- 2 



of Rubidium by the Spectroscope. 393 



volume to which such an unknown solution was brought in 

 the process of equalizing its spectrum with that of the stan- 

 dard its total contents in potassium was determined. In this 

 manner the strength of solutions of pure potassium chloride 

 proved determinable with accuracy, but when the effect of 

 intermixing pure sodium chloride with the potassium salt was 

 studied it transpired that the brilliance of the potassium spec- 

 trum was markedly increased by the presence of the sodium 

 in the flame, the maximum increase, which amounted to 

 twenty per cent, appearing when the ratio of sodium chloride 

 to potassium reached 100 : 1. It was found, however, that the 

 practical difficulty of determining potassium in presence of 

 sodium could be largely overcome by taking the precaution to 

 bring the test solution and the standard to apparent equality 

 as regards the potassium line, then to bring the solutions to an 

 equality in respect to the sodium line by addition of sodium 

 chloride to the standard, and finally to readjust the volumes of 

 test and standard until the potassium lines were again equal. 

 The accuracy of the determination of the potassium is, how- 

 ever, in spite of the precaution, somewhat affected by the 

 presence of sodium, — the error, though, sometimes falling as 

 low as y 1 ^- mg. in 10 mg., rising occasionally to 1 mg. in 15 mg. 

 With the excellent gravimetric method which we possess for 

 the determination of potassium recourse would, naturally, 

 never be taken to the spectroscopic method except in cases 

 when the determination of small absolute amounts are con- 

 cerned, but in such cases the spectroscopic method may prove 

 a convenience. 



In the work upon potassium the observations of the red line 

 were made in the ordinary laboratory in diffused light, but 

 preliminary experimentation upon the rubidium spectrum 

 immediately developed the fact that the blue lines are better 

 to work by in the case of this element, and that a dark room 

 becomes a necessity. For the experiments to be described 

 pure rubidium chloride was prepared by many fractional pre- 

 cipitations by alcohol out of aqueous solutions, and in settling 

 the question as to the coils which should be used the choice 

 fell upon the size holding - 02 grm. of water and made of the 

 No. 28 wire, the superior stiffness of these and consequent 

 constancy in capacity giving them the advantage over smaller 

 coils of finer wire, though the latter are capable of bringing 

 out greater sensitiveness of the reaction. We found, for 

 example, that under the most favorable conditions as to height 

 of flame and width of slit, 0-0002 mg. of rubidium chloride 

 produced the blue lines at the last limit of visibility when the 

 larger and heavier coil was in the flame ; with a coil holding 

 0*006 grm. of water and made of very fine wire the more 



