2. A Brief Review of Published Evidence

Sources for the study of hillforts and defended enclosures are sparse. The two main historical sources are the RCAHM inventories for Carmarthenshire (1917) and Pembrokeshire (1925), and the Ordnance Survey site record cards (copies of which are held by the Dyfed Archaeological Trust). In addition to these two sources, several thematic studies of hillfort/defended enclosure have been undertaken in southwest Wales, but none for several decades, and those that have been done are on a county rather than a regional basis. Crossley (1963) listed all the earthwork sites in Pembrokeshire, including some sites identified from vertical aerial photographs that had not been previously recorded. Cropmark sites were not included in his survey, and not all the sites identified on aerial photographs were visited, leading to some spurious sites entering the record. Carmarthenshire is better served with a register of sites published in 1935 (Lloyd 1935), followed by a second list in 1954 (Savory 1954) and a review of evidence in the late 1970s (Williams 1978 and 1979). In Ceredigion, Hogg published a list of sites in 1962, which he expanded in conjunction with J.L. Davies in 1994 (Hogg and Davies 1994).

It is clear from a recent review of evidence by Davies and Lynch (2000) that the earliest hillfort/defended settlements of Wales lie within the wider British tradition and originate in the late Bronze Age/early Iron Age transition of c. 800-550 BC. Most of our evidence comes from large sites in the north and east of Wales, and while no modern excavation has been undertaken on the largest hillforts in southwest Wales — from which late Bronze Age dates could be expected — investigative work on medium to large sites, such as at Merlin's Hill in Carmarthenshire (Williams et al. 1988) and Caer Cadwgan in Ceredigion (St David's University College 1984-86), has demonstrated at least early Iron Age occupation. Radiocarbon dates from two coastal promontory forts also indicate early origins: at Porth y Rhaw (Crane forthcoming) a series of dates indicate defence construction between the 8th century and 4th century BC, and at Dale (Benson and Williams 1987) dates of the 8th or 9th century BC were obtained from complex defences, which included palisade trenches pre-dating the rampart. Recent excavation on an inland promontory fort, Berry Hill near Newport in Pembrokeshire, indicates it was constructed during the 10th to 8th centuries BC (Murphy and Mytum forthcoming).

Inland, within the Llawhaden group of enclosures in Pembrokeshire, the pre-rampart occupation at Broadway, a medium-sized hilltop enclosure consisting of stake-holes and gullies, dates to between 700 BC and 400 BC and is considered to immediately pre-date the construction of the rampart, and at Drim a palisade trench concentric to and 50m from the rampart of a small defended enclosure returned a date of 762-398 BC (Williams and Mytum 1998, 53, 65). This pre-dates the later defences by several centuries, but has no stratigraphic connection with them. Bronze Age activity at other sites at Llawhaden — Woodside, Pilcornswell and Holgan — seems to have no connection with the later defended settlements. At Brawdy Camp radiocarbon dates from a hearth and other features indicate occupation during the 8th to 5th centuries BC.

Palisade trenches that pre-date construction of the main rampart are a common feature on large hillforts in Britain, with evidence repeatedly pointing to Bronze Age origins, often with a considerable time lapse between the disuse of the palisade and the construction of the main rampart. However, at Castell Henllys, a medium-sized hillfort in Pembrokeshire, extensive excavation has demonstrated that the defensive bank was rapidly constructed over the palisade, even incorporating palisade timbers into the earthwork. Radiocarbon dates are awaited, but a La Tène I brooch beneath the rampart indicates a 4th-century BC date for the main defences (Harold Mytum pers. comm.).

Limited excavation, few artefacts and the imprecision of the radiocarbon calibration curve have hampered the chronological interpretation and site development analysis for the later Iron Age. Nevertheless, broad trends are detectable. From the 4th century BC to the 1st century AD the trend is one of smaller, defended enclosures filling up the countryside in the gaps between hillforts and larger defended enclosures. This pattern is demonstrated by large-scale excavation at Woodbarn Rath (Vyner 1982 and 1986), which was constructed between the 4th century BC and the 1st century AD, and at the small Llawhaden enclosures of Woodside, Dan-y-Coed and Bodringallt (Williams and Mytum 1998, 142), as well as at Walesland Rath (Wainright 1971). A number of medium and small hillforts were constructed but evidence points to this happening in the early part of this period, in the 4th and 3rd centuries BC, as at Castell Henllys as well as Pilcornswell and Holgan, both part of the Llawhaden group of enclosures. Interestingly, excavation at the latter two sites indicates that they were short-lived, and were replaced by smaller defended enclosures/farmsteads in the near vicinity.

An increasing number of rectangular enclosures has been recognised in recent years. There has been some debate whether they are indeed of Iron Age date or whether they originate in the Roman period. However, excavations at Penycoed, Carmarthenshire (Murphy 1985), clearly demonstrate a late prehistoric origin for the construction of the rectangular defences, while at Llangynog II, Carmathenshire (Avent 1973 and 1975), and Troedyrhiw, Ceredigion (Murphy and Mytum 2005), the context of material providing Roman period radiocarbon dates and the stratigraphic location of Roman pottery are indicative of prehistoric settlements whose use continues into the Roman period.

Paucity of artefacts is a big problem when studying the Iron Age in southwest Wales. The area is virtually aceramic, and other finds are scarce. This problem hinders the establishment of chronological frameworks — resulting in the reliance on scientific techniques such as radiocarbon determination — and hampers the identification of trade networks, economic activity and social interactions.

PREVIOUS NEXT CONTENTS HOME