Determination of Rubidium by the Spectroscope. 399 



sium, it is possible to make the comparison between the 

 rubidium lines trustworthy. It was shown in the former 

 paper that in matching solutions of potassium by means of the 

 red line there is no difficulty ; but the convenience of being 

 able to use the spectroscope without readjustment throughout 

 the entire experiment made it desirable to see whether the 

 blue line of potassium might not serve sufficiently well in the 

 comparison. It is hardly necessary to reproduce bere in detail 

 the evidence bearing upon this point, but we found as the 

 result that the potassium may be determined by the use of the 

 blue line with an error amounting to 10 per cent or 20 per 

 cent, which, though far greater than that inherent in the use 

 of the red line, admits of the attainment of determinations 

 which should be accurate enough for the present purpose. 

 We proceeded, therefore, to make a determination of rubidium 

 in presence of potassium by the process referred to, the details 

 of which are given in Experiment TV. 



Experiment IV. 



c,, -. -, , , . ... ( 0*0005 mo-. Rubidium ) , ., 



btandard solution containing •< nn ~ 1c « -d . • Y to the 



& I 0*0015 " Potassium J 



coil-full. 



Test solution contained 8 mg. rubidium and no potassium. 



Stepl. Step 2. Step 3. Step 4. Step 5. 



I Preliminary 

 Preliminary test matching of Ki Rematching of 

 for Rb. line. Rb line. 



Readjustment F i n a 1 match- 

 of K line. ing of Rb line. 



Test at 20 cm 3 I Test at 20 cm 3 Test at 35 cm 3 .Test at 35 cm 3 Test at 35 cm 3 

 gave Rb line gave K line like gave Rb lines 'gave K line like gave Rb line 

 like standard, standard when, like standard, [standard when like standard. 

 1 mg. of K had j2 mg. werej 



been added. I present. 



Rubidium found 35X50X0*0005 = 0*875 mg. 



" taken =0*8 " 



Error ... — 0*075 " = 9*4 per cent. 



It is evident that the percentage error is considerable, but 

 inasmuch as the application of the process would naturally be 

 to the determination of small absolute amounts of rubidium 

 we thought it desirable to go a step further to see whether 

 means are at hand for separating large amounts of potassium 

 from small amounts of rubidium with an approach sufficiently 

 near to completeness to bring the potassium present within 

 the limits allowed by the spectroscopic method. After some 

 experimentation we fixed upon the simplest possible procedure, 

 viz : the solution of the salts in the least possible amount of 

 water, precipitation of the mass of potassium chloride by addi- 

 tion of alcohol, filtration, and the evaporation of the filtrate. 



Am. Jour. Sci.— Third Series, Vol. XLIV, No. 263.— November, 1892. 

 27 



