200 Mr. C. Bams on Maxwell's Theory of the 



for -t and for + t, both at 0=20° and 0=100°. In No. 3 

 the effect of — t and -f t is of different magnitude at 20° and 

 enormously more different in magnitude at 100°. In No. 4 

 the effect of +t following — r is even more phenomenally 

 pronounced, both at 20° and particularly at 100°. In No. 5 

 the wires nearly identical at 20° show differences at 100°. In 

 No. 6 this is true even in much greater degree, whereas in 

 No. 7 wires differing considerably at 20° show relatively small 

 differences at 100°. And so I might go through the series. 

 The element of vagueness asserts itself throughout, and is 

 almost serious enough to obscure the law. Nor is this wholly 

 due to the latent torsions mentioned at the outset of this para- 

 graph. Nos. 13 and 14 are wires originally free from torsional 

 strain ; but the said vagueness also appears in these. 



Careful inspection of the tables reveals the law that viscous 

 deformation takes place at numerically greater rates during 

 the even twists than during the odd twists which immediately 

 and respectively precede them. Irrespective of these oscil- 

 lations, the effect of twisting here, as in § 7, is pronounced 

 increase of viscosity. 



Maxwell's theory accounts for the stated vagueness of be- 

 haviour at once. In two samples of a complex substance like 

 steel, the distribution and relations of the unstable molecular 

 configurations will only in very rare instances be physically 

 and chemically identical. The previous paragraph shows that 

 such identity is rare even in pure homogeneous metal. 



The effect of twisting alternately in opposite directions is of 

 such great importance in its bearing on Maxwell's theory that 

 I made further special experiments. From these I select the 

 following typical instance, tabulating it in the way adopted in 

 Table I. The normal No. 1, An. 450°, has been described. 

 No. 18, An. 25°, or glass-hard, is carefully selected free from 

 latent torsion, having experienced no other strain prior to the 

 examinations in Table IV. than that incident to tempering 

 (quenching). There are 12 alternations of twist, indicated 

 by subscripts, and the current time in hours and minutes 

 of each is given. I also give, under m, the time in minutes 

 which refers specially to the duration of each twist. No. 1 

 being always of greater viscosity, t and (<f>— 4>')/t are alike in 

 sign, by agreement. 



