York property owners are used to the fact that their buildings are perched on the top of a heap of ancient rubbish. It comes as no surprise when structures settle into the soft debris below, or crack over long forgotten watercourse, or break their backs over resilient Roman halls pushing up through the mouldering debris of later generations. There was a philosophical tone, therefore, to the press notice which revealed to a dismayed public in 1974 that W. P. Brown's store, long established on the corner of Davygate and St. Sampson’s Square, was breaking at the spine and would have to undergo surgery.

Property owners in York have also grown philosophical about the interests of York Archaeological Trust in the rubbish below and the appearance of an archaeologist in the store one busy day in March, as work began to save the building, cannot have been entirely unexpected. The management stoutly accepted the demands of science; the architect reacted with sympathy and interest; the contractor was suspicious but kind; while those who were efficiently digging holes for the new foundations at first found the sight of archaeologists lovingly dissecting their spoil a cause of gentle ribaldry, the more so since the earth had a rather strong and suggestive smell.

The Trust had developed a code of practice over the years for such watching briefs. There are emergency kits for excavation and recording, insurance cover, procedures for safety, and staff specialists available at a few minutes notice. There is also a cardinal rule that contractors work must not be hindered or held up. The W. P. Brown's experience was therefore as much teabreak archaeology as keyhole archaeology. Even those digging were amazed by how much work can be done between 12.00 and 12.45 if needs must. The small square holes for the new foundation pillars were being dug down very fast, and the needs where often very pressing indeed.

In the event Trust staff observed three of the four main staunchion pits; and in 1974 we had observed a tunnelling operation under the pavement of Davygate nearby when new lavatories in W.P. Browns’ basement were connected to the Davygate sewer.

The information from the four excavations was remarkably consistent. At the bottom of each cutting there were traces of Roman occupation, lying directly on the natural clay. In three of the holes these included well-built limestone walls. The walls had been demolished to a more or less constant level two or three course above the foundations. Above them was mortary soil, destruction debris and levelling material. Thereafter there were thick deposits of dark soil, even, often strong-smelling, and for much of its very considerable depth containing fragments of wood, straw and other organic material. The cross-section drawing gives a visual impression of these deposits in one of the 1975 cuttings. There were clear interfaces between layers suggesting that if ever an area excavation is possible on the site of W. P. Browns, clear occupation levels will be revealed. Samples for biological analysis should in the meantime provide a hint of medieval landuse in the area. Near the top of the section was a layer of roof tiles, perhaps from the destruction of a nearby medieval building; big above it a layer of decomposed wood, perhaps a later planked floor.

Find from the section suggest dates for the events deduced from the stratification. Pottery and a jet bangle confirm the Roman date of the stone walls at the bottom. The black accumulations above seem to be medieval, two 13^th century potsherds having been found only 40cms above the Roman destruction levels. Medieval pottery was found in a similar position in the 1974 sewer excavation in W.P. Browns. There was some immediately above Roman surfaces in the 1972-3 excavations at the Church Street Roman sewer site nearby; and medieval levels at Blake Street, also in the pratentura (front part) of the Roman legionary fortress, similarly lay directly on the Roman in places. All these observations, and the almost complete lack of Anglo-Scandinavian artefacts from these areas, begin to argue for wholesale clearance, and the demolition of surviving Roman buildings within the fortress praetentura, in Norman times.

W. P. Browns lies in a part of the praetentura normally occupied clear by barracks. The walls discovered all run parallel with the axes of the fortress and could well belong to barracks. If so, the 1975 walls are likely to belong to part of the contubernia (soldiers quarters) of a block set out on the module proposed by Mr. Wen ham for the next blocks, discovered in 1956-58 during building in New Street and Davygate. The1974 wall would be an internal wall in the centurial quarters of the adjacent block. If this hypothesis is correct it raises important questions. The first is that it implies that all barracks on the praetentura front were act out on parallel axes, as at Caerleon; the second is more difficult to believe: it implies that there were only ten barracks, or perhaps more likely nine, between the via praetoria and the rampart, instead of the usual twelve. The RCHM were long ago aware of the problem (Eburacum, p40) and presumed 'the buildings in the S angle.... must have been different'. Keyhole peeping often leads to the viewer putting two all two together to make six and it might for the moment be better to leave the question open until the next time a construction's keyhole allows an archaeological peep at this corner of the fortress. The problem could have been solved long ago if archaeologists of a previous generation had been on their toes when the ladies convenience was constructed in St. Sampson's Square, where barracks, if they ever existed, must have been encountered.

P. V. Addyman