of Erbia and some other Earths. 305 



off at the sides, is different from that which distinguishes the spectra 

 of these metals when used as electrodes in the metallic state*. 



Roscoe and Clifton have investigated the different spectra presented 

 by calcium, strontium, and barium ; and they " suggest that, at the 

 lower temperature of the flame or weak spark, the spectrum observed 

 is produced by the glowing vapour of some compound, probably the 

 oxide, of the difficultly reducible metal ; whereas at the enormously 

 high temperature of the intense electric spark these compounds are 

 split up, and thus the true spectrum of the metal is obtained. In 

 none of the spectra of the more reducible alkaline metals (potassium, 

 sodium, lithium) can any deviation or disappearance of the maximum 

 of light be noticed on change of temperature "j\ 



As the experiments recorded in this paper show that the same 

 spectra are produced by the exposure of the oxides to the oxyhy- 

 drogen flame, Roscoe and Clifton's suggestion, that these spectra are 

 due to the volatilization of the compound of the metal with oxygen, is 

 doubtless correct. 



The similar character of the spectrum of the bright lines seen when 

 erbia is rendered incandescent would seem to suggest whether this 

 earth may not be volatile in a small degree, as is the case with lime, 

 magnesia, and some other earths. The peculiarity, however, of the 

 bright lines of erbia, observed by Bahr and Bunsen, that they could 

 not be seen in the flame beyond the limits of the solid erbia, deserves 

 attention. My own experiments to detect the lines in the Bunsen 

 gas-flame, even when a very thin wire was used, so as to allow the 

 erbia to attain nearly the heat of the flame, were unsuccessful. The 

 bright line in the green appears, indeed, to rise to a very small ex- 

 tent beyond the continuous spectrum, but I was unable to assure 

 myself whether this appearance might not be an effect of irradiation. 



It is perhaps worthy of remark that the chlorides of sodium, po- 

 tassium, lithium, csesium, and rubidium give spectra of defined lines 

 which are not altered in character by the introduction of a Leyden 

 jar, and which, in the case of sodium, potassium, and lithium, we 

 know to resemble the spectra obtained when electrodes of the metals 

 are used. Now all these metals belong to the monad group ; it 

 appeared therefore interesting to observe the behaviour of the other 

 metal belonging to this group. 



Chloride of silver when introduced into the Bunsen flame gave no 

 lines. The chloride was then mixed with alumina (which had been 

 found to give a continuous spectrum only) and exposed to the oxy- 

 hydrogen flame ; but no lines were visible. When, however, the 

 moistened chloride was placed on cotton and subjected to the in- 

 duction-spark without a jar, the true metallic spectrum was seen, as 

 when silver electrodes are used. 



* For the spectra of metallic strontium, barium, and calcium, see Phil. Trans. 

 1864, p. 148, and Plates I. and II. Both forms of the spectra of these substances 

 are represented by Thalen in his ' Spektralanalyse.' 



f Roscoe's Spectrum Analysis, p. 175, and Proc. Lit. & Phil. Soc. Manchester, 

 April 1, 1862. See also A. Mitscherlich, ' Ueber die Spectren der Verbindungen,' 

 S. 10. 



