CPAT logo
Cymraeg / English
Back Home
Projects index

Caerau Roman settlement


Between 2012 and 2015, the Field Services Section of the Clwyd-Powys Archaeological Trust (CPAT) have been investigating the fields around the Roman fort at Caerau, near Beulah, Powys. The work was designed to determine the extent, nature and dating of a range of sites thought to represent early activity associated with the fort, by means of excavation and geophysics. This formed part of the on-going programme of Cadw work which CPAT has been carrying out on Roman military sites in eastern Wales which is designed to enhance the management and protection of these important monuments.

The Roman fort

The fort at Caerau lies about half way between those at Llandovery and Castell Collen and around 1km south-east of Beulah in south-west Powys. Only conclusively recognised as a Roman military installation as the 1950s by Barri Jones, it has been recently described in some detail in Roman Frontiers in Wales and the Marches (2010). It lies on a low hill immediately adjacent to the main Roman road (RR623a) and overlooks the Afon Cammarch, a tributary stream of the Afon Irfon, to the north-east. The fort lies at an elevation of approximately 220m OD, in an area of rolling topography which covers a strip of land about 6km in overall width, sandwiched between the ranges of Mynydd Epynt and the Cambrian Mountains.

Plan of Caerau Fort


The location of Caerau fort in relation to other Roman sites in the vicinity and the modern settlement of Beulah, also showing areas examined by geophysics (indicated by pecked lines)


The Roman settlement

The presence of a vicus to the north-west of the fort was first recognised in 1958, and small-scale trial excavations and a ground survey were conducted by CPAT in 1990 in response to an application for scheduled monument consent. The ground survey identified the road running from the north-west gate of the fort towards the main north-south Roman route (RR623a), as well as a number of other, potentially Roman, earthworks and several linear features which were assumed to be modern field drains. To the north-east of the road a large platform was identified, cutting into the slope and measuring around 40m by 35m. Fourteen test pits were positioned either to investigate the earthwork features, or to provide data on the depth of archaeological deposits within the field as a whole. Two test pits were positioned along the road, both revealing a solid metalled surface composed of compacted fine gravel. Pottery and/or brick and tile fragments were recorded in ten of the test pits, including the rim of a mortaria which had been stamped DOCI, representing the name Doccius or Docilis, who was one of the most important potters working at Wroxeter in the period AD 100-150 (identification by Kay Hartley). Two of the three test pits located within the area of the large platform produced evidence for occupation in the form of a compacted surface, possibly a floor, and fragments of brick and tile.

In 2004, geophysical survey was carried out in the same area as the 1990 excavations and this revealed a regular layout of buildings with hearths on either side of the Roman road exiting the north-west gate of the fort, although no clear building plans could be deduced (Silvester, Hopewell and Grant 2005). Evidence of rectangular buildings was also found outside the north-east rampart of the fort, though an adjacent earthwork bank suggests the possibility that these buildings were located in an annexe to the fort, rather than in the vicus.

CPAT 3709-0102


Surface of the assumed Roman road.

A range of excavations were carried out over the area surrounding the fort by CPAT in 2012 and 2013 (Hankinson 2012, Hankinson 2014), with the intention of defining the extent of the extra-mural settlement. To this end, some forty test-pits were dug and these have clearly shown that Roman settlement was limited to the north-west and north-east sides of the fort. Most of the pits revealed little more than layers associated with occupation, although one confirmed the route of the Roman road as it ran down the slope towards the river after exiting the north-east gate of the fort. Work also included the cutting of a trench across the course of a linear earthwork which runs parallel to part of the north-west side of the fort and continues to the north-east. The excavation conclusively proved that the feature was of Roman origin and comprised a bank with a ditch on its north-west side, the two separated by a level berm, and measuring some 10.6m in overall width. Overall, the function of the earthwork appeared to be defensive, perhaps representing the defences of an annexe on the north-east side of the fort or some form of defended extra-mural settlement; it is believed that the fort was reduced in size in the Trajanic period and this may be a contemporary feature.

The results of the first season of test-pitting suggested that there were areas to the north of the known vicus and also to the north-east of the fort where geophysics additional to that carried out in 2004 could potentially add to the known picture of the settlement; these additional areas were surveyed in 2013 and 2014. Clear evidence of Roman activity was recorded to the north-east of the fort, including a strong thermo-remanent response of approximately rectangular shape, measuring about 13m north-west/south-east by 6m wide, initially thought to indicate the presence of a bath-house. Between there and the fort a building was identified, apparently defined by slots for sill-beams, which seemed to comprise a courtyard surrounded by at least eight rooms; it covered an area measuring some 27m by 26m. Possible evidence of a return for the defences of the probable annexe, in the form of a ditch, was also recorded in one test pit. The rectangular thermo-remanent anomaly was investigated by the excavation of a small trench in 2014 and is now thought most likely to represent a series of parallel Roman ovens set into the north-east-facing slope at right-angles to the contours. In contrast, the ground to the north of the known vicus revealed little evidence of Roman activity and it seems that the limits of this part of the Roman settlement have been identified.

In 2014-15, further geophysics and excavation (Hankinson 2015) was carried out by CPAT on other Roman features, located to the west of the main Roman road and unrelated to the civilian settlement. These are probably all relatively early features of Roman occupation in the locality and include a large marching camp and at least two practice camps. Geophysics at the former has now given its extent as almost 15ha and revealed a series of probable Roman ovens that appear to have been set into the inner face of the camp ditch at its southern corner. A possible additional camp was investigated by geophysics in 2014-15, comprising a disjointed series of cropmarks recorded by aerial photography; no firm conclusions were drawn but it could also form part of a field system. There seemed to be a clear association with evidence of further Roman ovens so it seems likely that these features were extant in the Roman period. Possible evidence of a larger field system on a similar alignment to the fort was also noted on aerial photography taken by RCAHMW in 2006.

Copyright RCAHMW


The 2006 RCAHMW aerial photograph of the superimposed practice camps to the west of the fort at Caerau, showing the approximately parallel marks that may define a field system of possible Roman date, from the east. Crown Copyright RCAHMW

The colour bands of the digital image have been rebalanced to highlight the archaeology .

Some 200m to the west of the fort, a sub-rectangular ditched enclosure was first recorded from aerial photographs, although it is also evident as an earthwork. It was clear on the photographs that this represented at least two camps, the earlier overlain by a later camp of the same approximate dimensions (see right). A section was excavated across the ditches in 2015 and it was actually found to comprise three superimposed practice camps, each measuring about 65m north/south by 55m east/west. A small amount of Roman material associated with the latest camp was recovered, but they remain otherwise undated.


Privacy and cookies