This grand project is one of the most remarkable ventures conducted by the Society throughout its long history. The systematic recording of Roman mosaics began in the eighteenth century, but it is not until the past decade that it has come to fruition, thanks to the single-minded devotion of David Neal and Stephen Cosh. Their work, aided by the editor, designer and printer, combines scholarship of the highest level, the hallmark of all our productions, with the beauty we associate with a coffee-table book – only any coffee table which bears the second part of the corpus, let alone all four volumes, will require sturdy legs.

Many of the illustrations are by the authors, but several are the contemporary work of Luigi Thompson, while others are by well-known mosaic artists of past centuries, amongst them Samuel Lysons (whose exemplary work in some ways anticipated the present project), John Lickman and Edmund Artis. A great virtue of Volume IV is an appendix compiled by the authors with Martin Cosh, giving short biographies of the mosaic illustrators, past and present, whose work appears in the volumes. In addition to paintings, lithographs and drawings, photographs are employed copiously as well as maps, plans and diagrams. Very little has escaped the eyes of the authors and the Corpus is as complete as it is humanly possible for such a work to be.

Internationally, Roman Mosaics of Britain has to take its place alongside mosaic corpora dealing with Germany, Gaul, Iberia and elsewhere. These are – or were – mainly the productions of well-funded state enterprises while, by contrast, Cosh and Neal, working from a redundant nonconformist chapel on the outskirts of Milton Keynes, had neither university nor state support for their labours. Nevertheless, the British corpus is by far the most monumental, by far the most complete and by far the best of these surveys and, unlike them, it covers an entire Roman province.

One particular virtue is that, although the authors scatter their well-informed opinions throughout the pages (and whatever they say about dating, style, workshop groups or iconography is always worth reading), they have also tried to be fair to the opinions of others – both predecessors and contemporary scholars. While nobody who has an interest in Roman mosaics will accept all their opinions, that is not the point: it is important that all such debate is conducted with courtesy and they certainly do so.

As a result of the new mosaic corpus, and taking into account the corpus of sculpture (Corpus Signorum Imperii Romani, Great Britain), it is now possible to survey two of the major arts of Britannia and thus discredit the view that Romano-British art is both uniform and uniformly bad. This is a point that comes out clearly in these final two volumes (for my assessment of the two earlier volumes, see J Brit Archaeol Ass, 156 (2003), 178–80, and 159 (2006), 309–10).

Volume III is so bulky that the book has had to be divided into two; it deals with the south east: Essex, Hertfordshire, London, Kent, Surrey and Sussex and also Hampshire. The majority of the mosaics here are of the first or second century, except for the Central Southern Group, focused in Sussex and Hampshire, of the earlier fourth century. This reflects the early Romanization and concomitant prosperity of this region. By contrast, many of the most important mosaics in Volume IV, dealing with Oxfordshire, Gloucestershire and Wales, are fourth century, a reflection of the late flowering of Britannia Prima. Such an observation in itself is of value. Indeed (perhaps a minority view, but) this reviewer believes there is mileage in assigning Hampshire and Sussex to Prima and that would make the contrast even more striking.

It is in the south east (Volume III) that we find the early (first-century) mosaics so familiar from Gaul and Italy, especially at Fishbourne, though the evidence for such mosaics at other sites in the area, notably Angmering (site no. 391) and the temple on Hayling Island (314), is noted. In addition, there were early mosaics in London and very possibly Colchester (Essex). Several mosaics of the early second century reflect the emblema devices so familiar from continental mosaics and wall paintings. Amongst them is a very fragmentary mosaic from Eccles (Kent), depicting a pair of gladiators; although it was believed by the excavator to be first century, it is more plausibly ascribed by Neal and Cosh to c 120 (mosaic no. 354.1). From Britain, they can only point to two fourth-century parallels, which are certainly symbolic: in Britain, scenes from the arena seem never to have been part of native high culture, which instead was attracted to mythological scenes or striking patterned designs. Interestingly, one of the best parallels in subject is a contemporary red jasper intaglio from the Classis Britannica fort at Dover (information from Brian Philp). Whatever the context of this floor, other mosaics with emblema-like devices are assigned to mosaicists native to south-east Britain. Themes include wrestling cupids on the Middleborough mosaic, Colchester (291.66), and a lion holding the head of a stag in its bloodstained jaws from Verulamium (348.40). Other floors of the highest quality from Verulamium include the scallop shell (348.6) and – especially for its original approach to its subject – Oceanus/Neptune (348.22). There are links with mosaics from London, though the very best mosaics here, including the Bucklersbury pavement (370.88), date from the earlier third century. Doubtless Severan stimulation is responsible, but where did the mosaicists of the Acanthus officina come from?

