In 1966 the owner of a small earthwork enclosure, referred to in this paper as Walesland Rath, notified the Ministry of Public Building and Works of his intention to remove the bank and ditch and to bulldoze the interior of the site. The earthwork occupied one corner of a much larger field, and on account of the difficulties involved in cultivating the area in a proper fashion had become overgrown with trees, gorse and bracken. The Ministry concluded that excavation was preferable to preservation in this case, and the author was requested to undertake the work. Previous excavation of the numerous earthworks in Pembrokeshire had been carried out in a desultory fashion by earlier antiquaries, or in more recent years with limited resources. It was therefore decided to take the opportunity of totally excavating the interior of this small enclosure, which in terms of situation and size is typical of many such earthworks in the county, in order to obtain evidence for its date and history and to establish the complete ground plan of the settlement. Such an operation is only rarely undertaken in British archaeology and it was hoped that the results would provide a datum to which other work in the area could be related.

One of the best-known inscriptions from Roman Britain was found in 1723 at Chichester. It is a dedication on a slab of Purbeck marble of which the left-hand portion is lost, and the remainder was broken into four pieces during recovery. These are now bound by concrete and built into the wall of the portico at the west face of the Council House in North Street at Chichester (PL. IX). There the inscription is protected by a sheet of plate glass.

A tombstone removed from its context loses much of its evocative nature; disassociated from the remains it was intended to commemorate its original function is obscured. The tombstone continues, nevertheless, to communicate through text and also images. Its removal may mute the impact of the message but if the stone remains un-shattered and un-concealed it is not easily silenced.

In the received history of Britain in the fourth century, a special place is occupied by the Saxons. It is asserted that the defensive system of the Saxon Shore existed to prevent them from landing; and it would seem that, unlike any of the barbarian tribes with whom Rome maintained uneasy relations along the Rhine and the Danube, unlike even the Goths, the possibility of their peaceful settlement within the boundaries of the most accessible Roman province could not be entertained. Their attacks, apparently, were so unpredictable, and their nature so savage, that in this case alone the policy of the Roman authorities was one of total exclusion. In 367, moreover, after forming an alliance with the tribes of the north and the west, they overwhelmed the Roman defences and ravaged Britain; and this date has been accorded the status of a turning-point in fourth-century history. But afterwards the Saxon Shore forts resumed their function of keeping the Saxons out. Finally, the Roman garrison was withdrawn and the Shore forts abandoned; and this permits historians to assign an early fifth-century date to the beginning of the Anglo-Saxon settlement.

Bradley Hill, 8·8 km (5½ miles) north-west of Ilchester, is a low hill projecting northwards from the Lower Lias plateau between Somerton and Langport and is capped by a thin band of Rhaetic clay.

The site of the Romano-British farmstead, 53 m (175 ft.) above sea level, is in a prominent and exposed position at NGR ST 4800 3034 (FIG. 1; PL. XVIII A). There are extensive views in all directions, particularly over the valley of the River Cary to the north and over the great expanse of King's Sedgemoor to the west. On the hilltop plateau, the calcareous ploughsoil was 0·20 to 0·25 m in depth, except around the sites of the Romano-British buildings.

The careful grid-patterns which are such a characteristic feature of Roman towns and camps are one of the most vivid illustrations of the Roman capacity for efficient organization. It is obviously interesting to know what units of size were used. The clearest demonstrations of the units employed in land-measurement occur in centuriation. As its name indicates, centuriation was the division of land into centuriae, squares measuring one centuria in area. Their side was 2,400 Roman feet. Considering the distances involved, the surviving evidence shows remarkable accuracy: centuriae that can still be measured range from 711 m in Emilia to 703 m in the Chott el Djerid, a variation of ±0·6 per cent. Shorter measuring units listed by Roman metrological writers are the actus of 120 feet, the pertica or decempeda of 10 feet, and the passus of 5 feet.

The Wentlooge Level in SE Wales represents an extensive area of some 35 km2 of alluvium reclaimed from estuarine saltmarsh from the Romano-British period onwards (FIGS 1-2). As was noted by Allen, the landscape is characterised at its southern and northern extremities by a pattern of small irregular fields, often fossilizing the meanderings of natural drainage channels (FIG. 2). This arrangement is typical of many of the reclaimed alluvial wetlands that fringe the Severn Estuary. The remainder of the Wentlooge Level is distinguished by a very different landscape, comprising regularly planned blocks of long, narrow, and generally straight-sided fields, quite unique among the wetlands of the Severn Estuary. At Rumney Great Wharf, north east of Cardiff, part of the latter field-system can be seen cut into a clay-peat shelf in the intertidal zone (see below), thus indicating a major episode of coastal retreat and the repositioning of the sea-wall across it; similar evidence from the intertidal zone can be recognised as far to the north-east as Peterstone Gout. An extensive spread of Romano-British pottery and primitive iron-making slag was associated locally with this field-system in the intertidal zone at Rumney Great Wharf, while survey of the adjacent mud cliff revealed at least one ditch, sealed by a buried palaeosol, which yielded stratified Roman material. Further erosion of the mud cliffrevealed more ditches with Romano-British material, as well as other indications of settlement, and prompted a programme of survey and excavation grant-aided by Cadw and the National Museum of Wales in the spring of 1992.

