of the Electric Spark. 289 



metallic calcium electrodes, that often there appear to be two 

 well-marked velocities commencing from the electrode at 

 the beginning of the spark (see fig. 5). The larger of these 

 is produced by the first stage of the first metallic streamer, 

 and the smaller often consists of an envelope formed by the 

 union of several streamers. Now photographs of the spectrum 

 have shown that all the calcium lines have the same velocity; 

 so that this is not a case of different velocities from different 

 lines as observed with bismuth and the metals. In fact, two 

 velocities can be seen in a single spectral line. All the lines 

 show both velocities when seen at all. The two simultaneous 

 velocities may be seen at the positive or negative electrode 

 or at both. Since they start at the commencement of the 

 spark, it seems probable that each represents a velocity of 

 metallic vapour. Thus, in the preceding table, two values are 

 given for the velocity of calcium. 



It should be clearly understood that the velocities in the 

 above table are those of the vapour which starts from 

 the electrodes at the commencement of the sparlc. The slope 

 of streamers which set out from the electrodes subsequently, 

 probably indicates not the velocity of the vapour, but shows 

 the velocity with which a pulse of luminosity travels through 

 vapour which has been previously formed. Milner * has, 

 however, compared the inclination of the streamers in the 

 spark with self-induction with that obtained in the induc- 

 tionless spark where the oscillations could not be separated, 

 believing that the former gave the velocity of the metallic 

 vapour. 



Summary of Results. 



1. It is shown that the initial air-discharge is always 

 present and remains instantaneous in character and simul- 

 taneous at all points of the spark-gap, even when the 

 period of the oscillation is increased to 2 X 10~ 4 sec. The 

 initial air-discharge occupies but a narrow thread in the 

 spark. 



2. The discharges through the air which follow the initial 

 one, and are of longer duration, are seen under certain 

 circumstances to be subdivided even in a single half- 

 oscillation. 



3. It is shown convincingly that the vaporization of the 

 metal electrodes is simultaneous with the initial discharge 

 and takes place equally at both electrodes, although after the 



* Milner, Phil. Trans. A. ccix. p. 71 (1908). 

 Phil. Mag. S. 6. Vol. 19. No. 110. Feb. 1910. U 



