158 



BKIDGMAN. 



cannot speak of the electrode resistance of a two dimensional conductor. 

 Applied to the present case, this means that if the electrodes are not 

 situated near to each other on opposite sides of the isthmus, the total 

 resistance will not be merely the resistance of the isthmus and practi- 

 cally nothing else, but the rest of the sheet will make a finite con- 

 tribution. But the rest of the sheet will not make a finite contribu- 

 tion to the departure from Ohm's law, since the departure increases 

 very rapidly with increasing current density, and it is only in the 

 isthmus that the density is high. 



The experiment was tried of making a specimen with two sets of 

 electrodes, one pair close to the isthmus, and the other pair much 



Figure 6. Results obtained with gold 8XlO~^ cm. thick. Abscissae, 

 current density in 10^ amp/cm^, ordinates, extrapolated difference between 

 A.C. and D.C. settings in cm. of bridge wire. The two sets of points were 

 obtained with the same specimen, but with different positions of the electrodes. 

 The points shown by the crosses are the observed points corrected by the 

 ratio of the total resistance to the resistance of the isthmus. If the effect 

 measured is a genuine departure from Ohm's law, the corrected points should 

 fall on the same curve with the others. 



more remote. The total resistance between the second pair of elec- 

 trodes was 1.27 times that between the first. Now if the effect found 

 is genuine, the extrapolated value of AR' for the same value of the 

 current density in the isthmus should have the same absolute value 

 irrespective of the position of the electrodes, and should not merely 

 be the same fraction of the total resistance between electrodes. That 

 this demand is met is shown in Figure 6. The points for the two posi- 

 tions of the electrodes lie on the same smooth curve within the errors 

 of the measurements, and this error is evidently much less than a 

 factor of 1.27. 