We know more about the years 406–410 than we know about any other quinquennium of Romano-British history, apart from the periods that Tacitus describes for us. The standard account of what happened is that which C. E. Stevens published in an Italian journal in 1957. In the first part of this article I propose to argue that Stevens's paper contains a number of fundamental errors, that its chronology is untenable, that it ascribes monstrous mistakes to our authorities when they are not mistaken at all, and that its influence on later discussion of the problems has been unfortunate. It has left an unhappy mark, for instance, on the last chapter of Sheppard Frere's monumental Britannia, where it is generously described as ‘a very valuable discussion of sources and chronology’. Having considered the chronology of the period I shall attempt in the second part of this article to describe the nature of the events which brought about the end of Roman rule in Britain.

The Roman villa at Stanwick (GR SP 972716) is situated on the west side of Stanwick village and the newly constructed A405 bypass. It occupies a gravel terrace on the east bank of the River Nene close to a point where the river swings west to join a secondary canalized channel on the west side of the floodplain. The area immediately west of the villa was partially disturbed in 1845 by the construction of an embankment for the now disused Blisworth-Peterborough railway. The south side of the site is bounded by a stream which rises as a spring east of the village: other springs issue from the valley slope close to the villa where the gravels give way to a bedrock of clay and cornbrash, a decomposed oolitic limestone. Beds of limestone occupy the higher slopes and were quarried in Roman and medieval times. Both the availability of fresh water and of building-stone probably influenced the location of the site.

Areas of coastal marshland formed an important and distinctive part of the landscape of Roman Britain, and current work is showing that different wetlands were utilised in very different ways. Some areas, for example in Essex and Kent, were simply exploited for their natural resources to produce salt and support seasonal grazing. Parts of Fenland were also used in this way, though the higher coastal siltlands were modified through the creation of drainage systems in order to improve agricultural opportunities within a landscape that was still liable to tidal flooding. A third strategy towards wetland exploitation is reclamation: a major transformation of the natural environment, involving the construction of a sea wall along the coast to keep tidal waters out and a system of drainage ditches cut into the surface of the former saltmarsh to lower the water table and remove surface run-off from the surrounding uplands.

At a time when the conjectured 1900th anniversary of the founding of York has recently been celebrated, and a medal has even been struck commemorating the founder, Petillius Cerialis, it may be appropriate to take a closer look at this man, certainly one of the most interesting persons to figure in the pages of Tacitus. He is best known to students of Roman Britain from the brief lines in the Agricola:

But when, together with the rest of the world, Vespasian recovered Britain also, there were great generals, splendid armies, the hope of the enemy was diminished. And Petillius Oerialis at once induced terror, attacking the civitas of the Brigantes, which is regarded as the most numerous in the whole province. There were many battles, and sometimes they were not without bloodshed; and he embraced a great part of the Brigantes either with victory or with war. And Oerialis, indeed, would have eclipsed the administration and the renown of any other successor: but Julius Frontinus took up and sustained the burden, a great man, in so far as greatness was permissible (Agricola 17.1-2).

Studies of the presence of Christianity in Roman Britain occasionally utilise a form of ‘negative evidence’: the destruction of pagan images. That this destruction is iconoclasm carried out by Christians is too readily assumed. Christian iconoclastic attacks did occur elsewhere in the Roman Empire, the question is whether we have been uncritical in automatically applying such evidence to the damaged Romano-British images. The apparent preference for certain parts of statuary, coupled with the occurrence of pieces in deposits that would, under any other circumstances, be regarded as ‘votive’ or ‘ritual’, suggests that these images may have been broken for reasons other than that of straightforward ‘destruction’. It is possible that some of these images may instead have been deliberately fragmented, with the resulting pieces retaining meaning and receiving special treatment and deposition. The aim of this paper is to question the validity of interpretations that suggest iconoclasm and to put forward an explanation based upon the theories of fragmentation as an alternative explanatory framework. By examining the high proportion of certain ‘body parts’ in the archaeological record, the contexts in which these occur, and the possible differing treatment of statuary of different materials and sizes, it is hoped to demonstrate that iconoclasm cannot account for all of the instances of image ‘destruction’ in Roman Britain.

