TUNGSTEN. 257 



The data furnished b}^ Bernoulli* differ widely from those just given. 



This chemist undoubtedly worked with impure material, the trioxide 



having a greenish tinge. Hence the results are too high. These are the 



l)ercentages of W : 



Rediiction Series. 



79.556 

 79.526 



79-553 

 79.558 

 79-549 

 78.736 



Mean, 79.413 



Oxidation Series. 



79-558 

 79.656 



79-555 

 79-554 



Mean, 79.581 

 Mean of all, 79.4S0, ± .056 



Two reduction experiments by Persozf give the following results 



1.7999 S'™- WO3 gave 1.4274 grm. W. 79.304 per cent. 



2.249 " 1.784 " 79-324 " 



Mean, 79.314, zb .007 



Next in order is the work done by Roscoe. % This chemist used a 

 porcelain boat and tube, and made six weighings, after successive reduc- 

 tions and oxidations, with the same sample of 7.884 grammes of trioxide. 

 These weighings give me the following five percentages, which, for the 

 sake of uniformit}^ with foregoing series, I have classified under the 

 usual, separate headings : 



Reduction Series. 

 79.T96 

 79.285 

 79.308 



.0146 



* Poggend. Annalen, in, 573. i860. 

 •1 Zeit. Anal. Chem., 3, 260. 1864. 

 I Ann. Chem. Pharm., 162,368. 1872. 



