These gateways have double portals, which were arched over, and were closed by two-leaved wooden doors, swinging on pivots shod with iron. The doors shut against a stone set up in the centre, or else against a stone threshold. The pivot-holes can very {86} often be seen, and sometimes the iron sheath is still in the pivot-hole, although the wooden door has perished. In the case of this fort of Cilurnum, as also at Amboglanna, there were six gateways altogether, two smaller ones to the south of the main eastern and western gateways; these had only one portal instead of two. [Illustration: Fig. 9.--Fortification Turrets from Trajan's Column. (After John Ward.)] A guard-chamber was always set on either side of each main gateway; there was a turret just within each rounded corner of the fort, and intermediate turrets were set along the walls. The appearance of these turrets may be surmised from the fortification-turrets shown on Trajan's column. {87} Chambers above gateways are also shown on Trajan's column, which suggest to the mind's eye a possible reconstruction of the single gateways of the Wall-forts. [Illustration: Fig. 10.--Fortification Gates from Trajan's Column. (After John Ward.)] It is generally accepted that the twin flanking-towers of the double fort-gateways, and also the Wall turrets, were carried up from one and a half times to twice the height of the wall (whether fort-wall or Great Wall). The rampart-walk is reckoned to have run at a height of from 13 to 15 feet from the ground, so that it would pass over a gate quite horizontally, and without steps. It would be continued right through the towers and turrets, passing through doorways in their side walls, and across the floor of their upper storey. Mr. John Ward (in _Romano-British Buildings and Earthworks_, p. 70) calls attention also to a gate {88} figured on a mosaic in the Avignon Museum, as suggesting the possible character of the double gates on the Wall-line. Frequently the great Wall aligns itself with the northern wall of the forts, as at Borcovicium, Æsica and Amboglanna; but sometimes, as here at Cilurnum, it strikes the fort about one-third way along the eastern and western walls, leaving one-third, including a gateway, to project northwards into the enemy's country. This looks as if the garrison could not have lived in a constant state of warfare. [Illustration: Fig. 11.--Gate of a Fort on a mosaic in the Avignon Museum. From _Collectanea Antiqua_. (After John Ward.)] Streets run between the north and south gateways and the east and west, in every case. Where they cross are the central buildings, which were called the "Forum" by Mr. Clayton. But that is a civil term. Probably the more correct term is "Principia," to indicate the H.Q. building of a military unit. {89}