the Rubidium Compounds. 47 



strontium. In order to obtain the average composition of this 

 somewhat unequally mixed mass, 550 grms. of it were dissolved 

 in 1975 grms. of water, and this solution analysed as follows : — 



(a) 7*1057 grms. of the liquid yielded 1*4065 grm. of mixed 

 anhydrous chlorides. Hence the original saline residue con- 

 tained in 100 parts — 



Water 10*92 



Chlorine compounds . . . 89*08 



100*00 



(b) The amounts of chloride of potassium and chloride of ru- 

 bidium can be found by precipitating the double chlorides of 

 these metals and platinum, and by weighing the platinum which 

 the double salts yield on reduction in a current of hydrogen. 

 This method is, however, open to the objection that commercial 

 platinum, unless purified by a series of long and tedious pro- 

 cesses, is quite unfit for any accurate quantitative estimation, as 

 it invariably contains impurities which cause its atomic weight 

 to vary several per cent. It is therefore much easier and sim- 

 pler to wash out the mixed chlorides of potassium and rubidium 

 from the reduced mass, then to weigh them, and to determine 

 their yield of chlorine as chloride of silver. 



If we call the weight of the chloride of potassium x, and the 

 weight of the chloride of rubidium y, A being the sum of the 

 weights of the mixed chlorides, and B the weight of the chloride 

 of silver obtained, we have 



Ag + C1 -, and Ag + C1 -3- 

 K + Cl ~^ ana Rb + Cl ' 



6A-B 



sc= -= , ana y = A--#, 



b — a 9 ■ 



*= 1-3601 B- 1-6143 A (1) 



4*0017 grms. of the solution, corresponding to 0*7921 grm. 

 of the dry chlorides, gave 0*4723 grm. chlorides of rubidium 

 and potassium = A ; and these yielded 0*7787 = B grm. of chlo- 

 ride of silver. 89*08 parts of the above chlorides therefore con- 

 tain 33*37 parts chloride of potassium, and 19*75 parts chloride 

 of rubidium. This latter salt was found, on examination with 

 the spectrum apparatus, to contain a little cassium. 



It may not be superfluous to mention that the formula (1) is 

 not applicable in the case of very slight differences in the values 

 of x and y, or for very small values of A. Hence it is advisable 

 to calculate the probable error of x and y for each of the values 



hence 



or 



