Band and Line Spectra of the same Metallic Elements. 89 
I have recently returned to this line of investigation in order to ascertain 
the connexion between these bands and lines which undoubtedly belong to the 
same element, and are a characteristic feature of its spectra. It is necessary to 
eliminate the spectra of compounds, and to avoid confusion with hydrides and 
halides, which on the one hand might be formed in the oxyhydrogen flame, or, in 
the case of chlorides, left undecomposed. 
In the continuation of former experiments, made with the object of deter- 
mining the actual weights of salts necessary for the formation of bands and lines 
respectively in the same flame, results were obtained which had not been 
anticipated. First, the spectrum reaction proved to be much less sensitive in 
several instances than it is generally believed to be. Secondly, the lines were 
not found to gradually merge into bands, as at first appeared to be the case, for 
the band spectra appeared independently of the lines, rather as if they belonged 
to entirely different substances. The lines did not invariably appear before the 
bands: that is to say, with a smaller quantity of substance in the flame, sometimes 
they appeared simultaneously with them, and with some substances they appeared 
after the bands were extinct. In other words, the first evidence of the presence 
of a substance in the flame was the appearance of a banded spectrum, and then 
subsequently, when the quantity of a substance had been increased, particular 
lines, probably the fundamental vibrations of the element, began to appear. 
Thirdly, a background of continuous rays, observed in certain of the flame- 
spectra, have been proved to be traversed by a number of bands. 
The following is an account of observations on the quantitative relations 
between the substances examined and the spectra photographed or otherwise 
recorded, Experience showed that only pure substances should be used when 
it is possible to obtain them. 
Reference may be conveniently made to a communication recently submitted 
to the Royal Society on this subject. Part I. dealt with the banded spectra of 
alkali and alkaline earth metals, and the proofs based upon chemical grounds 
that these are the spectra of the metals. 
Part II. consisted chiefly of a discussion of the thermo-chemistry of flame 
reactions and the colouration of the flame. In these discussions it was necessary 
to give evidence of the temperatures employed, and these, it may be remarked, 
were measured by the melting point of pure platinum. The temperatures of 
the flames were shown to lie between 1775° and 2000°. 
There were other facts taken into account, such as the facility with which 
tin, lead, iron, nickel, and cobalt are vapourized, the two first copiously ; and the 
boiling of silver and gold, with their subsequent condensation cn objects placed 
near to or in the flame; also that weighed quantities of iridium and palladium 
were volatilized in the flame in the same manner, and their spectra photographed. 
