Li r/ Jit Propagated in a Dielectric. 



343 



pronounced effect was observed when the magnet was excited. 

 This effect did not disappear when the plane of polarization 

 was shifted 45° on either side of this position, indicating the 

 Faraday effect. This result showed that even for a very 



^-&0^ 



narrow beam of light, the divergence was sufficient to intro- 

 duce this effect, notwithstanding the direction was shifted so 

 as to make the mean direction of the rays as near normal to 

 the field as possible. Difficulty was experienced both from 

 depolarization at the sides of the plate and from imperfect 

 annealing of the glass, so that this form of the experiment had 

 to be abandoned. 



An estimate of the Faraday effect for a small bundle of 

 rays will show the impossibility of eliminating this factor in 

 this w T ay. Assume for Faraday glass, w D = l'76, and a rotary 

 power of </> D = 2 x 10 -5 radians per cm. If H = 5250 the rota- 

 tion per cm. is p = 6° approximately. Allowing 200 cm. for the 

 total path of ray and 1 cm, aperture, we have, from Verdet's 

 law, the total rotation for the most divergent ray 



/> = G o x200x-005 = 6°, 



or an average of 3°. Taking the sensibility of the instrument 

 at '01°, the aperture must evidently be "003 cm. to eliminate 

 the Faraday effect in this arrangement. 



It is evident from Verdet's law, p=p Q cos 0, that p changes 

 sign when 6 passes through 7r/2 and 37r/2, hence if a ray is 

 reflected by a plane parallel to the lines of force, p will change 

 sign at each reflexion and the total Faraday effect will be 

 zero for an even number of passages. Hence this effect can 

 be eliminated for divergent rays. 



This condition was realized in the following experiment. 

 Two cylinders of Faraday glass, ?i D = l , 76, each 22 cm. long 

 and 2*4 cm. in diameter were placed end to end in the field 

 and polarized light was sent through the system and reflected 



2B2 



