In Fig. 113 is shown one of the celts from the hoard discovered in the Isle of Harty,[398] Kent, to which I shall have to make frequent reference. Besides eight more or less perfect unornamented socketed celts, various hammers, tools, and moulds, five celts of this type were found. Although so closely resembling each other that they were probably cast in the same mould, in fact in that which was found at the same time, there is a considerable difference observable among them, especially in the upper part above the loop. In the one shown in the figure there are three distinct beaded mouldings above the loop, and above these again is a plain, somewhat expanding tube. In one of the others, however, there are only the two lowest of the beaded mouldings, and the upper half-inch of the celt first mentioned is absolutely wanting. The three others show very little of the plain part above the upper moulding. As will subsequently be explained, the variation in length appears to be connected with the method of casting, and to have arisen from a greater part of the mould having been “stopped off” in one case than another. [Illustration: Fig. 114.—Harty. ½.] [Illustration: Fig. 115.—Dorchester, Oxon.] It will be noticed that the “flanches” on these celts are placed below the loop and not close under the cap-moulding. The beads which form them are continued across the sides. Running part of the way down inside the socket are two longitudinal ridges which are in the same line as the runners by which the metal found its way into the mould. The vertical ridge above the topmost moulding shows where there is a channel in the mould for the metal to pass by. If the celts had been skilfully cast so that their top was level with the upper moulding, no traces of this would have been visible. In Fig. 114 is shown one of the plain socketed celts from the same hoard. The mould in which it was cast was found at the same time, as well as the half of a mould for one of smaller size. The five other plain celts from the same hoard were all rather less than the one which is figured, and appear to have been cast in three different moulds, as the beading round the top varies in character, and in some is double and not single. The two projections within the socket are in these but short, though strongly marked. In the British Museum is a celt of this kind, 4 inches long, found at Newton, Cambridgeshire, which on its left face, as seen with the loop towards the spectator, has a small projecting boss 1½ inch below the top. Five socketed celts of this plain character (2½ inches to 3¾ inches) were found together at Lodge Hill, Waddesdon, Bucks, in 1855, and were lithographed on a private plate by Mr. Edward Stone.