further mention of the site occurs in Roman literature, and some Possible references in Celtic poems are neither certain nor helpful.⁸³
   Let us now review this patchwork of shreds and tatters, and see whether anything like a convincing pattern can be made of it. One useful fact is clear at the outset. If we exclude the doubtful fragment of wall far to the north, at the end of the Folkestone road (No.11), and the timber quay or causeway to the east of the river (No.10), all the Roman structural remains noted by Mr. Amos and his predecessors in the valley of the Dour fall within a compact and fairly well-defined area. This area extends from the vicinity of Dowgate, Princes Street, and the eastern part of New Street on the west to Gaol Lane, the eastern half of the Market Square and the western part of Stembrook Street on the east; and from the northern side of St. Mary’s churchyard on the north to a line just north of Queen Street on the south. From east to west the area measures about 430 ft., and from north to south about 550 ft. It covers therefore nearly 5½ acres, which may be compared with the 5 acres of the Richborough fortress as originally laid out.
   It may be possible, with the help of Mr. Amos’s observations, to define this area of occupation yet more closely. The most southerly, or seaward, structure which may be ascribed to the Roman period is the 50-yard length of thick wall found to the north of, and nearly parallel to, Queen Street. In ancient times a great bank of blown sand had accumulated against its southern face. At what period this accumulation occurred we cannot say, but the southern face of the wall seems to have perished before it was thus covered, and a well and other structures thought to be of medieval date had been built into the sand, which had therefore preceded them. This heavy layer of sand has been observed at many points between Queen Street and the present beach—at the eastern corner of Queen Street itself, in Last Lane, King Street, Chapel Street, Snargate Street and Townwall Street—whilst, as a further indication of the comparatively early date of the drift, it is noted that a medieval crypt in Bench Street is built into it. The evidence thus enables us to say that the whole area between the Roman wall in Queen Street and the present beach lay under sand by the Middle Ages. Small patches of brown sand are occasionally observed to the north of the wall, as on the Carlton Club site in the Market Square; but there can be little doubt that the line of the wall marks a definite break in the drift, and that the barrier was therefore present at a relatively early date. Moreover, the wall itself seems to have borne certain of the characteristics of the Saxon Shore. Its original width, in the absence of its southern face, is not known, but it was certainly more than 6 ft. In Roman building such a thickness is abnormal save in defensive works. It was built of re-used Roman material, including a sculptured head of late second- or early third-century date. It was therefore constructed not earlier than the latter part of the Roman period; and the use of old building-material and sculpture finds an easy parallel in the Saxon-Shore defences of Richborough and Lympne (pp. 30, 58). The northern side of the wall retained its facing, which included, incidentally, much of the tufa
   ⁸³   Sir John Rhys suggested that Dover might be the Dybrys, Dybyr and Dyfrau mentioned in the Book of Taliessin (Skene, Four Anct. Books of Wales, ii, 198). If this conjecture be correct, we should have evidence of Irish seamen in the Channel, probably pirates, in the fourth or fifth century. But we should get no further evidence as to Dover itself.