FAILURE OF OHM's LAW AT HIGH CURRENT DENSITIES. 159 



Experimental Procedure and Data. 



After the preparation of the sample, and measurement of its dimen- 

 sions, it was adjusted in the holder, the fine copper wires which were 

 soldered to the metal leaf were in turn attached to heavier leads, 

 and these were connected to the proper gap in the bridge, and the 

 nozzle for the cooling water was adjusted to give the most efficient 

 stream over the specimen. As already mentioned, the cooling water 

 was distilled water. This was contained in one of two large carboys, 

 which were at the temperature of the room. On starting the water, 

 there was no change in the zero due to change of temperature of the 

 specimen. Because of the limited capacity of the carboys, it was 

 necessary several times in the course of the experiment to interrupt 

 the readings to turn water from one carboy into the other. 



Readings were bcgim at some small value of the direct current, 

 obtained l)y insertion of the proper shunt in the feeding circuit. 

 An initial reading was made of the A.C. zero, with no D.C. flowing. 

 Readings were now made of the D.C. balancing point with A.C. flow- 

 ing, and the A.C. balancing point with D.C. flowing, and the two 

 readings were repeated with D.C. reversed. This set of four readings 

 was repeated for each of several frequencies, beginning at the lowest. 

 The frequencies used were 320, 460, 680, 1140, 1530, 2450, and 3750. 

 The frequencies were calibrated with tuning forks. At each frequency 

 the mutual inductance was readjusted to give the sharpest setting. 



Originally more readings were taken. These were the balancing 

 point for small D.C, A.C. zero with no D.C, D.C. balancing point for 

 large D.C. with no A.C. flowing, balancing point for large D.C with 

 A.C. flowing, and balancing point for A.C. with large D.C. flowing. 

 After a number of runs had been made, taking all these readings, I was 

 satisfied that these precautions were not necessary, and the readings 

 mentioned above proved sufficient. 



The A.C. zero, which gives the actual resistance of the specimen 

 itself, may change slightly during a run at a single D.C. intensity 

 and a series of frequencies. This may be due to some mechanical 

 change in the specimen, because of friction by the cooling water, or 

 at high current densities may be due to incipient burning out. Any 

 such change during a run was always small however, and could be 

 disregarded in the computations. 



The series of readings at different frequencies was now repeated at 

 some higher value of D.C, and the series continued until so high a 



