58 BRIDGMAN, 



seem to me that there is a great deal of significance in the " coefficient 

 of specific resistance." However, it is of interest to note in the table 

 that the changes of dimensions of manganin and therlo are so large 

 compared with the tension coefficient of observed resistance that the 

 tension coefficient of specific resistance is negative, whereas the tension 

 coefficient of observed resistance is positive. For the other metals the 

 correction for change of figure is not large enough to change the sign 

 of the coefficient of observed resistance. 



It is in the first place to be remarked from the table that of the seven 

 substances which are abnormal with respect to the sign of the pressure 

 coefficient, only two, bismuth and strontium, are abnormal with 

 respect to the sign of the tension coefficient. This would seem to 

 indicate some essential difference between the conduction mechanism 

 of these two substances and that of the others. Let us discuss what 

 this difference may be in the light of the theory of metallic conduction 

 which I have previously developed. 



I have thought of conduction as due to a free path mechanism; 

 the classical theory was a free path theory. The difl^erences compared 

 with the classical theory are these. In the first place, the free paths 

 are thought of as long, because the free electrons are few in number. 

 In tiormal metals, the paths of the electrons are to be thought of as 

 through the substance of the atoms themselves. The path may be 

 terminated when the electron makes the jump from one atom to the 

 next. The chance of termination on making the jump will depend 

 both on the amplitude of atomic vibration and the distance apart of 

 the atoms. Now if the distance apart of the atoms varies little com- 

 pared with the changes of amplitude, the variation of free path may be 

 calculated in terms of the variation of amplitude only. The changes 

 of amplitude, neglecting the effects due to changes of dimensions, may 

 be calculated for changes of pressure and temperature, and so the 

 change of path, and hence the changes of resistance may also be cal- 

 culated. It is in throwing the entire burden of the variations on the 

 free path, and in the method of computing the changes of the free 

 path, that my theory differs mathematically from the classical theory. 

 Now as a matter of fact, the changes of dimensions under changes of 

 temperature are very small compared with the changes of amplitude, 

 and the calculated changes of resistance agree well with the observed 

 changes. Under changes of pressure the changes of dimensions are 

 several fold larger, but still are so small compared with the changes 

 of amplitude that an important part of the pressure coefficient 

 may be computed. There is left an outstanding eff'ect depending on 



