A Research Framework for the Archaeology of Wales

Southeast Wales – Early Medieval






This report is the product of one of several working parties set up to review the state of research in Wales, with the aim of establishing an archaeological research agenda. This preliminary review of SE Wales in the Early Medieval period in Wales has benefited from input from Steve Clarke (Monmouth Archaeological Society), Ray Howell (University of Wales, Newport), Jeremy Knight (formerly Inspector for SE Wales, Cadw: Welsh Historic Monuments), Alan Lane (University of Cardiff) and Mark Redknap (National Museums & Galleries of Wales).


The period under review has been defined as covering the post-Roman to pre-Norman period in SE Wales. As for the rest of Wales, the period is poorly understood, and remains heavily reliant on sparse documentary sources (for attempts to review this, see Lady Fox 1946, Edwards & Lane 1988; Arnold 2000).


In Roman times, Glamorgan and Monmouthshire belonged to the territory of the Civitas of the Silures, with its caput at Venta Silurum (Caerwent), After the end of the Roman occupation, the eastern core of the territory around the former caput became the small successor kingdom of Guenta deriving its name from the former Venta Silurum, by the normal sound change v > gu. The survival of the name suggests a measure of political continuity and two sub-Roman cemeteries, one intra-mural and one extra-mural, show that the town itself retained a measure of importance as a centre for the surrounding area.


The western part of the former civitas territory came to comprise a series of seven cantrefi (‘the seven cantrefs of Glamorgan’) which 12th-century tradition saw as named after the sons of an eponymous king Glywys (Glywysing). In the 10th century the area was renamed Morgannwg, after a king of Glamorgan, Morgan Hen, though the origin story was simply a learned construct. The cantrefi corresponded to natural geographical units and may preserve an earlier territorial reality; the cantrefi or their subdivisions the commotes provide convenient units for archaeological and landscape research. In some cases, a mother church of the commote or cantref, equivalent to an English hundredal minster, can be identified and sometimes has evidence of early origins in the form of early inscriptions and sculpture. At times, a high status secular centre can be identified to go with the ecclesiastical centre (e.g. Llandough/Dinas Powys, Llantwit Major/Llysworney, Caerwent/Portskewett), though the secular centre was usually far more fluid in form and function than its ecclesiastical equivalent. Within each unit, the pattern of churches probably mirrors that of secular settlement, and a pattern of ecclesiastical centres can be identified, ranging from churches whose early foundation is attested by the presence of pre-Norman sculptures and inscriptions pre-dating the developed parochial system of the 10th and 11th centuries, to Anglo-Norman manorial churches, often with a parish whose boundaries and form show that it was carved out of a larger, earlier unit.


These territorial units – cantref, commote and parish – provide a framework which may help to overcome the difficulties in identifying secular settlements of the period, and within which the multi-period landscape archaeology can be used to identify that of the early medieval period.


A number of historical works provide a valuable insight into early medieval Glamorgan (e.g. Glamorgan County History); the equivalent work on Monmouthshire is scheduled for publication in September 2003.


A fundamental pre-requisite for research in the Early Medieval period is the need for:


1. Greater success in site characterisation, including establishing their extent [97 definite military/settlement Roman sites are listed in the GGAT Resource audit (and 72 potential sites; total of 1119 Roman archaeological sites), compared with 12 ‘key ‘ sites and 5 uncertain for the early medieval period];


2. A more precise chronological framework;


3. A multi-disciplinary approach, with projects that may involve a number of partner institutions. In what follows, strengths, weaknesses and threats are summarised, and some future opportunities suggested.


Taken together, these should provide a foundation on which to construct a sustainable research agenda.


Settlement and earthworks



The hillfort settlement at Dinas Powys, excavated between 1954-58, was fully published in 1963, heralded a period of optimism in early medieval archaeology in Wales. The recognition of a settlement rich in artefacts, Mediterranean and Continental pottery, and evidence for metalworking and other aspects of its economy suggested that at last parallels for the rich early medieval settlement evidence from Ireland was being found in Wales. The site has been the subject of several important reassessments, and its relationship with the recently excavated cemetery at Llandough is now being assessed. Re-assessment of the post-Roman evidence for Caerwent has suggested a secular basis for much of the activity represented there, in addition to monastic activity. Offa’s Dyke, previously studied by Cyril Fox and Frank Noble, is the subject of an ongoing research programme by David Hill and Margaret Worthington, who have re-examined Buttington Tump. Re-use of caves has been well documented and published, some with radiocarbon dates.



