68 SPECTRUM ANALYSIS OF FLAMES. [MEMOIR II. 



where the combustion is incipient, and from which red 

 light issues; then follow orange, yellow, green, blue, 

 indigo, and violet circles in succession, the production 

 of each of these tints being dependent on the rapidity 

 with which chemical action is going forward that is, 

 on the amount of oxygen present the tints gradually 

 shading off into one another and forming, as I have said, 

 a circular rainbow. An eye placed on the exterior of 

 such a flame sees all the colors conjointly, and from their 

 general admixture arises the predominant tint. 



An examination of the flame of a candle vertically 

 confirms this conclusion, for the red projects on the top 

 of the flame, and the blue towards the bottom. 



From this, which may be regarded as the normal 

 flame, the flame of cyanogen differs. It must consist of 

 as many concentric shells as the prism separates it into 

 regions of definite refrangibility. Its interior part is 

 tlu lefore divided into four red layers, followed by one 

 of orange, one of yellow, seven of green, etc. There are 

 two great inactive spaces towards the outside of the 

 flame, corresponding to the two great groups of fixed 

 lines. Perhaps through all these inactive parts the in- 

 combustible nitrogen chiefly escapes. 



VI. Effects of tlie introduction of air into the interior 

 of a flame, producing the destruction of the red and 

 orange strata, and converting them into violet. 



It now becomes a curious subject to determine what 

 takes place when an ordinary flame is disturbed by the 

 introduction of air into its interior. When a blow-pipe 

 jet is thrown through the flame of an oil-lamp, the sharp 

 blue cone which forms indicates, on the principles here 

 set forth, that the combustion is much more active. But 

 if the colors of the common flame come from different 

 depths, the red being the innermost, it is clear that the 



