46 THE EXCAVATIONS 



of a bank of earth, about thirteen feet wide, faced on the 

 outside with a four-foot wall, on the inside with one some- 

 what thinner.*^ The inner retaining wall was probably- 

 necessary there on account of the looser nature of the 

 earth. No inner retaining wall has been found at 

 Melandra, though Mr. Garstang mentions that " a row of 

 flat stones placed vertically, forty feet within the outer 

 wall may possibly have been designed to assist the align- 

 ment and construction." ^^ The defence of the Saalburg 

 fort is described^" as consisting of "a battlemented wall 

 which served on the inner side as revetment to an earthen 

 wall. . . . The rampart, 2| metres high, had a fortified 

 platform 3 metres broad, up to which a gentle incline 

 led." ^^ The Saalburg wall was about 1'9 metres thick. 



There is one other possible parallel to the Melandra 

 rampart, but it is in the defences of a city and not a fort. 

 The wall of the Roman settlement at Cirencester, known 

 as Corinium or Durocomovium, may still be seen on the 

 bank of the little river Chum, that flowed round and 

 possibly through it. Leland (Y. pp. 64, 65) speaks of 

 "the cumpace of the old waul" as "nere hand ii myles," 

 and adds " A man may yet walking on the bank of Chume 

 evidently perceyve the cumpace of foundation of towers 

 sumtyme standing in the waul." When the Bristol and 

 Gloucestershire Archaeological Society visited the site 

 some years ago [Proc. II. pp. 13, 14), there was still to be 

 seen " a perfect earthen bank which supported the Roman 

 wall." A correspondent informs me (April, 1906) that 

 this remains, and that in the course of the last three 

 months draining operations have uncovered another por- 



48. Bom. Fort of Gellygaer, plate iii., p. 32. 



49. Interim Refort. We have not seen these stones. 



50. Das RomerkasteU Saalburg. A von Cohausen and Jacobi, p. 24. 



51. See p. .37 and note 31. 



