Skip to main content area


Cookie settings
Main Content Area

Buckton Fort

Excavations at Buckton Roman Fort, 1959

(Information taken from S.C. Stanford, "The Roman Fort at Buckton, Herefordshire - Excavations, 1959", Transactions of the Woolhope Naturalists' Field Club, Volume XXXVI, 1959, pp. 210-218)

In the summer of 1959 (during a drought), Mr. Arnold Baker observed and photographed a new Roman fort and temporary camp at Buckton Park, one mile west of the Roman town of Bravonium(Leintwardine). The sites are on level ground just above the flood plain of the Teme, which lies to the south, and on a hard yellow shale that forms a stiff buff clay subsoil.

The plan of the fort is a rectangle measuring approximately 560ft east to west by 460ft. The defences consist of ditches and rampart fronted by a stone wall. The east and west gates are central, the south gate lies some 2/5 of the distance along the south front, measured from the south-east angle. This identifies the main north-south street as the via principalis, so that the fort faced east. Each of the three visible gates was seen to be equipped with stone-built guard-chambers and central spina for a twin-portalled gateway. Inside the rampart is a wide intervallum road. In the central range of buildings, the outline of the principia (in stone) may be observed. To the south lay a granary, so that the space to the north will thus have been occupied by the commandant's house. Each half of the praetentura is divided by two north-south streets into one narrow and two broad building plots, and there seems to have been a similar arrangement in the retentura.

The Excavations, 1959

The Woolhope Club's excavations at Leintwardine in 1959 showed that the permanent fort built here soon after AD 150 remained the local military establishment until well into the 4th century. Until further work has been done it is impossible to the certain of the earlier history of this site, but the pottery finds suggest that it was also occupied in the late first century AD and then relinquished for the first half of the 2nd century. The discovery of the Buckton fort suggested that it may be the missing early 2nd century fort, and possibly too the 1st century precursor.

Summary of Results:

A 3ft trench was cut across the eastern defences and extended 3yds into the building area beyond the intervallum road. This limited excavation showed that, at this point at least, the defences are all a single period work associated with occupation between c. AD 120 and c. AD 160. The rampart wall and buildings were deliberately dismantled in Roman times, presumably c. AD 160 when a new fort was built at Leintwardine.

The following features were revealed during excavation:

  • A ditch 15ft wide and 4ft deep.
  • The foundations of a stone rampart retaining wall 4ft 10in wide and 1ft 6in - 2ft 9in deep.
  • The base of a clay and turf rampart bank 17ft wide and 6-10 in thick.
  • A marker slot 1ft wide and 9in deep.
  • A gravel intervallum road 24ft wide and 9in thick.
  • A drain 3ft wide and 1ft 6in deep.
  • The position of the robbed wall of a (?) timber building.

Details of the section:

The ditch, cut in the hard natural shale, has a steep outer slope and a gentle inner one reaching right up to the wall. This inner slope, if projected, meets the top inner edge of the wall trench, and the difference in depth between outer and inner sides of this trench indicates that the ditch was originally cut from the line later used as the inner side of the wall trench. On the sloping surface the latter trench was dug. Trampled into the natural shale and clay subsoil are the masons' chippings scattered when the wall was built. These extend down to the inner slope of the ditch and into its bottom, proving that the wall and the ditch are contemporary.

The wall trench was dug on the inner slope of the ditch. In this were set the roughly-coursed foundations, bound with dumps of mortar and concrete. Although the stone used on the inner side was not dressed it was closely packed, completely filling the trench. At the front, well-mortared squared blocks were used, but it would seem likely that the single course exposed is the base of the dressed section of the wall. With the exception of some water-worn pieces in its core the wall was built of quarried purplish-red sandstone. The source for this is Coxall Knoll, an outlier of Old Red Sandstone in a district composed mainly of Wenlock Limestone and the massive Aymestrey Limestone. The Old Red provides better building stone than either of the limestones, but since Coxall Knoll is the nearest rock outcrop it cannot be argued that it was deliberately sought out by the Romans because of its quality.

Any estimate as to the height of this wall must be based on the evidence for the bank raised behind it. The wall foundations, with the facing stones removed to Roman ground level, show that the stone was removed deliberately for re-use elsewhere. The ditch section shows a thick spill of rubble and earth sliding down from the wall into the ditch and coming to rest on top of only a foot of primary clay silting. In this rubble there was hardly any large or dressed stone. This would have been taken away when the wall was demolished, and the low position on the ditch of the residual rubble shows the demolition took place soon after the neglect of the ditch.

Behind the wall the turf had been removed from a 17ft wide strip, exposing the firm clay subsoil and forming a 6in foundation slot in which the rampart bank was built. The base of this shows thin dumps of clay between retaining walls of turf and so suggests the method of construction used for the full height of the rampart. A 4ft front wall of turf would have been built to retain the clay and shale dug from the ditch and foundation trench, and afford a vertical face against which the stone wall might have been built. At the rear a sloping turf cover would have served to retain the core of the bank and so maintain a steep slope.

