LVENTINVM

Roman Fort & Gold Mine

Pumsaint, Cynwyl Gaeo, Dolaucothi, Dyfed

NGRef: SN6540
OSMap: LR146
Type: Fort, Aqueduct, Gold Mine.
Roads
N (10) to BREMIA (Llanio, Dyfed)
SE (8) to ALABVM (Llandovery, Dyfed)

The Roman Fort at Pumsaint

N.G.REFDIMENSIONSAREA
SN656406c.490 x 490 ft
(c.149 x 149 m)
c.5½ acres
(c.2.25 ha)

The auxiliary fort at Pumsaint guarded the nearby gold mine and workings at Dolaucothi. The almost square defences enclosed an area of just over 5½ acres (c.2.25 ha) by a rampart of turf and clay within double ditches. Seven phases of construction were found in the interior buildings, which were dated within the period c.75-150. It would appear that the fort was established c.A.D.75 and manned until about 140, when work was handed over to civilian contractors who would find plenty of skilled workers in the settlement that had grown up nearby.

There was a Roman watch tower five miles north along the road to Bremia (Llanio, Dyfed) at Careg y Bwci (SN6447), with two so-called "practice works" nearby at Pant-teg Uchaf (SN6449).

Luentinum - The 'Washing Place'

The only classical work which may possibly mention the name of the Roman gold-workings is Ptolemy's Geography, published in the early-secod century, who assigns the towns Luentinum and Moridunum (Carmarthen, Dyfed) to the Demetae tribe of south-west Wales.

"Below the peoples we have mentioned, but more toward the west are the Demetae, whose towns are: Luentinum 15*45 55°10 [and] Maridunum 15*30 54°40." (Ptolemy, Geography)

It has been suggested (by George Boon) that the place-name Luentinum, listed by Ptolemy in the territories of the Demetae, is connected with the Celtic word meaning 'washing, to wash', and this has been cleverly linked with the method of gold prospecting believed to have been carried out at the pre-Roman gold workings at Dolaucothi. Prior to the Roman period the gold was presumably collected in nuggets directly from the stream beds, and by 'panning' the local soils in these natural water-courses, but with the Romans came their superior hydrolic technology and this allowed them to channel copious water into the area from various sources as far as five miles distant to sluice the gold out of the auriferous quartz rocks, which were mined and crushed in a flurry of industrial activity powered, no doubt, by many slaves, driven, one may presume, by many whips.

The Dolaucothi Bath-House

Marked on the old Ordnance Survey maps as a "Villa", in 1831 the remains of a Roman building were discovered on the east bank of the Afon Cothi close to the confluence of the Twrch about 350 yards (c.320 m) south of Pumsaint village. This small building consisted of only two rooms, one of which measured 14 by 15 feet (c.4.3 x 4.6 m) and had a floor of tessallated brick tiles raised upon stone pillars about 2 feet (c.60 cm) high. The site yielded many flat tiles and bricks together with lime mortar and, in the words of a visitor to the site in 1845 "many ashes of wood and burnt stones were scattered among the pillars"; the same observer also reported "a brick pipe for the conveyance of water" within the structure. Although nothing now remains of the site, which was left open to visitors after excavation, is is almost certain that this building represents a Roman military bath-house fitted with a typical hypocaust heating system.

Finds from the excavation of the "villa"/bath-house included a box-flue tile measuring 19 inches long, 6½ inches wide and 5 inches deep (c.48.3 x 16.5 x 12.7 cm) with two diamond-shaped openings in the side; these tiles were used to line the walls of Roman bath-houses and conducted the hot gases from beneath the floor, up through the walls of the building to escape through holes set in the roof, thus the room was heated from below the floor and through the side walls. Various pieces of pottery were also recovered, mostly local "black-ware", but some pieces of samian ware also turned-up, the most diagnostic being a small red-ware bowl unearthed in April 1931 which was 2¼ inches (c.5.7 cm) in diameter and stamped on the inside base with the potter's stamp ADVOCISIO, easily translated as "[From the] workshops of Advocisus".¹ Other finds included oyster shells, pieces of glass, animal bones, a small vase containing the cremated remains of an animal, and "a stone palette with colour still adhering to it".

  1. The potter Advocisus worked at Lezoux in the Allier district of southern France during the Antonine period (i.e. mid-late 2nd century A.D.).

The Roman Aqueducts and Reserviors at Dolaucothi

SN664394

At least five aqueducts serve the gold-workings at Dolaucothi, all routing water from the Afon Cothi and the Afon Annell in the north-east, the longest captured water from the Cothi over five miles away as the crow flies, while the shortest routed water from the same stream less than 1½ miles distant. The main aqueduct ran from a dam on the Afon Cothi just west of Bwlch y Rhiw farm, and consisted for the most part a contour-hugging open channel cut into the bare rock to the south of the stream, but it is thought that this was supplemented in at least one place by a raised wooden trough in order to cross a narrow valley and thus to avoid a lengthy detour.

