60 L. B. Tuckerman 



Also: 



de o de 2 



~dd~dj~~ SeC ~ W ° (117) 



de de n 



de x ° dirN x s • 



The results in the two special cases are almost completely com- 

 plementary. Placing them together for comparison : 



1. ('/'-0,)=45 O %=o 



de dc de bo . 



-d&rw == ~^ 1 = m =:0 (II2) 



de n de n ., 



- = - — SPC" U> 



de x dirN x 

 30 30 



der l ~^ =sec2M ° 



2 . w =o N=% 



de o _ de o _d0„_ _90 o _ 



de~~ dN~ dT~ l dj~° (u6) 



de de ., 



~^=— sec2w (117) 



a# 



(113) 



dirN x 



The choice of a compensator to give greatest accuracy in the 

 measurement of elliptically polarized light depends therefore upon 

 the relative accuracy of the measurement of e x and X , the ellip- 

 ticity and azimuth of the compensated light. If e x can be meas- 

 ured with the greater accuracy, a quarter-wave plate gives the 

 more accurate determination of 6 o , while a plate of lower order, 

 ttJV==(d o approximately, gives the more accurate determination of 

 e u . If X can be measured with the greater accuracy, the reverse 

 is the case. 



216 



