294 Harrtey—The Action of Heat on the Absorption Spectra and 
from the spectra of other saline solutions, such as those of cobalt, nickel, and 
copper, the following deduction was drawn :— 
The absorption spectra of different salts of the same metal, whether solid or in 
solution, are not identical, even when the spectrum is a marked characteristic of the 
metal. 
From the examination of organic substances, whether salts of bases or other- 
wise, it was concluded that— 
Molecules of compounds—that is to say, molecules composed of dissimilar atoms— 
vibrate as wholes or units, and the fundamental vibrations give rise to secondary vibrations, 
which stand in no visible relation to the chemical constituents of the molecule, whether 
these be atoms or smaller molecules.* 
This conclusion is generally applicable also to metallic salts of inorganic 
origin, as well as to organic compounds. 
When anhydrous compounds are examined the same effect of heat is observed, 
but the only substance experimented upon was potassium permanganate, the 
absorption spectrum of which was first described by Vierordt. As there are 
seven bands in the yellow, green, and blue, of diminishing intensity, and not 
very sharply defined, it was necessary to examine solutions of different strengths 
so as to bring into prominence some of the weaker bands. Of course, the 
stronger bands increase in width, and even become confused as the weaker gain 
in intensity when stronger solutions are used ; hence the measurements are not all 
comparable with those of Lecoq de Boisbaudran, who used one solution of fixed 
strength, containing 0:1 gr. of KMnO, per litre. 
Potassium Permanganate, KMnO,.—A freshly prepared solution. Thin wedge- 
cell. 
ABSORPTION BANDS. ABSORPTION BANDS. 
At 16°. At 100°. 
r IN 
470 to 478 471 to 481 
485 — 495 488 - 501 
505 - 5138 505 — 519 
529 — 534 Confused and not measurable. 
The mean of the numbers at 16° are 474, 490, 508, and 531°5. The figures 
of Lecoq de Boisbaudran, representing the maximum intensity of absorption, 
and for apparently the same bands, are 469°4, 486:1, 504:5, and 524°6. That 
they are not more nearly in agreement with the above is due to the fact that the 
maximum of intensity is not in a position intermediate between the two edges of 
a band, but lies a little to one side or the other. This particular solution shows 
* «Researches on the Relation between the Molecular Structure of Carbon Compounds and their 
Absorption Spectra.” Part VII.—Trans. Chem, Soc., vol. 47, p. 685, 1885. 
