Skip to main content area


Cookie settings
Left Navigation
Main Content Area

New Weir

In the gardens of New Weir, a National Trust property some 6.5km west of Hereford, remains of walls and an octagonal cistern have been uncovered and provisionally identified as Roman.

The Site

New Weir is almost 1km south of the Roman town of Magna at Kenchester and is 0.5km west of the line of the Roman road which travels south from the eastern gate of Magna towards a presumed bridge crossing over the River Wye. The site is on a terrace of land, close to the Wye, at the south-eastern corner of the New Weir gardens. The terrace is 120m long and has a gentle slope from the steep river bank for about 30m, after which the ground rises steeply again. Springs occur at the base of the steep slope, above the river level.

Some 20m north-west of a modern pump house are the remains of an octagonal stepped cistern. The cistern, which was repaired early in the 20th century after its initial discovery in 1891, retains its original form and dimensions although some stones have been misplaced. The cistern is of six steps with each level shaped as an octagon, except for the base which is a single block with a 15cm hole in the middle.

Approximately 50m upstream of the cistern, two stone revetments support the river bank. The most distant one from the cistern is the best preserved, with masonry standing 4m high. The river elevation has a base of large blocks of cut stone with a well-coursed wall of smaller stones above. Similar masonry continues at right angles into the bank to the north-east, where the stonework rises higher and has a plastered surface. The large stone masonry foundation continues north-west from the main block of the revetment, and at right angles to it, an upper wall continues into the bank, and the remains of a mortar floor can be seen over 1m below the present ground level with building debris above. Traces of other walls and fragments of Roman tile and mortar are visible in the river bank immediately upstream. South-east of this upper revetment, and about 3m forward from its face, is a second, lower revetment. Only the large stone footings of this remain. Between the two revetments, in the river bank, is a mass of fallen roofing stone.

Discovery of a cistern, 1891

The spring supplying the hydraulic ram of New Weir had dried up, and a trench was excavated parallel to the river bank to seek a fresh supply. A spring was found and in following its course workmen found their work obstructed by enormous stones at a depth of between 4 and 9ft (1.2-2.7m). The stones were broken and moved until it was noticed that the stones were carefully carved and dressed and the remainder were excavated with more care.

What was revealed was a mass of masonry forming a series of steps, leading - in diminishing diameter - to a single large stone with a 15cm circular hole in the centre. When the hole was cleared numerous tesserae were discovered. The position of the hole was found to be on the course of a streamlet, the overflow of which was conducted to the river via a stone trough.

At this time a photograph and drawing of the cistern were sent to Professor Middleton at King's College, Cambridge, who replied that in his opinion, judging from the drawing, the cistern was likely to be medieval rather than Roman. 

The 1977 Excavations by Ron Shoesmith and M. G. Boulton

Ten trial holes of 1m squared were positioned along the terrace.

Trench 1: Located to the north-west of the main terrace. Below the turf were silt and gravel layers 0.7m deep and these sealed a stony layer, which in turn sealed a layer of clean alluvial silt (hill-wash). A fragment of Roman tile was found in the upper silt layer.

Trench 2: This trench was 11m south-east of Trench 1 and on the main terrace. Hill-wash material predominated and several fragments of Roman tile were also found.

Trench 3: 6m north-east of trench 2, close to the steep bank rising from the terrace. Excavated to a depth of 0.9m through hill-wash silt and gravel; no Roman material was found.

Trench 4: 14m south of trench 2 and 10m east of the upper masonry revetment. The upper levels of the trench contained much stone plus many fragments of Roman tile, several grey and white tesserae, a few sherds of Roman pottery and a piece of painted wall plaster. This stony layer was covered and filled with a grey silty material (hill-wash). About 0.8m below the surface, part of a mosaic pavement was found in the northern corner of the trench. It appeared that the mosaic was the corner of a pavement, possibly with robbed-out walls surrounding it on the south-west and the south-east. Both grey and white tesserae were used in the portion of the mosaic visible, apparently arranged in a geometric design. The pavement was laid on mortar which was also only present in the northern corner of the trench. Over 80 tesserae were still in place with 40 more loose in the overlying area, suggesting there had been little disturbance to this feature.

