Plate VI. depicts a rectangular mirror, 15 inches high by 10-1/2 inches wide, and weighing 5-1/2 lbs. It is the only one of this form that he has seen or heard of; though smaller square mirrors of 3 to 4 inches in the side are not infrequent. One of the latter, in his possession, an old mirror covered at the back with Chinese characters, has no magic qualities at all. The large rectangular mirror is slightly convex, but more so in its longer direction than in its breadth. The bamboos in the pattern, though not very highly raised, are in very sharp relief, the mirror being, apparently, an _ichi mai buki_, or artist’s proof. There are two raised lumps on the back, apparently the remains of the parts where the metal was poured into the mould; and, curiously enough, neither these nor the rims of the rocks in the foreground yield any image, though they are more highly raised than any part of the bamboos. [Illustration: PLATE V.] To complete the proofs that the effects are due to differences of curvature the author has made the following observations. By holding a magic mirror very obliquely to the light one can discern traces of the pattern in the face, especially if one is accustomed to examine optical surfaces for small inequalities of curvature. For example, on examining (at oblique reflection) in such a mirror the image of a horizontal window-bar or of the roof-line of a house, one sees the straight line slightly curved downwards out of the level wherever the image is made from a slightly concave (or less convex) part of the surface. Acting on this hint the author found that if one chooses as an object to be viewed in a mirror a pattern of narrow parallel straight lines, such as a finely-striped blind, the pattern on the back of the mirror can be dimly seen in the face, resembling that species of line-engraving, sometimes used for medallion portraits, in which the whole picture is crossed by lines from side to side, the lines being bent toward one another or widened out to give effects of light and shade. Another variety of the same experiment is to place a ruled diffraction-grating of 100 lines to the inch (ruled in a silvered surface on glass) close to a bright light, and with a short-focus lens project the lines upon a screen. Then interpose the magic mirror to throw the luminous lines upon another screen, when they will be seen with the usual magic image; the bright lines concentrating into the bright parts and avoiding the darker parts of the luminous pattern.