The Glamorgan-Gwent Archaeological Trust
Historic Environment Record

The following information has been provided under the terms and conditions of access as detailed on GGAT’s website  Copyright is reserved on all data supplied by the GGAT HER Charitable Trust. All output resulting from the use of the data must acknowledge the source as follows:-
Derived from information held by the GGAT HER Charitable Trust copyright.

The data below is intended to be used for information and research only and is not for use as part of a commercial project.  If you wish to use information derived from material held by the GGAT HER Charitable Trust for publication in printed or multimedia form or to compile resources for commercial use, prior permission must be obtained in writing. For further information or to arrange a visit to the Trust please send an enquiry form to


Primary Reference Number (PRN) : 00526w
Trust : Glamorgan Gwent
Community : Onllwyn
Unitary authority : Neath Port Talbot
NGR : SN85901073
Site Type (preferred type first) : Roman Fort
Status : Scheduled Monument

Summary :
A Roman fort with ramparts of earth, clay and timber; the internal buildings included at least one with timber sill-beams, but the interior of the fort is very poorly understood. There is no evidence that any of it was rebuilt in stone, and the pottery indicates that it was abandoned in the second quarter of the 2nd century AD.

Description :
The fort at Coelbren is situated on a low ridge overlooking the Afon Pyrddin immediately adjacent to Sarn Helen. It is almost square, measuring approximately 149x142m over the ramparts, which enclose an internal area of 2.25 hectares. The site has been continuously waterlogged since Roman times, with consequent preservation of organic material. It was excavated in for the Cambrian Archaeological Association c1906 by Col Llewelyn Morgan, who sectioned the defences on all sides and trenched the interior (Morgan 1907).

The ramparts were made of earth, clay and timber, with the composition varying along their length. On the south-west corner, south side, south-east corner, north-west corner it had been constructed over a corduroy of oak logs and planks and supported in some places on piles. A second layer of smaller birch logs was noted, less closely placed and separated from the lower layer by 0.3m of black soil which certainly contained much decomposed vegetable matter. On the north side, the timber was replaced by a layer of black vegetable matter, and in an incomplete section through the western side where the subsoil was hard clay and gravel, the only foundation noted was a band of stones 9in (0.22m) wide under the inner edge. There were two ditches, between 7ft and 15ft apart, separated from the rampart by a berm of 16ft on the south side, but narrower on the north. Upcast clean clay from the outer ditch had been spread over the turf on the outer lip. None of the sections was completely excavated because of flooding. The width of the inner ditch was 9ft-11ft (or possibly more on the north side) and the depth can be estimated at 3m; the outer ditch was 7-8ft wide. Two episodes of filling were identified, the lower silty clay or gravel, with or without organic matter, and the upper usually decayed plant material, although in the outer ditch at the west it is described as alluvial soil. The gates were not examined. At the west side a metalled causeway ran between the ends of the ditches, and is still visible outside the fort as an earthwork.

In the interior of the fort, Morgan identified timber sill-beams for a building just inside the south-east corner, a number of surfaces of stone, red brick earth (possibly to be identified as daub) or fragments of tile over organic material, as well as areas of burning and deposits of decomposed peat. He recorded some evidence for stratigraphy without attempting to understand it. Amongst his finds were slag and ironstone, as well as part-melted glass and lead, but he was unclear whether this industrial debris was connected with the Roman occupation or whether it was associated with the post-medieval pottery found on the site.