Despite success in examining some of the more obvious sites, few new sites have been identified on the ground, and even fewer excavated: some discoveries, such as Hen Gastell (Wilkinson 1995), were the fortunate outcome of a programme of rescue excavation. There is an assumption that many sites are poor in artefacts – true for some key fieldwalking indicators such as ceramics, but not necessarily the case for metalwork. Publication has focussed on two site types - small multivallate hillforts and caves – and the extent, nature and character of documented sites remain uninvestigated. Historically interesting areas such as Flatholm have been greatly modified by post-medieval development, making the search for early medieval evidence of activity very difficult. Estate boundaries have not been archaeologically tested, and the transition with the preceding period is poorly understood. The relationship of the Dyke to contemporary settlement/landscape/communications on both sides of it is poorly understood.



The piecemeal nature of infill development, particularly in rural villages such as Llancarfan and Llandough, makes site identification difficult; site evaluations and watching briefs have variable success rates. There may be a reluctance to initiate long-term research on an area, where results are unpredictable, at best slow, and often disappointing. Many earthworks are threatened by erosion (natural, weather, animal, people), agriculture/forestry and tourism



·         Identification of settlement. As has been recognised elsewhere in Britain, the increase in metal detecting over the last 20 years has resulted in a wealth of early medieval metalwork being recognised. The reporting of find spots, now under either the Treasure Act 1996 or the Portable Antiquities Reporting Scheme, is providing evidence for the location of so-called ‘productive’ sites. Future funding for investigative fieldwork would be valuable in ‘ground-truthing’ these clusters. No high-status site dating to the 8th-11th centuries has yet been excavated in SE Wales.

·         With a rich legacy of late Roman sites in SE Wales (in comparison with other areas), there is a need to investigate the late Roman/early medieval transition – working from the known to the unknown.

·         Comparative characterisation of land use and settlement. Land units and estate boundaries and their relationship to settlement merit further investigation.

·         The location and nature of hillfort settlement – whether re-used earthworks or founded de novo – deserves more work.

·         Available evidence suggests that some early medieval sites are complex, often multi-period, with ecclesiastical associations. The examination of complex sites with a high likelihood of an early medieval component would enable investigators to place the evidence within a broader chronological and topographic pattern. A sample of sites where this approach might prove fruitful is given below. In some cases, the preliminary work of assembling the currently available evidence has been carried out, e.g. in Glamorgan County History Vol. 2. An assessment of available evidence, both documentary and archaeological, for each site complex, followed by a carefully focussed modest research programme, including limited excavation and the use of scientific geophysical and dating techniques would probably prove fruitful.


Castell Arnallt


Castell Arnallt is situated near the River Usk approximately 4 miles SE of Abergavenny. The oblong mound, over 150m long on its NS axis, has traditionally been interpreted as the llys of Seisyll ap Dyfnwal, which had been destroyed by William de Braose in 1175. Contour and geophysical surveys conducted in 2000 suggest a number of structures on the site as well as possible multivallate defences (Phillips 2000). Surface stone in some areas may indicate an inner wall or palisade base. The general appearance of the site suggests parallels with known llysoedd although the apparent defences would be more in keeping with a castle. It is possible, therefore, that Castell Arnallt represents an important transitional site perhaps best described as a fortified or developed llys. It is also possible that the origins of the llys were in the early medieval period. This is clearly a site where additional investigation including excavation could pay substantial dividends.




Roman Villa evidence in recent excavations by GGAT and others, with possible Iron Age beginnings.


Late Roman temple on Portskewett Hill (“Farthing Hill”) above villa site. Large unpublished coin collection from site from Wheeler’s excavations (in possession of the landowner) and many coins are being found by metal detectorists. Former comprise one 3rd-century radiate and large numbers of 4th-century issues.


