﻿the Magnesium High-frequency Arc. 99 



the presence of two pairs of lines at wave-lengths 4385,4391 

 and 4428, 4434 *. 



These lines are ill-defined in the ordinary spark, but 

 appear distinctly when a small amount of inductance is in- 

 troduced into the discharge circuit, disappearing, however, 

 when the inductance is increased. In the high-frequency 

 arc these lines do not appear at the lower frequency of 

 1'lxlO 5 per sec. when the inductance is 163,200 cm., 

 but when the frequency is raised to 3*9 X 10 5 by reducing 

 the capacity only slight traces of the lines appear. On 

 increasing he frequency by decreasing both capacity and 

 inductance the lines become still more prominent. If the 

 frequency is kept constant and capacity and inductance 

 varied, the lines are strongest for the smallest value of the 

 inductance. 



Brooks f in his high- and low-frequency experiment found 

 that if in the high-frequency discharge the thin electrodes 

 were replaced by thick rods, the lines (F and P) were still 

 present, but not the hydride spectrum, except that after long 

 running traces of the head of the band 5210 appeared. 

 Comparing the conditions of the high- and low- frequency 

 discharge in Brooks's experiment, the introduction of the 

 leyden-jar increases both frequency and current density. 

 Brooks considers the current density to be the factor govern- 

 ing the presence of the lines. With the undamped high- 

 frequency arc the lines were absent at the low-frequency 

 and high-current density, but when the frequency was raised 

 by decreasing the capacity, and consequently decreasing the 

 current density, the lines appeared. This suggests that the 

 presence of the lines is governed by the frequency rather 

 than the current density. 



The only indication of the hydride is a trace of the head 

 5210 (lens), which, like the F and P lines, is to be seen in the 

 spectra given by the higher frequencies (PLY. fig. 2,/, g, h } i) . 

 If the arc is taken in an atmosphere of coal-gas the band 

 becomes very intense, and the carbon fluting at 4315 is 

 strong. No connexion between the line 4571 and the hydride 

 spectrum was observed. In view of the fact that Brooks 

 associates the presence of this line with the formation of a 

 nitride, it is of interest to notice that the conditions in the 

 oscillatory circuit which give this line with the greatest 

 intensity at the low potential are those which give the best 



* A. Fowler and H. Pavn, Proc. Roy. So<\ lxxii. p. 253 (1003). 

 t E. E. Brooks, Proc. Roy. Soc. lxxx. p. 223 (1908). 



H 2 



