THE EXCAVATIONS 57 



Trajan were found, as well as a large amphora with 

 pointed base, besides whetstones, and fragments of pottery, 

 lead and glass. It will be seen that the position of these 

 posts corresponds pretty nearly with that of the posts, lines 

 of which were found fronting the barrack-buildings at 

 Gellygaer, and which (as Mr. Haverfield suggested the 

 search for them) were known to the excavators there as 

 ■'Haverfield's posts." The excellent preservation of those 

 already found suggests that if a systematic excavation of 

 the northern area were undertaken, the plan of the build- 

 ings there might be recovered. It is possible to draw 

 inferences from the position of the other finds, especially 

 where there happens to be an accumulation near one spot. 



One of the interesting cases is that of the millstones, 

 of which a number were found together some years ago. 

 We found several more in the same place last year, and 

 no doubt others are there. (I also rescued a perfect speci- 

 men from the valley below, where I learnt it had been 

 rolled by boys at play.) It was disappointing, when we 

 had taken some pains to collect the millstones for a photo- 

 graph (see p. 8) to be told afterwards that three perfect 

 specimens were lying at a cottage in the neighbourhood. 

 As two of the Eoman millstones seemed to be composed 

 of a volcanic tufa I submitted one to Professor Boyd 

 Dawkins, who has identified it as having come from the 

 banks of the Rhine. One of these appears in the photo- 

 graph, in the foreground. 



In the early part of the year several sections were ex- 

 amined for finds, bvit they were quite unproductive, and 

 it is a question whether the more profitable method of 

 excavation would not be to set about recovering the 

 original plan of a large section of the interior. In the 

 late summer the sections numbered 136, 137 and 162 to the 

 W. of the central building were examined by Professor 



