Excavations of sites spanning the Beaker to early Roman periods at Stackpole Warren, Dyfed, are described. The sites are in an area of blown sand which enhanced their preservation and led to the separation of several horizons. The earliest is a buried soil beneath the blown sand which contained Mesolithic to Bronze Age artefacts. At site A, there was a roundhouse associated with Early Bronze Age pottery and dated to 1620±70 and 1400±70 BC uncal., and two other roundhouses, one possibly of Beaker age. After a period of soil formation, a ritual complex of Later Bronze Age date was established, this contemporary with the earliest besanding of the area; it included a stone setting of more than 2000 small stones, an alignment of small water-worn stones and a standing stone. A cremation gave a latest date of 940±70 BC uncal. Other Later Bronze Age activity is recorded at site G/J in the form of a rectangular enclosure, possibly unfinished.
Late Iron Age to early Romano-British settlement was present at sites A and B, consisting of scatters of occupation debris, burnt mounds, cooking pits, hearths and houses, some of stone, some of timber, all taking place in an area being intermittently besanded.
Peripheral to the religious and domestic sites, a field system was excavated. The earliest phase was a linear earthwork from which a C14 date of 400±70 BC uncal. was obtained from charcoal in the ditch. After the decay of this, rectangular fields with stone walls were laid out, one along the line of the erstwhile earthwork, this taking place around the end of the Iron Age as dated by C14 of charcoal directly beneath a wall to 90±70 BC uncal. Some of the fields had been cultivated by a succession of cross- and one-way-ploughing, others used for cattle.
An assemblage of 763 flints included a few Mesolithic artefacts but was mostly of Late Neolithic and Bronze Age date. A succession of ceramic assemblages included a small Middle Neolithic group (4 vessels), two distinct Beaker groups, one early (Lanting and van der Waals steps 1–3 (8 vessels), one late (steps 3–6) (45 vessels), an Early Bronze Age group of collared urns (43 vessels) and a Later Bronze Age group (26 vessels).
Environmental data was not prolific but there was a small quantity of animal bone, mostly cattle and sheep, and cereal grain, mostly barley with some wheat. Marine molluscs were present but sparsely utilized and there was no other indication of the exploitation of the coastal resources such as seals, birds, fish andiseaweed. Land Mollusca indicated open country from the Iron Age onwards when the record begins.
The importance of the site is in the ritual complex from site A, the succession of Iron Age/Romano-British occupation horizons, the succession of ceramic assemblages, the field system and the fact that blown sand horizons have allowed the preservation and separation of the sequence much of which would have been at best conflated in to a single horizon or at worst destroyed. Otherwise, there is no evidence that the site was in any way special with regard to the relationship of human activity and sand deposition until the Middle Ages when the area was used as a rabbit warren. Nor was the coastal location important, at least as could be determined by the results. This was a representative of a succession of later prehistoric farming communities and their various domestic, ritual and sepulchral activities in lowland Dyfed.