424 MR WILLIAM SWAN ON THE PRISMATIC SPECTRA OF THE 



tions equal weight, and to take simply the arithmetical mean of the whole. The 

 mean results of the two series, and the number of observations in each, being 

 tabulated separately, the reader will be able to form some judgment regarding 

 the probable accuracy of the final determinations obtained from the tAvo series 

 combined. In Fig. 1, which is a graphic construction of the observations in 

 Tables II. and III., the lines were drawn by the engraver through points laid 

 down by me on the copper to a scale, — adopted to suit the size of the plate, — of 

 one inch to 2200". I have ascertained the errors in the positions of these lines 

 to amount, in one case only to 01 inch (corresponding to 22"), and to be generally 

 much less; so that the spectra are represented in the figure with tolerable 

 fidelity. 



In addition to the observations of the carbohydrogen and solar spectra con- 

 tained in Table II., where the deviation for er.ch line of either spectrum was 

 separately determined by the theodolite or micrometer, I have also made simul- 

 taneous observations of the spectra of sun light and of olefiant gas. The gas, which 

 was prepared by heating alcohol with sulphuric acid, was conducted through 

 wash bottles containing caustic potash and sulphuric acid, to a gas holder ; from 

 which it afterwards passed, through a tube filled with pieces of quicklime, to a 

 platinum jet where it was burned. 



The lines in the spectrum of olefiant gas are very distinct, being well seen 

 without using the blowpipe ; but like the lines in the other carbohydrogen spectra, 

 they are not sufficiently luminous to be seen when projected on the solar spectrum, 

 unless the latter is made so faint, that its lines have disappeared. I succeeded, 

 however, in observing the spectra simultaneously, by intercepting the sun light 

 which fell upon one half of a narrow slit, and illuminating the whole slit with 

 the flame of olefiant gas. The gas spectrum then appeared immediately over that 

 of the sun, and the brighter lines in it were well seen, especially when the flame 

 was urged by the blowpipe. The intervals between the lines of the gas spectrum 

 and the nearest lines of the solar spectrum, given in Table V., were measured by 

 the micrometer, with a magnifying power of 21 ; and the observations for the 

 brighter line /?, 7, and 8, agree well with those of Table III. 



The line a was rarely visible in the spectrum of olefiant gas, and its appear- 

 ance was only momentary, which confirms the opinion already stated, that it does 

 not properly belong to the carbohydrogen spectra. To the proof already adduced 

 in support of this opinion, I may here also add, that I have found it permanently 

 absent in the flames of carbonic oxide, and of light carburetted hydrogen.* The 

 continued invisibility of so brilliant a line of the spectrum, coupled with its 



* I have found that the column of heated air rising from the flame of a spirit lamp with a salted 

 wick, is most energetic in communicating yellow light to the exterior envelope of the flame of the 

 Bunsen lamp. This effect is apparently confined to the outer, or oxidizing portion of the flame, 

 where there is no excess of hydrogen, to decompose the chloride of sodium ; and the experiment is 

 interesting, as tending to prove that the yellow light may be caused by simple incandescence, without 

 the actual combustion of sodium. 



