15ft) between two massive banks 4 - 5ft or more above the down. The pathway was sunk - usually the banks were several ft above ones head. Sometimes the top of a bank appeared level with the downs on one side, but not on the other.

This continued along the ridge and into the wood, along the side of it, but now with only 1 bank (the edge of the wood), but while this was still massive there were few other features.

Then it appeared that a fairly large scale cutting/terrace had been made, shortly after which the present track veered to the left (between 2 smaller banks) and on the right (S) of it there appeared what seemed to be an enormous agger - 35 - 40' wide, 100yds long, lying clear of vegetation in the wood. It seemed so large as both to be un-noticeable and unreal. Its height in places was 3 - 4' or more, and gently cambered. An obvious cut wall lay to its right (S) in places, with apparently ditches on either side which were either full of water or marshy. These spilled water across the road at one point making it impassable - but it continues on for another 100yds through the wood - now some way off the present track.

A channel had been cut through the agger near where we first saw it, in an attempt to drain the swampy ditch. A clear layer of medium sized stones an inch or two below the earth could be seen stretching right across the mound. At other points in the agger the metalling could be seen on the surface. Since the land hereabouts is very free of stones, there seemed little doubt that we were looking at a Roman agger, of not just a small side road thro' the hills, but a massive main road. The possibility of the main SW route continuing round the hills was revived. Visited June 2011, this ditch found - silted up (and another more to east). Ground fairly clear, this huge agger visible for long distance, becoming high above the track. Probably hollow way making it look like 2 aggers before becoming 1 huge agger again as enters bushes or small trees.

The OS map shows the route continuing down the hill in a straight line. The agger appears to turn (& terrace/cutting ends) at the required point, but the line is cut by modern banks & trackways, and is confused. [Later addition, June 73: Line continues back, first as bank and hedge, then as remnants of ditch & bank heavilly metalled, with traces of metalled agger all the way to the continuous section of agger. The parish boundary which cuts the line at this point itself changes direction here.] June 2011 - could find these remains - but did not locate parish boundary change of direction. July 11 - found this parish boundary change - the line of track from Stoke Knap & line of agger meet here. The way down the hill is joined again by the modern track - but is extremely sunken (10ft deep), muddy + running water. However again banks appear on the sides raised about field level in places - and even appear to accurately smooth the gradient where the natural contours change steepness sharply, or build up on one side if ground sloping. If we imagined the whole sunken way filled up to the level of the banks on either side we would obtain a massive agger/embankment of similar width as previously seen, which would smooth out the steepness of the assent. In short, these banks appear to be the remnants of a large massive agger 4' high, 40' wide which has been worn away in the middle. Map shows the line of this track continuing across B3162. Where track forks there is a huge build-up to S. To E of B 3162 buildings built on the terrace with hug build-up continuing to S - can see the S edge rising steeply to W where it would have gone before the B3162 cut its hollow way through it. Likely to be spur to Roman Fort.

Retracing our steps back the way we had come, past the agger through the wood, to the section where the route runs by the edge of the wood. Here we could now recognise something of a sunken way, now overgrown (the present path goes above it). If we imagines this filled in up to the level of the massive bank by the edge of the wood, then one would obtain a terrace way of generous proportions as we've seen on the modern roads. Coming to the ridge of the hill, if imagined the space between the 2 banks being filled, we would obtain a massive agger 4 - 5 ft high, 30 - 40' wide, and accurately altering its shape to the level of the ground.

Not much appears now to be visible going down the hill we first came up. The supposition is that it ran in the narrow field to the S, the two closing it at the top, which appears to be the actual course taken by parish boundary. Possible indications of metalling appear where the road to Wall Farm cuts deeply through the line just before the junction with B3164 (ST 424 017).

[June 1973 Route down Burstock Down to ST444 015]

The description 3 paras above continues all the way down to near the bottom. Traces of mettalling can be made out at some points. At one point the bank is 5' from the level of the field on W side (roadway is ~ 15' from bank's top) and here the foundations stones can be made out over a stretch of several yards - 2.5' of limestone blocks, 9' at the bottom with smaller stones on top with 1ft of earth on top of that. At the bottom it has reached a saddle (at Stoke Knap ST444 015).

We were unable to find traces of it on our proposed route (to the E). The actual route was far too steep; the road down to Stoke Abbot was a possible - its deeply sunken and no traces of agger or other features could be seen. Other routes which could achieve the same course N & S of Waddon & Gerrard's Hills) drew a blank.