Although the bulk of the mosaics from the south east are early, there are fourth-century pavements, one of the most important being that at Lullingstone, Kent (261.1). No other late mosaics are known from the vicinity and only one or two of different style are recorded from London, one of them from Old Broad Street (370.76), assigned to the Saltire Group. While the detailing is different, I would still maintain (against Neal and Cosh) that the rich iconography, and especially the use of the Bellerophon myth, the slug-like dolphins and the manner that both Pegasus and the bull kick back their hind legs, relates the composition to mosaics from the south west, although not to identifiable mosaicists from Dorset or Somerset.

Here and elsewhere, whereas this reviewer, and perhaps other readers, would centre their interests on over-arching themes of iconography and interpretation, the authors are most engaged by aspects of craftsmanship and compositional detail, and by doing so have provided absolutely invaluable information that would often have been ignored by art historians who were not practising artists. And if something of David Smith's schemas remains, both with regard to the earlier (second-century) and later (fourth-century) schemes, the question of workshop groups becomes much more complicated by the detail than it was when one could simply posit a small number of discrete schools on the basis of broad generalities. But then, the organization of ancient craft, like ancient society as a whole, was always going to be more complicated than many art historians would like it to be.

Some of the work on the fourth-century mosaics of Hampshire (the Central Southern Group) has now to be seen alongside the work done under the auspices of Professor Barry Cunliffe in the Danebury Environs project, in which the present reviewer was involved in commenting on the mosaics (Cunliffe 2008). The mosaic from Fullerton villa, Hampshire (311.3), with its fascinating centrepiece of Mars, re-laid in Fullerton Manor, has recently been re-photographed and subsequently lifted for display. Here, and for the most part elsewhere, there are no major clashes in interpretation, but Cunliffe did find a fragment of the border of the Thruxton mosaic (324.1) still in situ and also showed conclusively that the mosaic chamber was entered from outside; and, together with the most obvious explanation of the dedicatory inscription, this very much strengthens the case for this being a semi-public cult room. Likewise religious in theme, but still defying obvious attribution in terms of parallels and agreement on their meaning, are the mosaics from Brading on the Isle of Wight (331), but at least the photographs and careful descriptions will be a valuable starting point for subsequent work on them. Technically, the finest mosaics in the volume – and perhaps in the whole of Britain – are those from Bignor, Sussex (396). While distancing themselves from David Johnston's attribution to the Central Southern Group, the authors have no suggestion to put in its place and can point to no obvious continental predecessors.

Apart from modern photographs, we have, in the case of Bignor, the benefit of Samuel Lysons's superb draughtsmanship. Lysons was also responsible for making known what is the largest mosaic in Britain and perhaps north of the Alps, and the most original design, the Great Pavement at Woodchester, Gloucestershire, which is of course the high point of Volume IV (456.1), together with the smaller pavement from the extra-mural villa at Barton Farm outside Cirencester (461.63), self-evidently by the same mosaicists. As pointed out by this reviewer as long ago as 1985 (Henig 1985, 15–17), these mosaics provide a measure for the startling originality of Romano-British art at its best, as well as demonstrating the introduction of sophisticated Neoplatonic ideas. It is probable that the same mosaicists were involved in laying mosaics at Chedworth (418.2) and the Bacchus mosaic at Stonesfield, Oxfordshire (465.1), with a somewhat similar but not identical scroll. Cirencester, as a probable capital of Britannia Prima (if that was not Gloucester), was a major centre for craft and, apart from the so-called Orpheus Group (though the British ‘Orpheus’ may owe something to a local hunting deity, perhaps called Apollo Cunomaglus), there were mosaics based on saltire crosses, the best of which can be seen at North Leigh, Oxfordshire (462.1) – or can it? A fascinating story of neglect, rescue and incorrect reconstruction is revealed here. The Chedworth and North Leigh mosaics are linked artistically, in that the brilliant paintings are by Luigi Thompson, who has produced the finest mosaic illustrations in the book by techniques that require astounding patience. It might be suggested that the strange ‘urns’ in the North Leigh pavement are in fact lamps of a fairly unusual form.

