This collection of papers originated in a session at the 2012 Limes Congress in Bulgaria. It ranges widely across the Empire, examining developments in late Roman military architecture and especially the uses to which buildings were put. Two of the editors (Collins and Weber) introduce the collection and then Whately, in ‘a historian's perspective’, takes the distribution of army units on the Lower Danube as an example of the difficulties posed for this period by the literary and epigraphic sources. The following papers fall into two groups. First, Arce (on the limes Arabicus) and Rushworth (the North African frontier) examine fort types and their dispositions, while Symonds looks at a single type (fortlets in Britain) and Collins analyses changes in the use of fort granaries on Hadrian's Wall. The second group studies particular sites: the legionary fortresses at Vindobona-Vienna (Mosser) and at Novae in Lower Moesia (Lemke), the coastal fort at Oudenburg in Belgium (Vanhoutte) and the fort at Binchester in northern England (an interim of excavations in progress by Petts). Breeze in his concluding summary discusses some of the themes common to many of the papers.

Architectural study in its narrower sense — development in the design of buildings —features strongly in Arce's and Rushworth's valuable studies of the smaller forts, which also examine their positions in the overall systems of defence and supply. Symonds’ survey of the British fortlets in Wales and northern England, similarly concerned with their form, is on a smaller scale but of equal interest; he makes a bold attempt to show that the masonry remains at Burrow Walls in Cumbria, an earlier fort site, might have been part of a late Roman defensive system rather than a medieval building. In the other papers the emphasis is more on the forts and fortresses as settlements. Lemke and Mosser describe the conversion of their fortresses during the course of the fourth century a.d., partly resulting from reductions in the size of the legionary garrisons. Some of the largest buildings at Novae, such as the baths, were already out of use, but the principia still stood and at the end of the century accommodated dedications by civilian primipilares. A garrison was now combined with a large civilian population living in ‘arguably more of a fortified town than a military base’. Similar changes have been seen at Vindobona: the Severan barracks stood until the end of the Roman period, with some eventually occupied by civilians but others continuing to house legionaries. Vanhoutte demonstrates very clearly that Oudenburg also had a mixed population, but circumstances there were very different. The excavated part of the fort, built in c. a.d. 260 with an area of c. 2.8 ha, seems never to have contained barracks and though it contained a small bath-house and a small rectangular building, it was mostly an open space. The contrast with forts in northern England could not be starker, where their interiors were still densely packed with buildings throughout most of the fourth century a.d. and the only extensive open spaces were the streets. The demolition of one of the two barracks excavated at Binchester, its site then remaining vacant, is exceptional, though Petts, relying on ‘the admittedly circumstantial dating’, sees this as a result of a reorganisation in the third century a.d.

Common to all these papers is an acceptance that the numbers of soldiers in the garrisons were far smaller than in previous centuries. Collins sees this as one of the reasons for reduction in the storage capacity of granaries in forts on Hadrian's Wall, but his argument is complicated by the bald assertion that by the fourth century a.d. ‘the non-soldier population … had moved into the fort[s] from the now abandoned extra-mural settlements’ (28). The structural history of the barracks and the types of finds at Housesteads and South Shields lend no support to this assertion, while the mixture of military and civilian populations seen in legionary fortresses and at least some of the coastal forts in Britain and Germany might not have taken place in frontier forts until the later fourth century a.d.

The chief merits of this collection are its geographical range and its comprehensive treatment of sites, ranging from fortlets to fortresses. Some will agree with the editors that it demonstrates ‘the dynamism and ingenuity of the late Roman army’, but perhaps fewer that the army in the fourth century a.d. ‘provides a breath of fresh air after the suffocating uniformity of its forbears’ (4). Breeze is surely right to remind us that at least in the western provinces the army was a shadow of what it had been, with a limited offensive capacity and a much more defensive stance (140).