354 Mr. R. T. Glazebrook on 



angles to the plane of the paper, and the section of the wave- 

 surface by that plane will be two circles of radii A and C, 

 these being the ordinary and extraordinary wave-velocities 

 respectively. Hence a ray falling on the face AB in any 

 direction in the plane of the paper will be divided into two, 

 which will both undergo ordinary refraction, so that if the 

 incident ray be normal to the face A B, the extraordinary and 

 ordinary rays in the prism will coincide in direction, both 

 being normal to the same face. The extraordinary ray is not 

 deviated by the refraction ; so that no lateral displacement of 

 the extraordinary image is produced by the prism. The 

 ordinary ray is incident at about 70° on the face P Q ; it is 

 therefore totally reflected, and the emergent light is plane- 

 polarized. The prism differs from one described by Prof. S. 

 P. Thompson (Phil. Mag. Nov. 1881) only in the fact that 

 its ends are normal to its length instead of being inclined 

 obliquely to it. But this form of prism has other and more 

 important advantages. 



Let M, N (fig. 3) be two extraordinary wave-normals, 

 and A the optic axis. Pass a plane p. o 



M A through M and A, and in 

 this plane draw OP at right angles 

 to M ; then P is the direction of 

 vibration in the wave which travels 

 along M. Similarly, if N A be a 

 plane through N and A, and Q 

 a line in it at right angles to ON, 

 Q is the direction of vibration for 

 the wave along ON; and it may 

 happen, clearly, that P and Q are 

 inclined to one another at a large 

 angle even when OM and ON are 

 close together. Suppose, then, that the 

 extraordinary pencil of wave-normals 

 which is traversing the spar is slightly conical, and that N, 

 M are two of the wave-normals ; the planes of polarization 

 are inclined to each other at an angle equal to P Q ; and this 

 may be considerable. Or, again, suppose that we have a 

 polarized pencil of parallel wave-normals incident on the 

 prism. We determine the position of their plane of polariza- 

 tion by turning the prism until no light passes through. Sup- 

 pose that, when this is the case, the incident light is parallel to 

 M. Now let the plane of polarization of the incident light 

 be rotated, and suppose we wish to measure this rotation ; we 

 turn the prism until the light is again quenched. Theoretically 

 the axis round which the prism has been turned should be 