C
We then investigated the footpath leading from Stoke Knap (where the decent from Burstock Down ends) to Clanden Hill (ST448 023). Although this was going in nearly the opposite direction from our proposed route, substantial remains were found. We approached it from Clanden Hill (off the B3163), where immediately a substantial terrace way, 40ft wide running for 100 yds down the hillside. The lands have probably been ploughed, and much stone scattered in the grass on the route, but not elsewhere in the field. Near the base of the hill it peters out, perhaps because of the ploughing (but still stony all along). But ahead, a very straight mark can be seen in the next field, which on examination appears to be a ploughed out terrace, for a flatish section near the end of the field appears to line up with a hedge in the next field. This is a level part at the base of the hill by the stream, but it appears to be built up slightly into a low broad agger. The hedgerow of the next field takes over, running in two straight lengths (the footpath runs to the W of the route probably - a broad level below existing on the other side of the hedge). The last field boundary on this footpath then takes over, with possibly a broad low agger/terrace to the E of it. This brings us to the farm yard of Stoke Knap Farm, which is on the line of the final short section of the lane down Burstock Down.

Visited June 2011. Very clear driving E along B3163. There are 5 fields from B3163 to Stoke Knap Farm. Field 1 by B3163: Terrace clear on the terrace as a Pure Terrace, from which 4 parallel lines visible in Field 2, which in Field 2 can be identified as Nos 1, 2, 3, 4 - start of build up, level way, base of cut-back, end of cut-back without optical assistance. The Foot Path meanders between them. Continues into Field 3 with optical assistance, and clear from high above its SW corner. No trace in Field 4, but plough has brought down a lot of earth -> its NW corner high above Field 3. 1970s Field Notes suggest W field boundary is the RR and FP goes to its W. Field 5 by Stoke Knap Farm: Agger Terrace comes out of farmyard, built up 3' to W, running near W field boundary - pointing to W boundary of Field 4 (likely RR). A direct line from B3163 would involve a steep decent and climb due to a cross valley - course taken avoids these climbs.

[Addition July 73: Just up Waddon Hill from the farm yard is the recently found Roman settlement - shown on the Exeter University publication as a fort. Possibly the reason for the 40' roadway linking it with London and the Fosse Way - the 40' roadway is found on this (and London) route just before (but not after) the Foss Way.]

D
This major feature, the terrace on Clandon Hill, is unlikely to be modern, although the map (OS 1" : 1 mile 1946) shows a road on it (its not a road now), and 40' seems too wide and elaborate for a farm road. Also the continuation of the terrace can be traced on the other side of the B3163, becoming a cart track with wide verges and massive hedges - 60' total verges.

The assumption is that the road continued round the edge of the hills, for which modern roads/tracks exist, rather than making a short 'dash' for it across low land. This last footpath is going across a bridge in the hills - probably the only one which exists, and it does not drop below 500ft contour.

July 1973
The terrace on the other side of the B3163 looks substantial from the road, but very overgrown. But investigation shows it clearly on either side of the sunken track over 40' wide, built up 4' above the fields on W, and sculptured out 10ft on E. This terrace shows the features seen so often further on: a sunken way bounded on either side by wide level portions of the old road. (This can often be concealed by dense vegetation, but shows as flat platforms above the modern way, each level with the other, and the whole may be engineered out of the hillside. These indicate that the original route to be much more elaborate than the present one, which has contracted to a sunken way perhaps occupying a small part of the whole. Again, filling in the sunken way - perhaps clearing the vegetation - would prove 40' roadway, for the most part of gentle gradients.)

Possible indications of metalling in the bank of sunken way, and at a point where drainage ditch has been cut thro' - small - medium sized limestones. The terrace develops into an agger terrace running along the side of the hill for 1/4 mile. [?where? - this track from B3163 is only 1/4 mile long B4 it turns to NW - this dealt with 2 paras on 1/4 mile estimate must be wrong] At first it is a wide drove way, rutted and slightly sunken, with dense hedge to the west, where the roadway is built up 3 - 4 ft. At various places metalling of lightly packed limestones outcrops along the track, which may be original.

The original terrace runs in several short but clearly straight alignments. On the last the present path becomes narrower, and the broad hedge to the W is a flat platform, which continues also on the E of the path - again forming a 40' roadway built-up 3ft on the W. (The path meanders somewhat, and the width of the platforms on either side expand and contract accordingly.)

