Transparent Inactive Crystal Plates. 187 



especially Jamin, 1 and found to be valid except for light 

 waves incident at Brewster's angle, in which case the light is 

 not perfectly plane polarized, but shows elliptic polarization. 

 In convergent polarized light, therefore, all waves except 

 those near the center of the field and along the principal sec- 

 tions of the nicols suffer from the rotating effect of the glass 

 surfaces of the crystal mount on the planes of polarization. 

 This effect increases in amount as the margin of the field is 

 approached, where total extinction under crossed nicols is 

 noticeably imperfect in the diagonal positions. The glass 

 surfaces of the lenses, both of the condenser and the objec- 

 tive, moreover, exert also a rotatory influence on transmitted 

 plane polarized waves, and for this reason, in convergent polar- 

 ized light, the field appears to be divided by a dark cross into 

 four less dark quadrants even when no crystal plate intervenes. 

 The amount of rotation and consequent intensity of illumina- 

 tion is greatest for waves contained in planes at 45° from the 

 principal nicol sections. The amount of rotation varies also 

 with the refractive indices of the isotropic substances exam- 

 ined and the enveloping media as indicated by equation (47). 

 It is, however, obviously impossible in practical work to take 

 into account the rotatory effects of all glass surfaces interven- 

 ing between the two nicols, and all methods based on the exact 

 measurement of extinction angles of obliquely transmitted 

 waves must suffer from this defect and can only give approxi- 

 mate results. This applies both to methods involving con- 

 vergent polarized light and to the universal stage methods. 

 The error produced from this cause alone is not great (in 

 unfavorable instances, 2° to 4°), but it precludes results of a 

 high degree of accuracy. In the universal stage methods the 

 use of the small hemisphere as recommended by Fedorow 

 materially decreases the rotatory effect both of the glass mount 

 and of the crystal plate, and tends toward greater accuracy as 

 well as increasing the angle of view of the field. The greater 

 the difference in refractive index between the two media, the 

 greater the amount of rotation as indicated by equation (46). 

 The glass sphere not only tends to offer more nearly horizontal 

 surfaces to the entering light wave than the inclined glass 

 mount, but at the same time renders the boundary surface of 

 the glass mount practically inert in its action ; the glass mount 

 in turn decreases materially the amount of rotation which the 

 surfaces of the crystal plate exert when unmounted. These 

 facts justify the application of Fresnel's rule for finding the 

 directions of total extinction in a tilted crystal plate to the 

 universal stage methods, as a close approximation to the truth, 

 1 Ann. Chim. Phys. (3). xxix, 263, 1850 ; xxxi, 165, 1850. 



