706 Prof. C Kinsley on 



includes the values obtained by Earhart and by Shaw, as 

 shown in fig. 9. There was no indication that values both 

 higher and lower could not also have been obtained. 



The constancy o£ the potential gradient throughout every 

 independent series of observations is good evidence that the 

 carrier of the electrical discharge is unchanged between 

 450 pfi and 3 pp. 



Since the relation between the spark-potential and spark- 

 distance gives a straight line passing through zero, it is 

 evident that throughout the range examined the discharge 

 occurs along the shortest path between the two surfaces. 



The sudden change of potential gradient depending on the 

 treatment of the surface, and the slow change while isolated 

 in the discharge-chamber, seems to show that the carriers of 

 the electrical discharge come neither from the gas in the 

 chamber nor the metal of the electrodes. 



The lag in the discharge, as shown in the tables, is the 

 same phenomenon so well known in the* case of long spark- 

 discharges. It points to the same mechanism of discharge for 

 short as well as for long sparks. 



Probably a determination of the cause of the change in the 

 potential gradient will settle the question of the real carrier 

 of the electrical discharge in the case of short sparks. 



I am pleased to acknowledge that I have been greatly in- 

 debted to Professor A. A. Michelson for his continued 

 interest and many suggestions throughout the course of this 

 investigation. 



Byerson Laboratory, University of Chicago, 

 December 23, 1904. 



Appendix to paper on " Short Spark-Discharge $ : }> 



Note on the Coherer. 



In the preceding paper it was found that the discharge 

 caused the surfaces to cohere. Although it has already been 

 convincingly shown that the coherence is caused by the for- 

 mation of a metallic bridge * between the surfaces, the following- 

 evidence on that point may be of interest. 



When two clean metallic surfaces are pressed together they 

 will form a low-resistance metallic contact. If now they are 

 slowly separated, the resistance of the contact will rapidly 

 increase. Under normal conditions the resistance will increase 



* For a resume see Gute, Proc. International Electrical Congress, 

 St. Louis, 1904. ° ' 



