136 



Profs. W. E. Ayrton and John Perry. [Dec. 12, 



This difference in molecular constitution might produce the re- 

 sults : — 



a. By causing the portion of the face corresponding with the pattern 

 at the back to attract more mercury, and so to become capable of 

 being polished more easily ; or 



b. By causing it to be harder, and so to acquire a better polish ; or 



c. By causing it to polarise light. 



This difference in molecular constitution might be produced : — 



a. By the inlaying of another metal ; or 



b. By portions of the surface being acted on chemically ; or 



c. By unequal density produced by inequality in the rate of 

 cooling ; but 



d. Not by stamping, Japanese mirrors being all cast. 



3. The phenomenon might arise from the face of the mirror having 

 intentional or accidental inequalities on its surface, in consequence of 

 which, the part corresponding with the pattern on the back might be 

 relatively concave, and so concentrate the light, or, at any rate, might 

 disperse it less than the remainder of the slightly convex mirror. 



The question then resolves itself into considering to which of these 

 three groups of causes is the apparent reflection of the back in some 

 Japanese mirrors due. 



To ascertain this, we tried Sir David Brewster's suggestion that 

 the light reflected by the thicker part of the mirror was polarised ; 

 but even with a fairly good polariscope, we could detect no marked 

 difference between the light reflected from the various portions of the 

 surface. This failing, we availed ourselves of a very simple method 

 of experimenting, but one that has apparently not suggested itself to 

 previous observers. On one occasion, when some of our students 

 were using lenses to endeavour to make the exhibition of the pheno- 

 menon more striking, it occurred to us that the employment of beams 

 of light of different degrees of convergence or divergence would fur- 

 nish a test for deciding the cause of the whole action. For while, if 

 the phenomenon were due to molecular differences in the surface, the 

 effect would be practically independent of the amount of convergence 

 of the beam of light ; on the other hand, if it were due to portions of 

 the reflecting surface being less convex than the remainder, a com- 

 plete inversion of the phenomenon might be expected to occur, if the 

 experiment, instead of being tried in ordinary sunlight, were made 

 under certain conditions in a converging beam — that is, the thicker 

 portions of the mirror might be expected to appear darker instead 

 of brighter than the remainder. Figs. 1 — 5, which are all much 

 exaggerated for the sake of distinctness, explain this better. MM, 

 fig. 1, represents an ordinary Japanese slightly convex polished bronze 

 mirror. SA, SB, SC, are rays of a parallel beam of light falling on 

 it, and reflected as AD, BE, OF, on to a screen DF ; then, if the areas 