Re-examination of the pottery by P V Webster from Morgan's excavations, indicates that occupation ceased cAD130-40. This is in accordance with the structural evidence that the fort was never rebuilt in stone.
Evans 2011

Some 200m to the southeast lies the site of the later, permanent, Roman fort. The principal investigation of the site was undertaken by Col. William Lloyd Morgan at the beginning of the 20th century, whose work concentrated on the fort itself and left the environs largely undisturbed. The defences were found to comprise a substantial earth and timber rampart, beyond which were two concentric ditches; these remain visible as earthworks in all but the southeast corner of the fort. Within the defences a number of significant features were identified. A building was present in the southeast corner, the foundations of which comprised sill beams retained by large stones and driven piles; elsewhere, other floors and hearths were encountered, but the plan and organisation of the buildings to which they related was not established. Between the structures a series of gravel roads was present.
The defences were never rebuilt in stone, and there is no evidence of modification or repair. However, Morgan did postulate two phases of occupation, consisting of initial military occupation followed by an industrial phase involving the manufacture of glass and iron. The meagre dating evidence from Morgan’s excavations suggested that the fort was a Flavian foundation, and that occupation continued to the mid 2nd century. This has recently been supported by Flavian to early Antonine pottery recovered from a drainage ditch cut through the northern defences of the fort (Sell 1983).
Pearson 2002, 17

To the east of the fort is an annexe measuring 200m east-west, 130m north-south at its western end, narrowing to 100m at its eastern, and enclosing an area of approximately 2.3ha. Geophysical survey results suggest that structures may be present within the annex at Coelbren, although the extent and nature of these are uncertain (Barker and Mercer 2001).

Sources :
Barker, P.P & Mercer, E.J.F , 2001 , Coelbren Roman Fort, South Wales
Evans, E. , 2010 , Coelbren
Morgan, W L , 1907 , Report on the excavations at Coelbren , Archaeologia Cambrensis : 7 : 129-74
Pearson, A , 2002 , Roman roads and vici in Southeast Wales. GGAT report no. 2002/061
Sell, S H , 1983 , Coelbren
Yates, A. , 2001 , Coelbren Roman Fort, Duffryn Cellwen, Neath Port Talbot: Archaeological Survey
10/PM Desc Text/Smith EG/1937-8/Neath Ant Soc/pp28-30
12/MM Record Card/OS/1982/SN 81 SE 2
13/MM Record Card/OS/1986/SN 81 SE 2
14/MM Desc Text/Cadw/Burnham HG/1985/AM7
11/PM Desc Text/GGAT/Sell SH/1983/Annual Report 1982-3 pp65-66
09/MM Photo/GGAT/Lewis W/1980
08/MM Record Card/OS/1977/SN 81 SE 2
16/PM Desc Text/Owen-John HS/1990/From Antiquarianism to Archaeology
07/PM Desc Text/1961/Journal of Roman Studies volLI pp119-135
06/PM Desc Text/Williams VEN/1969/The Roman Frontier in Wales pp81-3
05/PM Desc Text/1963/Arch Camb/CXII pp43-5
04/PM Desc Text/1939/Arch Camb/XCIV pp25-8
03/PM Desc Text/Morgan WLl/1907/Arch Camb/sixth series, volVII pp129-74
02/MM Record Card/OS/1956/SN 81 SE 2
01/PM List/RCAHMW/1976/Glam Invent/no 731 p83
17/MM Desc Text/Cadw/1991/AM107
15/MM AP/RCAHMW/Musson C/1988/88-MB-112
018/Desc Text/Cadw/Full Management Report /08/11/2007/Copy in further information file.

Events :
E000446 : COELBREN ROMAN FORT, EXCAVATION, 1904-7 (year : 1904-7)
E004470 : Coelbren Roman Fort, Duffryn Cellwen, Neath Port Talbot: SUR (year : 2001)

Related PRNs : 94581, 00527w

Compiled date :

Images :

March 25, 2019, 3:26 am - HTML file produced from GGAT HER Charitable Trust Ltd. Heathfield House, Heathfield, Swansea SA1 Tel. 01792 655208; Fax 01792 474469 website:  email: Registered Charity no. 505609
Data supplied by in partnership with Local Authorities, Cadw and the partners of ENDEX © GGAT HER Charitable Trust, 2019 (and in part © Crown, 2019).