Trench 5: 5m north-east of trench 4 and with a ground level 1.29m above that of trench 1, this trench contained many stones, in places tumbled on top of one another, all within a grey hill-wash material. A few stones were removed, exposing further masonry. Many fragments of Roman tile were found amongst the stones. This stonework was presumably the spread of debris from a building.

Trench 6: This trench was 7m north-east of trench 5 and close to the steep bank rising from the terrace. Part of a stone cover of a water tank was found in the eastern part of the trench. The water tank was about 1m deep and 2.7m long, and was full of water. In the surrounding soil pieces of Roman tile were found.

Trench 7: This trench was 14m south-east of trench 5 at a point where river erosion has caused the terrace to become narrow. The ground surface was 1.85m above that in trench 1. This trench contained a thick layer of grey-brown silt sealing stone and river pebbles. At 0.6m deep white, black and grey tesserae and fragments of tile were discovered. At 0.8m deep many large stones were found to be sitting on a layer of clean gravel (undisturbed ground). The presence of building debris but no floor suggests that this trench lay just outside the limits of any building.

Trench 8: Situated 14m south-east of trench 7 and 18m north-west of the cistern. Modern disturbances were found to have removed most of the Roman levels. Along the north-eastern side of the trench, at a depth of 0.8m, a disused channel had been constructed from semi-circular field drains. The south-western section was also cut by modern disturbances (fragments of modern drainpipes). A narrow ridge in the centre of the trench contained some stone, Roman tile and pottery.

Trench 9: 16m south-south-east of trench 8 and 8m south-west of the cistern. The area had suffered from some erosion. Footings of a substantial wall running parallel to the river were exposed. The stones were left in situ. Overlying layers included hill-wash and loose stones (debris from stone robbing). Roman pottery, tile and tesserae were found in the upper levels but there was no definite evidence to indicate that the wall was Roman.

Trench 10: 2m north-west of the cistern. Below the topsoil was a layer of large stones, including one ashlar block, all within grey silty hill-wash. The stones and silt sealed a layer of broken stone roofing tile. Both layers contained Roman tile and some pottery, and the stone tile layer contained several chalk and sandstone tesserae. Under the roofing stone was a mixed calcareous orange layer, 20mm thick, on top of a clean white layer.

The Finds

Pottery: A total of 24 sherds were found, of which five were modern. The Roman sherds included:

  1. A rim of a shallow bowl with out-turned rim. Soft sandy oxidised fabric with brown inclusions (Oxfordshire kiln?) of 4th century date.
  2. A base of a bowl, soft sandy oxidised fabric with traces of a dark red colour (Oxfordshire kiln?) of 3rd or 4th century date.
  3. A hooked flange from a bowl in a micaceous, fairly hard sandy buff ware. Probably late 3rd or 4th century in date.

The amount of pottery found was too small to form any definite conclusions, but it can be used to suggest occupation on the site in the late 3rd and 4th centuries AD.

Tile and Brick: Fragments were found in most trenches and include fragments with wavy, combed decoration. Identified pieces include hypocaust tiles, roof tiles and box flue tiles.

Metalwork: Seven nails and one possible handle were found.

Glass: Two small fragments - both could be of Roman date.

Plaster and mortar: Several fragments of opus signinum were found with small pieces of plaster. Opus signinum is a type of Roman hydraulic concrete partly composed of crushed brick. It was used for covering walls and floors.

A sample of plaster, found in situ on the south-eastern elevation of the upper revetment, consisted of a fine gravel mortar up to 47mm thick with a thin (2mm), fine, hard lime skin on the outside. This surface had apparently been whitewashed and may have had some dark red painted decoration.

