N Britain air-photography has proved to be one of the more important tools of the historian and archaeologist working in the Roman period. The principal results of aerial reconnaissance since 1945 have been described in a notable series of papers by the writer's colleague, Professor J. K. St. Joseph, published in the Journal of Roman Studies.' Most of the discoveries reported had been made on flights sponsored by the Committee for Aerial Photography of the University of Cambridge. A fuller treatment of the photography of walled towns has also appeared;z the present paper likewise is devoted to a single class of site in greater detail than is feasible in a periodic general review. Some of the sites to be described are hitherto unpublished; others, previously noted in brief, now receive a more extended commentary.3 Roman villas are numerous in Britain,4 but are relatively seldom detected from the air. Positive identification of a previously unsuspected site calls for the recognition of at least one building, whether of stone or timber, whose plan is distinctively Roman. Only in the most favourable conditions do crop-marks afford detail of the quality required. The post-holes and narrow foundation-slots of timber houses are only revealed by crop-marks of particular fineness and clarity, and their delicate traces are all too easily lost in a damaged, mottled or windblown crop. The lines of stunted growth produced by stone foundations are normally slow to appear, and it may take conditions of drought to bring out enough of a building's plan to establish its character beyond doubt. On a site already identified by fieldwork and provisionally dated to the Roman period the criteria are less stringent. It may be sufficient to observe in a bare field the vague shapes assumed by areas of building debris pulled to the surface by the plough. Again, the crop-marks of a rectangular enclosure-ditch could be supposed to mark the nucleus of a known villa whose actual buildings remained unseen. Yet even on known villa-sites apparently suitable crops can yield no information for years together. One of the ways in which this can come about is well illustrated by the site at Welton Wold (East Yorks., SE 972282), where a small corridor-house was exposed during emergency excavations on the edge of a large chalk quarry.5 Continued ploughing had removed all remains above the