July 18, 1902.] 



SCIENCE. 



93 



these conditions are fulfilled and will be 

 reflected ofi: parallel to its former direction. 

 This relation will be maintained as long as 

 the above conditions hold, which will be, 



distant mirror M may be concave and of 

 the proper focus for the system; or two 

 lenses / and F may be used so as to obtain 

 a greater aperture to the beam. These two 



approximately, through a considerable frac- 

 tion of the circumference. This will be 

 best satisfied when the mirror m is normal 

 to the radius, the incident ray being thus 

 nearly normal. The arrangement as finally 

 adopted is shown in the figure. Light from 

 the slit S is converged by a lens L to a half 

 silvered plate /, one beam being reflected 

 and the other passing through and being 

 reflected by the mirror 7'. Both converg- 

 ing beams strike the mirror m at the same 

 point and are then reflected, the first beam 

 to the adjacent face of the rectangular 

 mirror m' and the second to the opposite 

 face, where they form images of the slit 8. 

 The first beam is reflected to M' then to M 

 and finally to a focus on m', while the sec- 

 ond ray passes over the path of the first to 

 a focus on m'. Thence the two rays trace 

 each other's paths and are reflected and 

 transmitted respectively by the plate I 

 through the spectral system P to the eye, 

 where interference bands are formed. 

 Thus, aside from other losses by reflection, 

 one fourth of the light reaches the eye. The 



rays will in general travel over slightly dif- 

 ferent paths and hence give bands which 

 may be conveniently analyzed by means of 

 channeled spectra. If now the disc rotates 

 the path of one of the rays will become 

 greater than the other and the interference 

 bands will shift. If a spectrum is used the 

 bands will move across the fleld, increasing 

 or decreasing in number. If the adjust- 

 ment is initially made so that the paths are 

 the same, no bands will appear until the 

 disc is set in motion. By counting the num- 

 ber passing any point we can obtain the 

 order of the interference for that wave- 

 length, and from the dimensions and speed 

 of the disc determine the wave- velocity for 

 that color. From the position of the other 

 bands at this instant we can calculate the 

 velocity of that color. Thus we have the 

 means at hand for obtaining the wave- 

 velocity for all colors, from which the group- 

 velocity for the same can at once be ob- 

 tained. The radius to the disc m' is 15 cm. 

 and a speed of 500 per second is assured. 

 The concave mirror M has an aperture of 



