that was always dear to the heart of the Roman builder; and the contrast between the well-preserved northern and the perished southern face of the wall at once recalls the common contrast between the inner and outer faces of a Roman fort-wall the inner comparatively sheltered and further preserved by the accumulation of occupation-earth against it, or even by a protective bank; the outer exposed to the weather and (in this case) the sea, and more readily attacked by the stone-robbers of later ages. Only the seeming inadequacy of the mortar in the core of the wall might appear to suggest a medieval rather than Roman builder. On the other hand, medieval mortar was normally good, whereas Roman mortar, as in the walls of Caerwent, was sometimes spread in intermittent layers in such a manner as to leave much of the rubble of the wall ungrouted. Such a wall, exposed to sea-weather, would rapidly disintegrate. On the balance of probability—we cannot say more— the Queen Street wall is more likely than not to have been the seaward rampart of the Saxon-Shore fortress.
   On the eastern side of the area, one definite landmark seems to be vouch safed to us. The inchoate mass of masonry, some 12 ft. thick as preserved, found at the junction of Church Street with the Market Square, can scarcely be other than a fragment of the Roman defences. Mr. Amos thinks that it was a part of a bastion. He is probably right, since the thickness of the unfaced fragment is considerably greater than that of the normal Saxon-Shore rampart. Its central position in the eastern side of our Roman area suggests that it may even have been a part of a gateway. Like the Queen Street wall, the ‘bastion’ contained re-used building material, and, if the tufa-facing observed by Mr. Amos near the bastion was a part of it, or of the adjacent rampart, we have a further slight constructional link between the two groups.
   A line drawn through the ‘ bastion,’ at right angles to the Queen Street wall, passes through or close to the substantial wall found under the eastern wall of the Duchess of Kent public-house on the south side of the Market Square. Little is known of this wall; but the ground immediately to the east slopes away from it, was apparently under water (see above, pp. 44—5), and was presumably therefore outside the defences.
   On the northern side, the Roman defences cannot have stood far to the north of the Roman building under St. Mary’s church (No. 2), for the low-lying fields between the churchyard and the Dour were marshy up to1800. The only structural evidence which may have a bearing upon the problem is the tough fragment of masonry found long ago near the Biggin Gate (No. 1). If this was not a part of the medieval wall, it was probably a relic of its Roman predecessor. The details, however, are too badly recorded to support definite conclusions.
   On the western side, the limit of Roman building (as known to us) is marked by the tufa-faced wall, substantial, but of unascertained thickness, at the south-west corner of the former ‘ Royal Oak’ yard (No. 3). The wall had an offset on its eastern side and ran approximately north and south. It was thought at the time of its discovery to be town-wall, although it lay off the medieval line. But further details of it are badly needed. Meantime, the apparent absence of Roman structures to the west of it, and its suitable distance from the Church Street ‘ bastion,’ suggest the possibility that it may be a relic of the western rampart but fall far short of proof.