to produce Discharges in Gases at Low Pressures. 487 



decreased, tlie minimum sparking potential rises for the 

 first direction of: the field, and becomes still lower for the 

 second, the change becoming more and more rapid as we 

 approach the small diameters. But finally a critical diameter 

 is reached for the negative and a different one for the 

 positive, and when the diameter is decreased below these 

 critical values the effect on the minimum sparking potential 

 is reversed, so that a sharp drop appears in the positive and 

 a rise in the negative curve. 



Fig. 9 also shows curves for hydrogen which are some- 

 what similar to those for air, with a maximum in the positive 

 and a minimum in the negative. The number of points for 

 the hydrogen curves is not such that we can consider the 

 latter as definitely established, but they are probably ap- 

 proximately correct. If the curves in fig. 9 are produced, 

 as seen by the dotted lines, they come to horizontal positions 

 at approximately parallel plate values of the minimum 

 sparking potential, and at the diameter of the outer cylinder. 

 Additional readings might cause the curves to be drawn 

 somewhat differently, but this ought to hold true in any 

 case, since the field between concentric cylinders approaches 

 that between parallel plates as a limit, and the minimum 

 sparking potential for parallel plates is constant. 



The phenomena of the discharge between concentric 

 cylinders may be explained along lines suggested by 

 Prof. Townsend in connexion with the collision theory*. 

 Above the critical pressure, all the main effects may be 

 explained on the supposition that the ions are produced by 

 the collision of positive and negative ions with the molecules 

 of the gas. But below r , and possibly in the region of the 

 critical pressure, some other process of ionization in addition 

 to ionization by collision evidently comes into play, and 

 Prof. Town send suggests that some form of non-penetrating 

 radiation, due to the impact of negative ions against the 

 positive inner electrode, may be the cause of the effects 

 obtained. Such radiation, dependent upon the velocity of 

 the negative ions, would affect only the | ositive curves in 

 figs. 3 and 4, since only when the inner cylinder was positive 

 would the negative ions approach the metal under a large 

 force and impinge on the electrode with a high velocity. 

 The radiation would naturally be absorbed at the higher 

 pressures, but when the pressure first reached the point at 

 which radiation would be present, the curve for positive 

 inner electrode would begin to fall below the position which 



* ' The Theory of Ionization of Gases by Collision,' pp. 67 ei seq. 

 and 76. 



