^{page 229 note 1} See Archaeologia, xxxvi, pp. 418–37; lvii, pp. 295–316; lviii, pp. 119–52, 391–406; lix, pp. 87–124, 289–310; lx, pp. 111–30, 451–64; lxi, pp. 565–82; lxii, pp. 1–20, 405–48; lxiv, pp. 437–52.

^{page 230 note 1} For the dating of the separate fragments see the detailed list below, p. 240.

^{page 251 note 1} See especially Archaeologia, lviii, pp. 119–52; lix, pp. 87–124; lx, pp. 111–30.

^{page 251 note 2} Archaeologia Cambrensis, 1916, pp. 1–36.

^{page 252 note 1} Notably near no. 4 bastion, where it is also freely employed in the quoins of the bastion itself. The point is perhaps not irrelevant to the question of the relative dates of the wall and the bastions.

^{page 252 note 2} During the summer of 1930 a further bastion was found attached to the north wall of the town in close proximity to the north-west corner (see Archaeologia Cambrensis, 1931). It now seems likely therefore that all the Caerwent walls were equipped with bastions.

^{page 254 note 1} The Breaks are numbered from west to east.

^{page 254 note 2} A similar Break was found in the Roman wall at York immediately north-west of the east corner. ‘… There was a distinct join in the masonry from top to bottom. At the far side of that join the facing was rougher than on the other, and the foundation was not laid so deep.’ Journal of Roman Studies, XV (1925), p. 187.Google Scholar

^{page 257 note 1} = the South Gate.

^{page 257 note 2} or 286 ft. 6 in. to approximate internal centre of SE. corner.

^{page 258 note 1} The exterior face of the projecting gate-piers not improbably marks the forward limit of these returns. See p. 259, note 1.

^{page 259 note 1} It is possible to reconstruct the destroyed front of the gateway with some degree of certainty. If the outer returns be allowed the same dimensions (i.e. width and inward projection) and offsets as the inner, they can be shown to occupy the whole of the projecting piers between their forward edges and the existing (destroyed) front of the gate-jambs (see fig. 6). This suggests—and the present condition of the gateway is entirely consonant with the suggestion—(1) that the gate-front was deliberately dismantled so as to eliminate from the face of the defences an otherwise enclosed and covered space dead to the defenders; (2) that, therefore, the dismantling was contemporary with the blocking of the gateway. In the case of the North Gate, which also was blocked, there is definite evidence that the gate was already partially destroyed or dismantled before blocking, the blocking here being carried over the ruined lower half of the inner eastern pier.

^{page 260 note 1} i.e. beyond the plinth course.

^{page 261 note 1} Archaeologia, lx, p. 117.

^{page 261 note 2} Not improbably this debris, which included much painted wall-plaster, formed a pocket in the ramp similar to those behind nos. 3 and 5 counterforts. It was impossible to excavate behind no. 6 counterfort, owing to the risk of a fall.

^{page 262 note 1} Counterforts of similar character have been found attached to the interior face of the north wall also. ‘At a distance of 65 ft. to the east of the gate a counterfort 15 ft. long projects inwards, to the amount of 1 foot at the west end and 2 ft. 2 in. at the east end. It lies in both cases right over the rough stone foundation of the wall.… At the west end the lower part for 3 ft. 6 in. above the top of the upper set-off is not bonded into the main wall.… At the back of the counterfort a mass of stones and mortar has been placed to give additional strength, breaking into the slope of the [primary ?] mound, which has been cut away to give room for it.… There are three weep-holes in the counterfort similar to those in the main wall. Another similar counterfort 200 ft. to the west of the gate is 13 ft. 6 in. wide, projecting 2 ft. inwards, while a third, 253 ft. to the east of the gate, is 13½ ft. wide and projects inwards 18 in.’ Archaeologia, lix, pp. 94 f. and Archaeologia Cambrensis, 1916, pp. 12 f.

^{page 261 note 2} This interval covering as it does the South Gate makes it reasonably certain that the counterforts were sited along the wall without reference to the gate (cf. the case of the North Gate above).

^{page 261 note 3} or 222 ft. 6 in. to the approximate internal centre of the SE. corner. On the analogy of the SW. corner there should be a seventh counterfort in the SE. corner (now covered by the Norman motte), probably at the mean interval of 180–90 ft.

