60 ' Spectra of Metallic Compounds, [January, 



in the mixture, but that of the compounds which had been 

 formed from their decomposition and subsequent re-combina- 

 tion, according to the strength of their affinities at elevated 

 temperatures. For example, a mixture of baryta and calcic 

 chloride gave not only the spectra proper to these two com- 

 pounds, but that of barytic chloride as well. The careful 

 study of these changes would remove a source of embarrass- 

 ment in spectroscopic analysis. Moreover, it would probably 

 furnish some information upon the chemistry of compounds, 

 which we are wont to study in the solid state or in solution, 

 when converted into vapours, and upon the phenomena of 

 disassociation. 



We already possess a very laborious and extensive series 

 of determinations of the spectra of compounds by Prof. A. 

 Mitscherlich (Pogg. Ann., No. 3, 1864, and Phil. Mag. [4] , 

 September, 1864). He found that compounds of the first 

 order, in so far as they are volatile and remain unde- 

 composed when adequately heated, always exhibit spectra 

 which differ completely from those of the metals. He 

 obtained the spectra in a variety of ways. 1st. By 

 evaporating solutions in a narrow flame of coal-gas or 

 hydrogen. 2nd. By bringing the substances into the flame 

 of an oxygen-coal-gas burner. 3rd. By bringing them into a 

 hydrogen-chlorine burner. 4th. Evaporating bromine and 

 iodine in hydrogen, and volatilising the substance in the 

 flame produced by the burning of this mixture in air or 

 oxygen. 5th. By passing the gas to be examined either 

 alone, or, in case it is not combustible, along with carbonic 

 oxide or hydrogen, through the middle aperture of an oxy- 

 hydrogen burner, and burning the mixture in air or hydrogen. 

 6th. By volatilising the substance in a current of hydrogen, 

 and igniting the jet thus charged with the substance for 

 examination. 7th. By passing the electric spark between 

 electrodes of the metals or of their salts when surrounded by 

 an atmosphere of various gases. 8th. By using solutions 

 of metallic salts as electrodes, and passing the spark from 

 liquid to liquid. 



It will be seen, by a comparison of these methods, that 

 they differ greatly with regard to the temperature at which 

 the spectra are formed. It is much lower when, as in the 

 first method, the solutions of the salts are volatilised than 

 when the fused salts themselves are used. In the latter 

 case, the spectra are much more brilliant and persistent, 

 and the lines are more numerous. 



The third method was improved by Diacon, in that he 

 surrounded his hydrogen-chlorine burner with a hood in such 



