Indices of Doubly -refracting Crystals. 



835 



The results o£ this table are reproduced in fig. 3, the lower 

 curves showing the individual indices of refraction, and 

 the upper curve giving their difference. When z = 50° or 

 r = 30° (nearly), the rays pass too near the edge, so that 

 measurement is uncertain, because some of them are excluded. 

 For large r, moreover, the interpretation is complicated by 

 the occurrence of rotary polarization along the axis. The 

 data for £//, fluctuate to within a few units of the fourth 

 place, which is to be expected for the given thickness of 

 plate used. 



Fig. 3. 



ZOO 



i / 



X;. 





i 



/ 



1 



400 



?> o. 



/ 







k l / 











tG~ 



sa 





4-5S 



t 





M. c 





5 TH? 



P>o 



m\ 



T-+ 1 





0° 10° ZD° 50° 



5. Results for Different Spectrum Lines. — In § 3 the method 

 for computing the approximate dispersion constant b was 

 indicated. There is, however, a difficulty inherent in this 

 method inasmuch as for the case where /j, is to be found 

 for four or more places of decimals, the abbreviated Cauchy 

 equation is insufficient. 



Leaving this for the present, it is interesting to compare 

 the results obtainable by the interferometer with the standard 

 data of Mascart and de Lepinay and others *. For this pur- 

 pose the quartz of Table II. was mounted at normal incidence 

 and the micrometer readings for the C, D, E, b, F lines of 

 the spectrum taken in succession, both for the presence and 

 absence (air) of the quartz plate. The differences of corre- 

 sponding values (quartz and air) of SN for the same line were 

 then reduced to the value of AN for the D line in Table II., 

 so that the AN for each spectrum line was obtained in 

 succession for the computation of /jl. In other words, the 

 values of SN are shifted so that for the sodium line SNd = 0. 



The results show that the value of /a thus computed differs 

 from the value of Mascart, //,m> by fourteen units in the fourth 

 place in case of the extreme C and F lines. The reason of 

 this is the approximate value of b taken. To find what the 



* See Landolt and Bornstein's Tables. 



