96 Prof. F. L. 0. Wadsworth on the Optical Efficiency 



for the purpose. It would be better in general to use a 

 short, thick mirror with an optically worked surface, rather 

 than a longer thin one of ordinary quality*. 



In the ordinary method of using a mirror the only possible 

 way of increasing the ultimate measuring power is to in- 

 crease its aperture. Some years f ago a simple method 

 suggested itself to me, by means of which the angular devia- 

 tion of the image for a given movement of the mirror might 

 be doubled, and the ultimate measuring power thus increased 

 without changing the dimensions of the reflecting surface. 

 I have since learned that IVlichelson and Morley had pre- 

 viously proposed to employ the essential feature of this method 

 (that of multiple reflexion) as a possible means of solution of 

 a different problem ; but as far as I am aware no one has 

 suggested the use of it in ordinary measurements of deflexion 

 by the mirror method. If it has been suggested, it is at least 

 not generally known, and a description of it may therefore 

 not be without interest. The plan is simply to place close to 

 the revolving mirror m a second mirror m' (fig. 5), which 

 receives the rays after the first reflexion and returns them to 

 m, from which they are again reflected in nearly the original 

 direction. 



Fm. 5. 



M 





\ 

 \ 

 t 

 * 







i 



• 



\ / 



\ / 

 \ / 



/ \ 





(, 



^^^5-?- 





. M 







Any angular movement 6 of the mirror m then causes an 

 angular deviation of the doubly reflected ray of four times 

 that amount. In order to realize the effect of the full aper- 

 ture of the mirror it is necessary that the two mirrors should 

 be as nearly parallel as possible, and as close together as 

 possible. These two conditions evidently conflict with each 

 other. If we need only read deflexions on one side of the 

 middle position the mirror may be placed as in fig. 5, so that 



* An accurately cylindrical surface is, however, nearly as good as a 

 " flat " if the axis of the surface is parallel to the axis of rotation. 

 t The entry in my note-hook is dated November 12th, 1889. 



