364 SMITHSONIAN MISCELLANEOUS COLLECTIONS VOL. 54 



Soon after Schneider's paper appeared, another set of determinations 

 by Shinn ' was published from Smith's laboratory. Attempts to verify 

 the results obtained by Smith and Desi having proved abortive, and other 

 experiments having failed, Shinn resorted to the oxidation method and 

 gives the subjoined data. The percentage column is added by myself: 



.22297 grm. W gave .28090 WO,. 79.377 



.17200 " .216G4 " 79.394 



.10989 " .13844 " 79.377 



.10005 " .12598 " 79.417 



Mean, 79.391, ± .0066 



Hence W = 184.908. 



This figure is very close to that found in Pennington and Smith's series. 



The great discordance between the determinations so far cited, led 

 Hardin* to a very careful investigation of tungsten trioxide. The sub- 

 stance was prepared from various sources, and manipulated by various 

 methods; and although concordant results were sometimes obtained in 

 succession, the discordance between different series of experiments was 

 very great. Hardin therefore concluded that a discussion of his figures, 

 with reference to the atomic weight of tungsten, would be useless. Never- 

 theless, partly for the sake of completeness, and partly because this 

 calculation is in great measure a study of the compensation of errors, I 

 prefer to cite Hardin's determinations, in order that they may be com- 

 pared with others. For this purpose I give his sixty-four determinations 

 as one series. The letters o and r indicate oxidation and reduction ex- 

 periments, respectively. The atomic weights found were as follows : 



^Thesis, University of Pennsj-lvania, 1896. "The atomic mass of tungsten." 

 * Joum. Amer. Chem. Soc, 19, 657. 1897. 



