MELANDRA CASTLE. 95 



dimensions are identified as the actual imposts from tliese 

 pilasters. Two are complete, and fragments of others were 

 found in the enclosing wall of this field, and near a farm- 

 house in the neighbourhood. Other large squared stones had 

 undoubtedly been masoned into these pilasters. A mortice and 

 a stop in the chamfer of one of the imposts, readily give its 

 alignment with the tower wall, and accordingly its projection, as 

 represented in the detailed diagram. The spans of the voussoirs, 

 too, corre.spond satisfactorily with the width of the entrance,' 

 and with what is otherwise theoretically required to invite and 

 render possible a restoration of the whole gateway up to the 

 spring of its arches. It seems probable that on the right hand 

 was the main arch spanning a broad way for traffic, while on the 

 left was a narrower gate for the entrance of those on foot. This 

 type of gateway is very unusual in northern forts, and seems only 

 to have been known at Hardknott Castle, but it is still to be seen 

 in the Roman gate at Lincoln,* and Mr. Haverfield speaks of a 

 similar construction in some of the .smaller Roman forts of 

 Northern Africa. 



[In the restoration of the pilaster c, which was actually effected 

 on the ground, by some oversight the line of the stop and 

 mortice was built a little way from that of the tower wall, thus 

 giving the whole pilaster a projection somewhat too great. 

 Owing to the weight of stone, it was deemed inadvisable to 

 place it aright until the restoration of that opposite to it is 

 taken in hand.] 



None of the other gates are so well preserved, and it is 

 possible that this one to the east, leading to more level land, 

 was the most used and best built. But both that opposite to it 

 across the enclosure, and that to the north also, seem to have 

 been similar in plan. The fourth may have been smaller and 

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

 of which there is some suggestion in various stones found in a 

 cursory examination of it. 



* The Lincoln gate, however, is not really analogous. It is interesting 

 to compare the later work of the western gate at Richborough, excavated 

 last year. Arch" CaiWi, XXIV. (J.G., 1901.) 



