112 RESEARCHES UPON ATOMIC WEIGHTS. 



HISTORICAL DISCUSSION. 



In discussing the bearing of this investigation upon the results of earlier work 

 by other chemists, the experiments of Millon upon silver and potassium iodates ^ 

 and of Eerzelius ^ and Dumas,^ who converted silver iodide into silver chloride, 

 may be disregarded, since at the time the analyses were made, quantitative 

 analysis was in its infancy. Marignac's ^ value for the atomic weight of iodine, 

 126.79, obtained from titration of weighed amounts of silver with potassic 

 iodide, and from syntheses of silver iodide from weighed quantities of silver, 

 may be accounted for by the supposition that the iodine used in the experiments 

 was not pure. To explain Stas's ^ low value, 126.79, is a difficult matter. His 

 iodine was purified by two different methods, i. e., by once precipitating or dis- 

 tiUing the iodine from a strong solution of potassic iodide, and by precipitation 

 of nitrogen iodide. A third sample was purified by both methods. Since the 

 material purified by each of the two methods gave identical results with that 

 purified by both methods, it is inconceivable that either method of purification 

 should not have been effective. Impurity in the silver or loss of silver iodide are 

 improbable causes of the discrepancy, for the weight of the silver iodide produced 

 was equal to the sum of the weights of the silver and iodine employed. Rich- 

 ards and Wells have recently shown, however, that Stas was not infallible, and 

 in fact was capable of m.aking serious mistakes, such that his value for the atomic 

 weight of sodium was 0.2 per cent too high, and that for the atomic weight of 

 chlorine 0.03 per cent too low, so that it is not at all surprising to find that here 

 also his work was faulty in some undiscovered particular. 



Ladenburg's result," when calculated from the true value of the atomic weight 

 of chlorine, becomes 126.988, which agrees very closely with the value deduced in 

 this paper. His determinations were affected by several small errors, so that the 

 close agreement is somewhat the result of chance. In the first place, porcelain cru- 

 cibles, as has been pointed out before, gain in weight when used for the conversion 

 of silver iodide into silver chloride, so that the weight of the silver chloride is 

 somewhat uncertain. Furthermore Ladenburg did not fuse the silver iodide 

 before weighing it. Although Stas states that silver iodide may be completely 

 dried without fusion, his experiments show a loss in weight on fusion of 0.002 

 per cent, while the average loss on fusion as given on page 97 is about 0.004 per 

 cent. Moreover, Ladenburg's method of purifying silver iodide, by washing the 

 precipitated salt with ammonia, could hardly be expected to remove last traces 

 of silver chloride and silver cyanide "included" by the precipitate. These 

 errors are all so small that they would not affect the second place of decimals. 



Shortly after the publication of the first portion of this research there ap- 

 peared the complete paper of Kothner and Aeuer upon the same subject.^ These 



^ Ann. dc Chem. et de Phys. (3) 9, 400 (1843). ^ Ibid., (2) 40, 430 (1829). 

 ' Ann. Chem. Pharm., 113, 28 (i860). * Eerzelius' Lchrbuch, 5th ed., 3, 1196. 



^ (Eiivres Completes, i, 548. ^ Ber. d. d. Chem. GeselL, 35, 2275 (igo2). 



' Liebig's Ann., 337, 123 (1904). 



