the Castle was built; after that it became the south gate of the Castle, while the public road was diverted eastwards and left the city by a new exit opened for the purpose, called Wincheap Gate. About 1540—50 it was blocked up. About 1793 it was reopened, the Castle ditches were filled up, and Castle Street was carried straight out along its present line. At the same time the old arch was removed. But it is known through earlier accounts and sketches, of which perhaps the most accurate are those of Thorpe, made about 1771 It was a single gate, opening 12½ ft. wide and some 14. ft. high, with jambs composed of squared ragstone blocks and a round head turned in a double row of Roman bricks, and measuring 28 in. in thickness. This arch has been thought Roman since Somner, and perhaps longer, and the tradition is probably correct. The extant sketches of the arch certainly suggest Roman rather than Saxon or Norman work. Brent urged that it might be post-Roman work built with Roman materials, like so much else in Canterbury. But this is one of those easy explanations which must be used with caution, and the sketches show an arch built with more understanding of what an arch is than Saxon builders usually possessed. Unfortunately, the position of the gate does not help. A Roman gate must have stood somewhere here to admit Stane Street. But Worth Gate is 850 ft. outside the Roman inhabited area; part of the interval is filled with a large Roman cremation cemetery, while the street leading to the gate, when trenched in 1867—8, revealed no continuous and clearly ancient roadway beneath it. Two further pieces of evidence may be cited, though contradictory and of uncertain value. On the one hand, Pillbrow in 1868 found a wall of Roman character close to the gate (13). On the other hand, Brent in 1877, excavating at Wincheap Gate which pierces the town-wall just east of Worth Gate, found under it (as he asserts) only the flint and concrete foundation of the medieval wall, resting on the native brick-earth (Cant. Olden Time, 13 n). If he was right in thinking the flint and concrete medieval, there was no Roman wall here. But medieval concrete is hard to distinguish from Roman, and further search is needed.
   (15) We pass to the district east of Castle Street and St. Margaret’s Street. St. Mary’s Street supplies only a bronze latchkey (Cant. Olden Time, 49, p1. xviii). In St. John’s Street, Pillbrow met foundations, potsherds, and coins at 120 ft. from Castle Street, and again at the turn in the lane, but he gives no details (Arch. xliii, 159, nos. 38—40). Where St. John’s Lane meets Rose Lane and Wading Street he found Roman structural remains—painted stucco, burnt masonry and wood, and (at 10 ft. deep) a tessellated floor, and by it a cornelian intaglio afterwards pronounced modern by Fortnum (ibid. 157, no. 37; Arch. Cant. xv, 348). From the western part of Watling Street he records only a gold coin of Vespasian: here no ancient road lay beneath the modern street. But under the eastern part of Watling Street was a hard road of strong ballast, large flints, chalk and concrete, which sank gradually from 3 ft. to 6 ft. in depth as it approached Riding Gate. At one point near an Independent Chapel 140 ft. from Riding Gate, he found Roman potsherds, oystershells, charcoal and abundant ashes 10 ft. below the street-level and directly under the buried road. The depth of these remains is the same as that of the adjoining tessellated floor, and the charcoal and ashes are probably the marks of conflagration which occur all over Canterbury at the Roman level. About here the hard road stopped. Instead, a vast deposit of black vegetable matter, exceeding 14 ft. in depth, extended almost to the Gate. This deposit contained Roman urns with ashes or burnt bones inside, a Roman silver spoon, boars’ tusks, etc., while a rude cobble road crossed it at 3 ft. deep (Arch. xliii, 157 and plate). The vegetable matter probably represents an old watercourse, like those noted in the Parade and Iron Bar Lane (secs. 6, 18). The burials in it suggest that it, or some part of it (we are not told their precise position), had become dry in the Roman period. Pillbrow suggested that the road-surface noted under the eastern part of Wading Street was connected with that under Beercart Lane and belonged to a Roman Street which crossed the town to Riding Gate. But the road under Beercart Lane cannot be accepted as proven Roman (no.10), and the traces in Watling Street, with their shallow depth (3 ft. to 6 ft.) and with Roman remains actually underlying them, are equally doubtful. The name Watling Street is, of course, no evidence; it is the invention of an antiquary (p. 134).
   (16) Watling Street ends in Riding Gate. This, like Worth Gate, has long been imagined to stand on the site of a Roman gate. Some facts may be cited in support of that view. Riding Gate is ancient: a gateway stood here in A.D. 1040. It is also the starting point of the Old Dover Road, which is thought to represent roughly the line of the Roman highway to Dover, though Pillbrow, when trenching it in 1867—8, found no old road underlying it (Arch. xliii, 158). Moreover, early sketches of it show two brick arches (destroyed in 1782) which are pretty clearly Roman workmanship (Somner, Canterbury, ed. 1640, p. 19 Stukely, Itin. ed. 1724, p. 115, plate 96 here repeated) There are, it is true,. certain difficulties. Its position, with burials and a broad water channel behind it, is unsuitable; it lies some way from the Roman inhabited area; the traces of a Roman Street leading to it are doubtful, and the line of the Roman road to Dover is at this point unknown. Once more the final proof of excavation is required. If excavation should prove the gate, or the ramparts near it., to stand on the site of a Roman gate and town-walls, we should draw