94 BELL SYSTEM TECHNICAL JOURNAL 



tion: the well-known action of free electrons striking molecules of the 

 gas, and a complementary process, which may (for instance) be the 

 ejection of fresh electrons from the cathode by positive ions striking 

 that electrode.^^ Were it not for the latter process (or some other) the 

 direct-current discharge could never develop; for though at a given 

 instant there might be some electrons in the gas, they and all the 

 other electrons which they might liberate would steadily drift off 

 toward the anode, and ionization and current-flow would cease after 

 all of them had reached it. Now if the voltage is oscillating instead 

 of constant, the electrons in the gas may rush to and fro and ionize all 

 parts of it, and the importance of the complementary process will be 

 reduced; though it can never be annulled, since the electrons will 

 sooner or later get to the anode or the walls, and must be replenished 

 from the cathode. 



Again, we know that a factor in the advent of breakdown by a 

 steady voltage is the distortion of the field in the gas by space-charge, 

 which arises chiefly near the cathode, because the positive ions formed 

 there by electron-impact drift only slowly toward the cathode while 

 the electrons which should balance their charge drift rapidly off toward 

 the anode. If the voltage is oscillating there will also be a positive 

 space-charge due, in the last analysis, to the fact that electrons drift 

 faster than positive ions; but it will be distributed symmetrically 

 about the middle of the gap. 



These remarks may have given the impression that the differences 

 between breakdown at high frequencies and breakdown by steady 

 voltage have been successfully explained. As a matter of fact, there 

 is no quantitative explanation, and I have little to say except to present 

 the data. 



Breakdown of Air at Atmospheric Pressure 



For air at atmospheric pressure, for which breakdown is spark- 

 over unless one or both of the electrodes be sharply curved, the latest 

 data are those of Lassen. 



Curves of sparking-potential versus distance, (Fg-vs-c? curves), 

 obtained with spherical electrodes of 2.5-cm. diameter, over the range 

 of distances from 0.05 to 0.5 cm., appear in Fig. 12. The voltage was 

 adjusted to a chosen value, and the distance gradually lessened until 

 sparkover occurred. The straight line is fitted to the points obtained 

 with frequency 1.1-10^ and the points obtained with frequency 50. 

 Fifty-cycle A.C. is practically the same, with regard to the processes 



""Electrical Phenomena in Gases," pp. 280-297. 



