332 Professors Kirchhoff and Bunsen on Chemical 



this portion yielded 1*6070 grm. chloride of silver. By a repeti- 

 tion of this process on the salt C, a product D was prepared, of 

 which 1*9486 grm. gave 2*3091 grms. chloride of silver. The 

 quantities of chloride of silver obtained from one part by weight 

 of chloride of rubidium after each of these purifications, are 

 therefore,— A 1*2308 



B 1*1873 



C 1*1873 



D 1*1850 



These numbers prove that the products of the three last prepa- 

 rations possess a constant composition. The bright spectrum- 

 lines of caesium and lithium were almost invisible in the last of 

 these preparations; and the lineKa of potassium could not be seen 

 at all in our spectrum-apparatus ; so that we may fairly conclude 

 that the productof thelast preparation is pure chloride of rubidium. 

 In order to obtain a still further proof of the purity of the 

 chloride thus prepared, a method was employed of which we 

 shall again have to speak when discussing the mode of separa- 

 tion of csesium from potassium and rubidium. This consists 

 in treating the three caustic alkalies with carbonic acid until 

 about one-fifth by weight of the whole mass is converted into 

 carbonates, and then extracting the anhydrous salt with alco- 

 hol. If an alkali possessing greater or less basic properties 

 than rubidium, and having a different atomic weight, were 

 present together with this metal, the alcoholic solution must 

 possess a composition differing from that of the residue ; the 

 portion of caustic alkali which dissolved in alcohol yielded, 

 however, a chloride possessing a composition identical with that 

 obtained from the portion of alkali undissolved by the alcohol, 

 0*5116 grrn. of the former yielding 0*6078 grm. chloride of silver, 

 or 1 part of chloride giving 1*1830 of chloride of silver, closely 

 corresponding with the previous results. If we only consider the 

 precipitation of the products possessing constant composition, 

 and if we take, according to Stas, the atomic weight of silver to 

 be 107*94, and that of chlorine 35*46, we obtain the following 

 numbers for the atomic weight of rubidium on the hydrogen 

 scale:— B 85 . 31 



C 85*32 



D 85*55 



E 85*24 



or a mean of Rb = 85*36. The weight of the atom of the new 

 metal is therefore more than twice as great as that of potassium. 

 Although the numbers thus obtained do not coincide with the 

 degree of accuracy which may be desirable in determinations of 

 atomic weights, we believe that the mean experimental number 



