If too much tin be added, the lustre diminishes. The correct proportion is, therefore, attained when a further small addition of tin produces no apparent increase of brightness or fineness of grain. About three-quarters of the tin may be added at first, and the other quarter added with testing as described. The alloy is allowed to cool until on skimming the surface the metal appears bright and remains so without losing its lustre by oxidation for a sensible time; it will still be quite red-hot. Fig. 59. Fig. 60. As the speculum alloy is too difficult to work with ordinary tools, it is best to cast the speculum of exactly the required shape and size. This is done by means of a ring of iron turned inside (and out) and on one edge. This ring is laid on a plate of figured iron, and before the metal is poured the plate (G) (Figs 59 and 61) is heated to, say, 300 deg. C. In order to avoid the presence of oxide as far as possible, the following arrangements for pouring are made. A portion of the lower surface of the ring is removed by radial filing until a notch equal to, say, one-twentieth of the whole circumference is produced. This is cut to an axial depth of, say, half an inch. A bar of iron is then dovetailed loosely into the notch (Fig. 60, B), so that it will rest on the iron plate, and half fill the notch. The aperture thus left forms the port of ingress for the hot metal (see Fig. 61, M). A bit of sheet iron is attached to the upper surface of the ring, and lies as a sort of flap, shaped like a deep shovel, against the outside of the ring overhanging the port (Figs. 59 and 61 at F). This flap does not quite reach the iron plate, and its sides are bent so as to be in contact with the ring. A portion of a smaller ring is then applied in such a manner as to form a pouring lip or pool on the outside of the main ring at E, and the metal can only get into the main ring by passing under the edge of the flap and up through the port. This forms an efficient skimming arrangement. The process of casting is carried out by pouring steadily into the lip. To avoid air bubbles it is convenient to cause the metal to spread slowly over the chill, and Mr. Nasmyth's method of accomplishing this is shown in the figure (61). The chill rests on three pins, A B C (Figs. 59 and 61). Before pouring begins the chill is tilted up off C by means of the counterpoise D, which is insufficient to tilt it after the speculum is poured. It is important that the chill should be horizontal at the close of the operation, in order that the speculum may be of even thickness throughout. This is noted by means of levels placed on the ring (at K for instance). Fig. 61.