286 Mr. T. Royds on the Constitution 



The air-discharges are o£ two kinds. First, there is an 

 intense and almost instantaneous luminosity, showing in the 

 photographs as a narrow vertical line passing almost straight 

 from one electrode to the other (see figs. 1, 2, 4, 5, PL IV.). 

 This marks the commencement of the spark. Its time of 

 duration is quite inappreciable, and is not increased when 

 the period of the oscillations is made as great as 2 x 10~ 4 sec. 

 By comparison of the photographs with and without pris- 

 matic dispersion (with small self-induction) it is obvious that 

 this initial discharge gives the air spectrum. In a few 

 photographs there is a suspicion that this discharge is 

 followed by a second but much fainter one, in which the 

 duration is also short (see fig. 1, PL IV.). 



The second kind of air-discharges (see fig. 1) have a 

 duration comparable with the half period of the oscillations. 

 One of these follows the initial discharge almost immediately, 

 and the subsequent ones occur during each half-oscillation. 

 With periods smaller than 2 X 10 ~ 5 sec. the velocity of the 

 photographic film was not sufficient to separate the initial 

 <iir-discharge from the first air-discharge of the second kind. 

 Very frequently a single half-oscillation contains several 

 air-discharges (see fig. 2). Their time of appearance is 

 generally seen to coincide with that of metal streamers to be 

 described later. 



The initial air-discharge occupies but a narrow thread in 

 the spark itself, as is evident from direct photographs taken 

 without the interposition of a slit. The air-discharges of the 

 second kind are produced in a wider region of the spark, for 

 if the electrode-box is so placed that the line joining the tips 

 of the electrodes is a little to one side of the slit, the initial 

 discharge is not seen in the light that traverses the colli- 

 mator, but those of the second kind are still as strong as 

 before. When the period of the oscillations has been in- 

 creased to 2*0 X 10~ 4 sec. (fig. 3) the central thread only 

 is seen ; discharges of the second type are absent from 

 the photographs on the moving film. 



Metallic Streamers (figs. 2, 4, 5, PL IV.). — After the 

 initial air-discharge the light from the spark is chiefly due 

 to the incandescence of the metallic vapour which is pro- 

 duced. The moment of vaporization of metal is found, as 

 in the former experiments with small self-induction, to 

 be practically simultaneous with the passage of the initial 

 air-discharge. It is important to notice that metallic 

 vapour is formed at the commencement of the spark at 

 both electrodes and advances from both electrodes towards 



