[5] Plan in Allcroft, op. cit., p. 686; the camp is described fully pp. 682-97.

[6] See Bruce, Hand-Book to the Roman Wall, 5th ed., 1907 (ed. R. Blair), pp. 19-21.

[7] The list from the Notitia Dignitatum is given, ibid., pp. 11, 12.

[8] The bank is, strictly speaking, the agger, the vallum being the rampart on the top of the bank.

[9] The large villas of Romano-British landowners, as at Bignor (Sussex), Chedworth (Gloucestershire), Horkstow (Lincolnshire), were within easy reach of the military roads, but were not directly upon them.

[10] The topography of Roman Lincoln is described by Dr E. M. Sympson, Lincoln (Ancient Cities), 1906, chapter I.

[11] See Archæologia, vol. liii., pp. 539-73.

[12] See [below] as to the blocking of the main gateways at Cilurnum after the building of the great wall. The small single gateways at Cilurnum are on the south side of the wall. At Amboglanna both gateways were south of the wall.

[13] Borcovicus is described by Bruce, u.s., pp. 140-60.

[14] Plan in Besnier, Autun Pittoresque, 1888. The north-west and north-east gateways of the Roman city remain, but the centre of the city was shifted in the middle ages.