Documentary evidence for pre-Norman llys of kings of Gwent on site of Harold Godwinson’s hunting lodge of 1063. Early Norman church with substantial adjacent earthworks (“Harold’s Palace”) which are probably the medieval manorial complex of the Deneband family, but with possible earlier elements. There are associated wetland structures nearby, including fish traps, with dates from 6th century onwards.




Extensive excavations by Stephen Clarke and the Monmouth Archaeology Society have revealed a sequence comprising a pre-Flavian Roman fort, the Roman settlement of Burrium with a strong early 4th-century phase, followed by evidence for early pre-Norman settlement associated with timber defensive structures (on the basis of stratigraphic sequences). Pre-Norman Chester ware jar known from site – only one in south Wales. Important evidence of Norman settlement, with castle and Priory church. Only brief summaries of these excavations have been published. Full publication an urgent need.


Caerwent area


History of the Roman town is being re-assessed by Richard Brewer in light of fresh excavations. Post-Roman metalwork from old excavations and intra-mural and extra-mural post-Roman cemeteries now published. Need for carefully targeted research programme to take forward the early medieval element of this.


Wetlands Edge Sites


The recent research programme on the Gwent Levels has emphasised that the period between the late Roman marine transgression and the 12th-century resettlement, the levels were exploited from wetland edge settlements. The potential of this has been shown in GGAT excavations at Pencarn Farm (Coedkernew) where a late Roman fen-edge settlement underlay a medieval site with documentary evidence for early medieval occupation associated with Bassaleg and Llancarfan in the Vita Cadoci. Bishton (Llancadwalladr) is another site with potential – a pre-Norman ecclesiastical estate with charter evidence for the site of medieval episcopal manor.




Upland complex of prehistoric and medieval sites, studied by Cyril and Aileen Fox (Forts and Farms on Margam Mountain). Thought to have early medieval element because of presence of the ‘Bodvoc’ stone, but Fox’s attempts to identify an early medieval element not successful.


Several Group I Latin- / ogam-inscribed stones in area with known findspots. Medieval Cistercian Abbey has an important group of pre-Norman sculptured crosses but no early inscriptions / cross-slabs. A fragment of early medieval beaker has been found within the park nearby. Cluster of important disused church sites in area. Eglwys Nynidd (with Group I stone) threatened by golf course extension etc (excavated by Pre-Construct Archaeology); Cae’r hen Eglwys being destroyed by ploughing.


Other sites


A range of earthworks with possible early medieval origins may provide a fertile focus for targeted survey and other investigations.  A case in point is represented by dykes including the short lengths of rampart found on many upland ridges in Glamorgan.  Some sites in this category have ramparts forming a possible defensive line including Tor Clawdd, Ffos Toncenglau, Bwlch Garw, and Bwlch y Clwydd.  Dry stone walling or revetting are incorporated in the dyke at Ffos Toncenglau and Bwlch y Clwydd. (RCAHM 1976, I, iii, 6)  With notable exceptions such as work by Hill and Worthington on Offa’s Dyke (Hill & Worthington 2002), fewer sites have been investigated in Gwent where there is considerable scope for new initiatives.  Place names such as Pen y Clawdd, while possibly ‘head of the ridge’ rather than ‘head of the dyke’, may provide a useful starting point for new research.


In many cases, earthworks may have been associated with ecclesiastical sites.  There is an assumption that early monastic sites, accommodating members of the claswyr, would have been enclosed by a bank or wall, usually curvilinear in plan. Shallow ditches may also have been associated, at least on some sites. (RCAHM, 13)  Similar smaller scale enclosures are associated with many llan churches and references in Liber Landavensis, the Book of Llandaff, may provide a useful starting point for possible field survey.  In some instances earthworks may also be associated with domestic occupation sites.  A rubble bank at Thornwell Farm near Chepstow  (Hughes 1992, 8-13) and a shallow rubble bank at Caerwent Quarry (Vyner & Allen 1980, 67 - 122), while presumably late Roman in date, are suggestive.  An interesting possibility is that shallow banks may define some early llysoedd sites. For example, the tradition which places a llys of Caradoc ap Ynyr near Portskewett (Vita Sanctorum, 270 - 287) and pronounced field marks at the site, encourage archaeological investigation there.


