194 C. jffarus — Viscosity of /Solids. 



If the numerics of A (w—y') I t be regarded in their depend- 

 ence on time, the results are seen to oscillate around a mean 

 line of equilibrium. The ordinates of this mean line decrease 

 with time at a gradually retarded rate, until a definite inferior 

 limit is eventually reached. It is curious to note that the 

 largest observed ordinate (time = 0, nearly), is at least 3 times 

 the limiting ordinate (time = oo ). After 12 twists oscillation 

 has considerably subsided, but it has not ceased ; in the same 

 degree the viscosity of the glass-hard rod has reached a fixed 

 maximum. 



This complicated phenomenon is at once elucidated by Max- 

 well's theory. The ordinates of the line around which oscilla- 

 tion takes place, are an index of the degree of instability of 

 molecular configuration, at the time given by the abscissae. 

 The oscillations are the result of strain (latent shear, I called 

 it) imparted to the configurations by the successive twists to 

 which the wire is subjected. Thus if z be the impressed twist, 

 and At the mean strain left in the configurations at the instant 

 when t is removed ; and if n be the original relative number 

 of unstable configurations, and An the number broken up during 

 the period of the strain r ; then (apart from subsidiary consich 

 erations) Maxwell's theory .analyzes the effects of alternate 

 twisting in accordance with the following scheme : 





Strain. 





Molecular instability. 



First twist 



._ — T 





+ n 



Second twist 



_. +r-Ar 





+ n—An 



Third twist 



__ —r + Axr— A'r 





+ n— Ajw— A'w 



Fourth twist 



-. + T + A 2 T— A\t + A"t 





+ n—A 2 n—A' 1 n—A"n 



Fifth twist 



.. -t + A»t—A'iT + A" 1 t- 



-A'" 



r +n—A 3 n—A' ti n—A" 1 n—£ 



The variation which A undergoes in passing from one twist 

 to the next is indicated by subscripts. Thus At, A x t, A 2 t, .... 

 is probably a decreasing series; whereas An, A x n, A 2 ?i, .... is 

 an increasing series because reversal of the sign of the twist 

 must be supposed to reconstruct some of the configurations 

 broken up by the preceding twist. The first part of the scheme 

 indicates that the strain in the 2d, 4th, 6th . . . twists is neces- 

 sarily greater than the strain in the immediately preceding 1st, 

 3d, 5th . . . twists respectively, at least at the outset of the ex- 

 periments. Hence the observed oscillation. Again the num- 

 ber of unstable configurations must continually decrease, 

 according to the second half of the scheme. Hence the mean 

 line about which the observed viscous deformations oscillate. 

 Finally experiment shows that the accelerating effect of At on 

 viscous deformation is greater than the retarding effect of — An. 

 After this, however, the accelerating effect of A^t— A\t-\-A"t, 

 and the succeeding r- quantities, is always less than the retarding 



