16 C. Bar us — Viscosity of Steel and 



to 13 clearly show that the viscous detorsion <p exhibited by 

 steel is very much more influenced by temperature when the 

 steel is hard than when it is soft. In other words, viscosity 

 decreases with temperature (ca3t. par.) at much greater rates 

 in hard steel than in soft steel. Again, at the same tempera- 

 ture T, the differences of viscositjr are very large when the 

 temper of steel lies between glasshard and annealed 350°. In- 

 termediate differences within this interval are larger in propor- 

 tion as mean hardness is greater. When temper lies between 

 annealed at 350° and soft, differences of viscosity at the same 

 T are relatively small. In general therefore the variations of 

 viscosity clue to temper are marked occurrences during the first 

 phase* of annealing, and nearly vanish during the second 

 phase, a result which corroborates the close analogy between 

 the viscous effects of temper and the thermoelectric effect of 

 temper already pointed out (cf. p. 5). 



For the same degree of hardness viscosity increases at an accel- 

 erated rate with temperature T. In figure 2, where T= 100°, 

 the large viscous interval between An. 100° and An. 360° is in 

 striking contrast with the smaller viscous interval An. 360° to 

 An. 1000°, notwithstanding the fact that to avoid erroneous com- 

 parisons the latter interval has rather been chosen too large 

 than too small. Indeed, the values for An. 1000° in table 6 

 would place the locus for soft wire even above the curve An. 

 450° in figure '2, suggesting the occurrence of a maximum vis- 

 cosity for T=100°. Passing from figure 2 to figure 3 (7 1 =190°), 

 the interval between An. 190° and An. 360° is phenomenally 

 increased. The interval An. 360° to An. 1000° is also in- 

 creased, but only slightly. Again, passing from figure 3 to 

 figure 5 (1 7 =360°). the interval of the second phase of anneal- 

 ing (An. '360° to An. 1000°) is largely increased.^ The marked 

 tendency of a steel wire annealed from hardness at t° to suffer 

 viscous deformation when exposed to t°, is the chief result of 

 the present paper. This tendency decreases with great rapidity 

 in proportion as the temperature of exposure falls below t°. 



As temperature increases, glass seems to lose viscosity much 

 less rapidly than hard steel, but probably not less rapidly than 

 soft steel. The magnetic instability of glass-hard steel is prob- 

 ably due to its extreme susceptibility to temperature, since 

 every change of temper is the cause of loss of magnetic mo- 

 ment. This close relation between hardness and magnetism is 

 good evidence in favor of the essentially strained character of 

 hard steel. 



* This Journal, vol. xxxi, p. 443, 1886. 



f Of course the degree of hardest temper to be exposed to T for viscous com- 

 parisons is An. T. Harder wires would be annealed at this temperature and the 

 concomitant effects due to changes of temper erroneously confounded with simple 

 viscous motion. 



