42 THE EXCAVATIONS 



on tlie Ober-germaniscli-raetisclie Limes nine have an 

 area of between 6,000 and 7,000 sq. yds., ten have an area 

 of between 24,000 and 26,000 sq. yds. (Melandra would 

 come balf-way between the two groups), the rest are much 

 larger. 



The variation in the dimensions of the forts suggests 

 the question as to how far these were determined by the 

 number of men to be accommodated, a point which it 

 would be out of place to discuss here. Apparently each 

 of these forts was garrisoned by an ala of cavalry or a 

 cohort of infantry,*^ both auxiliary troops. There is 

 reason to suppose that the forts at Manchester and 

 Melandra were both garrisoned by infantry. The cohort 

 of Tungrians at Borcovicium is supposed to have numbered 

 1,000 men. Mancunium covered the same area as Bor- 

 covicium. It is probable that the garrison at Melandra 

 did not much exceed half that number. 



Without doubt the fort that most resembles Melandra 

 is that of Hard Knott. The plans are almost identically 

 the same and apparently at both stations all but the 

 official buildings were of wood. Unfortunately, a plan 

 of Hard Knott to the standard scale has not been pub- 

 lished. I have, therefore, for purposes of comparison, 

 placed the plans of Melandra and Gellygaer,*^ both drawn 

 to the same scale, on opposite pages. An examination of 

 the two plans side by side will show the striking points of 

 resemblance, and perhaps it is not unreasonable to assume 

 (at least until the further excavation of Melandra has 

 disclosed the plan of the interior) that the arrangement 



42. Except the smaller forts. Mr. Haverfield estimates that some of 

 the smaller forts on the Danubian frontier may have been held by as 

 few as 50 men under a beneficiarius. {Athenauvi, October 22nd, 1892.) 



43. As explained above, I am indebted to Mr. J. Ward, F.S.A., for 

 permission to reproduce the plan of Gellygaer from his memoir on that 

 fort. 



i 



