the north —eastern corner, the second-century courtyard—house had  likewise been laid low, for the great rampart sprawled across it. But the traditions of this much—built—on site were still kept alive. Here facing up with and slightly overlapping the abortive foundation of the eastern defensive wall, was now erected a small bath—building, consisting of only four rooms, with the usual furnace and hypocausts. A coin of Tetrictis I (A.D. 268—273) found beneath one of the floors shows that the building was not earlier than that date. How long the building remained in use there was no means of ascertaining.
   Apart from some uncertain traces in the north—western quarter, there is at present no other stone building which can definitely be associated with the fortress. of slighter evidences of fourth-century occupation—burnt clay floors, vestiges of huts and the like—there are many. But the contrast between the solidity and permanence of the defences and the meagreness and evanescence of these structures within them is nothing less than astonishing. It would seem that the gaunt shell of Richborough was always a shell; that it was a corral into which men and materials could he herded in an emergency, rather than a permanently garrisoned stronghold either in the full sense of the earlier Roman forts or even in the more restricted sense of the later medieval castles. As a temporary refuge, the absence of bulk stone buildings would increase its capacity, whilst the towering walls would keep timber hutments tolerably sale from hostile firebrands. Along lines such as these it may ultimately be found possible to explain the paradox; but the solution cannot approach certainty until other Saxon-Shore forts, such as Burgh Castle, have been called to witness by the excavator.
   It remains to notice certain other structures of various Roman dates on and about the hill of Richborough. Within the lines of the fortress itself have been found, close to the western side of the platform, the remains of an oblong building, resembling in plan a small classical temple with cella and eastern portico. The foundations, which were in bad condition, were probably of third or fourth century date, and the date of the structure is doubtful. About 80 yards north-west of it was a small hexagonal fountain, also of late period.³⁴ Outside the fortress many miscellaneous remains have been noted to the west and south. Long ago Camden recorded marks of crossing streets as visible in the growing corn. His statement is obscurely worded. Probably he refers to the ‘cross’ inside the fort.³⁵  But he has been generally taken to mean streets outside the fort, and this has at any rate prompted others to observe indications there. Stukeley thought he saw the lines of streets in the corn outside the fort; Boys actually mapped some roadways in front of the west gate; others have noted similar indications more recently. These surface indications are not, perhaps, safe guides, since they are sometimes due to natural masses of gravel. Certainly, trial-trenches both in 1887 and more recently have only partially confirmed them. Dowker, in the earlier year, found abundant traces of Roman occupation of a kind, both west and north of the fort—one or two bits of roadway, scattered foundations, traces of buildings destroyed by fire, a brick floor, a rubbish pit, much
   ³⁴ First Richborough Report, 19.
   ³⁵  Camden, Britannia (ed.1590), 205-6;  ‘hodie arvum est, in quo platearum tractus cum seges succreverit, se intersecantes videas. (Ubicunquc enim plateae duxerunt, rarescit seges) quas vulgo  S. Augustines Crossere appellat. Et semiruta quaedam arcis moenia solum supersunt, quadrata forma  sabulo tenacissimo materiata’ Camden seems, at first reading, to distingush the fort from the remains outside it, and to place the marks in the corn among the latter. But his reference to the 'cross' definitely us inside the walls.