ox A ROMAN FORT AT GELLYGAER, 453 



Buildings of a similar shape, but with stone pillars instead of wooden 

 posts, have been found at Chesters (Cilurmim), in Northumberland, 

 As these were divided into a number of narrow apartments opening on 

 to the veranda, it is probable that the Gellygaer examples were similarly 

 divided, only by wooden instead of stone partitions. These buildings 

 were prob.ably used for barracks. 



The other ' long ' buildings differed from one another in shape and 

 size, and it is likely enough that they were used for different purposes. 



A complete Memoir of the exploration, with plans, etc., drawn up by 

 Mr, John Ward, F.S.A., for the Cardiff Naturalists' Society, has now 

 been published. Application for copies should be made to the hon. 

 secretary of the Society, Dr. William Sheen, Cardiff, 



Silchester Excavation. — Report of the Committee, consisting of Mr. 

 Arthur J. Evans (Chairman), Mr. J. L. Myres {Secretary), 

 and Mr, E. W. Brabrook, appointed to co-operate with the 

 Silchester Excavation Fund Committee in their Excavations. 



The Committee have to report that the excavations at Silchester in 1901 

 were begun on May 10, and continued without break until November 13, 



The work was confined to the northern half of the town, on a strip of 

 ground lying to the east of insiilte. XXI and XXII (excavated in 1899), 

 and extending northwards from the modern road traversing the site to the 

 town wall. The area examined was nearly 6 acres. 



The ground in question was found to contain, in its southern half, 

 a square insula (XXVII) of large size, while the triangular piece to the 

 north proved to be an extension eastwards of insula XXII, of which the 

 larger portion was excavated in 1899. The extension contained only two 

 more buildings ; one towards the north ; the other towards the south, 

 perhaps a small house with an eastern apse. 



The westei'n half of insula XXVII contained the foundations of three 

 houses. Two of these are especially interesting on account of the addi- 

 tions that have been made to their plans. The northernmost house, which 

 by these additions was more than doubled in area, was oi-iginally a com- 

 plete example of the courtyaid type, with mosaic floors in most of the 

 rooms, and a number of interesting features. Two of the mosaic pave- 

 ments were sufficiently well preserved to justify their removal. The house 

 was evidently a half-timbered building, and in the added portion some 

 remarkable evidence was found of the method of construction and the 

 ornamental character of the half-timbered work. 



The second house was also of the courtyard type, but apparently of 

 less importance than its neighbour, and perhaps of later date. The addi- 

 tions to it included a number of winter rooms warmed by an elaborate 

 series of hypocausts, and a building of doubtful use exhibiting some un- 

 usual features of construction. 



The third house was of the corridor type, but its more interesting por- 

 tion had been to a lai'ge extent destroyed. 



The extensions to the houses noted above stretch so far across the 

 insula as to leave little room for any other buildings, and those that were 

 Hjicovered were not of an important character. 



