﻿100 Spectrum of the Magnesium High-frequency Arc. 



nitrogen-band spectrum, when by the introduction of a 

 solenoid the higher voltage is used. 



At the high voltage an arc of several mm. length can 

 easily be maintained, but the intensity o£ the light dies out 

 rapidly with increase of length. The length used in these 

 experiments was between 2 and 3 mm. 



The lines of the series, 5528, &c, are sharpened but much 



weakened in intensity, and die out as the frequency is raised 



(PL V. fig. 2, j, k,l, m, n). The trace of the positive-band 



spectrum of nitrogen which was only slightly visible at the 



lower potential becomes much more prominent. At the 



lowest frequency used (1*1 X 10 5 per sec.) it is not easily 



obtained. It comes up intermittently and dies out leaving a 



continuous background with only a few magnesium lines and 

 © j © 



the oxide fluting at 5007. The intensity of the nitrogen 

 spectrum is roughly in inverse proportion to the intensity of 

 the other lines. This is especially applicable to the oxide 

 fluting, which was never found to be present at the same 

 time as a good nitrogen-band spectrum. At the higher 

 frequencies, when the bands have become the most prominent 

 part of the spectrum, all traces of the oxide have disappeared, 

 and the spectrum consists almost entirely of the nitrogen 

 bands, the b triplet, H*, and the line 4481. At these fre- 

 quencies the current through the arc is very small and 

 unsteady. The light is consequently feeble and the spectrum 

 can only be seen intermittently, the lines not all disappearing 

 at the same time. It will be noticed that the most refracted 

 line of the b triplet is very faint (m, n). Visual observations 



showed that during the fluctuations of the light this was the 

 © © 



least persistent component of the triplet. At the same time 

 the presence was noted of several fine lines in the neigh- 

 bourhood of the triplet, some of which appear on the photo- 

 graphs (n). These may be some of the lines of the fluting 

 already measured by Eder and Brooks, although their 

 appearance suggests a possible connexion with the F and P 

 lines and the associated ultra-violet lines. The wave-lengths 

 could not be determined accurately enough to enable a 

 connexion between the frequencies to be established with 

 certainty. 



There are also seA'ernl lines which could not be identified 

 between the oxide fluting at 5007 and the b triplet, which, 

 like the above pairs, are seen in the high-frequency spectra 

 at low voltages. At both high and low voltages a line was 

 visible though not strong at 6160. It is intensified by the 

 high voltage. Several variations in the mode of producing 

 these spectra have been tried, including an air-blast directed 



