Page:Roman Manchester (1900) by Charles Roeder.djvu/25

 a constant and considerable variation in the substructure of the wall, and it appears that the builders were mainly guided by the nature of the ground they had to deal with, adapting their material to the varying ground and levels. At one point they might have strong gravel, at another loose sand or again marshy, yielding ground and swamp, or the outcropping red rock. For the same reason the rough outline wall varied and oscillated between 5 feet to 7 feet in thickness.

—The fort on the eastern side was bounded by a fosse, of which Whitaker has left a description. On the south-east angle, upon the lower margin of the Medlock, a deep and narrow gully, cut through the solid rock, existed in 1765. From this point, he says, the ditch seems pretty plainly to have mounted directly up the little garden that now lies on the eastern side of Castlefield, the rocks of the right side of the garden appearing evidently cut away and sloping towards the west.

On the south it was naturally protected by the curve of the Medlock. The southern part of the Roman stronghold stood on the rocky ledge of the ridge or plateau, which commenced at Knot Mill, and extended along the northern banks of the Medlock. It may still be viewed in Castlefield Street, at the coal wharf, and along the towing path of the Rochdale Canal. The southern side rose on the crest of the rock and stood on higher ground than the northern part of the castrum, towards which the rock gently dips away. Green's Map of Manchester (1789) gives a good representation of the appearance and