This article continues the series on air reconnaissance in Roman Britain published by Professor J.K. St Joseph in the Journal of Roman Studies between 1951 and 1977. It is at once an affirmation of the undiminished value of this mode of archaeological prospection and an acknowledgement of the notable contribution made by that means to the study of Roman Britain by Professor St Joseph over a period of thirty-five years.

Brandon Camp lies just south of Leintwardine in the central sector of the Welsh Marches. A Roman road running south from Wroxeter through the Church Stretton gap crosses the river Teme at Leintwardine and skirts the east side of Brandon Hill before traversing the Aymestrey gap and running on to Kenchester; thence branches lead to both Gloucester and Usk. The hill projects into the flood-plain to form a notable feature in this sector of the Teme Valley (PL. XI A), rising to a height of 165 m above O.D.; it has a broad almost level summit which is easily ascended from the south but is protected around the northern edge by low precipitous cliffs. The north-western end is occupied by a pear-shaped hillfort of about 8 acres (3.2 ha) with an excellent outlook west and north. An upstanding rampart 3 m high marks the south and east sides of the Camp, but along the steep scarp it is only just discernible.

Historically dated pottery assemblages have been widely used in the establishment of ceramic chronologies, especially those currently used for samian ware.2 The methods employed in the construction of these chronologies have hitherto been comparatively unsophisticated and the results have rarely been subjected to critical scrutiny. Further, in most of the development and use of these chronologies it has been the presence of specific types (for instance stamps or decorated sherds) that has been considered significant rather than the overall composition of the assemblages. This is despite the fundamental assumption on which the chronologies are based, that assemblages deposited at or near the same time, within the same distribution network, will be similar in composition. This assumption has itself rarely been tested to establish the limits within which other factors determining pottery supply, use and deposition, may obscure these underlying chronological structures, and render them unreliable for the precise dating upon which Roman archaeologists rely.

The renewed excavations in the pre-Hadrianic area of the Vindolanda site have continued to produce large numbers of writing-tablets. The inventory numbers under which the finds of 1985 and 1986 are catalogued run as far as 670; many include pieces of more than one tablet but there is, of course, a large number of scraps which are either blank or have only exiguous traces of writing. It is gratifying to report that the 1985 tablets have now gone to join their predecessors of 1973–5 in the British Museum, with an agreement that subsequent discoveries will also reach the same destination. It is envisaged that there will be two further seasons of excavation in 1987 and 1988. A rough estimate of the bulk of the significant written material which emerged in 1985 suggests that it is approximately the equivalent of the corpus of texts published in 1983. The yield of 1986 was somewhat smaller but there are good grounds for believing that this was not because the progress of the excavation was coming to the end of the tablet deposit and there is every reason to hope for significant quantities of tablets in 1987 and 1988.

Although Deva was primarily a military centre, the installation of a garrison was quickly followed by the growth of a neighbouring civilian community. Where such extramural settlements grew up beside auxiliary forts they were classed as vici and their occupants as vicani, while those which developed outside legionary fortresses were, from the time of Hadrian at least, known as canabae and their inhabitants as canabenses. This distinction could become blurred, however, as some canabae are known to have been divided into a number of constituent vici. The population of these settlements included purveyors of the modest luxuries and entertainments which appeal to soldierly tastes as well as a wide variety of other traders and craftsmen, some of the latter perhaps being under contract to the army while the remainder were attracted to the vicinity of military camps by the opportunities for financial gain which their garrisons afforded. In addition, there were the wives and families of serving soldiers (although, until the early third century, it was a condition of service for the ordinary soldier that he be single, this did not prevent large numbers from taking common-law wives), veterans and their families, and any servants or slaves employed in the households of such people. The only civilians allowed to reside within military installations were the families of the garrison commanders and, in the case of legionary fortresses, those of the other senior officers.

The 1983 excavations at 6–12 Fenchurch Street produced several thousand pieces of painted wall plaster, which were recovered from the debris of a clay and timber domestic building destroyed in the Hadrianic fire. A full report on the plaster is to be prepared in due course, but two portions are of exceptional interest, and deserve immediate publication.

