Strains Mechanically Applied. 157 



a decided increase of viscosity amounting to almost one half 

 of the total viscous interval, hard-soft *. This marked incre- 

 ment of viscosity is accompanied by an equally striking 

 electrical elFectf. For it has been shown that the specific 

 resistance of hard steel diminishes as much as 15 per cent, as 

 the result of annealing at 100°. Finally, since the electrical 

 effect is a sufficient indication of the changes of volume 

 (decrement) and of carburation, it appears conclusively that 

 the underlpng cause of the increase of viscosity here in 

 question is a disintegration of the unstable carbide molecule 

 of steel. 



4. Again, an increment of the viscosity of glass-hard steel 

 comparable in magnitude with that of annealing at 100°, 

 may be obtained without heat. It is merely necessary, for 

 instance, to apply to the glass-hard wire large enough rates 

 of twist, a sufficient number of times alternately, in opposite 

 directions. This method of increasing viscosity has no elec- 

 trical concomitant comparable with the electrical effect of § 3. 

 Hence the underlying cause of the observed increase of 

 viscosity in this case is probably not a disintegration of the 

 carbide molecule of steel ; or at least a disintegration quite 

 unlike that of the foregoing instance. 



D. In the following tables (I. and II.) I give the data 

 necessary to substantiate § 4. The method J of experiment 

 being identical with the one described in verifying § 3, I need 

 only call to mind here that my normal steel wire (maximum 

 viscosity) and the fresh glass-hard wire (minimum viscosity) 

 were countertwisted ; that the ends of the vertical system 

 were fixea, and a mirror-index placed near the middle, at 

 the junction of the two wires. Cf. figure 1, § 12. Table I. 

 contains the results for viscosity, both wires being at the 

 same temperature 6. The impressed rate of twist is given 

 in radians under r, and the permanent torsion observed at 

 the close of each experiment under 2 (0 + (^'). Since the 

 low^er end of the system w^as twisted 360°, t + 2 (^ + (/)Q 

 = 27r/L, where L is the length of the two wires. Finally 

 {(f> — (j)')/T is the viscous motion at the index, per unit of t, 

 per unit of L, at the time specified. In other words 2(j> and 



* Am. Journ. Sci. [3] xxxiii. p. 26, 1887 ; PhU. Mag. [5] xxvi. p. 188 

 etseq., 1888. 



t Bull. U.S. Geological Survey, No. 14, p. 49, 1885. 



X Phil. Mag. [5] xxvi. p^. 189 to 191, 1888. The use of difiereutial 

 methods premises that the viscous deformations of all the wii'es to be com- 

 pared are similar time functions. That this is the case must be verified 

 preliminarily by some absolute method. 



