This latest volume in the ongoing Winchester Museums' programme of backlog publication is an important addition to knowledge of the Roman town. Although the result of long-drawn-out processes of production, the eventual decision to combine two projected volumes, one each on the suburbs and cemeteries, was a sound one, since so much of the key evidence from extramural excavations in the period covered relates to the cemeteries, which represent the most significant contribution to knowledge of Roman Winchester made by these excavations. Furthermore, among the more interesting aspects of Victoria Road, one of the key sites reported, is the successive use of adjacent parts of this site for burial and settlement activity, which is clearly brought out in this combined volume. A less fortunate, if unavoidable, result of this publication programme is that details of associated grave goods are not found here, ‘small finds’ having been published previously (H. Rees, N. Crummy, P.J. Ottaway and G. Dunn, Artefacts and Society in Roman and Medieval Winchester. Small Finds from the Suburbs and Defences, 1971–1986 (2008)), while the volume on pottery is yet to appear. The compilation of the present volume took place in the period 2007–2010 (preface), but inevitably some of its components were prepared much earlier.

The report discusses a range of sites of varying size, complexity and level of examination. Cumulatively the most important are in the northern suburbs, where 326 (out of a total (table 55) of 661) burials came from the two Victoria Road sites alone, but there are also significant groups of burials from the western and eastern suburbs, and all the groups have slightly different characteristics. Victoria Road East is the only early Roman cemetery represented; burial may have commenced in the pre-Flavian period but the cemetery was largely out of use by the later second century. Some 118 cremation burials were present, and 85 inhumations were also recorded, of which 69, remarkably, were of infants. The remaining burial groups, including that at adjacent Victoria Road West (the two sites lying on each side of the Roman road to Cirencester, possibly disused by the late Roman period), are essentially late Roman inhumation cemeteries. Those in the western suburbs at Carfax and New Road are notable for having been contained almost entirely within the ditch of the later prehistoric Oram's Arbour enclosure. These were not casual burials, but represented reuse of quite constructed areas (i.e. located within small excavations) over an extended period through the late third and fourth centuries. Almost 43 per cent of burials from the western suburbs (total 82 (table 34)) were of neonates and a further 15 per cent were of infants.

The atypical location of the Oram's Arbour burials may help explain their unusual age profile, but it is clear that, within certain broad parameters, local variability is likely to have been the norm rather than the exception. This is accepted in relation to Oram's Arbour (342; might the reuse of Dyke Hills at Dorchester on Thames be a possible parallel?), and with regard to the marked contrast in furnishing of late Roman graves between the sites reported here and the Lankhills cemetery (370). Elsewhere, however, the tendency seems to be to presume uniformity rather than diversity of practice. The population of late Roman Winchester presumably incorporated numerous different communities, so why should this not have been reflected in conscious or unconscious variation in burial practice with regard to definition of cemetery plots, alignments and other physical characteristics of graves and the provision of grave goods and so on? A recurring and related theme picks up Clarke's idea of declining standards in the execution of burial in the latest phases at Lankhills, which seems to be based on a teleological view of ‘decline and fall’ (and a further aspect of this is a potentially very short chronology for the very latest phases of burial, a question that still needs much further work, as does the chronology of burial in the third century; both could be addressed with systematic radiocarbon dating programmes). There is certainly evidence for change in places, in respect of grave depth, alignments and extent of intercutting, but these developments are, as is rightly pointed out, not universal at Winchester, and might as well reflect the different priorities of different burying communities as a premonition of things falling apart.

This volume therefore provides much food for thought in relation to interpretation. It is extremely welcome for its systematic presentation of important evidence (not least in the gazetteer of burial and other sites supplementing the material published in detail). One can but hope that some of the key questions outlined in the conclusions by Ottaway (376) can be pursued by work that builds upon this substantial foundation.