1894. J 060 [Pennington and Smith. 



enced much difficulty in obtaining pure tungsten trioxide. He found by 

 considerable investigation that Rose's method for the separation of molyb- 

 denum and tungsten was the most convenient. Traube {Jahr.fur Min- 

 eralogte, etc., Beilageband. 7, 233) and more recently Friedheim (Z. filr 

 anorg. Chemie, 1, 76) has shown that this procedure does not entirely 

 eliminate the molybdenum. Recent investigations, in this as well as in 

 other laboratories, upon artificial and supposedly pure, as well as natural 

 tungstates have proved that these all contain molybdenum in appreciable 

 amounts, and it may well be questioned whether the earlier determinations 

 of the atomic mass of tungsten were not affected by the presence of traces 

 of molybdenum. Its complete and absolute removal would tend to raise 

 the value of this constant of tungsten. An effort to this end appeared to 

 be desirable. 



An attempt was first made to free the trioxide from impurities through 

 the sodium salts. This course was abandoned inasmuch as traces of alkali 

 persistently adhered to the tungsten. This fact is not new ; Schneider was 

 fully aware of it, and in consequence he had recourse to another method, 

 which was also adopted in this investigation with some modifications. 



A quantity of tungsten trioxide, derived from wolframite, was heated 

 in a porcelain dish for three days with concentrated nitric acid. The lat- 

 ter was then decanted and the yellow oxide was well washed with water. 

 It was next subjected to the action of boiling aqua regia for the same 

 length of time, the acid solution, however, being removed from time to 

 time. The washings and acid liquor were tested for iron. When this 

 was no longer detected the washed trioxide was dissolved in yellow am- 

 monium sulpliide, air being excluded as much as possible. The solution 

 was filtered from insoluble material, and concentrated almost to the point 

 of crystallization, when hydrochloric acid was added. The liquid was 

 removed from the resulting precipitate, which was ignited, then boiled 

 with nitric acid, and later with aqua regia. The resulting trioxide no 

 longer showed the presence of iron and manganese. It was then washed, 

 brought into a porcelain dish, covered with distilled water and am- 

 monia gas conducted over the liquid until it was saturated. Several 

 days were required for this purpose. Only a very small residue remained 

 undissolved. It contained a trace of silica and a little greenish-yellow ox- 

 ide. Hydrogen sulphide gas was next introduced into the ammoniacal 

 solution, which was then digested at 80° C, for several hours. On the 

 addition of pure dilute hydrochloric acid to this solution tungsten trisul- 

 phide was precipitated. This wa5 filtered out, washed and roasted in a 

 large porcelain crucible with air access. The trioxide prepared in this 

 manner no longer contained silica, iron or manganese. If tin had been 

 originally present it would have gone out with the aqua regia, and any 

 columbic acid would have remained when the trioxide was subjected to 

 the ammonium sulphide treatment. The next step was to prove the pres- 

 ence or absence of molybdenum. A portion of the purified tungsten tri- 

 oxide was converted into the ammonium salt and the sulphocyanide test 



PROC. AMER. PHILOS. SOC. XXXIII. 146. 2 q,. PRINTED DEC. 29, 1891. 



