MR SWAN ON THE CONSTITUTION OF FLAME. 27 
flame. There was, however, an inconvenience to be anticipated. Most of the 
hydro-carbons which burn brightly, such as turpentine, are so rich in carbon that 
they produce a very smoky flame; and when they are employed to give light, as in 
the common camphine and paraffine oil lamps, special arrangements are needed 
for consuming their smoke by directing a powerful current of air against the flame. 
It was not convenient to adopt any of these arrangements; and it occurred to me 
that I should obtain a smokeless, but sufficiently brilliant flame by burning a 
mixture of two liquids, the one rich, and the other poor, in carbon, such as tur- 
pentine (C,H,) and alcohol (C,H,O,). The liquids I selected were common coal 
naphtha and wood spirit; and I found that a mixture of about three volumes of 
commercial wood spirit to one volume of coal naphtha produces a flame neither 
less brilliant, nor more smoky, than that of an ordinary wax taper.* 
I may remark, that, for the purpose I had in view, there seems no objection to 
the use of such a compound liquid as I have described; for I have shown else- 
where that all the hydro-carbons in burning produce similar spectra,—the only 
difference being, that the spectrum produced by the incandescent solid carbon, or 
by the envelope of the flame, is brighter in some cases than in others. 
The following tables contain the observations made on the flame of a mixture 
of coal naphtha and wood spirit :— 
OBSERVATIONS OF THE FLAME OF CoaL NAPHTHA AND Woop SPIRIT.t 
First Set of Observations. 
Line B. Line 0. 






130 138 
135 II. 135 
134 136 
138 140 
135 V. 139 
136 133 
Means 134-07 136°83 



Mean observed breadth of spectrum=134°72. 
* This flame might be used for household purposes, and probably also with advantage in experi- 
ments where an invariable source of light is wanted. The liquids, although both smell disagreeably, 
emit no perceptible odour in burning. 
} Each number in this and the following table is a single micrometer reading. 
¢ The observations in this and the following table, on the various lines of the spectrum, were 
made in the order of the numbers I, II.....VI., in order to eliminate the effect of any gradual 
change in the dimensions of the flame. 
