72 Wright and Downs — Spectrum of the 



great as in the first case owing to the fact that the flame was 

 somewhat disturbed by the jet of steam. Nevertheless, when 

 the plate was developed, the lines upon the lower half came out. 

 much more strongly than upon the upper, the only difference 

 being in the intensity, which shows that if the lines are not 

 directly due to water vapor, the latter greatly facilitates the 

 development of lines due to the several gases of the air, or 

 combinations of these gases, probably both. 



Tables of wave lengths of the different constituents of the 

 atmosphere were next examined to see if any of the lines 

 given coincided with the lines studied. In a note* regarding 

 a very careful search which he made for the line spectrum of 

 hydrogen in the oxy-hydrogen flame, Liveing states that he 

 failed to find the slightest trace of any one of the hydrogen 

 lines. As oxygen is a very large constituent of the air, it was 

 very important to make a comparsion of the lines due to it, which 

 were catalogued, with these lines. There were found, how- 

 ever, to be no coincidences, which indicates that none of the 

 lines are due to oxygen. 



Just before the completion of this work an article by G. 

 Berndt appeared. f While studying the spectra of several 

 metals, Berndt had noticed in all of them a banded spectrum 

 extending from w r ave lengths 5100 to 2000, a portion of which 

 covers the region in which the banded structure previously 

 described was located. He concluded from his observations 

 that the banded structure in the visible portion of the spectrum 

 of different metals is dependent upon the presence of oxygen 

 and also suggested that the banded spectrum in the ultra-violet 

 is due to air. At the close of his article Berndt publishes a list 

 of nitrogen lines obtained by an induction discharge in nitrogen 

 free from oxygen. Upon comparing them with the lines 

 studied in this investigation, a number of apparent coincidences 

 were seen, making it probable that some of these lines at least 

 were due to nitrogen. 



The last experiments made were a continuation of the 

 attempt to ascertain if possible to. which one of the intra-polar 

 gases the lines belonged. As the solar light in this region is 

 very feeble, the light from a voltaic carbon arc was passed 

 through bulbs containing nitrogen tetroxide, and an attempt 

 made to secure the absorption lines due to it. The lower half 

 of the photographic plate was exposed to the light which 

 passed through the nitrogen tetroxide and the upper half to 

 the light from the discharge alone to see if any of the lines 

 coincided with the absorption lines due to the nitrogen tetrox- 

 ide. Although several exposures of nearly two hours were 



* Phil. Mag., vol. xxxiv, p. 371. 



f Drude's Annalen der Physik, No. 4, p. 788, 1901. 



