Scott Archaeol Rev 6

Stopford (see 88/514) underestimates the complexity of the Danebury data set. The interior looks devoid of houses, but very large numbers of postholes on an eroded site strongly suggest that houses were there. It remains a huge, mainframe computer problem, and some behavioural patterns may never show up.

Not all hillforts have the same function. Scratchbury is positioned more for ceremony than for defence, and Whitesheet is similar: its interior is visible from below, it has adjacent barrows and a causewayed enclosure. Both these sites are on boundary zones, perhaps for seasonal transfer of resources. The same kind of interior visibility is seen in some Northumbrian hillforts. Until a large number of variables have been examined the norm cannot be defined.

A Haltern Type 70 Roman amphora at Gurness is well outside the normal distribution: is it best explained by Eutropius's (4th century) statement about the Claudian submission of the Orkneys? It dates at least 20 years before the Agricolan conquest and had held either liqueur wine or olives. Only our rigid views of the unsophistication of the northern Iron Age stand in the way of accepting Eutropius at face value. Alliances with other parts of Britain could have occasioned the arrival of this piece in the north.

Examines the use of architectural space in expressing social differences within monastic settlements, using data from a modern contemplative community of Anglican nuns. The analysis is directed towards the evolving perception of the concept of community in med English monasticism. The thousand-year evolution occurred in a context of social ranking, gender relations, and a growing desire for privacy (as in secular dwellings and hospitals).

Problems in using the Harris matrix arrive when users wish to assess life spans of site entities; by contrast, the Carver system is hierarchically based (context-feature-structure) and involves interpretation from the start. The two systems are really complementary. The problem of deposits truncated by levelling is discussed: dating and correlation of entities depends on contained artefact assemblages, with concomitant problems of derived material. The method Carver used at Sidbury (Worcester) to phase the 'site impacts' needs care in application but can be useful. This methodology has considerable potential for diagrammatic representation. (But see next abstract.)

Enquires into the 'intractable' problem found by Collingwood and Myres for the end of Roman Britain. A less empirical approach is offered to the two main issues, the magnitude of social change, and the emergence of a single cultural system seemingly independent of either western Roman or Germanic traditions. Lines of enquiry have been too fragmented by individual specialisms: instead, Giddens's theory of social structuration - which involves human action - helps to unravel 4th-5th century power structures, variously based on land, charisma, violence, etc. Thus both indigenous RB people and incoming Germanic settlers had equal chances to develop kingship. Town survival is unimportant compared to how fields of social discourse were altered by the new situation, which itself laid the basis of feudalism.

Objections are raised to Armit's reinterpretation (89/423) of this site: Dun Cuier is not a broch, as shown by several features including the entrance passage and the stratigraphy of the central court. Its primary inhabitants had Dark Age artefacts; the structure could be an unskilled copy of a ruined broch.

Applauds this 'fundamental contribution to British prehistory' despite some faults in cross-referencing and unreliable maps. We now need a fresh look at swords in Ireland and Atlantic France.

Two separate review-articles, one from a medieval historian and the other from an archaeologist, of this collective volume.