Curves of the Oscillating Spark. 715 



two inductances Lj and L 2 ; the spirals thus obtained show 

 alternations o£ light and dark all round at uniform intervals, 

 indicating that the p.d. between the plates X and Y does not 

 vary uniformly with the time but in a series of steps. The 

 period of these agrees approximately with that calculated 

 for the circuit formed by the parallel plates X and Y and 

 the coil L 2 , and the effect is evidently due to oscillations in 

 this circuit superimposed on those of the main discharge. By 

 this means oscillation periods as small as 5 x ]0~ 9 sec. can be 

 rendered evident and measured. Such periods are much 

 smaller than can be measured in any other way. 



Mechanism of Spark Conduction. — The general principles 

 which govern the formation of the spark have been explained 

 by Sir J. J. Thomson * in 1900, but they have not been 

 applied to the condensed spark, which differs in a number of 

 ways from the ordinary spark. We see that there are three 

 characteristic potential differences which occur in the con- 

 densed spark, viz., the initial sparking p.d. of some thousands 

 of volts, the glow p.d. (300 v.) between the oscillations, 

 and the arc p.d. (35 v.) which lasts throughout the greater 

 part of each oscillating discharge. Each of these is evidently 

 associated with a distinct phenomenon, and the three phe- 

 nomena necessary are to hand in the known theory of 

 ionization. 



Consider the conditions in the gap just before the spark 

 occurs. In the absence of a supply of ions at the electrodes, 

 a permanent current cannot be set up until the field reaches 

 a value at which the poLdtive as well as the negative ions 

 produce ionization of the air molecules on collision. This 

 field determines the initial sparking p.d.; and once it is 

 reached, if it could be kept up, the ionization would increase 

 extremely rapidly until it was practically complete. In the 

 actual spark, however, as soon as any appreciable ionization 

 has been produced, the field cannot be kept at its initial 

 value, but must rapidly fall. The rate of fall of the p.d. 

 between the spark terminals will depend on two things, 

 viz.: (1) the rate at which the charge is carried away from 

 the poles by the ions in the gap; (2) the rate at which the 

 charge is brought up to the poles along the wire from the 

 leyden-jar, and this depends on the inductance in the circuit. 

 We have seen that in sparks in free air the p.d. falls to the 

 arc p.d. value in a very small fraction of the oscillation 

 period. The air is now almost completely ionized, but the 

 high initial p.d. has another function to perform than 

 ionizing the air in the gap; for even when the ionization 

 * Phil. Mag. 1. p. 278 (1900). 



