224 COMPTON— ARC SPECTRA AND 



maintained at a voltage lower than this, which is the minimum ionizing 

 potential of the molecule. 



The shape of these curves differs from that of the hydrogen arc 

 curves in that the current rapidly increases with increasing voltage 

 above the arcing voltage. This is probably due to the greater com- 

 plexity and stability of the molecule, since it seems to be related to 

 the rapid increase of intensity of the negative band spectrum as the 

 voltage is increased. This negative band spectrum is due, as shown 

 by its behavior in a positive ray apparatus and its relation to exciting 

 voltage, to excitation of the positively charged molecules N,, which 

 can evidently be partially or completely ionized without breaking up 

 into atoms. 



In the neighborhood of 70 volts the arc current began to increase 

 rapidly again, a brilliant " flare " setting in and the current becoming 

 as large as 15 amperes. Simultaneously with this increase the lines 

 5006 and 5003 of the nitrogen line spectrum appeared. At about 90 

 volts the lines 5680 and 5667 were visible. If the filament is hot, 

 this " flare " could be maintained at voltages as low as 25 volts, after 

 setting in at 70 volts, but no lines of the line spectrum were seen 

 below 70 volts. Lines of the tungsten spectrum were always ob- 

 served in the flare. 



Comparison with observations by Strutt^* shows that this flare 

 was due to the presence of atomic nitrogen, and that the presence of 

 the tungsten spectrum was due to the burning of tungsten in atomic 

 nitrogen. It is significant that the flare could be produced at 40 volts, 

 instead of 70 volts, if the outer tungsten tube were very hot. This 

 is the only evidence of dissociation of nitrogen by the hot tube. It is 

 also of interest to note that Langmuir discovered an anomalous be- 

 havior of thermionic emission from tungsten filaments in nitrogen at 

 low pressures and voltages above 70 volts, and attributed this to a 

 chemical action which did not occur at lower voltages. ^^ It is not 

 easy to explain why the formation of atomic nitrogen should increase 

 the current through the arc so enormously. Multiple ionization, 

 ionization by chemical action on tungsten, or heat developed by atomic 

 recombination at the surface of the filament may be important factors. 



^* Loc. cit. 



15 Phys. Rev., 2, p. 450, 1913. 