Twelve pieces of leather waste from various sites in the City of London and Southwark bear stamped or incised inscriptions, the majority of which seem to stand for personal names. Some occur on shoe sole off-cuts, others on waste pieces of thinner leather, and one on a discarded piece of hide from the head of a calf.

Windridge Farm lies approximately three-quarters of a mile to the south east of the Roman town of Verulamium (FIG. I) and about 400 yards from the eastern edge of Prae Wood. The finds of lead which form the subject of this paper were recovered over a number of years by several individuals with metal detectors and it is no longer possible to identify precisely and with confidence their exact find-spots. Certainly their distribution is fairly localised, the principal area of their recovery being centred on TL 124060. Sixty four lead objects have been recorded from the area and at least 50 more are thought to have been recovered. Thirty-seven are now in the possession of the Verulamium Museum (Accession Numbers SABMS 79.2960; 82.571–2; 84.1–25, 644–5, 1421–6,1435). The finders reported that the lead objects were found in small groups rather than individually, though no large hoards have been discovered. Although the area was field-walked by the local archaeological society, no further lead finds were recorded.

In recent years there has been much discussion of the ancient units of measurement used in both civil and military planning in Roman Britain, without great consensus being achieved. Clearly, this is not a subject for dogmatic assertion, but prolonged re-examination of the evidence has convinced the writer that the units of measurement employed in planning the town houses of Silchester and Caerwent were 7 1/2 and 3 3/4 pedes Monetales. It will be argued that the 7 1/2 and 3 3/4 p.M. units (hereafter referred to as units and half-units) were used not only to lay out the houses themselves, but also to calculate their positions within their respective insulae and, indeed, the overall sizes of the insulae themselves. If right, these suggestions clearly have far-reaching implications for the whole subject of Roman urban planning.

A building inscription from the fort at Ribchester (Bremetennacum) in the Pennines is preserved in the following drawing (RIB 589):

A fragment of a large building inscription was found in the courtyard of the principia at Ribchester in 1908 (RIB 591). It is restored: [Imp. Caes….]│Avg. [et … ]│Avg. [et …]│ Cafesari … vexillatio legionis sextae Vict]│ric[is … leg.]│Av[gg ….]. The repeated Avg. and the erased Ca in line 4 show that it was set up in 198–209.

Of the senior Roman officials who took part in the administration of Britain perhaps the most intriguing is Severus' procurator, M. Oclatinius Adventus. His service in the province is recorded for us on two inscriptions erected in the governorship of L. Alfenus Senecio between 205 and 207. The first was found at Chesters on Hadrian's Wall:

Between the usurpation of Constantine III and the death of Aetius there are a handful of dateable events associated with the British Isles. Of these the best recorded are the two visits of Germanus to Britain in 429 and c 435. The narrative account of these visits, however, is contained in Constantius's Vita Germani, a work in which literary and theological concerns may well triumph over factual accuracy. Speculation about sub-Roman Britain on the basis of Constantius's account is a fascinating, but dangerous, pastime. If we wish to stay on relatively firm ground, we must consider the voyages of Germanus, as well as subsequent contacts between Britain and the continent, in the context of events south of the Channel.

The archaeological evidence for the period spanning the end of Roman Britain and the start of Saxon England needs careful examination to avoid unwarranted speculation. The testimony of Gildas cannot be lightly disregarded and it is time that the only archaeological evidence produced to support his account of massacres in the towns, namely that cited from Caistor-by-Norwich, was scrutinized, particularly since the excavations in 1930 by the late Professor Donald Atkinson which produced the evidence have never been published. This has led to descriptions of scenes of horror as the Romano-British inhabitants of Caistor were put to the (?)sword by Saxon raiders in the early fifth century, the details of which are at least questionable (a building burnt down at the same time, unburied bodies, perhaps decapitated, associated with coins etc.).