The wealth of mosaics in the west reflects the wealth of Britannia Prima in the fourth century. Most are from villas in the country or town houses in Gloucester or Cirencester. But at Lydney, we have a richly appointed temple with baths and a guest house that, with all its mosaics, would have been regarded as a rich villa if it was not located next to the Temple of Nodens. The mosaic from the temple cella (440.21) had an inscription that (with the Thruxton inscription) provides the important evidence for patronage in connection with Romano-British mosaics. Both Chedworth (418) and Great Witcombe (433) have also been proposed as sanctuary sites and, while nothing in the mosaics themselves would either support or deny the suggestion, it does allow us to consider that mosaics do occur in non-domestic contexts.

Wales and the Marches are included in this volume and there are mosaics from the fortress baths at Caerleon, one of them identified by Boon as a thyrsolonchus (482.3), though I am taken by the new identification by Cosh and Neal of an amphora. From the principia comes the well-known labyrinth mosaic, though the centre is missing so we do not know whether Theseus and the minotaur were figured. Chester also had mosaics (471), though rather cruder, and, indeed, as bizarre as Chester's Roman sculpture.

Apart from Caerleon, Wales is represented virtually entirely by a rich assortment of mosaics from Caerwent and a string of mosaics from the edge of the Bristol Channel. Some of the mosaics have close affinities with those of Gloucestershire and Somerset, though others are harder to parallel. Although, of course, some of the wealth of Caerwent may have come from Caerleon, as, by the time of its floruit in the fourth century, Caerleon was no longer a thriving fortress and has to be seen as a successful civilian town. Essentially, surely, the Severn and the Bristol Channel were to be seen as a highway and not as a barrier, and the south Wales coastal plain was part of the Gloucestershire and Somerset culture zone; and that is the explanation for the existence of mosaics at Langstone (486.1), Llantwit Major (488.1) and Oystermouth (489.1).

There is so much more here to study, debate and ponder. The relatively few mosaics from Shropshire, even from Wroxeter, tell of its remoteness from the main channels of Roman life and culture. As with sculpture, here one was on the ‘edge of Empire’. More generally there is the question of availability; in some places, mosaics were available in the second century but not in the early fourth century; in others (as in Hampshire), there are plenty of early fourth-century mosaics, but few later. What about survival? The secondary patching of mosaics demands greater study, for in some cases this might reveal they were somehow valued in the fifth century or even beyond. I am glad Fawler in Oxfordshire is mentioned (459); there are, indeed, records of mosaics here from the nineteenth century, but before that in the name itself, derived from the Anglo-Saxon ‘fagan flore’. There is so much left for students to research, and the four-volume corpus is set to spawn doctoral theses for years, perhaps centuries, to come.

An appendix presents some very recent finds, at Pillerton Priors, in Warwickshire (491.1), and Queen Camel (492.1) and Yarford (493.1), in Somerset. The extreme plough damage to the first and a ditch dug through the second demonstrate that this is a heritage at risk. But these and other mosaics are surely harbingers of fresh discoveries, and it is essential that these wonderful volumes are not regarded as the last word but encourage the prompt reporting and illustration of new discoveries at the same superlative standard achieved by Drs Neal and Cosh.

Cunliffe, B (ed) 2008. The Danebury Environs Roman Programme. A Wessex Landscape during the Roman Era, 2 vols, Oxford Univ Sch Archaeol Monogrs 70 and 71, Oxford: Oxford University School of Archaeology

Henig, M 1985. ‘Graeco-Roman art and Romano-British imagination’, J Brit Archaeol Ass, 138, 1–22