FLAMES OF COMPOUNDS OF CARBON AND HYDROGEN. 425 



almost instantaneous appearance at very long intervals, — for it did occasionally 

 appear for a moment, — satisfactorily proved it to be due merely to foreign matter 

 which had accidentally entered the flame. 



From an examination, either of Table IV., or of Plate VIII., fig. 1, it will be seen 

 that certain of the lines in the carbohydrogen spectrum occupy nearly the same 

 places in the scale of refrangibility with dark lines in the solar spectrum. These 

 are the lines a, 7, 8 3 , and £, which coincide more or less exactly with the lines D, 

 b 2 , F 1? and G. The first of these coincidences has been long known, having been 

 discovered by Fraunhofer ;* and similar remarkable relations have been observed 

 by Sir David Brewster to exist between certain lines in the spectrum produced 

 by " deflagrating nitre," and the corresponding lines of the solar spectrum.f 



From these singular coincidences occurring in so many different cases, the 

 inference might be drawn, that all bright lines in the spectra of flames coincide 

 with dark lines in the solar spectrum ; and the extremely close proximity of the 

 lines 7 and b 2 , d A and F l? Z, and G indicated in Table IV., might at first sight seem 

 to confirm such an opinion. For it might be argued, that so close agreements in 

 the ascertained deviations indicate absolute identity; the minute differences 

 observed being attributed simply to errors in the observations. It will be seen, 

 however, that the observed deviations of the lines b 2 and 7, differ by no less a 

 quantity than 40", which is quite beyond the sum of the probable limits of error 

 in the observations for these lines, which I have ascertained to be only about 5" ; 

 and thus their coincidence is shown to be highly improbable. 



But any remaining doubt on the subject is completely removed, by the simul- 

 taneous observations of the spectra of sun light and olefiant gas, given in Table V., 

 where the micrometrical measurement of the interval between the lines b and 7 

 differs only by 11" from that obtained by the theodolite observations. In fact, 

 the bright line 7 was seen when the spectra were viewed simultaneously, to coin- 

 cide, not with the dark line 6 , but with the clear space immediately beyond it. 

 If we omit the line a, which, for reasons already fully stated, I do not regard as 

 properly belonging to the carbohydrogen spectrum, not one of the other twelve 

 lines which I have observed in that spectrum occurs near any conspicuous dark 

 line of the solar spectrum, with the exception of the lines 7, 8 3 and £, which fall 

 near b 2 , F t and G. Now, of these, 7 has been proved beyond doubt, not to coin- 

 cide with b 9 , but with a bright space in its vicinity ; and from the simultaneous 

 observation of the spectra of sun light and of olefiant gas, as well as from the 

 results of the theodolite observations, I believe that the other bright lines of the 

 carbohydrogen spectrum also coincide not with dark lines, but with bright spaces 

 in the solar spectrum. 



* Schumacher's Astronomisclie Abhandlungen, 1823, p. 29. See also Brewster's Edinburgh 

 Journal of Science, vol. viii., p. 7. M. Foucault has lately verified this result with the double 

 yellow line seen in the spectrum of the voltaic arc, between charcoal electrodes. See De La Rive's 

 Electricity, vol. ii., p. 322. 



f Report of British Association, 1842, p. 15. 



