REVERSED AND NON-REVERSED SPECTRA. 



105 



between G and P about 10 cm. In this case the focal planes were nearly iden- 

 tical and the interferences easily found in the red region between the two C 

 lines. They appeared as small red pearls, very vivid on limiting the lateral 

 extent of the pencil L to about 5 mm., but, to my astonishment, they very 

 soon vanished on displacing M in a direction normal to itself i or 2 mm. 



55. Methods using prismatic dispersion. The small range of displace- 

 ment available in the prismatic reflection methods induced me to devise 

 corresponding refraction methods, to see whether these would show any 

 advantage in this respect. Accordingly the interferometer (fig. 75) was in- 

 stalled and the fringes found without much difficulty. Here P is the symmet- 

 rical prism, receiving the collimated beam of incident white light on the faces 

 meeting at the obtuse edge and refracting them in relation to the smaller 

 prism angle <p. This must be less than 45, for convenience in observation, 

 as otherwise the dispersed beams meeting the opaque mirrors M and N will 



c/J- 



76 



be too far apart for manipulation, supposing, of course, that the distance 

 PM and PN are over a meter. I used an equilateral 90 prism for want of a 

 better. The spectra reflected from M and N respectively impinge on the 

 grating G, concave or plane, and are viewed at T with a lens or telescope. 

 In consequence of the large angle 9, second-order spectra were used, without 

 apparent disadvantage. The dispersion of P and G being summational, the 

 total is very large. 



To return to the angles again, if < denotes the obtuse prism angle, and r the 

 angle of refraction, the angle of incidence is 90 </2, or 



(1) cos0/2=/i sin r 

 Again, 



(2) sin i' fj. cos (0/2 +r) 

 when i' is the angle of emergence. Hence 



M 2 cos 2 0/2 sin </>/2) 