Earthworks associated with re-occupied hillforts may also offer productive areas for new research.  Recent excavation at Lodge Wood Camp north-west of Caerleon, for example, revealed a series of narrow terraces with small post holes and a limited amount of late Roman pottery in an oval enclosure in the interior of the Iron Age hillfort.  The entrance had been re-cut during a materially impoverished phase.  A trench through the inner bank and ditch demonstrated a stone-revetted timber-laced rampart with a large collapse deposit of stone rubble.  Perhaps significantly, a later phase survived as a thin rubble spread overlying soil which had formed after the collapse of the primary rampart (Howell & Pollard 2000, 97-98) Given evidence of late Romano-British activity at several hillfort sites in Glamorgan (RCAHM 1976, I, ii, 8) and Romano-British pottery at other Gwent sites such as Llanmelin (Nash-Williams 1933, 237-315) and Sudbrook (Nash-Williams 1939, 42-79), additional targeted investigation of hillfort sites could shed considerable light on early medieval origins. 


Ecclesiastical Sites and Cemeteries



Recent excavations at the Atlantic Trading Estate, Llandough and at Llanelen (Gower) have shed more light on the nature of early ecclesiastical centres and cemeteries. A major step forward has been the reassessment of the evidence within the Llandaff Charters (Davies 1979), research into specific fields (pulled together in Edwards & Lane 1992) and models for the origin and initial development of early medieval ecclesiastical centres. The corpus of early medieval inscribed stones and stone sculpture published by Nash-Williams in 1950 is being revised. This is a multidisciplinary study (a collaboration between the National Museums & Galleries of Wales, the Universities of Bangor and Aberystwyth, and RCAHMW) which will include details where available of the original locations of many stones, geology, epigraphic, linguistic and stylistic analysis. It is hoped that it will form an important tool in future research.



There is no identified pre-Norman church fabric in SE Wales, and the early phases of many documented sites are poorly understood. While the early medieval inscribed stones and stone sculpture are individually and collectively important, few remain in situ, and the contexts of those that do remain poorly understood. Skeletal data from early excavations such as Caerwent are not published in detail; bone does not survive well in some areas; chronology remains a problem, with difficulties in establishing reliably dated burial sequences and horizontal phasing. A low number of radiocarbon dates for some sites. As for other periods, some excavations have yet to be fully published (e.g. Burry Holms), while other sites would merit further study.



As with settlement, the small-scale nature of much work undertaken limits the results; a recent problem has been insufficient funding during the post-excavation stage, especially if the excavation has been developer funded. Increasing church redundancy and limited funding poses threats to both church sites and monuments. Early medieval inscribed stones at many sites are under threat of damage (natural or human), and lack protection. Resources will be required to implement the recommendations of the newly established National Committee for the Recording and Protection of Early Medieval Inscribed Stones and Stone Sculpture. This initiative could help to direct further research in this field.



·         The work of the Early Medieval Ecclesiastical Sites Working Group set up by Cadw: Welsh Historic Monuments and the review by GGAT of ecclesiastical sites and their morphology in SE Wales, will focus ideas for future management plans and related research.

·         The development, role and function of ecclesiastical sites, and their wider context.

·         The demography of early medieval Britain. The retention of the complete skeletal collection from Llandough (over 850 burials) by NMGW will provide a valuable ‘key-site’ assemblage to support future research.

·         As in the case of settlement, programmes of aerial photography, geophysical surveying and trial excavation could lead to further site identification (e.g. Llangan)


As with settlement, carefully focussed modest research programmes on complex sites, including limited excavation and the use of scientific geophysical and dating techniques would probably prove fruitful:


Llantwit Major


Roman Villa with undated post-Roman cemetery cut into ruins. Radiocarbon determinations urgently needed.


Major monastery of St Illtyd, presumably on the site of the early church. Abbots recorded from 7th century onwards; series of crosses and inscriptions referring to abbots and kings of Glamorgan. Extensive earthwork complex and buildings around church, belonging to the medieval grange of Tewkesbury Abbey. Partly excavated by V. E. Nash-Williams.


