TUNGSTEN. 143 



Closely following Dumas' method, he reduced molj^bdenum 

 trioxide to metal, finding in it 66.708 per cent, of the latter. 

 This figure comes within the limits of variation of Dumas' 

 experiments, and therefore gives them additional confirma- 

 tion. Its introduction into the general mean, however 

 would exert too little influence upon the latter to justify the 

 labor of recalculation. 



TUNGSTEN. 



The atomic weight of tungsten has been determined from 

 analyses of the trioxide, the hexchloride, and the tungstates 

 of iron, silver, and barium. 



The composition of the trioxide has been the subject of 

 many investigations. Malaguti * reduced this substance to 

 the blue oxide, and from the difference between the weights 

 of the two compounds obtained a result now known to be 

 considerably too high. In general, however, the method of 

 investigation has been to reduce WO 3 to W in a stream of 

 hydrogen at a white heat, and afterwards to reoxidize the 

 metal, thus getting from one sajnplc of material two results 

 for the percentage of tungsten. This method is unquestion- 

 ably accurate, provided that the trioxide used be pure. 



The first experiments which we need consider are, as 

 usual, those of Berzelius.f 899 parts WO 3 gave, on reduc- 

 tion, 716 of metal. 676 of metal, reoxidized, gave 846 WO3. 

 Hence these percentages of W in WO 3 : 



79.644, by reduction. 

 79.905, by oxidation. 



Mean, 79.7745, it .0880 



These figures are far too high, the error being undoubt- 

 edly due to the presence of alkaline impurity in the trioxide 

 employed. 



* Journ. fiir Prakt. Chem., 8, 179. 1836. 

 j Poggend. Annal., 8, i. 1826. 



