Mr. E. T. Glazebrook on NicoVs Prism. 



'I do 



to the possible directions of vibration : that is, with the same 

 notation as previouslv, 



>|r = 45 C . 



And if we suppose the axis of rotation to be inclined at 5° to 

 the incident light,] 



0=5°, 



ft = 3° for mean rays. 

 Also 



«=65°. 



.". substituting in; 



cot 6' sin i/r = cos yj/ cos ft + sin ft cot ct } 



<9 / = 42 c 49 / , 

 , a tan 6 f 



tan V= j-z -rjr, 



cos (<i>-fty 



= 42° 5(K] 

 also 



cos <r= cos 6 cos i/r + sin 6 sin ijr cos $, 



o- = 4 17 / (in 



Now suppose that the prism and axis of coordinates are 

 turned round OX' until the plane xOy passes through OX'. 

 The amount of rotation will be rather more than 45°; and in 

 this case we shall have 



yjr, 6. 6\ and <r all zero. 



Thus the plane of polarization of the emergent light now co- 

 incides with the plane ^0/. whereas previously it was inclined 

 at 4° 17' to the plane xOy. Thus, in a rotation of about 45°, 

 the motion of the plane of polarization differs from that of the 

 Xicol by 4 C 17 v , or an error of 10 per cent, is introduced into 

 the measurement. The error, therefore, in a prism with ends 

 normal to its length may be greater than in one with oblique 

 ends. As before, if the axis of rotation be oblique to the length 

 of the prism the expressions would require slight modifica- 

 tions, but the numbers would remain of about the same mag- 

 nitude. 



But there is another way in which the measured rotation is 

 affected by the inclination of the axis of rotation to the length 

 of the Xicol. The angle we require is that through which a 

 plane passing through the optic axis of the Xicol and bisecting 

 one angle of the rhombic face of incidence has been turned. 

 Unless the axis of rotation is parallel to tjjie plane, the angle 



