36 THE EXCAVATIONS 



evidently Iiad upper storey s.^^ It is quite possible that 

 the upper parts of these turrets were largely constructed 

 of wood. Yitruvius expressly recommends this as a 

 precaution: "so that, if the enemy obtain possession of 

 any part of the walls, the wooden communication may be 

 promptly cut away by the defenders, and thus prevent the 

 enemy from penetrating to the other parts of the walls 

 without the danger of precipitating themselves into the 

 vacant hollows of the towers." ^^ 



To sum up, the excavations in 1905 (coupled of course 

 with those of 1899) would seem to show that the three 

 double gateways at Melandra were massive stone structures 

 consisting of two double arches of equal span springing 

 from six piers and flanked by towers which may or may 

 not have had a useful chamber on the ground floor. 



THE SOUTHERN GATEVi^AY. 



Mr. Garstang's conjecture that both the northern and 

 western gates would be found to be " similar in plan " to 

 the eastern entrance turned out to be correct. He proceeds 

 (loc. cit., p. 95) : " The fourth may have been smaller and 

 spanned by a single arch, or even enclosed in a wooden 

 frame." The excavation of this gateway, of which, again, 

 no indication existed but a slight depression in the bank, 

 was commenced in April. The plan is given opposite. 

 It will be seen that the entrance took the form of a single 

 gateway, flanked by towers, the dimensions of which are 

 greater than those of the other flanking chambers. The 

 width of the gateway was about 10 ft., and the outside 

 measurement of the towers is 12 ft. by 11 ft. 3 in. The 



25. Cumb. and Westm. Antiq. and Arch. Soc. Proc, vol. xii., p. 383. 

 29. Vitruv. De Architect, i., 5. 



