Trans Bristol Gloucestershire Archaeol Soc 86

ST 775649. One complete and two partial sections were obtained of the defences of a large univallate fort (formerly 80+ acres), and its relation to a field system was examined. The rather nondescript pottery obtained from the pre-rampart occupation level is best paralleled at Chalbury (Dorset). The rampart, of small rubble and clay faced in and out with limestone slabs, was only 9ft 6in wide and presumably non-defensive. There was no evidence for timber structures within it. An 11ft berm lay between it and the rock-cut ditch. No evidence to date the construction of the fort was found in either rampart or ditch, but one entrance was proved non-original. Banks of the field system overlie the S ditch at several points; within the defences and to the N the relationship is less clear, but again the fields probably post-date the fort. Earlier (unstratified) finds of pottery from this and other local sites are discussed; they include finger-printed situlate jars and furrowed bowls. A survey of the early univallate hillforts of 60+ acres in S England strongly suggests their use as stock enclosures.

NGR 960400. A complete cycle of cereal agriculture, using Iron Age techniques as far as possible, was attempted on a limestone site (Bredon Hill, Worcs). With human traction, a reconstructed Donnerupland ard easily formed furrows in a light soil. Modern seed (rika barley) had to be used in the absence of adequate supplies of older strains. The grown crop was harvested partly with sickles, partly with flints (which proved more efficient). The storage experiment was designed to discover: 1) how both moist and dried grain survived in basket-lined pits; 2) how grain reacted in an unlined pit; 3) whether a pit could be opened and then resealed; 4) whether grain could be thus stored in the ear. Several methods of sealing were tried and carbon dioxide formation was regularly monitored during storage. It was shown that a well-sealed pit with firm walls stored grain successfully, with germination rates between 60-77%. Even the unlined pit that was opened for inspection and then resealed gave satisfactory recovery. Further experiments should be undertaken on different soils.

The Roman city wall was uncovered for a distance of 42ft during redevelopment work and has been preserved in situ with access from the modern building above. Building phases are represented by: 1) Stone turret, probably contemporary with clay bank on corduroy timbers; 2) Wall 1, undated, of grouted rubble with small facing blocks on a chamfered plinth; 3) Wall 2 (seen in an earlier, more restricted excavation), of large grouted limestone blocks, a rebuild of Wall 1, with a single sherd suggesting a date after AD 290. The evidence from the SE angle tower of Glevum (1932 excavation) should assign the turret-and-clay-bank system to the first defence of the colonia, at a date suggested as AD 70-100 rather than the Claudian date postulated by Richmond and O'Neil from the 1961 and 1964 excavations.

Several sites which extend the distribution given on the 1963 map are briefly reported. A note on the Welsh sites expresses some concern about the accepted dating for production at the kiln.