An inspection of its course by Professors Haverfield and Bosanquet in 1909 found the rock-cut channel to range between 2½ and 4¼ feet (c.76-130 cm) in width but to be almost uniformly 19 inches (c.48 cm) in depth. They also recorded a "rock-hewn reservoir" measuring 120 feet long by 30 feet broad (c.37 x 9 m), in a field above the Dolaucothi-Cynwyl Gaeo road. The rock-cut tank and another similar tank further along the aqueduct were not the main Roman reservoirs, however, this distinction being reserved for a large natural hollow, now a marsh, situated at the terminus of the Roman aqueduct to the south of the Cynwyl Gaeo road and named Melin y Milwyr or 'The Mill[pond] of the Soldiers' on the old Ordnance Survey maps.

The Main Roman Mine Workings

Most of the auriferous quartz measures in the Dolaucothi area have been removed by opencast quarrying, but this method also entails the removal of many hundred of tons of the surrounding rock in order to follow the gold-bearing quartz veins into the ground. The Roman method, it seems, was to follow the general direction of the largest veins by cutting upward-sloping tunnels into the hillside with military precision. The highest and most accomplished tunnel is named Ogof y Cawgiau or 'The Cave of the Bowls', and still bears the marks of Roman picks and chisels upon its sides and roof. The distinguished antiquarian Mr. Horace Sandars visited the mines in 1910 and his paraphrased report is recorded in the 'Carmarthen Inventory':

"Mr. Sandars also deals with the so-called "Roman Adits or Levels" which seemed to be entitled to this name by their bold and regular workmanship. The upper one (Adit A) has a height and breadth of 6 feet, cut very square; it runs straight for almost 60 yards, and then branches right and left." (RCAHMCWM V, p.29)

The Principal Finds from the Area

The Cae Garreg Lwyd ('Hoarstone Field') Hoard

A number of pieces of gold jewellery were turned up by the plough during the late-18th and early-19th centuries in Cae Garreg Lwyd ('Hoarstone Field') on the Dolaucothi estate. The existing antiquarian reports associated with these items are vague and conflicting, but whatever their history since recovery the pieces described included a number of rings, two of which were decorated with serpents, and a number of torques or armlets, one of which was decorated with a dolphin, another fashioned like a serpent. Of particular note are two massive gold chains, one 44½ inches (c.107 cm) in length and weighing in excess of two pounds (c.1.2 kg), the other about 21 inches (c.53 cm); the longer of the two possibly consisting of two conjoined lengths. One of these chains was associated with large, wheel-shaped clasp or pendant about 8½ inches (c.21 cm) in diameter with eight spokes, and the other with a crescent-shaped pendant, both icons being used extensively in the classical world as love charms and as emblems of the sun and moon. The pendant pieces, although utilising Roman workmanship and technologies, seem to be particularly Celtic in design and symbolism.

The fashioned gold items recovered from Cae Garreg Lwyd have all been dated to the 2nd-3rd century A.D. and are now held at the British Museum as part of the Payne Knight Collection.

Epigraphic Evidence from Luentinum

The only entry recorded in Volume I of the R.I.B. for the Dolaucothi area is an inscribed stone found in 1811 along the line of the Llandovery to Llanio road which reads: P CXXV "One-hundred and twenty-five paces." (RIB 406); possibly a construction record from one of the many Roman aqueducts.

Close by the find spot of the distance marker in 1836 was found an unfinished onyx seal only 1/16th of an inch (c.1.5 mm) thick, bearing an image perhaps of Meleager, although other interpretations are Diana with hound or Dionysius with panther. The piece had been attached with cement to a river pebble to facilitate cutting, but had been evidently abandoned incompleted for reasons unknown.

Other Excavations and Surveys at Luentinum

SN665403 - Evidence for the processing of auriferous quartz within a timber structure with a beaten clay floor was found during excavations in 1972 at Cynwyl Gaeo, Dolaucothi.

SN656405 - A magnetometer survey conducted across the village cricket pitch in 1999 revealed details of the east side of the Roman fort and its associated vicus settlement. The survey also revealed a substantial boundary feature where the fort defences were located, overlain by what appears to be a linear bath-house.

In 1999 and 2002 a full topographic survey of the site of the Dolaucothi gold workings was undertaken by Cambria Archaeology on behalf of the National Trust which recorded evidence of Roman water tanks, aqueduct channels and building platforms across a wide area (see address below).

Dolaucothi Gold Mines Survey 2002 - from Cambria Archaeology

References and Bibliography

See also: Inventory of the Ancient Monuments in Wales and Monmouthshire - V County of Carmarthen by the RCAHMCWM (HMSO, London) pp.25-32 & fig.32;
Britannia xxxi (2000) p.372 & fig.1 p.373;
Roman Crafts and Industries by Alan McWhirr (Shire, 1982) pp.7-10;
Britannia iv (1973) p.271;
Britannia ii (1971) pp.244/5 & fig.3;
Air Reconnaissance in Britain, 1965-1968 by J.K. St. Joseph in J.R.S. lix (1969) p.126;
The Roman Inscriptions of Britain by R.G. Collingwood and R.P. Wright (Oxford 1965).

GoTop

This page was last modified: