538 Royal Society ;— ^ 



nitrogen gives only one spectrum; (2) that this is the line-spec- 

 trum ; (3) that the fluted spectrum of the first order is due to oxides 

 of nitrogen, formed under the influence of the electric spark. 



2. First experiment. — The first experiment which I made with 

 respect to the spectrum of nitrogen, was a repetition of an experi- 

 ment of Secchi, who found that in difl^erent sections of the same 

 tube three diff'erent spectra of nitrogen might be obtained. A 

 vacuum-tube was made exactly according to Secchi' s description, 

 filled with nitrogen and exhausted. To my astonishment the tube 

 showed, even in its widest parts, only a spectrum of Hues. No accu- 

 rate measurements were taken at the time, but the spectrum was no 

 doubt that of the second order described by Pliicker. Suddenly, and 

 while I was looking through the spectroscope, the spectrum changed, 

 and the well-known fluted bands appeared. The first spectrum 

 could now easily be obtained by introducing a Leyden jar in the 

 circuit. The spark very soon ceased to pass, and it was then found 

 that the tube was leaking. 



3. The behaviour of this tube at once suggested the idea that the 

 presence of air was necessary for the formation of the fluted spec- 

 trum. It is well known that the oxides of nitrogen are formed on 

 passing the electric spark through air, and the resemblance which 

 this spectrum bears to the spectra of the oxides of metals rendered 

 this view probable. In order to test it, a series of experiments were 

 made, showing that, — 



(«) Whenever the fluted spectrum appeared, it could be shown 

 that traces of oxygen were present ; 



(6) Whenever there- was a certainty of no oxygen being present, 

 the spectrum of the second order appeared under all pressures and in 

 all temperatures. 



In order to free the nitrogen from every trace of oxygen, I adopted, 

 at Dr. Stewart's suggestion, the plan of heating a small piece of 

 sodium placed in the vacuum-tube. This proved in each case per- 

 fectly satisfactory ; for when every trace of oxygen had thus been 

 absorbed, the line-spectrum alone was invariably obtained*. 



4. Wave-length of the two spectra. — There is no possibility of 

 confounding the two spectra. The fluted spectrum is well known 

 by its beautifully shaded violet bands ; but in order to exclude any 

 possibility of error, their position was read off* on the reflecting scale 

 of the spectroscope ; the measurements were reduced to wave-lengths, 

 and the following numbers obtained for the least refrangible end of 

 the bands in tenth metres f: — 



Fluted Spectrum. 



5129 



4436 



4981 



4390 



4649 



4318 



4556 



4237 



^ The formation of the flsied spectrum does not im^Dly that all the nitrogen in 

 the tube has been oxidized ; it has been remarked by different observers, and espe- 

 cially noticed by Pliicker, that when the spark passes through a mixture of two 

 gases, the spectrum of one only is often seen. 



t A tenth metre, according to Angstrom, means a metre divided by 10^°. 



