Viscosity of Solids and its Physical Verification, 209 



temperature. The reasons for this phenomenon, though not 

 far to seek, are exceedingly significant. In hard steel two 

 causes of molecular instability here produce superposed effects. 

 The first is the chemical or carburation instability already 

 discussed ; the second cause is purely thermal. Cf. § 17. 



The explanation of the diagram for 100° (fig. 1) is now 

 clear. Viscous deformation is marked in all the rods examined 

 from An. 500° to An. 100° ; but the deformability increases 

 at a rapid pace in proportion as we pass from softer to harder 

 steel, because in such a march the carburation instability 

 superimposed upon the thermal instability increases rapidly. 

 Molecular configurations on the verge of instability are 

 encountered in continually increasing numbers. 



12. The line of argument followed out for 100° applies, mut. 

 mut.j at 200°. Results of this kind I published elsewhere*. 

 The character of the evidence bearing on all the points in 

 question is here even more pronounced and conclusive. Steel 

 An. 200° is in a state of incipient annealing at 200°. Thermal 

 and carburation instabilities of high degree being encountered, 

 the superimposed effects are correspondingly large. 



Finally, above 300° the molecular instability is largely 

 thermal. The behaviour of hard steel, therefore, approaches 

 that of other metals more nearly. The effect of the carbu- 

 ration instability ceases to predominate, and finally vanishes 

 altogether in proportion as the march is made from lower to 

 higher temperatures of annealing. 



13. I have finally to touch upon the series of phenomena 

 in which pronounced annealing occurs simultaneously with 

 pronounced external viscous deformation. If, for instance, a 

 glass-hard rod is twisted and then suddenly heated to 100°, the 

 rod is both annealed and suffers deformation in virtue of the 

 applied twist at the given temperature. Conformably with 

 the excessively greater amount of molecular instability 

 which characterizes these experiments, the observed viscous 

 deformation must be proportionately large. This prediction 

 of Maxwell's theory is fully verified by experiment. In the 

 case of the twisted rod postulated, the motion of the image 

 across the field of the telescope is so rapid that Gauss's method 

 of angular measurement is no longer satisfactorily available. 

 I may say, without exaggeration, that during the small in- 

 terval of time within which appreciable annealing occurs, a 

 glass-hard steel rod suddenly heated to 300° is almost a viscous 



* Amer. Joura. [3] xxxiv. pp. 14^16, 1887. 

 Phil Mag. S. 5. Vol. 26. No. 159. Aug. 1888. P 



