A number of methods have been invented to demonstrate the reversibility of spectra; among these methods we will cite two which are very easily carried out. In Bunsen's method sodium chloride is put into an apparatus for evolving hydrogen (the spray of the salt is then carried off by the hydrogen and colours the flame with the yellow sodium colour), and the hydrogen is ignited in two burners—in one large one with a wide flame giving a bright yellow sodium light, and in another with a small fine orifice whose flame is pale: this flame will throw a dark patch on the large bright flame. In Ladoffsky's method the front tube (p. 561) is unscrewed from a spectroscope directed towards the light of a lamp (a continuous spectrum), and the flame of a spirit lamp coloured by a small quantity of NaCl is placed between the tube and the prism; a black band corresponding to sodium will then be seen on looking through the ocular tube. This experiment is always successful if only there be the requisite relation between the strength of light of the two lamps.

The absorptive capacity is the relation between the intensity of the light (of a given wave-length) falling upon and retained by a substance. Bunsen and Roscoe showed by direct experiment that this ratio is a constant quantity for every substance. If A stand for this ratio for a given substance at a given temperature—for instance, for a flame coloured by sodium—and E be the intensity of the light of the same wave-length emitted at the same temperature by the same substance, then Kirchhoff's law, the explanation and deduction of which must be looked for in text-books of physics, states that the fraction A/E is a constant quantity depending on the nature of a substance (as A depends on it) and determined by the temperature and wave-length.

Heated metals begin to emit light (only visible in the dark) at about 420° (varying with the metal). On further heating, solids first emit red, then yellow, and lastly white light. Compressed or heavy gases (see Chapter III., Note ²²⁷), when strongly heated, also emit white light. Heated liquids (for example, molten steel or platinum) also give a white compound light. This is readily understood. In a dense mass of matter the collisions of the molecules and atoms are so frequent that waves of only a few definite lengths cannot appear; the reverse is possible in rarefied gases or vapours.

Brewster, as is mentioned above, first distinguished the atmospheric, cosmical Fraunhofer lines from the solar lines. Janssen showed that the spectrum of the atmosphere contains lines which depend on the absorption produced by aqueous vapour. Egoreff, Olszewski, Janssen, and Liveing and Dewar showed by a series of experiments that the oxygen of the atmosphere gives rise to certain lines of the solar spectrum, especially the line A. Liveing and Dewar took a layer of 165 c.m. of oxygen compressed under a pressure of 85 atmospheres, and determined its absorption spectrum, and found that, besides the Fraunhofer lines A and B, it contained the following groups: 630–622, 581–568, 535, 480–475. The same lines were found for liquid oxygen.

If the material of the whole heavenly space formed the absorbent medium, the spectra of the stars would be the same as the solar spectrum; but Huyghens, Lockyer, and others showed not only that this is the case for only a few stars, but that the majority of stars give spectra of a different character with dark and bright lines and bands.