50 PROCEEDINGS AT MEETINGS OF

lime and sand, and a third chamber, laid as far we could observe with square red tiles, of fine and brittle material. " Whether these remains had any connexion with Musbury Castle, an ancient encampment about two miles distant, I am not competent to say." The floor presents considerable elegance and variety in its design. The introduction of a multangular figure in the centre occurs in other examples; but the four interlaced circles over which it is laid, n^ithout combining with them, form a feature of more rare occurrence. The irregularity and inferior design of the central hexagon, lead to the conjec- ture that it may not have been part of the original work, but inserted possibly, to repair some injury which the floor had suffered. The looped-chain pattern surrounding the whole is not uncommon ; it occurs at Woodchester and other Roman sites.' The roofing-tiles, above-men- tioned, usually of stone, the form being in this instance a long irregular pentagon, have been found elsewhere in Roman villas in England, and more commonly of an irregularly hexagonal shape, as those at Bisley, Glou- cestershire, figured in this Journal, at Mansfield Woadhouse, Notts, and at Woodchester.-' In every instance they are perforated near one end, and ■were thus attached to the woodwork of the roof by means of iron nails. During the past year some further remains have been found, described by Mr. Tucker as those of a bath, and situated about twelve or thirteen yards south of the pavement, in a direct line Avith the eastern wall of the room. The form is octagonal, the dimensions are as Aillows, — depth, three and a half feet ; width from side to side, where there arc no benclies, eleven and a half feet ; where the benches occur, ten feet. They measure two feet in height. The floor is laid with tessera) of pale fawn colour, and it is almost perfect. Tlie same roofing-tiles occurred, as before des- cribed, and red floor-tiles were also found. Mr. Tucker reported that the tesselated pavement had become soft and had lost much of its colour. These discoveries supply an interesting addition to the list of vestiges of Roman occupation on the confines of Dorset and Devon ; it is, however, highly probable, that so agreeable and salubrious a part of the southern coasts was not neglected by the colonists from Rome, with the facilities also of access by the British Ikcneld Street, running westward from Dor- chester, scarcely a mile north of the spot where the remains found by Mr. Tucker are situated, as also by the branch of the Fosse-way crossing the Ikeneld at Axminster, and passing at about a mile west of Uplyme, on its course towards .Seaton, the supposed Moridinuun of the Romans. Roman coins have been found at Axminster, and in several places in the vicinity. An urn containing a large Tiumber of Roman coins was found in llolcombo Bottom, in Uplyme parish, in removing a heap of stones provincially called a "stone barrow," and other vestiges are described by Mr. Davidson in his " British and Roman Remains " near Axminster. A remarkable discovery near tlie Ikeneld way, in Uplyme parish, deserves notice. In 1817, a labourer digging a hole for a gate-post turned up an ornament of pure flexible gold, about ftnnteen inches long, rather more than an eighth of an inch in (liametcr, except towards the ends, where it gradually became ' LyH(>u'» W'ooilclicBtcr, j)I;it<'H xv. imd llicy urc (icscrilitMl as slates ; Lysoii's xxi. fi^. "j;'!. Woodclu'stcr, pi. xxviii, lijf. (i, tlicw' last - Arcliji.Mil. .Jiiurii. vol. ii. ji. 11 ; an- of tin- f^ritty sIdik- IouikJ near Jh'islol Archoiol. viii. pi. xxii ; in this iiistiiiici; or tliu l'"orcNt ol' Diaii.