ATOMIC WEIGHTS 365 



The mean of all is W = 184.105, ±.0337. This gives a percentage of 

 W in WO3 of 79.320, It .0185. The discordances were shown by Hardin 

 to be due partly to impurities in his material, such as nitrogen retained 

 by trioxide prepared from ammonium tungstate, and partly to volatility 

 of the oxide at high temperatures. In a later memoir ' he discusses 

 these errors at some length, and gives a few other determinations which 

 are even more discordant, and therefore not worth citation now. 



Taylor's thesis," representing work done in Smith's laboratory, is es- 

 sentially a study of errors. He found that constant weight could not 

 be secured during reduction experiments with the trioxide, and he also 

 found, like Hardin, that the oxidations generally gave the higher values 

 for the atomic weight of tungsten. Furthermore, he ascertained that 

 tungstic oxide derived from colloidal ammonium tungstate gave different 

 values dependent upon whether the latter compound was dialyzed or 

 undialyzed. Oxide from the dialyzed salt gave the highest atomic 

 weights. Some of the discrepancies were ultimately traced to the pres- 

 ence in the material studied, of a complex salt containing manganese and 

 iron, and the influence of these impurities was studied. Iron, and also 

 molybdenum, tend to lower the apparent atomic weight of timgsten; 

 manganese, and in much greater measure, raises it. The errors are in 

 opposite directions, but do not absolutely compensate one another. 



One new method for measuring the atomic weight of tungsten was 

 tested by Taylor, but the results were not satisfactory. Sodium car- 

 bonate was heated in a glass bulb with tungsten trioxide and water, the 

 latter was distilled off after effervescence had ceased, and the residue 

 was then heated to 300° in a vacuum. The weights of carbonate and 

 oxide being known, the loss in weight represented carbon dioxide. The 

 ratio between WO3 and CO, was thus determined. I cite the weights, 

 and also the values for the ratio WO3: COo" 100: x: 



Weight WO,. Weight CO.. Ratio. 



2.0802 .3952 18.998 



2.1937 .4173 19.023 



4.0818 .7762 19.016 



3.3629 .6394 19.013 



Mean, 19.0125, ± .0034 



Hence W = 183.45; a determination which Taylor regards as worth- 

 less, while admitting that the method is one of some promise. 



Several of the investigations so far described were carried out under the 

 direction of, or in cooperation with Professor Edgar F. Smith. The 



1 Joura. Amer. Chem. Soc, 21, 1017. 1S99. 



''Thesis, University of Pennsylvania, 1901. "Atomic weight of tungsten." 

 24 