A sample of the floor exposed in the erosion of the north-western elevation of the upper buttress consisted of a fine lime mortar with some glacial pebbles, occasional angular stone fragments and charcoal flecks.

Stone - Buildings: A few very small pieces of white Bath stone with traces of a carved decoration were found.

The masonry of the upper revetment is in two parts. The lower four courses are of large, squared blocks each of which contains a lewis hole, generally 0.5-0.6m squared. The upper courses are regularly laid using well-shaped stones in courses 0.13m thick, pointed and in places covered with good quality mortar. The lower revetment is more ruined and consists only of the larger, squared blocks.

The cistern is made of carefully shaped stones, each tier forming an octagonal shape.

Stone - Tesserae: Loose tesserae were found in several trenches and a small section of mosaic in situ was uncovered. The tesserae were all grey, white or black and most were c.15mm cubes, though a few were oblong.


The terrace was occupied during part of the Roman period by buildings of some stature. The nature and method of construction indicates that the complex was built into the river bank. The buttressing suggests that the course of the river in this particular area has changed little.

The physical limits of the terrace would not allow for a large courtyard, but the complex of buildings is at least 70m long.

The Use of the Buildings

The excavations and survey show that there is a complex of rooms in the vicinity of the revetments which includes at least one mosaic. Some 50m south-east there is apparently a further building complex close to the cistern, and the tesserae found in the central hole and in the trenches indicates the presence of one or more mosaics.

The cistern or pool is perhaps part of a water shrine or nymphaeum and as such would probably have had a prominent position within the villa complex.

The remains are apparently those of a medium-sized villa. The complex is sufficiently large to have incorporated two separate residential units, presumably with at least one bath building. It is possible that the site had religious significance and that the building complex incorporated a temple or a shrine.

The buildings could also have been the home of a merchant, supplying goods which had been transported up the river Wye.

Protection work, 1991

In 1991 the Cotswold Archaeological Trust (CAT) was asked by the National Trust to assist in the development of a scheme of riverside protection for the remains at New Weir. Initial preparatory work comprised standing structure, topographic, river bed and documentary studies of the site. In July 1995, with the river at its summertime low point, CAT returned to the Weir Gardens to assist with the execution of the riverbank protection works. This involved the insertion of a natural blockstone revetment along the riverbed, to the rear of which many tonnes of soil were to be dumped, turfed over, and the damaged bank re-profiled. In the immediate vicinity of the upstanding Roman remains a blockstone apron would replace the turf to ensure complete protection of the structures.

Before any aspect of the work could take place the line of the revetment had to be cleared of collapsed Roman masonry. Some 241 blocks and fragments, some weighing nearly a tonne, were recorded over a 100m stretch of the riverbed. All were examined and recorded.

A small-scale investigatory excavation was conducted on top of the upstream buttress prior to consolidation works. This revealed the presence of a small concrete-floored room or vestibule with a blocked doorway, once opening onto a possible flight of steps leading up from the river, but now only surviving three or four rises high. The adherence of painted plaster to one of the external faces of the buttress revealed an original decoration of white stucco onto which a design of rectangular blocks had been traced to create the effect of ashlar facing, the first time this had been found in situ in Britain. If the entire complex was treated in this way the total effect must have been grandiose when viewed from the south bank or from the river itself.

No further excavation could be carried out on the terrace, but it was possible to examine the eroded river bank in detail, thus revealing further walls between the buttresses and evidence that at least part of the area was once roofed. Understanding of the lower buttress was more problematical, but clearly it originally extended further into the river than previously thought and might have functioned as a landing stage or breakwater, or possibly supported rooms above in the manner of the upper buttress.

Despite many unanswered questions about form and function, we can say that this complex of elaborate and imposing buildings was undoubtedly once very grand.

(Information from Graeme Walker, Cotswold Archaeological Trust)

[Original author: Miranda Greene, 2004]