470 SCIENCE PROGRESS 



accounts for the recovery represented by d e ; and finally there 

 is a viscous flow, also included in ac, from which there can 

 be no recovery. If this third element be absent, as it apparently 

 is in certain cases, then a c and d e are identical in shape, and 

 complete recovery occurs after a finite time. But if part of the 

 strain is viscous in nature, then we cannot expect the recovery 

 to correspond exactly to the deformation, nor will the original 

 shape be ever regained. In some materials this viscous flow 

 forms a very large part of the strain ; indeed, in the cases of 

 pitch 1 and lead the curve ac (fig. i) after a short time becomes 

 practically a straight line, indicating a uniform flow similar to 

 that exhibited by liquids under constant stress, and the recovery 

 following release is comparatively small. 



Of all those who have conducted research on this question 

 of " Elastische Nachwirkung," probably Kohlrausch 2 has been 

 responsible for most of the information acquired. In 1863 he 

 began an investigation of the effect of torsional forces on glass, 

 indiarubber, and various metal wires. He has shown that the 

 after-effect — i.e. that portion of the recovery represented by d e 

 in fig. 1 — is nearly proportional to the initial deformation ; or, 

 in other words, the rapidity with which a twisted wire untwists 

 bears a constant ratio to the initial angle of torsion. He also 

 found that the after-effect was greatly increased by rise of 

 temperature. An over-strained wire may be caused to recover 

 more quickly by raising its temperature. But perhaps the most 

 suggestive effect discovered by Kohlrausch is the result of the 

 following experiment carried out by him : 



A wire, clamped at one end, was twisted at the other end, 

 and, after being held for some time, released. The gradual 

 recovery followed the immediate one; but before it was complete 

 the wire was again twisted through a small angle, this time 

 towards the original equilibrium position, i.e. in the direction 

 in which motion was at the moment taking place. What now 

 happened is of great interest. The wire, instead of continuing 

 to untwist towards the original unstrained position, first turned 

 in the direction opposite to that of the second twist applied, 

 or in the direction of initial twisting, came to rest, and then 

 continued its first unwinding course. 



Apparently it is possible to superpose two of these after- 



1 Prof. Trouton and E. S. Andrews, Phil. Mag. April 1904. 



3 F. Kohlrausch, Pogg. Ann. 119 (1863) ; 128 (1866) ; 158 (1876) ; 160 (1877). 



