A. W. Ewell — Rotatory Polarization. 387 



for other colors. These facts, coupled with the irregularities in 

 location of the plane of polarization of yellow light in different 

 azimuths, as illustrated above where full sets of readings are 

 given, make it natural that the dispersion should not be very 

 clear. 



Theory. 



With hesitation I suggest the following imperfect theory. 

 In a section of a twisted cylinder, perpendicular to the axis, 

 along any line, whether straight or curved, there is nothing to 

 distinguish one direction of motion from the opposite direction 

 of motion. Therefore for two rays circularly polarized in oppo- 

 site directions, traversing a twisted cylinder parallel to the axis, 

 the conditions are identically the same in the wave' front. 

 Therefore the only difference can be in the lines along which 

 the two opjDosite disturbances are propagated. The successive 

 points disturbed by a positive circularly polarized ray lie at any 

 instant on a negative spiral and vice versa. Is there anything to 

 distinguish two opposite. spirals parallel to the axis of a twisted 

 cylinder? If the side of a spiral farthest from the axis of the 

 cylinder has the same direction of twist as the adjacent part of 

 the cylinder, the side of the spiral nearest the axis must have 

 the opposite direction, and so there would be no distinction 

 between the spirals were it not for the important fact that in a 

 twisted cylinder the stresses and strains increase as we leave 

 the axis radially, and therefore the outside of a spiral has 

 a predominating influence. 



Suppose the twist of a cylinder is positive, counting direction 

 the same as for the rays. Any tangential ( perpendicular to a 

 radius) square in the cylinder with two of the edges perpen- 

 dicular to the axis, will be distorted into a rhombus, the axes of 

 strain being the diagonals. Therefore there will be stretching 

 along the outside of the positive spiral and compression along the 

 outside of {he negative spiral. We may expect compression 

 to decrease the velocity, for, in general, the denser a body, the 

 slower light is transmitted through it. If this be so, we 

 should expect to see the ray corresponding to the positive spiral 

 to be transmitted faster than the ray corresponding to the nega- 

 tive spiral, or the cylinder having a positive twist and the positive 

 spiral, corresponding to a negative circularly polarized ray, a 

 negative circularly polarized ray will have the greater velocity 

 and the plane of polarization will be rotated in a negative 

 direction. Or, torsion should produce rotatory polarization in 

 the opposite direction to the twist, which agrees with the 

 experimental facts. 



The effect of pressure upon the refractive index of jelly was 

 investigated. Pressure was applied to jelly between two glass 

 plates, forming a prism, by varying the angle of the plates. No 



