Strains Afechanically Applied. 173 



13. Returning to the case of simple traction, the changes 

 of sign of ({/) — 0')/t suggesting increase of viscosity for loads 

 slightly greater than 5 kg., indicate that here, as in the case 

 of motional annealing due to torsion, the initial effects are an 

 increase of viscosity. Nevertheless the obscurely small vis- 

 cous variations due to traction do not admit of interpretation, 

 unless supplemented by data for very much larger loads. 

 Again, it is desirable and quite feasible, by aid of the apparatus 

 discussed above, to operate cyclically ; in which case the results 

 must be such as to bear on the lag-quality of solids under 

 stress. 



In the following Table (IX.) I have therefore inserted some 

 data for larger pulls. Traction increases successively as far 

 as 90 per cent, of the breaking stress. It will be superfluous 

 to give more than a few typical examples ; and the tables 

 can be further abbreviated, because in the stress-decreasing 

 phase of the cycles the viscous effect of loads P below the 

 maximum employed is nil, § 10, and therefore sufficiently 

 given by the final load zero. To retain a fixed rate of tor- 

 sion, T, it is necessary to twist the wires anew after each of 

 the larger loads ; in other words to bring back the same scale- 

 division into the telescope at the outset of each experiment. 

 Otherwise the torsion seriouslj' decreases in consequence of 

 the rapid viscous deformation of the loaded wire. 



In adding the successive loads care was taken to avoid 

 vibration and jarring ; but without special machinery it is 

 impossible to protect the system completely against it. To 

 this cause I attribute certain irregularities of sequence which 

 these and others of my results exhibit. Their general sig- 

 nificance is none the less definite. The effect of traction is 

 diminution of viscosity, increasing at an accelerated rate in 

 proportion as stress approaches the limit of rupture. Indeed 

 by sufficiently increasing stress, viscosity may be diminished 

 in any degree whatever. The singularly curious feature of 

 these experiments is this, that with the removal of load the 

 viscous effect of traction almost entirely vanishes. It is in 

 this respect that the present experiments bear directly on 

 the truth of Maxwell's theory ; for it is only during the 

 interval within which conditions favourable to molecular 

 break-up are forcibly maintained that the wires exhibit a low 

 order of ^^scosity of a sufficiently marked degree to compare 

 with the viscous effect of drawn strains and of temper, 

 where instabilities are structurally retained. 



It follows, in general, that slight applications of mechanical 

 treatment (twisting, traction, &c.), inasmuch as they decrease 

 in number the motionally unstable configurations of the soft 



