14 THE DETERMINATION OF LTTHTT^M. 



other clilorids was from a solution containing only ab(jut 0.0001 gram 

 of lithium chlorid per cc, it was not to be expected that much lithium 

 would be found in the insoluble jjortion. It seems probable that 

 Waller might have extracted all the lithium in liis experiments by 

 two more extractions. Although his figures for total lithium sul- 

 phate agree well, it is possible that the figures obtained by Carnot's 

 method are liigh on account of the correction, which assumes that the 

 filtrate and wash water were saturated with litliium fluorid, which 

 may not always be the case. 



In discussing the difficulties attending the complete separation of 

 the litliium, sodium, and potassium from the other substances, 

 Waller called attention to the necessity for careful cleaning of the 

 platinum mres used in testing, as after holding a wire in the flame 

 until the lithium hne is no longer observed in the spectroscope, the 

 line may reappear when the wire is moistened ^\ath hydrocliloric acid 

 and again brought into the flame. He noted that the barium car- 

 bonate and barium sulphate precipitates may carry down perceptible 

 amounts of Htliium. To lessen the error from this source he advises 

 the precipitation of the sulphate in dilute solution followed by thor- 

 ough washmg to remove the lithium and recommended the solution 

 and reprecipitation of the carbonate. He made no quantitative 

 experiments on the subject, but considered that his experience con- 

 firmed Mayer's remark that under the influence of heat in the presence 

 of w^ater, the htliium chlorid had a tendency to exchange clilorin for 

 oxygen. In liis tests of the phosphate method he found Hthium even 

 in the first filtrate upon taking up the soluble material from the 

 precipitate. 



Ranzoli ^ in a comparison of different methods of separation of 

 lithium obtained the following errors: By Mayer's ^ phosphate 

 method when working with quantities of from 0.2452 to 0.5200 gram, 

 the losses were from 0.0097 gram to 0.0176 gram, equivalent to an 

 average of 3.05 per cent. By Carnot's ^ method he obtained 0.4780 

 gram and 0.4797 gram instead of 0.5200, showing losses of 0.0470 and 

 0.0403 gram, or equivalent to 7.9 per cent. By Gooch's * method 

 he obtained 0.4907 and 0.4969 gram instead of 0.5200, losses of 0.0293 

 and 0.0231 gram, or equivalent to 5 per cent. He said that all the 

 gravimetric determinations were carried out wdth all the precautions 

 recommended by the authors. Of course in the analysis of mineral 

 waters one rarely has to deal wdth quantities of lithium sulphate as 

 great as 0.5 gram, and the quantity of mixed chlorids from a natural 

 water which would contain such an amount of litliium would be 

 entirely too large to handle as a matter of quantitative analysis. 

 Nevertheless, from experience with the Gooch method it is the opin- 



1 Gaz. chim. ital., 1901, 31 (I): 40-48. » Bull. soc. cMm., Par., 1880, (3) 1: 280. 



* Ann. Chem. Pharm., 1856, 98: 19.3. * Amer. Chem. J., 1887, 9: 33-51. 



