Viscosity of Solids and its Physical Verification. 199 



data of Table III. It follows that the results of this table 

 are not a direct heat-effect subsidence of accumulated heat 

 agitation. 



Again, the sign of the twist, T,inTable III. is the same through- 

 out. It must therefore be asked whether in these experiments 

 an earlier stage of viscous subsidence simply overtakes a later 

 stage. To throw light upon this point it is sufficient to reverse 

 the sign of the twist alternately, without fresh annealing : or 

 to reverse the sign of the twist with each alternate annealing 

 of the upper and lower platinum wire. In such a case a latent 

 strain * favourable to motion is imparted to the wire not an- 

 nealed. Experiments which I made in some number showed 

 that, even in this case, the results of Table III. hold good, the 

 character of the motion being diminished in degree, but not 

 in sign. 



All these results are such as follow at once from Maxwell's 

 theory. They show that the above viscous effect of twisting 

 is to be referred to the motion of molecules which accompanies 

 it ; molecules and atoms are thus placed in new positions and 

 relations to each other; unstable configurations during the 

 course of such motion are therefore continually broken up into 

 configurations of smaller potential energy and greater stability. 

 Hence, finally, the observed increase of viscosity. By re- 

 versing the sign of the twist, the original configurations can 

 only be partially restored, at best, even for small permanent 

 set, such as is here in question. Finally the effect of prolonged 

 and repeated twisting is stiffness, because all the molecules 

 have collapsed into configurations of maximum stability, and 

 the intrinsic molecular energy is the potential minimum com- 

 patible with the given conditions. In the next paragraph (8) 

 this is shown to better advantage with steel. 



8 a. Steel wires were used in our earlier work t free from 

 torsional strain. The hard steel wires of the present paper, 

 employed in other work, may contain twists stored up like 

 permanent magnetism. This produces a kind of unilateral 

 symmetry, so far as torsions are concerned ; but it is not 

 otherwise objectionable. In critical cases wires free from 

 lateral torsion are selected. 



Turning to Table I., the individual wires enumerated are 

 found to show wide differences of viscous behaviour. In No. 

 2 the viscous subsidence takes place at nearly the same rate 



* Cf. § 8, closing remarks. 



t B. and S., American Journ. [3] xxxii. p. 448, 1886; ibid, xxxiv. 

 p. 4. 1887, 



