V.E. Nash-Williams' The Roman Frontier in Wales appeared in 1954; a completely revised edition followed in 1969, under the direction of M.G. Jarrett and with 19 contributors. This book is an equally radical revision, with two editors (also authors of a large part of the text) and 33 contributors, three of whom (Davies, Casey and Manning) were responsible for sections of Jarrett's book 41 years earlier. Camps are not included, having been dealt with separately in 2006 (J.L. Davies and R.H. Jones, Roman Camps in Wales and the Marches), and this gives space for the comprehensive treatment of topics such as coins, pottery, environmental evidence and Roman roads which were scarcely considered by Nash-Williams and Jarrett. More than half of the book is taken up by the Gazetteer, which is graced by many superb aerial photographs and clear plans of sites, though comparisons of the latter are hindered by a variety of scales, few of them natural (that is, at 1:500, 1:1000, 1:2500, and so on). The first four chapters deal with the history of the military occupation, the military installations and their associated units, and the communication systems. Under the rubric of ‘the military-civilian interface’, the other two general chapters cover areas such as extramural settlements, religious beliefs, local interaction and military supply.

Forty years of research, particularly rescue excavations and geophysical survey, have developed a much more detailed narrative than Jarrett and his collaborators were able to offer. Of the three stages of the Roman military occupation — conquest, stabilisation (with a diminishing military presence), and a limited renewal of the defensive system in the later Roman period — it is perhaps the last where the least progress has been made. Casey provides an admirably succinct summary of what is known, discarding wishful theories about scribal omissions from the Notitia Dignitatum and showing that the coin evidence is consistent with the final withdrawal of effective units before the date at which that document is thought to have been put together. For the other two stages, there is a wealth of new information: for example, 51 forts and campaign bases as opposed to the 35 examples Jarrett listed, in addition to newly discovered fortlets and watch-towers. Of all the campaigns of conquest in Britain, the historical sequence in Wales is best known, thanks mainly to Tacitus, but much has been learnt about the disposition of forces in the aftermath, though there are gaps or uncertainties in our knowledge of the locations of the legions until well into the Flavian period. Excavations described by Manning have added a new Neronian fortress at Usk (still regarded as a large fort in 1969) together with about 10 per cent of its detailed plan, which, however, has some puzzling features. On the other hand, we have apparently been deprived of a full-size fortress of the same period at Kingsholm where, as Holbrook shows, the defences might have enclosed an area no larger than 6.9 ha. What seems to be an intermediate stage in the disposition of forces is illustrated by the remarkable geophysical survey at Llanfor which is described by Hopewell. The complete plan of a base with an area of 3.86 ha, apparently of early Flavian date, has emerged with wonderful clarity: there are barracks for cavalry and infantry, and a principia with the seven rooms in the rear range which are characteristic of legionary examples (in passing, Latinists will be upset by ‘principiae’ used elsewhere on 75–7). Though further analysis of the plan is in progress, there seems to be a strong possibility that the infantry accommodation was for a legionary cohort and that, amongst the numerous pre-Flavian and Flavian sites of a similar size, there are others where legionaries were brigaded with cavalry. Burnham and Davies see these large forts as a short-lived stage in the transition from active campaigning, with a fluid disposition of forces, to a system of smaller forts, some of them built on the sites of the larger forts.

This system of smaller forts was gradually thinned out. The process, often involving a reduction in the size of forts or their replacement by fortlets, and its chronology are much better understood than in 1969. By the second half of the second century, about half a dozen forts were left in Wales. B. and D. take their continuing presence as an indication of ‘a real concern for internal security in the heart of Wales’ but are uncertain whether this was because its population posed a substantial threat to neighbouring areas or because of a general tendency towards internal feuding and brigandry.

In addition to the historical narrative, the introductory chapters provide admirable summaries of wide-ranging research, most notably the geophysical surveys of the settlements outside the forts. The particular value of this work has been to show the extent and character of these settlements in the later first and earlier second centuries. The well-known surveys of the more extensive military vici on Hadrian's Wall show them in their third-century state, with their stone buildings perhaps concealing earlier settlements of timber construction comparable in extent to those in Wales. One discovery about settlement in the hinterland of forts of which more perhaps could have been said is a tablet from a Roman will found in the mid-nineteenth century at Bodyfuddau 5 km from the fort at Tomen y Mur and rediscovered in 1991 during a house clearance in West Kensington. Here, according to Tomlin's original publication, is evidence that a Roman citizen, presumably an auxiliary veteran, acquired land, possibly a ‘native settlement’ known near the find-spot of the tablet, and could make testamentary disposal of it.

All these chapters draw on the Gazetteer, the entries in which, although by many authors, maintain a generally uniform approach and high standard of clarity. Perhaps the greatest contrast is in the treatment of the two permanent legionary fortresses. Evans provides a detailed structural account of Caerleon, particularly welcome because the last such essay (by Boon) was published more than twenty years ago. There are more recent detailed accounts of Chester which allows Mason to say more about the function and history of the fortress. Whether there is any real substance in the suggestion that its large size (paralleled at Neuss and Bonn, for example) can be explained by the inclusion of accommodation for the governor is doubtful. No mention is made of Fulford's suggestion that the Elliptical Building was a macellum (Journal of Roman Archaeology 18 (2005), 685–9).

The book as a whole has been an ambitious undertaking, triumphantly accomplished by B. and D. and their collaborators; all those interested in Roman frontiers and frontier provinces will be deeply indebted to them and the organisations which have sponsored the work.