54 THE EXCAVATIONS 



the result of several measurements) to be roiigUy tlie arc 

 of a circle of 32 foot radius. This was afterwards found 

 to be exactly the figure obtained at Brougb.^^ Tbe walls 

 of the corner tower at Brough, however, were splayed. The 

 two best preserved towers at Melandra were excavated by 

 Mr. Garstang, and he records the interesting fact that in 

 one or two instances he found that the mound was piled 

 against the walls of the towers (p. 92). At the two other 

 corners we found only the core remaining, and this may 

 account for the apparent inequality of the Melandra tur- 

 rets, as shown by the plan. These structures are, however, 

 unequal in other forts. ^' The photograph opposite shows 

 the rounding of the wall at the N.E. corner, where, though 

 the walls of the tower are missing, two courses of the outer 

 rampart remain. ^^ 



THE CENTRAL BUILDING. 



No important work has been done here during the year. 

 The clearing of the floor of the central room brought to 

 light a circular stone lying a few inches below the surface 

 of the floor in the middle of the room. The western half 

 of the courtyard has yet to be examined. 



The Via Principalis, which is in good preservation, had j 

 already been uncovered. The excavation of the north 

 gate brought to light the remains of a hard concrete road 



66. Proc. Derb. Arch. Soc, 1904, p. 10. The radius of the curve at 

 the Saalburg was 12 metres. (Op. cit., p. 25.) 



67. Cf. e.g. Hardknott, where the side measurements vary from 8ft. 

 Sin. to 13ft. Sin. The turrets at Borcovicium show the same irregularity. 



68. It will be interesting here to refer to the fact that the recent ex- 

 cavations at Castlecary on the Antonine vallum have brought to light 

 " the first Roman wall-tower met A^-ith in Scotland." Proc. Soc. Ant. Scot., 

 Ap.,1903, p. 11. 



