A. W. Ewell — Rotatory Polarization. 379 



applied to the jelly tube, probably in the form of greater con- 

 straint upon the jelly by the rubber tube. From the previous 

 results upon the effect of an envelope, the greater rotation 

 would be expected. 



For example : 

 July 21. Tube D 3. 



Negative rotations greater. 



80 g pull on cords gave a twist of 64 



— 80S" « « "■ —65 



200& " " " 178 



— 200 s " " " —183 



Here we see greater rigidity for positive twists, which pro- 

 duce negative rotations. 



Tube D4. 



80s 



gave 



47 



-80S 



u 



— 49 



150s 



it 



91 



-150S 



a 



— 96 



!• giving a similar result. 



The dissymmetry is not due to the natural rotation, as the 

 excess is as likely to be for one direction as for the other. 



As neither rotation is more important, we will study the 

 average, or sum of the two rotations. The observations with 

 tube D 3 are plotted in fig. 4. The observations upon tubes 

 D 4 and F 1 are given in fig. 5 and for tubes 11 and 43 in fig. 6. 

 The abscissae are twists and the ordinates rotations (sums). 



The curves resemble exponential curves of the general form 

 y=cx n or log y—log o+n log x. To determine whether they 

 are such, and if so, of what power, I have plotted in fig. 6a 

 the logarithms of the twists and rotations. All of these curves 

 approximate to straight lines, at least for the greater twists, 

 and the inclination of these straight lines is very close to four, 

 giving the very interesting result that, when other conditions 

 are kept constant, the rotation is closely proportional to the 

 fotcrth power of twist. 



The curves of the sums of two rotations will have the nature 

 of each one separately, for if the law of one be y^kjF and of 

 the other y 2 =k 2 x n , the equation of the sum will be the same 

 power of the twist, or, y 1 +y 2 =(k 1 -\-'k^x n . 



The law could not be tested with independent cylinders since 

 they would not sustain the weight of the lower clamp under 

 twist long enough to determine the rotation. 



Plane Parallel Light. — Since no lenses are used, the light 

 passing through the jelly will not be perfectly plane parallel. 

 To make sure that the phenomena are not caused or influenced 

 by this fact (for instance, through reflections from the interior 

 of the rubber tubing, which however is quite dark), a jelly 

 tube was clamped under twist and examined with strictly plane 



