64 BRIDGMAN. 



depressed. (The compressibility and Young's modulus which enter 

 this relation have not been determined experimentally.) To explain 

 the failure of a single sharp transition point, we invoke the same sort 

 of mechanism that we suppose to be responsible for elastic hysteresis. 

 Modern theories of the structure of metals explain elastic after effects 

 and hysteresis by the fact that the microscopic crystalline grains are 

 unequally exposed to changes of stress. Certain grains are unfavor- 

 ably situated, and the elastic limit of these is exceeded before that of 

 the average. Just these grains would first reach the transition point 

 under an increase of tension, and other grains not until later. Thus 

 the transition and also the change of resistance would not be expected 

 to take place discontinuously at a single tension, but to be spread over 

 a range instead. If now the resistance of the new phase is less than 

 that of the original phase, we have a reason for the sign of the observed 

 effect. The minimum of resistance with increasing tension is reached 

 at that point where the rate at which new grains are being trans- 

 formed has been so decreased by exhaustion that the decrease of resist- 

 ance brought about by the transition is equalled by the increase of 

 resistance due to the normal tension effect on the transformed grains. 

 This view of the phenomenon demands in the first place that the 

 effect of tension be normal on that phase which is stable above 360°. 

 So far as I know this effect has never been measured, but it would be 

 going out of one's way to assume that it is abnormal. It is also 

 demanded that the resistance of the high temperature phase when sub- 

 cooled into the low temperature region be less than that of the low 

 temperature phase, at least under the tensions which are found inside 

 the grains. This again is subject for further experimental investiga- 

 tion. So far as I know, no measurements have been made on nickel of 

 high purity. The measurements of Werner ^° are the only ones which 

 I know with respect to the behavior of resistance through the transi- 

 tion point. I find from his data that the temperature coefficient of 

 his hard drawn wire between 31° and 110° was 0.0041 of its value at 

 0°, and for soft wire the corresponding coefficient between 18° and 111° 

 was 0.0054. The pusest nickel should give under the same conditions 

 a coefficient of about 0.0062. Werner finds no discontinuity in the 

 resistance at the transition point, but does find a change in the direc- 

 tion of the curve with temperature. It is highly probable that there 

 was actually a discontinuity, which was masked by the effect of 

 impurity. It has almost always turned out that a phenomenon which 



20 M. Werner, ZS. Anorg. Chem. 83, 275-321, 1913. 



