(]g SPECTRUM ANALYSIS OF FLAMES. [MKMOIK JJ. 



except, perhaps, the yellow ; but the green, the blue, and 

 the violet greatly predominate. The interior lilac flame 

 is the source of the bright spectrum with fixed lines just 

 described. 



V. Continuation of the same principle in the case in 

 which combustion is carried on in oxygen gas instead of 

 atmospheric air. 



If the principle that high refrangibility is connected 

 with intense chemical action be true, it must hold good 

 when the nature of the atmosphere in which the burning 

 is carried forward is changed. If, instead of being the 

 common air, it is oxygen gas, we ought to be able to 

 foresee the result. Carbonic oxide, when made to burn 

 in that gas, should not change its tint ; because if the air 

 can carry on the process to its maximum effect, oxygen 

 can do no more. But the result should be just the re- 

 verse with cyanogen, which, if made to burn in oxygen, 

 should be capable of emitting rays of higher refrangi- 

 bility. 



Foreseeing this result, I submitted the two gases to 

 experiment, and first arranged the carbonic oxide so 

 that its spectrum might be examined in the telescope 

 as already described; then causing a clean bell -jar full 

 of oxygen to be inverted over it, the flame diminished 

 somewhat in size, emitted a slight crackling sound, but 

 retained its color unchanged. Its spectrum appeared pre- 

 cisely the same both as respects extent and the distribu- 

 tion of color, whether the burning took place in oxygen 

 gas or in atmospheric air. 



If cyanogen be made to burn in oxygen, we should 

 expect that it would lose to a great extent its character- 

 istic lilac tint, and emit a whiter light. It was therefore 

 very interesting to find that the moment the flame was 

 immersed in oxygen it lost much of its pinkish color, 



