64 PROCEEDINGS OF THE AMERICAN ACADEMY 



state, when the loss of weight corresponds to the sum of the water 

 and ammonia. 



As in the case of the phospho-tungstates, the quantitative determi- 

 nation of phosphoric oxide is a matter of considerable difficulty. The 

 method of separation by means of magnesia mixture has been carefully 

 studied by Dr. Gooeh, to whose paper I have already referred.* Dr. 

 Gooch found it necessary to precipitate the ammonio-magnesic phos- 

 phate a second time, a single precipitation giving an error amounting 

 sometimes to 6 or 7% of the phosphoric acid present. After re-solu- 

 tion and pi'ecipitation by ammonia, the mean error amounted to only 

 0.65^, which makes an almost insensible correction when, the quan- 

 tity of phosphoric oxide is small. In a few instances I have applied 

 this correction after a double precipitation, but I prefer to employ the 

 following method, which gives an almost perfect separation from 

 molybdic teroxide. The phosphoric oxide is first precipitated from a 

 hot solution as ammonio-magnesic phosphate, the supernatant liquid 

 after complete subsidence carefully decanted upon an asbestos filter, 

 the precipitate washed with magnesia mixture and ammonia, then 

 redissolved in the least possible quantity of hot dilute chlorhydric acid 

 and reprecipitated with ammonia. After complete subsidence and 

 decantation, the precipitate is boiled with successive portions of a 

 solution of ammonic sulphide. A more or less dark orange-red so- 

 lution of ammonic sulpho-molybdate is always obtained at first, but 

 after two or three repetitions of the process the ammonic sulphide 

 added remains colorless on heating. The ammonio-magnesian phos- 

 phate is then filtered upon the asbestos filter already employed. In 

 place of tliis method I have sometimes employed the following modifi- 

 cation, which gives, I think, equally good results. After the first 

 precipitation the phosphate is to be redissolved, and the hot solution 

 precipitated at once by ammonic sulphide in excess. The precipitated 

 phosphate is then to be boiled two or three times with ammonic 

 sulphide as above. Whatever inaccuracy is inherent in this method 

 depends, in my judgment, upon the fact that, as Dr. Gooch has 

 shown, the determination of phosphoric acid by means of magnesia 

 is, under the most favorable cu-cumstances, a less accurate process 

 than has been supposed. 



The determination of ammonia and the alkalies was effected by the 

 methods already described in the case of the phospho-tungstates. 

 Water must be estimated by ignition with sodic tungstate, as there is 



* Proceedings of American Academy, Vol. XV. p. 63. 



