ATOMIC WEIGHTS 367 



1. Berzelius 79.7745, ± .0880 



2. Schneider, 18-50 79.320, ±: .0068 



3. Marchand 79.3205, ± .0073 



4. Borch ■ 79.293, ± .0108 



5. Dumas 79.312, ± .0090 



6. Bernoulli 79.480, ± .0560 



7. Persoz 79.314, ± .0070 



8. Roscoe 79.264, ± .0146 



9. Waddell 79.339, ± .0069 



10. Pennington and Smith 79.392, ± .0004 



11. Schneider, 1896 79.311, ± .0018 



12. Shinn 79.391, ± .0066 



13. Hardin 79.320, ± .0185 



14. Smith and Exner 79.3169, ± .0007 



General mean 79.3706, ± .00034 



In this combination only two values carry much weight; the tenth 

 and the fourteenth. The series by Pennington and Smith is evidently 

 much overvalued, and exerts an undue influence upon the general mean. 

 In reality the series by Smith and Exner is by far the most trustworthy 

 of all, and the figures given by Schneider, Marchand, Dumas and Persoz 

 are in harmony with it. The other series are more doubtful. The weighted 

 mean of twelve series, omitting ISTos. 10 and 14, is 79.31G0; a value almost 

 identical with that of Smith and Exner. The latter, therefore, is 

 abundantly confirmed. 



In 1861 Scheibler ' deduced the atomic weight of tungsten from 

 analyses of barium metatungstate, BaO.4WO3.9H2O. In four experi- 

 ments he estimated the barium as sulphate, getting closely concordant 

 results, which were, however, very far too low. These, therefore, are re- 

 jected. But from the percentage of water in the salt a better result was 

 attained. The percentages of water are as follows : 



13.053 

 13.054 

 13.045 

 13.010 

 13.022 



Mean, 13.0368, ± .0060 



Hence W = 184.05. 



The work of Zettnow," published in 1867, was more complicated than 

 any of the foregoing researches. He prepared the tungstates of silver 

 and of iron, and from their composition determined the atomic weight 

 of tungsten. 



1 Journ. prakt. Chem. , 83, 324. 



2 Poggend. Annalen, 130, 30. 