Villa and Monastery set in extremely large medieval parish on excellent corn land with good microclimate. Large medieval monastic estate (and possible Roman villa estate?) requires investigation. Llysworney, within parish, identified as llys of cantref of Gorfynydd. Possibility of paired early medieval religious/secular high status settlements, as Dinas Powys / Llandough.


Documentary evidence (Vita Illtuti) of refortification of Castle Ditches hillfort during a Welsh attack on Anglo-Norman settlement (Glamorgan County History Vol. 2).




Major early monastery of St Cadoc and later manorial centre of St Peter’s Abbey, Gloucester. Exceptionally rich documentary evidence – two saints lives, that in British Library MS Vespasian A XIV accompanied by charters and detailed account of pre-Conquest / early post-Conquest settlement and land use in large parish. Much other material e.g. in Gloucester Cartulary.


Monastery possibly scattered along valley in same manner as, for example, Glendalough in Ireland, rather than within a nucleated enclosure. The whole valley bottom is presently disappearing under housing. Traces of earthworks in vicinity of church are as yet uninvestigated. The whole area around the church has exceptionally high potential for geophysics. During the only archaeological investigation under the car park of the Fox and Hounds Public House next to church, an aceramic dark layer was observed by Howard Thomas. The church itself is of considerable interest. (Glamorgan County History Vol. 2)


Burry Holms / Llanmadoc / West Gower


Medieval extra-parochial area in West Gower. Very rich archaeological sequence from prehistory onwards. Major early ecclesiastical site associated with Llanmadoc (group I Latin-inscribed memorial stone and two Group II cross-slabs). Douglas Hague’s excavations on the associated medieval hermitage site on Burry Holms, with pre-church timber structures, still unpublished. Elizabeth Walker’s recent excavations, which revealed later prehistoric timber structures including roundhouse, may throw fresh light on the nature and date of the early timber features under church.


Extra parochial area probably representing pre-Norman ecclesiastical estate, later passing to Abbey of St Taurin, Evreux (France) and then to All Souls College, Oxford. Llangennith church is a Norman foundation, of interesting plan, replacing a pre-Norman establishment.




Abernant at Kemeys Inferior near Caerleon has produced a grave grouping with inhumations in stone-lined cists. These burials may be late Roman in date, although the excavators have detected a curvature to the burial site which is not dissimilar to that found at Llandoc (Llandough) and other early medieval ecclesiastical sites. There are no grave goods and the burials range in orientation between 42 and 73 degrees. An industrial site with evidence of smithing located 180m to the NE has produced a calibrated radiocarbon date of AD 690-990 (Tuck 2001). Given the historical references to a court of Ffernfael near this site, further investigation of a potentially important early medieval site is worthwhile.


Economy, Land-use, Natural Resources



Much more is known about early medieval activities in the coastal zone of the Bristol Channel, where waterlogged features are exceptionally well-preserved. Fish traps from the Severn Estuary, the Newport boat fragment have been published, and many issues summarised in Cadw’s rapid coastal survey and The Coastal Archaeology of Wales (Davidson, ed. 2002). Imported wares are well documented, their sources have been refined and residues analysed. Under the Portable Antiquities Reporting Scheme, important examples of early medieval metalwork are being recorded and their provenances assessed: most have archive level reports, and many are in the process of being published. Little is known of field system patterns and subsistence economies.



The mudflats of the Severn levels are under constant threat of erosion. The evidence for water transport is very limited, and there has been little work on communication routes. Evidence for trade/commerce/industry during the 8th-11th centuries poorest. Artefact studies rely on the random nature of new discoveries; some old finds are no longer available for re-examination. Apart from work on the Severn Levels, there has been little palaeoenvironmental analysis; there are no large animal bone assemblages to compare with the well-studied animal bone from Dinas Powys.



The intensification of housing development threatens areas high in potential for archaeological evidence for early industry. Non-reporting of finds, and lack of resources to investigate their contexts.



·         The excellent survival of waterlogged features in the Severn estuary provides rich resource unparalleled on land, in particular for organic and environmental evidence and accurate dating.

