a soft speculum-tool, or by taking an exact cast from the mirror, and seeing whether this also had magic properties. Govi thus challenged, at once disproved Brewster’s position, and established his own by showing that the character of the image, white lines on dull ground, is changed to that of dull lines on a brighter ground, by simply interposing in front of the mirror a convex lens. This effect can only be produced by difference of curvature of surface. Further, Govi performed the following striking experiment. Taking a Japanese mirror which ordinarily showed no magic effects, he heated it behind with a spirit lamp, when at once it acquired magic properties. He further improvised with a piece of daguerreotype plate, and a metal ring which he soldered to the back at as low a temperature as possible, a mirror which showed the same effects when, by heating, there arose an unequal expansion between the thin and the thick parts. [Illustration: PLATE III.] (_iv._) 1878-80. Researches of Ayrton and Perry, and those of Bertin, of Laurent, and of Muraoka. First, Ayrton and Perry set about procuring mirrors from the shops in Japan, and found that the vendors were in many cases wholly ignorant of the existence of the magic property. Further, the mirror makers could not say how the mirrors became magic, nor which of their mirrors were such; or they gave explanations subsequently found false, declaring the effect to be produced by application of an acid paste to etch the face prior to polishing. The investigators then, having procured some good mirrors, proceeded to test the various possible suggestions as to the origin, such as the supposed difference of density, or of molecular constitution, or the supposed inlaying of the face with inferior metal, or to supposed concealed scratches on the surface, or, finally, to differences of curvature. By simple experiments in which bright light was caused successively to fall upon the mirror in diverging, parallel, or converging beams, it was demonstrated beyond question that the last-named cause was the true one. The whole phenomenon is explained if once it could be established that the surface over the thick parts is flatter (less convex, or even actually slightly concave) than the surface over the thin parts. Using a large convex lens to converge sunlight that had been reflected from a mirror, they showed that by merely altering the distance of the screen on which the image was received they could make the image either positive or negative at will; a result impossible on any other hypothesis than that of differences of curvature of surface. This being established, they then investigated the process of mirror manufacture, to ascertain how these delicate inequalities of surface arise; and they found it to be due to an accident, or incident, of manufacture. All the finished Japanese mirrors may be observed to be slightly convex in the face. They are