164 Mr. C. Barns on the Viscous Effect of 



8. The first result given by these data is obtained by com- 

 paring 4> — (j>' and (^ — 0')/t, showing that the latter quantity 

 is not independent of r. Nor can it be asserted that the 

 change of (0 — 0')/t with increasing t is retarded. It follows 

 that comparable values of (0 — c^Q/t are only obtained by 

 keeping t nearly constant, as has been done in all my 

 experiments. 



Again, the large variations of (cfi — cf)') in the direct or 

 stress-increasing phase of the cycles, as compared with the 

 corresponding variations in the retrograde or stress-decreasing 

 phase of the cycles, together with a repetition of the whole 

 phenomenon on a smaller scale for succeeding cycles of the 

 same sign, are features of these experiments. When the sign 

 of the twist is reversed, cycles which exceed the original one 

 in magnitude are obtained. These in their turn dwindle on 

 repetition of like signs of r. Finally, the influence of residual 

 action in case of retrograde cycles appears more clearly in 

 proportion as stress approaches zero. On the other hand, the 

 circumflex contours of the earlier retrograde curves (t de- 

 creasing from '10b) is not a mere error of observation, but 

 results from superposition of direct and residual phenomena. 

 This is specially marked in the third cycle, where stress, after 

 T = *087, was applied in a zigzag way, viz. : 



TX 10^ = 87, 70, 96,87, 105, 96. 



9. The clue for the interpretation of the above complex 

 phenomena as a whole is suggested by the data for A {<f> — (ji'). 

 It then appears that the viscous behaviour of the wires is 

 intimately connected with the amount of permanent set 

 imparted during the period of action of stress. From this 

 point of view the phenomena become not only strikingly 

 analogous to thermal annealing in case of temper, but of 

 special importance as regards their bearing on Maxwell's 

 theory. The following description "^ apiplies, mutatis mittandis, 

 to both classes of phenomena: — 



(1) The viscous deformation (annealing eff'ect) of any 

 stress (temperature) acting on glass-hard steel increases 

 gradually at a rate diminishing through infinite time ; 

 diminishing very slowly in case of low stress (temperature) ; 

 diminishing very rapidly at first and then again slowly in case 

 of high stress (temperature) ; so that the limit of permanent 

 deformation is approached asymptotically. 



(2) The ultimate viscous deformation (annealing effect) of 

 any stress t (temperature ^), is independent of preexisting 



* Cf. this Mi^gazine [o] xxvi. p. 214, lo88. 



