III. made use of it in certain parts of his temple at Medinet Habu; but, as a rule, it was preferred for stelae and small monuments.  The last-named method lessened not only the danger of damage to the work, but the labour of the workman.  It evaded the dressing down of the background, which was a distinct economy of time, and it left no projecting work on the surface of the walls, the design being thus sheltered from accidental blows.  The intermediate process was, however, generally adopted, and appears to have been taught in the schools by preference.  The models were little rectangular tablets, squared off in order that the scholar might enlarge or reduce the scale of his subject without departing from the traditional proportions.  Some of these models are wrought on both sides; but the greater number are sculptured on one side only.  Sometimes the design represents a bull; sometimes the head of a cynocephalous ape, of a ram, of a lion, of a divinity.  Occasionally, we find the subject in duplicate, side by side, being roughly blocked out to the left, and highly finished to the right.  In no instance does the relief exceed a quarter of an inch, and it is generally even less.  Not but that the Egyptians sometimes cut boldly into the stone.  At Medinet Habu and Karnak—­on the higher parts of these temples, where the work is in granite or sandstone, and exposed to full daylight—­the bas-relief decoration projects full 6-3/8 inches above the surface.  Had it been lower, the tableaux would have been, as it were, absorbed by the flood of light poured upon them, and to the eye of the spectator would have presented only a confused network of lines.  The models designed for the study of the round are even more instructive than the rest.  Some which have come down to us are plaster casts of familiar subjects.  The head, the arms, the legs, the trunk, each part of the body, in short, was separately cast.  If a complete figure were wanted, the disjecta membra were put together, and the result was a statue of a man, or of a woman, kneeling, standing, seated, squatting, the arms extended or falling passively by the sides.  This curious collection was discovered at Tanis, and dates probably from Ptolemaic times.[38] Models of the Pharaonic ages are in soft limestone, and nearly all represent portraits of reigning sovereigns.  These are best described as cubes measuring about ten inches each way.  The work was begun by covering one face of a cube with a network of lines crossing each other at right angles; these regulated the relative position of the features.  Then the opposite side was attacked, the distances being taken from the scale on the reverse face.  A mere oval was designed on this first block; a projection in the middle and a depression to right and left, vaguely indicating the whereabouts of nose and eyes.  The forms become more definite as we pass from cube to cube, and the face emerges by degrees.  The limit of the contours is marked off by parallel lines cut vertically from