The modelling of the figures was done by hand in the case of the earlier fabrics, and of small objects such as the toys and dolls; the clay was worked up into a solid mass with the fingers, and the marks of these, left while it was wet, may still be often seen. Subsequently the use of moulds became universal, the final touches being given to the figure either with the finger or with a graving-tool, traces of which are often visible on the faces and hair of the Tanagra figures. These were invariably moulded, and the finer ones show traces of having been most carefully touched up. There is a pretty epigram in the Anthology,[404] which seems to imply that the wheel was sometimes brought into use for modelling figures, perhaps for the first rough outlining. A statuette of Hermes is supposed to say: The rolling circle of the potter’s wheel Me, Hermes, formed, of clay from head to heel. Mud-made, I lie not: the poor potter’s art, Stranger! was ever pleasant to my heart. (MACGREGOR.) The process of moulding gave scope for reducing the “walls” of the figure to the smallest possible thickness, thereby avoiding the danger of shrinkage in the baking; it also rendered them extremely light, and allowed of great accuracy in detail. A model (πρότυπος) was made in terracotta with modelling-tools, from which the mould (τύπος) was taken, also in terracotta, ------------------------------------------------------ PLATE V [Illustration: MOULDS FOR TERRACOTTA FIGURES, WITH CASTS FROM THE MOULDS. 2, 3. ARCHAIC, FROM RHODES; 1, 4. ARCHAISTIC, FROM TARENTUM (BRITISH MUSEUM). ] ------------------------------------ usually in two pieces, which were then baked to a considerable hardness. From this mould the figure was made by smearing it with layers of clay until a sufficient thickness was reached, leaving the figure hollow. The back was made separately, either from a mould or by hand, and then fitted carefully on to the front, the join being concealed by a layer of wet clay. The base was usually left open, and a vent-hole was left at the back which may have served a double purpose—first to allow the clay to contract without cracking, and subsequently in some cases for the suspension of the completed figure. The heads and arms were usually moulded separately and attached afterwards, and altogether the average number of moulds employed—say for a Tanagra figure—was four or five. M. Pottier[405] quotes an instance of an Eros from Myrina which is made up of no less than fourteen; yet it is not a specially complicated figure.