798 Prof. Townsend and Mr. Edmunds on the 



considerations the difference between the sparking potentials 

 for positive and negative discharges when the potentials are 

 large, as the values of fijp corresponding to the smaller 

 values of X/p, from X/p=40 to 150, have not yet been 

 determined accurately. 



The fact that the positive discharges are produced by 

 smaller potentials than the negative discharges in these cases 

 is not contrary to the theory, since it is quite possible that 



1 d6 , „ . . .,, 1 da 



— ~- mav become small in comparison with — =. when 

 (3 rfX a dX. 



X/p is between the limits 40 to 150. 



The above investigations apply to cases in which the ions 



are generated in a space near the inner cylinder, and the 



force X at the surface of the outer cylinder is less than 



40 xp. 



9. The effect on the sparking potential when the outer 



cylinder is so small that — is greater than 40 is shown by a 



measurement of the sparking potential obtained with a pres- 

 sure of 1'35 millimetre between cylinders of radii '063 cm. 

 and 198 cm. TVhen the wire was positively charged the 

 sparking potential was found to be 640 volts. This gives 



^ = 69. Hence it is to be expected that the value of aX, 



P . 



should be higher than that obtained in other cases where 



— - is less than 40 : for the outer cvlinder cuts off part of 



P . . 



the region where the force is large enough to produce a con- 

 siderable amount of ionization. The value of aXj obtained 

 from the above result is 186, which is, in fact, considerably 

 higher than the value 150 taken from the curve for the same 

 value, 0*85, of a/ p. On the other hand, the negative spark- 

 ing potential for the same cylinders and air-pressure was 

 found to be 420 volts, giving «X 1 = 122. which agrees with 

 the value 118 taken from the curve. This is quite consistent 

 since the value of X fl ^ is only 45. 



10. Some observations were also made of the relation of 

 the rise in potential to the current and also of the effect 

 of moisture. It was found impossible to work with very 

 damp air without completely reconstructing the apparatus as 

 the moisture settled on the surfaces of the insulators. How- 

 ever, some experiments were made with air taken from the 

 room and freed from dust without being dried. This 

 amount of moisture had no appreciable effect on the sparking 

 potentials, but the increase of potential required to maintain 