^{page 263 note 1} It is noteworthy that counterforts almost identical in character with those at Caerwent and at intervals nearly the same occur along the town wall of Silchester. ‘At intervals of about 200 ft. along the whole course of the wall are what look like internal buttresses. These are formed by carrying up the full thickness of the masonry, 9½ ft., from the bottom. The breadth of these buttresses is usually 12 ft.’, but some are as much as 15 ft. in breadth. Victoria County History, Hants, i, p. 354

^{page 266 note 1} See above, p. 252, note 1.

^{page 267 note 1} i.e. taking No. 1 Break to be approximately the centre of the south-west corner.

^{page 267 note 2} But see below, p. 273, note 2.

^{page 268 note 1} Bastions were attached to the walls of other Romano-British towns (as distinct from forts), e g. London, Caistor-by-Norwich, and Cirencester (?). ‘Leland, who visited Cirencester in the day of Henrv VIII mentions … “a fundation of towers” as “sometyme standing on the waul”, along the east front These “towers” were presumably projecting bastions, such as were common on Roman town-walls in the later Empire. None are now visible, but the bases of one or two near the Workhouse, on the west side of the town were … destroyed within living memory’ Archaeologia, lxix, pp. 166 f.

^{page 269 note 1} Ward (Arch. Camb., 1916, p. 16) states with reference to the bank inside the north wall of the town that ‘the summit of the long stretch of mound [= bank] exposed behind the north wall exhibited the curious feature of two irregular lines of longitudinal indents, roughly semicircular in section, and of varying widths. They had all the appearance of being the impressions of treetrunks, and it is not unlikely that these supported a timber breastwork or stockade’ No similar evidence was noted during the present work.

^{page 271 note 1} In this connexion the even distribution of the gaps along the line of the wall may be noted.

^{page 271 note 2} For this, see above, p. 252.

^{page 271 note 3} The buildings themselves do not necessarily antedate the wall, since buildings erected after the construction of the wall would naturally conform with its line. The only test of the relative date of a particular building must be whether it overlies the earthen ramp backing the wall or not.

^{page 272 note 1} Evidence has already been cited (above, p. 262, note 1) suggesting that the interval between counterforts on the Caerwent north wall also was 190 ft. Cf. the 200 ft. interval between counterforts in the case of the Silchester town wall.

^{page 272 note 2} Rectangular internal bastions (so called) which may have served such a purpose were attached to the Colchester walls. See S., J. R., IX, 1919, p. 141.Google Scholar

^{page 273 note 1} Ward (Arch. Camb., 1916, p. 29) states that the bastions ‘were floored with concrete on the surface of the natural ground within’. He is mistaking for concrete the layer of spalls and mortardroppings left by the builders.

^{page 273 note 2} Ward's investigation of (2) was not exhaustive. The point could not be generally elucidated during the present work, but in one case (no. 1 bastion) where it was fully investigated bastion and wall footings were found to correspond (fig. 8). As to (3) the difference is very slight.

^{page 274 note 1} Except for the coins and potters’ stamps this has not been published. For the coins see the Bulletin of the Board of Celtic Studies, ii, pp. 92 ff., and iv, pp. 99 ff; for the potters’ stamps ibid., V, pp. 166 ff.

^{page 274 note 2} Quantities of animal bones also occurred in and below the floors of all the newly explored bastions.

^{page 277 note 1} The date of the blocking of the South (and of the North) Gate in default of evidence must remain obscure. Reasons have already been adduced for assigning it to the final phase of the town's history (p. 259, note 1, above). Dr. Ashby in a foot-note to his report on the excavation of the South Gate in 1904 (Archaeologia, lx, p. 112, note) states: ‘From the evidence of coins it [i.e. the blocking] must be attributed, at earliest, to the time of Valentinian I, one of whose coins was found 4 ft. below grass level, 12 ft. north of the east pier of the gate. A coin of Helena was also found here, and one of Constantine the Great, 7 ft. below grass level, 5 ft. north of the west pier; and on the road level, 15ft. north of this pier, a coin with the legend URBS ROMA.’ But the evidence, so stated, scarcely appears conclusive. It is perhaps more to the point that the blocking of the North Gate, which presumably is contemporary with that of the South Gate, includes much re-used material.