530 Prof. It. Bunsen's Spectral- Analytical Researches. 



The carbon points of the spark-apparatus, when dry, or when 

 moistened with hydrochloric acid, gave no spectrum, provided 

 they had been thoroughly cleaned by the method already 

 described ; this was shown by the fact that the sparks passing 

 between them in an atmosphere of hydrogen showed only the 

 less characteristic lines of the latter. 



The sparks which passed in the air during the spectral obser- 

 vations showed therefore only the air-lines of oxygen, nitrogen, 

 and hydrogen. 



This air-spectrum is given in the first line of each of the 

 spectral Tables (Plates III.-V.), so that no confusion may be 

 made between it and the lines which show the presence of 

 the individual elements. 



In the Plates / indicates the flame-spectrum, e the electric- 

 spark spectrum, a the absorption -spectrum. 



It Spectra of the Elements of the Alkali Group. 



The pure chlorides, whose spectra are shown in Plates III. and 

 IV., nos. 1 to 6, were prepared as follows : — 



Sodium chloride, prepared from hydrochloric acid and sodium 

 carbonate, was purified by washing with concentrated hydro- 

 chloric acid and repeated recrystallization. 



Pure potassium chloride was prepared by strongly heating 

 potassium chlorate which had been six times recrystallized. 



The material with which I determined the atomic weights of 

 caesium and rubidium, and which was prepared in a state of great 

 purity by a method described elsewhere, was used as the source 

 of the salts of these metals. 



Thallium chloride, precipitated by hydrochloric acid from the 

 sulphate, was purified by repeated crystallization. 



The exceedingly pure preparation of lithium with which the 

 atomic weight of that metal was determined by Mr. Diehl in my 

 laboratory, served for the production of the lithium-spectrum. 



The whole of these chlorides were easily and completely 

 volatilized when supported on platinum wire in the non-luminous 

 gas-flame. Test quantities of them, prepared by fractional 

 crystallization, showed not the smallest trace of impurities when 

 tested in the spectrum-apparatus. A glance at Plate III. is suffi- 

 cient to convince one that all of the members of the alkali group 

 may be recognized by their flame-spectra much more easily than 

 by their spark-spectra. Not a trace of the potassium-chloride lines 

 appears in the spark- spectrum ; those of rubidium and csesium 

 chloride are very faint; while, on the other hand, the flame- 

 spectra of these substances are produced with the greatest clear- 

 ness and beauty. 



The elements of the alkali group are therefore always to be 

 recognized by means of their flame-spectra. 



