252 REPORTS ON INVESTIGATIONS AND PROJECTS. 



are somewhat empirical, and any information relating to infra-red 

 lines predicted by them will aid in establishing their validity. 



In the present work the same apparatus was employed as in the 

 investigation of absorption spectra. The salts of the alkali metals 

 were examined in the carbon arc. An examination was also made 

 of the arc between metal electrodes, but in neither case could emis- 

 sion lines be detected beyond 2 fi. This is caused in part by the 

 radiation from the oxides, which is sufficiently intense to obliterate 

 any weak lines if there be any in this region. The vacuum-tube 

 radiation does not contain that of the oxides; hence was better adapted 

 to this work. Since there is but little heat radiated from a vacuum 

 tube, a very sensitive radiometer was necessary — in fact, the most 

 sensitive one yet constructed. 



The following gases and vapors were examined : CO, C0 2 , NH 3 , 

 H, O, N, C 2 H 5 OH, and H 2 0. In the preliminary work an emission 

 band was found in common at 4.75 fi for all gases except H and water 

 vapor. After eliminating all traces of C0 2 this band was not found 

 in any gas except C 2 H 5 OH (vapor), C0 2 , and CO. Since this band 

 is more intense for CO than for C0 2 , other conditions being equal, 

 it is quite probable that it is due to CO. 



Nitrogen has strong emission bands just beyond the red, but none 

 of the gases show lines beyond 2 fi, except CO and C0 2 , for the region 

 of 4.75 fi. These emission bands were examined for constant current 

 and varying pressure of the gas, and vice versa. The band at 4.75 a 

 behaves in an entirely different manner from all the rest. Its inten- 

 sity increases with increasing pressure (for constant current), but 

 never attains a maximum, becoming quite asymptotic at 5 to 6 mm. 

 pressure. On the other hand, the other emission bands increase in 

 intensity with increase in pressure, become a maximum at about 

 2 mm. pressure, then decrease in intensity with a further increase 

 in pressure, which agrees with observations in the visible spectrum. 

 All lines increase in intensity with increase in current, as found in 

 the visible spectrum. The whole shows that the bands near the 

 visible spectrum are related to those in the visible, while the 4.75 

 band is of an entirely different type. 



The decrease in infra-red radiation and the simultaneous increase 

 in the visible radiation, with decrease in pressure , explains the rise in 

 efficiency of vacuum tubes observed by others. 



The results of these investigations were published in October, 1905 

 as Publication No. 35 of the Carnegie Institution of Washington. 



