OP ARTS AND SCIENCES. 113 



loss of phosphoric oxide is, however, extremely small, and probably 

 never exceeds two or three hundredths of oue per cent. The tungstic 

 oxide was always determined by subtracting the phosphoric oxide as 

 directly determined from the sum of the two oxides. My various 

 attempts to determine the two oxides together by precipitation with 

 salts of lead or barium, after neutralization with an alkali, proved 

 failures in all cases. 



The precise quantitative separation and estimation of phosphoric 

 oxide in the phospho-tungstates is a matter of no small difficulty. 

 After many trials of various methods, the separation by means of 

 magnesia-mixture was found to give the best results, magnesic chloride 

 being employed. Only, in all cases in which this method is used, it is 

 necessary to redissolve the ammonio-magnesic phosphate first precipi- 

 tated, and to precipitate the salt a second time with ammonia. The 

 whole subject was specially investigated by Dr. F. A. Gooch, and 

 I shall do him and the process the fullest justice by referring to his 

 paper in Volume XV. of these Proceedings. Since the publication of 

 Dr. Gooch's paper, I have also used in many cases the following proc- 

 ess, which gives excellent results. The phosphoric oxide is to be 

 precipitated in the usual manner as ammonio-magnesic phosphate, and 

 well washed with magnesia-mixture and ammonia. The precipitate is 

 to be redissolved in chlorhydric acid, ammonia added in small excess, 

 and afterwards acetic acid, until the reaction is faintly but distinctly acid. 

 The phosphoric oxide is then to be precipitated from the last solution 

 by uranio-sodic acetate in excess. The precipitate must be allowed to 

 settle completely, and the clear supernatant liquid poured upon the 

 filter without disturbing the precipitate. A moderately strong solu- 

 tion of ammonic nitrate is then to be added, and the precipitate again 

 allowed to settle. After repeating this process two or three times, 

 the precipitate may be brought upon the filter and washed with solu- 

 tion of ammonic nitrate. In this manner there is no danger of obtain- 

 ing a turbid or milky filtrate. After drying, the precipitate is to be 

 ignited, moistened if necessary with nitric acid, and again ignited. 

 This process is somewhat longer than that by double precipitation and 

 estimation as magnesic pyrophosphate, but is, I think, rather more 

 accurate. In almost all the analyses, by either method, the filtration 

 and washing of the precipitate was effected by means of the asbestos 

 filters devised by Dr. Gooch* It is hardly possible to speak too 

 highly of this admirable Contrivance, which in a very large number of 



* Proceedings of American Academy, Vol. XIII. p. 342. 

 vol. xvi. (n. s. viii.J S 