It is postulated that the Roman road is the modern series of roads and tracks on the edge of the downs, which encircle Beaminister. But at the point we have just reached the track takes a deviation which appears unnecessary. At this point the path and its verges begin to contract until the whole is very narrow and featureless. It meets the road/track at the bottom of a short hill, at which point that road's verges expand to 40 - 60ft, and it appears built up in places. At the top of the hill there is a terrace quite alot wider than the present road; it runs for 50yds but curves. This point could however be reach directly from the point where the last Roman features were seen, turning to the E & going across a relatively level ridgeway. Although no obvious remains could be seen on this route, it is the more likely route, and little modification to the ground would have been necessary. Air photo: show crop mark avoiding detour. Not conclusive, but possible indication. At this point (the curving terrace) some easing of the gradient would have been necessary, which may be the origin of this terraceway. July 2011 Surveyed from B3163. The next item thought to be the causeway remembered but not recorded.

Turn 455 029. This leads onto a very clear agger/terrace running st for 50m yds; built up on the N to 4 - 5 ft to match the contours of the ground, and running in undergrowth to S of the present road, to give 40' width. Several short alignments 1/3 mile, running almost due E - W. Towards the end the modern road becomes sunken, but both platforms on either side begin to rise up to take the next hillock.

The side of this hillock is sliced off by a short terrace, the greater part of its flat surface runs in undergrowth to the N, very clear once in undergrowth, giving 40' terrace. The line continues for 100yds as an agger over a depression which it renders level: builds up to 4 - 5 ft on both sides, 40' wide. Modern track runs on top, not much sunken, much of agger obscured by in undergrowth to N.

Slight change of direction as assent of Horn Hill begins in earnest, road wide verged to start with but becomes sunken - indications of agger terrace built up 3' or more on S where now escarpment falls away steeply. Then where gradient increases over short distance indications of a cutting with portion of original roadway raised up to N (cut out 5') and also probably a slight Zig-Zag. This merges into the final 100yds section along this line, and while now deeply sunken (10') a broad 'level' way exists above the present track to N (20ft wide) which is on the same level as the 3' bank to S. This again would provide a 40' roadway, which is going along the top of a narrow spur.

Where this meets the main body of Horn Hill (462 029) present track turns sharply NE. It becomes extremely sunken, but could be remains of a terraceway, remains of which appear about half way along to the W as a raised platform 18' wide, built up 5' above the field. This build up appears to follow the contours; filling in the hollow way would again produce a generous roadway, but being 15' deep at the end it has destroyed many of the features.

August 1973
Track now turns at rt Ls climbing steeply to summit of Horn Hill, but having regard to the fact that the original road would not have been sunken at the turn the original gradient would haven been much greater.

There is a very narrow bridge of high land connecting Horn Hill with Buckham Down, completing the circuit of the downs. (The A3066 tunnels underneath at its narrowest point.) The Roman Road crosses this bridge of land in 3 alignments from the summit of Horn Hill, which are clearly straight when viewed from outside their verges, which are for the most part 62' - 64' wide. These also form clear agger terraces, built-up 5' - 6' to the S in the first 2 alignments, and to the N in the last. The verges are v overgrown, but the 40' roadway is clear on investigation. The present roadway is narrow, fairly well made up track, which meanders a little within the 40' roadway, and probably outside it (to the S) in the last alignment. It is not generally significantly sunken all along here.

Just over the A3066 tunnel a section of the road to the N has been washed away (468 033), by what appears to be water draining off the present track just to the S. This has uncovered a carefully made dry stone culvert, covered over at the top with slabs of stone, and on the bottom. It is about 1/2 full of earth, but continues back as far as one can reach with a stick. It was found on the far extremity of the other side of the road, with water running into it. It was partly uncovered, or broken into, and appeared to end there. On a wet day it was found to be working, with water seeping through channels in the earth. Removal of earth indicated it was stone lined at the bottom, and its construction was as in the section below.

It was 1.5 - 2' below the ground, and where it is now visible is not a road but waste ground, so it is difficult to see it being a recent construction (although a few ft in front of its present opening a brick wall has been constructed to prevent further erosion. It looks very much like a Roman drainage culvert, partially still working (see above). This culvert was found again in 2010 by the Dorset Roman Occupation Group, but in a much more damaged state.

The 3rd alignment then merges with the modern road (branching off from A3066), continuing a series of straight alignments across Buckham Down (2010 note: Alignments 18 - 21 from Eggardon), and joins up with the route found in the second part of this investigation, leading back over the downs to Eggardon Camp (2010 note: This second part of the 1970s investigation is transcribed first in this reproduction of the 1970s field notes).


Sept 78 Stone blocked up drain, perfect channel, walls straight and true, to other side of road - half filled with mud