BUNSEN AND KIRCHHOFF 327 



flame allowed of, he was able to lay the foundation of 

 spectrum analysis. 



It was now possible to evaporate on a pure platinum wire 

 any substances whatever in the non-luminous gas flame, 

 and so cause them to emit light, and it was now a promising 

 step to continue to purify the substances used by every means 

 known to chemical science, until each of them developed its 

 peculiar flame colour fully and unmixed, a task which Bunsen 

 undertook with admirable persistence. In the course of this 

 work, however, the flame colouration was not observed only 

 with the naked eye, but also, following the methods of 

 Newton and Fraunhofer, in a state of spectral decomposition 

 by means of a prism. It then appeared, that the yellow 

 colouration of the flame, so often observed, and having a 

 yellow line already well-known to Fraunhofer, belongs ex- 

 clusively to sodium, and that the uncertainty hitherto felt 

 with regard to the origin of this yellow light was on the one 

 hand to be referred to the incredible minuteness of the 

 quantity necessary to produce this colouration quite visibly, 

 or to exhibit the spectral line, and on the other hand, to the 

 universal distribution of sodium in the form at least of traces 

 as an impurity in practically all substances found on the 

 surface of the earth. But we now had a means of recogni- 

 tion of the constituents of all substances which could be ob- 

 tained in flames, or in electric sparks, or arcs, or otherwise 

 in the state of glowing gases. This means allowed the certain 

 recognition of the smallest possible quantities, while being 

 easily and simply applicable; spectrum analysis by emission 

 was thus founded. 



This led Bunsen himself very quickly - five and six years 

 after the introduction of his burner - to the discovery of two 

 new elements, rubidium and caesium, which had hitherto 

 been unrecognised on account of the fact that they only 

 occur in traces, as a rule, and are very similar to the well- 

 known alkali metals. This spectrum analysis was just as 



