1904] SMITH AND EXNER — ATOMIC WEIGHT OF TUNGSTEN. 1-17 



tion of it was weighed out for the eighteenth experiment, the 

 remainder being heated for a day more in hydrogen. After stand- 

 ing over night a second portion was removed and used in experi- 

 ment 19. The remainder was exposed all of the third day to the 

 action of hydrogen, and was then oxidized for experiment 20. 

 Had not the first reduction been complete, the results would not 

 have agreed so well. 



The mean atomic value from the hexachloride is 184.04, that 

 from the oxidation of metal 184.065, or the average of the two 

 independent series is 184.OS, which probably approximates the 

 truth very closely and may be safely regarded as the atomic weight 

 of tungsten. 



Summary. 



Our study, extended over so long a period, has revealed — 

 I. That it is quite doubtful whether any chemists who in the past 

 occupied themselves with a determination of the atomic weight 

 of tungsten have worked with pure substance. Tungstic acid is 

 prone to form *' complexes." It was found that if the acid contain 

 no iron, for instance, but be digested with acids, e.g., hydro- 

 chloric or nitric acid, in which iron is present, the latter will enter 

 the tungstic acid. Iron and manganese are eliminated from the 

 acid with the greatest difficulty. In the earlier work there is no 

 evidence of their removal. Neither do we discover that vanadium 

 and phosphorus had been considered as present, yet in purify- 

 ing ammonium paratungstate by recrystallization alone it was found 

 that the tenth recrystallization showed vanadium. 



2. The slimy, greenish or bluish-white masses believed to be 

 " paratungstates " because of their great insolubility are probably 

 " complexes." 



3. The fourth method of purification can be relied upon to yield 

 pure tungstic acid. 



4. The use of pure sodium carbonate (29^) to dissolve tungsten 

 trioxide srives an excellent means of ascertainina: when the iron, 

 manganese and silica are fully removed, but that its development 

 into a method for the determination of the atomic weight of tung- 

 sten is not at all probable. 



5- The plan of digesting pure ammonium paratungstate with 

 nitric acid, then evaporating to complete dryness and gently ignit- 

 ing affords pure oxide. 