·         A co-ordinated approach to the targeting of possible early medieval horizons for environmental analysis.

·         A fast-track mechanism for obtaining radiocarbon dates for potential early medieval deposits (equally applicable to any site type) would be beneficial.

·         The identification of artefact clusters through the Portable Antiquities Reporting Scheme and the Treasure Act 1976 should provide pointers to early medieval areas of activity, as well as opportunities for scientific analysis and the identification of cultural trends.

·         Near Caerleon has produced a grave grouping with inhumations in stone-lined cists. These burials may be late Roman in date, although the excavators have detected a curvature to the burial site which is not dissimilar to that found at Llandoc (Llandough) and other early medieval ecclesiastical sites. There are no grave goods and the burials range in orientation between 42 and 73 degrees.

·         An industrial site at Abernant, Kemeys Inferior, has produced evidence of smithing (hammer scale) associated with a calibrated radiocarbon date of AD 690-990 (Tuck 2001).


Wider Issues


In short, our state of knowledge for early medieval Wales based on archaeological evidence is parlous, and many of the wider research questions will have equal relevance for the rest of Wales:


·         We know virtually nothing about the character of 5th-century settlement, and impact on settlement of the transition from ‘Roman’ to ‘early medieval Wales’ beyond what has been gleaned from historical sources.

·         We know little about the character and development of early medieval agriculture and land-use.

·         We know little about the exploitation of different landscape zones.

·         We know very little about the nature of ‘industrial’ processes and resource exploitation.

·         We know little about the layout and development of early ecclesiastical centres.

·         We know little about mariculture, and the relationship with the sea (ships, boats, fishing etc).

·         We know little about settlement types, development, location, continuity, and status.

·         We know little about the material culture of early medieval Wales, regional patterning, and cultural identities.

·         Beyond historical reconstruction, we know little about the changing structure of society in early medieval Wales.





Davies, W. (1979), The Llandaff Charters. Cardiff.


Edwards, N and Lane, A. (1992), The Early Church in Wales and the West, Oxbow Monograph 16.


Hill, D. & Worthington, M. (2002).  Offa’s Dyke: history and guide (Tempus).


Howell, R. & Pollard, J. (2000), ‘Caerleon, Lodge Wood Camp’, Archaeology in Wales 40, 81-83, 97-98.


Nash-Williams, V. (1933), ‘An early Iron Age hillfort at Llanmelin, near Caerwent, Monmouthshire’, Archaeologia Cambrensis 88, 1933, 237-315.


Nash-Williams, V. (1939), ‘An early Iron Age coastal camp at Sudbrook’, Archaeologia Cambrensis 94, 1939, 340-341.


Phillips, N. (2000), ‘Castell Arnallt Survey’, Archaeology in Wales 40, 107-109.


RCAHM (1976) Glamorgan Vol. 1, Part 2: the Iron Age and Roman Occupation (Royal Commission on Ancient and Historical Monuments in Wales).


RCAHM (1976) Glamorgan Vol 1, Part 3: The Early Christian Period (Royal Commission on Ancient and Historical Monuments in Wales).


Tuck, M. with Leaver, A. and Sorrell, J. (2001), ‘Abernant, Early Christian and Medieval report’, Archaeology in Wales 41, 151-54.


Vyner, B. & Allen, D. (1988), ‘A Romano-British Settlement at Caldicot, Gwent’ in Robinson, D., ed., Biglis, Caldicot & Llandough, three Late Iron-Age and Romano-British Sites in South-East Wales (British Archaeological Reports (British Series) no. 188).


Wade-Evans, A., ed. (1944).  Vitae Sanctorum Britanniae et Genealogiae (University of Wales Press.           


Wilkinson, P.F. 1995, ‘Excavations at Hen Gastell, Briton Ferry, West Glamorgan, 1991-92’, Medieval Archaeology 39 (1-50), 19-22.




Compiled by Mark Redknap (Chair)


For the Early Medieval Wales in SE Working Party (Steve Clarke, Ray Howell, Jeremy Knight, Alan Lane)




This document’s copyright is held by contributors and sponsors of the Research Framework for the Archaeology of Wales.