610 Scientific Proceedings, Royal Dublin Societi/. 



entirely remove the spectra of these gases, the gold and carbon alone 

 showing. With elements such as chlorine, bromine, and iodine, the effects 

 of self-induction are not so pronounced. In the case of carbon the effects 

 of self-induction are very pronounced and remarkable. First, with self- 

 induction, the three bands attributed to cyanogen are seen clearly defined 

 in almost any atmosphere except hydrogen, but without self-induction only 

 the one band at A 3883'6 is seen, and it is also seen in any atmosphere 

 except hydrogen. 



Adopting the theory of Schuster and Hemsaleoh, that the current 

 is conveyed across the gap at first by the air when there is no self-induction, 

 I conclude that under these conditions the temperature is too high for 

 cyanogen to be formed by the carbon electrodes and any trace of nitrogen 

 present. If this be so, the band which remains when self-induction is absent 

 is probably due to carbon, and not to cyanogen. 



In the tables of wave-lengths the values given are those generally 

 accepted, and are taken for the most part from Watt's Index, Eder and 

 Valenta's, or Exner and Haschek's Tables. New wave-lengths were 

 measured between known gold lines, and their wave-lengths determined by 

 an interpolation cui've. 



Nitrogen. 



Nitrogen was prepared from ammonium chloride and sodium nitrite, 

 and was purified and dried by means of potassium hydrate and sulphuric 

 acid. 



A great number of nitrogen lines show when there is no self-induction, 

 but on inserting the Hemsalech coil most of these lines disappear, and those 

 which remain are not so strong as before. The three cyanogen bauds, 

 X 4216-1, X 3883-6, A 3590'5, show very strongly with self-induction and 

 carbon electrodes, but one band A 3883-6 alone remains when no self- 

 induction is used. This band is probably due to carbon and not to cyanogen 

 at all. Schuster and Hemsalech showed that the introduction of self-iuduction 

 reduced the temperature of the spark, and that when no self-induction was 

 present the initial discharge of the current was conveyed across the spark- 

 gap by the intervening air. Adopting this view, it would appear that with 

 no self-induction the temperature was too high for cyanogen to be formed 

 by the carbon electrodes and the surrounding nitrogen, and consequently the 

 cyanogen bands do not appear in the spectrum, and the band \ 3883-6 is, 

 therefore, due to carbon. In the spectrum with gold electrodes the carbon 

 lines show ; no doubt this is due to a trace of carbon or hydrocarbons in the 

 flask. 



