2 On Double Spectm. [January, 



contains sodium, we obviously assume that no other sub- 

 stance but sodium ever gives a line of that precise refrangi- 

 bility. Or, again, if knowing that under circumstances similar 

 to those under which sodium vapour produces a dark line 

 in the yellow, potassium vapour gives a dark line in the red, 

 we conclude, from the absence of this line in the solar spec- 

 trum, that potassium does not exist in the sun's atmosphere, 

 we tread on still more dangerous ground ; since, in order 

 that our conclusion may be legitimate, it is necessary that 

 potassium vapour under no conditions that may exist in the 

 sun should ever give any other spectrum than that containing 

 this dark line. Or, to take still another example; if we 

 find from our experiments that incandescent solids produce 

 continuous spectra, while incandescent gases produce dis- 

 continuous spectra, and therefore conclude that the nucleus 

 of the sun is an incandescent solid, our conclusion becomes 

 no longer tenable when it is shown that, under high pressure, 

 gases also give off continuous spectra. 



It is proposed in this paper to describe some of the excep- 

 tions to first-enunciated broad principles which, though 

 unsuspected at first, have been shown to exist since spectrum 

 analysis has been known as a separate method of research. 



Professors Bunsen and Kirchhoff, in their first memoir on 

 spectrum analysis, describe the spectra of the metals of the 

 alkalies and alkaline earths. They endeavour to establish 

 firstly, that the spectrum is the same whatever compound 

 of the metal is employed. If a bead of sodium chloride 

 be brought into the Bunsen flame the spectrum will con- 

 sist of two yellow lines only, and will remain the same 

 if the sodium chloride be exchanged for sodium iodide, 

 or sodium sulphate, or sodium carbonate. A second con- 

 clusion arrived at in the same memoir is that the position 

 of the bright lines is independent of the temperature to 

 which the vapour of the substance is heated. Professors 

 Bunsen and Kirchhoff found differences of intensity only 

 when they employed, instead of the ordinary Bunsen flame, 

 the flames of sulphur, carbon disulphide, carbonic oxide, 

 hydrogen, or the oxyhydrogen flame. They also compared 

 the flame-spectra of sodium, lithium, potassium, strontium, 

 and calcium with those obtained when the spark from an 

 induction coil was taken between wires formed of the 

 respective metals, and convinced themselves that the bright 

 lines of the flame spectra were present in the same position, 

 although other lines were seen which they supposed to be 

 due to foreign metals present in the electrodes, and to the 

 nitrogen of the air. Professor Kirchhoff expresses his opinion 



