[15] Plan in Allcroft, u.s., p. 322. As Burgh Castle had the sea on its west side, it possibly had no west wall. Another tower, on the east side of the north gateway, has fallen away from the wall.
[16] At Pevensey the foundation of the wall is of chalk and flint, covered in one part by an upper layer of concrete, composed of flints bedded in mortar. Below the foundation is a layer of puddled clay, in which oak stakes were fixed vertically at intervals. See L. F. Salzmann, F.S.A., Excavations at Pevensey, 1906-7, in Sussex Archæol. Collections, vol. li.
[17] Cilurnum is described by Bruce, u.s., pp. 86-119, with plan. See also the description and plan in An Account of the Roman Antiquities Preserved in the Museum at Chesters, 1903, pp. 87-120.
[18] This was not invariable. At Cilurnum the main street was from east to west, and this was also the case at Corstopitum (Corbridge-on-Tyne).
[19] In this case, the first cohort of the Tungri.
[20] The tenth cohort of the legion had its quarters here: hence the name.
[21] Or the east and west gateways, as already noted, at Cilurnum. The forum occupied the centre of Cilurnum, the praetorium forming a block of buildings east of the centre. The first wing or squadron of the Astures was stationed at Cilurnum.
[22] Prof. Haverfield holds the view that this southern extension is post-Roman. See Archæol. Journal, lxvi. 350.
[23] The same thing happened at Lincoln, where the eastern wall of the city followed a line now covered by the eastern transept of the cathedral.
[24] Wat’s dyke, of which remains can be traced south of Wrexham and near Oswestry, was to the east of Offa’s dyke.