1897.] KEW YORK ACADEMY OF SCIENCES. 333 



work reductions and oxidations were continued to constant 



weight. 



B 

 182.34 

 174.63 

 181.87 

 181.79 

 174.87 

 177.87 

 176.79 



Another portion of the trioxide fractionated by a different 

 method (hydrogen chloride gas) gave with successive reductions 

 and oxidations of the same portion the following weights : 



Keductions Oxidations 



At. Wt. At. Wt. 



182.68 184.65 



177.25 185.26 



174.00 191.61 



A large number of reductions and oxidations were made using 

 porcelain, glass and platinum tubes, and porcelain and platinum 

 boats. After each reduction the tubes were examined and wher- 

 ever there was the slightest indication that a loss had occurred, 

 either mechanically or by volatilization, the results were re- 

 jected. As tungsten forms an alloy with platinum, any vol- 

 atilization which took place could not be detected in a plati- 

 num tube as in a porcelain or glass ; and for this reason a 

 certain amount of suspicion attached to the results obtained with 

 the platinum tube. A record, however, was kept of all losses in 

 both the accepted and rejected reductions, so that the amount in 

 excess of the theoretical loss was known for the total material 

 used. An accident to the tube enabled us to make an analysis 

 which showed that only a third of the actual loss could be ac- 

 counted for by the quantity of the tungsten allo^^ed with the tube. 



The atomic weights obtained varied from 154.064 to 191.61, 

 the total loss being ascribed to oxygen. The numbers are re- 

 ferred to 0= 16, but are not reduced to vacuno. 



Attempts were made to determine the water formed in reduc- 

 tion, but such discordant results were obtained that the method 

 was temporarily abandoned. 



A series of experiments* was made showing that in what- 

 ever way tungsten trioxide was prepared nitrogen was present, 



*Made in part at University College, London, in Dr. Ramsay's laboratory. 



