ATOMIC WEIGHTS 



363 



tion data are as follows, with the usual percentage column added. The 

 weights are reduced to a vacuum : 



Mean, 79.392, ± .0004 



Hence W = 184.92. 



The very high value for tungsten found by Pennington and Smith, 

 nearly a unit higher than that which was commonly accepted, seems to 

 have at once attracted the attention of Schneider.' who criticized the 

 paper somewhat fully, and gave some new determinations of his own. 

 The tungsten trioxide employed in this new investigation was heated in 

 gaseous hydrochloric acid, and the absence of molybdenum was proved. 

 The data obtained, both by reduction and by oxidation, are as follows: 



Reduction Series. 



2.0738 grin. WO3 gave 1.6450 W. 79.323 per cent. 



4.0853 " 3.2400 " 79.309 



6.1547 " 4.8811 " 79.307 



Oxidation Series. 



1.5253 grm. W gave 1.9232 WO3. 79.311 per cent. 



3.1938 " 4.0273 " 79.304 



4.7468 " 5.9848 " 79.314 



Mean of all, 79.311, ± .0018 



Hence W= 184.007. 



In order to account for the difference between this result and that of 

 Pennington and Smith, an impurity of molybdenum trioxide amounting 

 to about one per cent, would be necessary. Schneider suggests that the 

 quantities of material used by Pennington and Smith were too small, and 

 that there may have been mechanical loss of small particles during the 

 long heatings. Such losses would tend to raise the atomic weight com- 

 puted from the experiments. On the other hand, the losses could hardly 

 have been uniform in extent, and the extremely low probable error of 

 Pennington and Smith's series renders Schneider's supposition improb- 

 able. The error, if error exists, must be accounted for otherwise. 



iJourn. prakt. Chem. (2), 53, 28S. 1896. 