In the first decade of the fifth century Gaul and Britain, Spain and Italy all suffered from a succession of barbarian invasions and Roman civil wars. We possess not just one but several literary accounts of the crisis, along with chronicles, poetry, and imperial laws that all help to illuminate its course, but the sequence of events bristles with technical difficulties. These have not been satisfactorily resolved in the countless modern narratives of the period, which regularly depend on one another rather than the ancient sources. The most recent treatment of the topic, by John Drinkwater, appeared in the pages of this journal, and represents a tremendous advance on our understanding of the crisis. Drinkwater's analysis clarifies any number of crucial interpretative points, and should be the first port of call for any one interested in the study of the period. Problems nevertheless remain. Drinkwater's episodic treatment of the period encourages a certain obscurity that a more systematic narrative might dispel, and he relies heavily upon badly understood ancient slanders against the regent Stilicho to underpin important parts of his narrative. What follows is in no sense an attack on Drinkwater's magisterial interpretation of usurpation and crisis, which warrants wholehearted endorsement. Instead, I would like to suggest that the welter of modern literature on the crisis has blinded us to how little we actually know about it, and that only by maintaining a rigorous awareness of what actually appears in the primary sources can we hope to enjoy the fruits of more nuanced interpretation.

Two phases of excavation were carried out by Essex County Council Field Archaeology Unit in 1993 and 1994–5; the first was located in the immediate hinterland of a known Late Iron Age and Roman settlement and the second across the eastern half of the settlement itself. The project was necessitated by the imminent development of an area totalling c. 29 hectares (71 acres) by Bovis Homes Ltd of which some 21 hectares were subject to varying degrees of archaeological investigation. While the 1993 phase of fieldwork was developer-funded, the 1994–5 excavation and all subsequent post-excavation work have been financed by English Heritage.

This article details the first unambiguous evidence for occupation in the Late Iron Age, dating to around 10 b.c.-a.d. 25, at the site that was to develop into the Roman Palace at Fishbourne (near Chichester, Sussex). The collection of sealed and well-dated imported and local pottery, accompanied by food refuse and a copper-alloy scabbard fitting, suggests significant activity at the site a generation prior to the Roman Conquest of a.d. 43. The material was found in the bottom of a ditch that had been deliberately back-filled. As such this discovery opens a new chapter in the remarkable story of Fishbourne.

The Chesters villa is situated midway between Chepstow and Lydney on the right bank of the Severn Estuary (FIG. 1). The residential quarters (centred on ST 59709870) lie on the shoulder of the Kidderminster Terrace (sands, gravels), about 250 m from the modern shore between the 11 and 16 m contours, and face south across the estuary. Between 1932 and 1935 C. Scott Garrett carried out a series of excavations, mostly in the field known as Lower Chesters south of the railway line. These revealed, on the eastern side, a residential range with a bath-block at its southern end. From this a courtyard wall was defined on three sides, with traces of structures along the south, and a subdivided, rectangular building at the northern end of the western wall. No evidence was found of a northern range, although rubble spreads on the surface of the field, and now aerial photography, make it clear that such a range existed. Indeed, its size, coupled with the incidence of mosaic tesserae in the ploughsoil, suggest that it was the most elaborate quarter of the villa. The cropmark evidence for this range is plotted on FIG. 1. Apart from one reference to iron slag as a ‘material which can be found almost anywhere on the site', the excavations provided no evidence of the role that iron-making might have played in the economy of the villa.

During the past few years several late Roman churches have been excavated inside the towns and forts along the Rhine-Danube frontier. These churches were built in the fourth and early fifth centuries, and were designed for congregational worship and the celebration of the Eucharist; in this respect, they differ from the churches which developed from memorial chapels in the cemeteries outside the towns. Baptism was one of the functions of these congregational churches, and at several of the excavated sites fonts have been found. These churches and fonts provide a number of parallels to an unexplained structure at Richborough and to various unexplained features of the excavations, suggesting that there was a congregational church of this type with a nearby baptistery inside the fort there.

The material upon which this study is based is the ornamented architectural stonework of Roman Britain. The decoration of architecture was an accepted feature of the way in which the Romans built, and is one which appears in Britain at all levels of achievement—in the comparative grandeur of the Corinthian capitals of the basilica at Silchester, in the neatly-turned columns which adorned such a villa as Great Witcombe (PL. XIX, A), and in the humble column-bases of the auxiliary fort at Housesteads.

I shall assume that Gildas wrote towards the middle of the sixth century and that he composed the De Excidio much as Mommsen printed it. I shall also assume that, apart from the works of St Patrick, his book is the only British literary source of authentic information about the history of Britain in the mid-fifth century. To be sure, Bede's conjectures deserve careful attention, but scholars (apart from Dr D. N. Dumville) have unwisely disregarded Mommsen's words about the De Excidio: ‘unicus liber omnino qui eius saeculi condicionem insulae publicam aliqua luce illustret’.

This article is a speculative essay about one aspect of authority and power in the late Iron Age. Excellent narratives exist which discuss the political context of the rise of the polities of Verica and Cunobelin in the South East of Britain, but over the last few decades the place of druids in this story has been neglected. However druids did exist, and they need to be worked into our narratives of the past. What follows leads to a discussion of their nature and function in society, and of their decline in importance during the large-scale social changes which took place in the generations immediately preceding the Roman conquest. The article uses a mixture of ethnographic, historical, numismatic, and archaeological evidence.