﻿**54 Or the Rectification of the Spectrum. 



shift of ellipses as suggested in the preceding equation (9) or 



ANc = 36>B(l/\ 2 -l/\^>. 

 If we compute /3 in 



AN c =y3(l/\ 3 -l/X|) 



from the observed value of AN C , the mean value for the 

 B, 0, D, F, G lines is found to be /3 = '975 x 10" 10 . Since 

 e = -68 cm. and B = '18xl0- 10 in table 6, 3eB = 10~ 10 x '979, 

 which agrees very nicely with the computed value of /3 just 

 stated. 



The full line in fig. 7 contains the data computed for N e , 

 in case of 1 = 17° and the glass of the grating (B x 10 10 ='48). 

 Fig. 7 (upper curve, marked 17°) distributes the observation 

 on the computed curve for AN C; where AN C for the E line 

 is zero. 



11. Case of 1 = 90° nearly. — The further case of interest 

 corresponding to fig. 9, where the angle 1 = 72°, and as 

 large as it could be made in the given spectrometer. The 

 spectra were very fine and sharp, but too dark at the violet 

 end and red end. 



The computed graph with the observations located with 

 reference to it, is given in fig. 7 on the curve marked 72°. 

 The two curves together indicate the change of sensitiveness 

 on passing from a very small to a very large angle, the slope 

 of the curve, fig. 6, where 1 = 15° being intermediate. The 

 method of interference given in fig. 9 is useful when 

 observations are made with polarized light (as in the exami- 

 nation of doubly refracting media) and the observer finds 

 it convenient to be near the polarizer at the slit. 



12. Summary. — I conclude therefore that the observations 

 made at angles of incidence 1 = 15°, 17°, 73°, have verified 

 the equation 



-o/ X dti\ ^( 2B \ 



h c — e cos K La ^n — I = e cos It [ lu + r-5 ^ts I 



c v cos 2 ltd\/ \ \-cos J R/ 



from the shift of the centres of the ellipses satisfactorily. To 

 restore the ellipses from one spectrum line to another is 

 equivalent to a motion of the micrometer overN c — N c ' = AN C . 

 This constitutes a second method of interferometry of lower 

 sensitiveness ; i. e., a coarse adjustment on the number of 

 fringes which pass a given spectrum line. Per fringe or 

 per vanishing ring, obviously, the displacement of the mirror 

 on the micrometer will be a half wave-length of the part of 

 the spectrum in question. 



