1915-] IN RECENTLY HARDENED STEEL. 163 



is found as follows : The small steel rod spontaneously shrank 

 0.0175 inch. To spring it back to its original length required a 

 weight of 15 pounds hung below M, Fig. 4 (= 12.400 pounds strain 

 per square inch of cross-section). Hence, in longitudinally shrink- 

 ing 0.0175 inch, the rod had done work equal to lifting 15 pounds 

 half this distance or 0.00875 inch. The rod weighed about 1230 

 times less than the weight, so that the work done was sufficient 

 to lift the rod itself 1230 X -00875= 10.76 inches. But this rep- 

 resents one-dimensional shrinking only, and we must take three 

 times this amount of Hft, or, say 2^ feet, to represent the work 

 done in the three-dimensional shrinking which certainly occurred. 

 We have already seen that the internal work spontaneously done 

 in the steel bars of the first experiment, in generating the observed 

 amount of heat, was sufficient to lift the bars about 800 feet, which 

 is 300 times greater than the work done in spontaneously shrinking 

 the small rod. If spontaneous shrinkage was less in the large bars 

 than in the small rod, which is highly probable, then this ratio was 

 accordingly greater than three hundred to one. The disparity in 

 weight between the twelve large bars and the one small rod does 

 not count, because the work done in each case is computed for the 

 weight of steel which did it. 



It has been suggested that loss of the generated heat may per- 

 haps be regarded as a cooling process without change of tempera- 

 ture (which implies reduction in specific heat), and that this may 

 be sufficient to account for the spontaneous shrinkage. But this 

 hypothesis accounts for only a modest fraction of the shrinkage ; 

 while the implied change in specific heat is much too large to be ad- 

 missible. 



An attempt was made to measure Young's modulus of elasticity 

 in the small rod both in the hardened condition (after spontaneous 

 shrinking) and after annealing, by hanging various weights below 

 M, Fig. 4, and measuring with the microscope the distortions pro- 

 duced, — always far within the elastic Hmit. But I was unable to 

 obtain reHable results because of an interesting fact which was 

 brought to light, as follows : In the annealed condition the steel ex- 

 hibited a small amount of viscosity or internal friction which some- 

 what delayed full distortion and subsequent restitution ; but in the 



