73S Spectrum Analysis. [December, 
gases, the constancy and distinctness of the speCtra of the 
chemical elements, when made incandescent by electricity, 
had pretty clearly resulted, and the expediency of their 
adoption as a chemical test had thus become manifest. A 
powerful impetus in this direction — or, rather, to the use of 
the optical phenomena of substances when rendered incan- 
descent or self-luminous by whatsoever means, in order to 
their chemical discrimination — was provided by some re- 
searches on flames by Mr. Swan, first published in 1856. 
By this observer “ absolute identity was shown to exist be- 
tween the speCtra of dissimilar carbo-hydrogen compounds,” 
thus proving (not, however, in a manner altogether irre- 
fragable) the dependence of the constitution of the speCtra 
solely on the chemical composition of the compounds; and, 
adverting in particular to the colouration of flames by soda, 
he stated that it formed such a delicate test for sodium that 
1-2, 500, oooth grain could be detected by it. Under the cir- 
cumstances which we have attempted to describe — and 
among which we should include also the publication of some 
valuable researches on coloured flames by Prof. Miller, of 
King's College, in 1845 — Profs. Kirchhoff and Bunsen were 
scarcely justified in proposing, in i860, the optical charac- 
ters of glowing gases as the basis of “an entirely new 
method of qualitative chemical analysis.” Nor, on the 
other hand, did they place the trustworthiness of the pro- 
posed method, whether new or not, beyond a doubt. MM. 
Kirchhoff and Bunsen, in point of faCt, had investigated the 
speCtra of a number of binary and other compounds of the 
six metallic elements — sodium, lithium, potassium, stron- 
tium, calcium, and barium, as volatilised in flames ; and 
they found that these speCtra were distinct from one another, 
and always the same, whatever the temperature of the flame, 
at least between the limits of from 2000° to 8ooo° C. They 
asserted, further, that the speCtrum of any one of those 
metals was independent of the nature of the other elements 
with which it formed a binary compound or salt, except that 
“ those compounds of a metal gave the brightest speCtra 
which were most volatile ;” and that, in faCt, the speCtrum 
of any such compound was the same as that of the uncom- 
bined metal itself, volatilised by the eleCtric spark. All 
this, however, was yet very far from demonstrating, as 
alleged, that these speCtra might be relied on “ as absolute 
proofs of the presence of the particular metal ” in each 
special case. In the first place, it was strange, as asserted, 
that the speCtrum of any of the metals combined with 
oxygen or chlorine, for instance, should be altogether 
