A. W. Ewell — Rotatory Polarization. 375 



This asymptotic approach to equilibrium is similar to that 

 observed with metals, only more conspicuous. 



Relaxation disturbances can be avoided as described above. 

 There is, however, another effect of the previous history. A 

 prolonged preceding twist increases the rotation for an imme- 

 diately following less twist in either direction but has little 

 influence on a greater twist. 



(Relaxation effects eliminated, observations in chronological 

 order.) 



July 16. Tube C 3. 



Twist . , 90 120 100 120 90 120 



Rotation —9 —25 for 10' —29 —12 —27 



Twist (cont.) 90 —100 120 90 90 



Rotation" —11 for JO' —26 —10 -7 



July 17. Tube C,. 



Twist 90 120 —100 90 120 100 90 120 90 



Rotation. —7 —24 for 15' — 11 — 22 for 16' — 11 —24 —7 



Twist (cont.) 120 100 120 90 -100 120 90 90 120 

 Rotation " —24 for 18' —27 —10 for 21' —24 —9 —9 —26 



This effect is not due to a simple disturbance of the jelly ; a 

 prolonged strain is necessary ; alternate twists of short dura- 

 tion have little effect. 



July 16. Tube C 8 . 



Twist 90 16 alternate twists 90 90 



Rotation —9 of ± 120 in rapid succession —9 —9 



July 17. TubeC 3 . 



Twist. ._ 90 120 16 alternate twists 120 90 



Rotation —9—26 of ±100 in rapid succession —26 —8 



As illustrated above, this effect of a previous twist rapidly 

 dies away. Therefore to avoid this disturbance small twists 

 should precede greater and in general some time should elapse 

 between observations. 



If we make alternate twists in succession, counting the 

 twist from a constant zero, we notice both effects of previous 

 history. A typical case is the following : 



June 26. Tube F 1. 



Twist 100 —100 100 —100 100 



Rotation —19 10 —25 11 —28 



Twist (cont.)... —100 100 -100 100 —100 



Rotation " .. 12 —28 13 —28 13 



