72 



SPECTRUM ANALYSIS OF FLAMES. [MEMOIR II. 



Table of wave-lengths corresponding to the eight great fixed lines of 

 the solar spectrum, the Paris inch being supposed to be divided into 

 one hundred millions of equal parts. 



A = 26GO 

 B = 2541 

 C = 2422 

 D = 2175 



E = 1945 

 F = 1794 

 G = 1587 

 H = 1464 



An examination of this table proves that 



The wave-length of B is 119 parts less than A ; 

 C 238 



238 

 485 

 715 

 866 

 1073 

 1196 



and these differences of length are obviously very nearly 

 as the whole numbers 1, 2, 4, 6, 7, 9, 10. This coincidence 

 is far too striking to be merely accidental. Moreover, 

 it must not be forgotten that the observed numbers as 

 determined by Fraunhofer are wholly independent of 

 any hypothesis. 



If the relation of whole numbers were rigorously true, 

 the numbers in the foregoing table would stand as fol- 

 lows: 119, 238, 476, 714, 833, 1071, 1190. 



The wave-length of the most luminous portion of the 

 spectrum, the centre of the yellow space, is 2060 parts. 

 If we take this as an optical centre, it will be found that 

 the great lines are situated symmetrically in relation to 

 it. E and D are equidistant above and below it; the 

 same observation applies to G and B, and also to H and 

 A. The only departure from this symmetry is in the 

 case of F, which is not symmetrical with C. It will be 

 understood that I am here speaking of one of those 

 spectra which are formed when a grating or ruled sur- 

 face is used. In this the colors are arranged side by 

 side, according to their wave-length, the centre of the 

 spectrum, which is its most luminous portion, is occupied 



