Thus the former takes a predominating importance when the gas is rendered luminous by electrical discharges, and chemical transformations, especially, play a preponderant role in the emission of the spectrum of flames which contain a saline vapour.

In all the ordinary experiments of spectrum analysis the laws of Kirchhoff cannot therefore be considered as established, and yet the relation between emission and absorption is generally tolerably well verified.

No doubt we are here in presence of a kind of resonance phenomenon, the gaseous atoms entering into vibration when solicited by the ether by a motion identical with the one they are capable of communicating to it.
If we are not yet very far advanced in the study of the mechanism of the production of the spectrum,[47] we are, on the other hand, well acquainted with its constitution.

The extreme confusion which the spectra of the lines of the gases seemed to present is now, in great part at least, cleared up.

Balmer gave some time since, in the case of the hydrogen spectrum, an empirical formula which enabled the rays discovered later by an eminent astronomer, M.Deslandres, to be represented; but since then, both in the cases of line and band spectra, the labours of Professor Rydberg, of M.Deslandres, of Professors Kayzer and Runge, and of M.Thiele, have enabled us to comprehend, in their smallest details, the laws of the distribution of lines and bands.
[Footnote 47: Many theories as to the cause of the lines and bands of the spectrum have been put forward since this was written, among which that of Professor Stark (for which see _Physikalische Zeitschrift_ for 1906, passim) is perhaps the most advanced.