484 Mr. F. Twyman on Improvements 



whether it be an ordinary prism of 60° or other angle, or 

 a constant-deviation prism as in the present instrument. 



Considering the case where a beam o£ light, collimated 

 but initially unpolarized, is passed through the prism. Such 

 a beam is the upper beam in the present instrument, which 

 passes over the first nicol. The beam when it emerges 

 is partially polarized, owing to the unequal reflexion 

 of vibrations in and at right angles to the plane of 

 incidence. 



(In the case of a beam initially polarized in a plane neither 

 in nor at right angles to the plane of incidence, matters are 

 further complicated by the reflexion within the prism, which 

 causes elliptical polarization ; a state of things which has to 

 be avoided by very carefully setting up the first nicol, until 

 it is found that satisfactory extinction of the polarized beam 

 can be obtained by rotation of the second nicol.) 



Considering the appearance of the field of view as seen 

 in the eyepiece, in the zero position of the second nicol : the 

 position, namely, in which its plane of polarization is parallel 

 to that of the first nicol. One sees a slit of light wider or 

 narrower according to the width to which one may think it 

 advisable to open the shutters in the eyepiece which limit 

 the field of view. This slit of light is bisected by a fine 

 horizontal line, the edge of the rhomb in front of the slit, 

 which is almost invisible when the intensities above and 

 below are equalized by the adjustment of the wedge of 

 neutral -tint glass. The light -absorbing medium, whose 

 density it is required to measure, is introduced into the top 

 beam, and the second nicol is then rotated to obtain equality 

 of intensity. The lower half varies in intensity according 

 to the cos 2 law ; but owing to the polarization due to the 

 dispersion prism the top beam also varies in intensity, and 

 the absorption if determined simply by the application of 

 the cos 2 law will consequently be incorrect. The amount 

 of the error can be determined by applying Fresnel's 

 equations to any particular case. We will consider the 

 actual state of things in the original instruments made, 

 i. e. ; a prism of refractive index 1*658 for the light under 

 observation. 



Let 6 and 6 1 represent respectively the angles of incidence 

 and refraction at the surface of the dispersion prism. 



According to FresnePs equations, if unpolarized light of 

 intensity a 2 fall on a reflecting surface at an angle of inci- 

 dence #, and if 1 represent the angle of refraction, the 

 intensity of that part of the reflected light which is polarized 



