About the same time Mendenhall, then in Japan, made further observations, which were communicated to the American Association for the Advancement of Science at its meeting at Cincinnati. Mendenhall was followed by two Japanese observers,--both trained in physical research in Europe,--Goto and Muraoka, the latter of whom, while confirming the main propositions that the effects are due to differences of convexity, and that the convexity is acquired during the process of _mege_ or scoring over with the _megebo_, gave some further details learned from the mirror-makers of Tokio. He showed that any plate, if thin enough, whether of bronze, brass, copper, lead, zinc, iron, or glass, acquires the property of convexity on being scored over with scratches; and he came to the conclusion that this surface-expansion on the scored side of the metal arose from some sort of release of molecular tension across the lines so scored. Like Ayrton, he held that this production of convexity on the scored surface explained the phenomenon that a scratch made by a file or pointed tool on the back causes a corresponding bright line to be reflected from the face. In a second article Muraoka sought to prove that the differences of curvature are not really due to differences of pressure during the operations of scoring and surfacing, but are due to the unequal rising-up of the metal in the thick and thin parts when the surface is caused to expand by scoring it over with scratches. [Illustration: PLATE IV.] In May, 1886, Professors Ayrton and Perry, having discovered that the act of amalgamating with mercury the surface of a thick brass bar produces a powerful expansion of the amalgamated surface and bends the bar convexly, made the further suggestion that the employment by the Japanese mirror-polishers of a mercury-amalgam might assist the operation of the _megebo_ in producing the differences of curvature between the thinner and thicker parts. The author having from time to time picked up a number of Japanese mirrors from dealers in oriental curiosities, and having repeated all the researches enumerated above and added some others, is able to demonstrate very completely the phenomena in question.