In estimating the original height of this bank and, consequently, its stone revetment the sources of material available need to be considered. There is first the turf and earth upcast from the ditch and wall trench, insufficient to provide the counterscarp bank 6ft wide and 3ft high. This would only provide a relatively small bank; but the absence of an old turf line from most of the section suggests that the turf was stripped from the whole fort site. This would provide sufficient extra material from 6in turves to raise the bank 12ft above the original ground level, allowing for a 6ft rampart walk on the bank itself, plus possibly another 3ft from the wall thickness. This estimate allows the probable height of the wall to be determined as 17ft 6in externally from ground level to the parapet, 3ft 6in above the rampart walk. To the merlon top it would have been a little under 20ft.

The intervallum road began close behind the rampart, leaving only 3ft between. The road was made of 9in of rammed clean gravel laid on the bared subsoil. On each side was a shallow gutter now filled with grey silt and gravel. There was no sign of more than one period of construction.

Below the western, inner, gutter and overlain by the edge of the road was a V-shaped drain 3ft wide and 1ft 6in deep, filled with large gravel and grey silt. The careful grading of the road base shows that this drain was intended to take the run-off from under the road as well as that from the buildings to the west. On such an impermeable subsoil surface drainage would have been a major problem. This drain - cut before the road making and building had begun - indicates the thorough planning of the engineers responsible for building the fort.

The westernmost eight feet of the trench showed the start of the built-up area. Here again no turf line was present and occupation material was embedded in the top 3 inches of clay subsoil. This was covered by a 3in destruction layer containing roofing tiles, stone slabs and iron bolts, which spilled onto an irregular shallow depression running north-south. The depression would appear to mark the former line of a timber wall, but from the section available it is not possible to decide whether a sleeper beam or stone sill was removed when the building was dismantled.


There were no finds of pottery or other objects from below the rampart or road, or in the primary silt of the ditch. The Samian sherds are from the deliberate fill of the ditch and in a very soft condition. Although they would be in place in a Hadrianic-Antonine context, none is precisely datable.

The coarse ware was found unstratified in the ditch; on top of the rampart base; on the road surface; and in the destruction level in the building area. The limited pottery evidence agrees with the theory that the site was only occupied for a single, relatively brief, period.

The earliest material found here is the rustic ware. Such ware was common on late 1st century sites and appears to last until c. AD 130, so we can assume that the Buckton fort was occupied no later than this.


The Buckton fort was occupied by a garrison of up to 1,000 troops some time between AD 120 and AD 160. The 1959 excavations have produced no evidence of earlier occupation and it must be provisionally concluded that the stone-walled fort was the first on the site.

There is no evidence of any repairs to the defences before the wall was demolished and the site abandoned, and it has been shown that this was a methodical dismantling occurring soon after the neglect of the ditch, and certainly therefore in Roman times. The occasion for this must have been the transfer to the Leintwardine site on the other side of the River Clun in c.AD 160.

This raises two problems with the Buckton fort. Why should the garrison have been moved at all in the middle of the 2nd century AD, and why should this move have been associated with a change from a stone-faced rampart to a timber-laced one? It is not clear whether the move was for strategic or geographical purposes, but it is probable that the move was part of the general reorganisation of the frontier in the Central Marches - suggesting that the repercussions of the northern revolt were felt along the Welsh frontier as well. These sites may then show the military response to a situation that had been allowed to get out of control.

Differences between Buckton and Leintwardine:

Buckton, on the Welsh side of the Clun, may be characterised as a forward, aggressive, confident position, in keeping surely with the whole tenor of the Roman army's activities in Britain under Hadrian. Even its thick stone wall reflects the permanence intended by engineers with confidence in the maintenance of the status quo. The "retreat" to Leintwardine, east of the Clun, c. AD 160 anticipated the need to control the Clun ford for the safe keeping of the north-south communications along Watling Street West, and would have allowed immediate troop movement eastwards without a river crossing.

The Leintwardine site is, then, one of greater overall strength, a site for the cautious, defensive army that took on the task of restoring order and maintaining communications after the troubles of the middle 2nd century AD. This concern with communications - great enough to allow the Buckton fort to be abandoned - lends support to the idea that the forces which had given trouble were in the immediate neighbourhood of the fort; that the descendants of the hillfort communities of Brandon, Coxall Knoll, Croft Ambrey, Wapley and others remained hostile to the conquerors into the late 2nd century AD, if not right through the occupation.

The second difference is the change in defences from the stone wall at Buckton to the logs and clay at Leintwardine. One explanation may be that the Roman army were experimenting with more efficient defences and logs and clay overcame the biggest problem of stone walls - vulnerability to sapping.

[Original compiler: Miranda Greene, 2004]