68 PROCEEDINGS OF THE AMERICAN ACADEMY 



ments (7) to (10), in each of which 20 cm. 3 of the precipitant were 

 used, while theory requires ahout 1.1 cm. 3 , the error upon 0.0265 gr. 

 of phosphoric oxide varies from -(--0.0004 gr. to -[-0.0021 gr., or from 

 -{-1.51 to -(-7.92 per cent.; in experiment (11) in which 6 cm. ! of 

 the precipitant were used, while about 2.1 cm. 3 were theoretically 

 required, the error is -j-0.0007 gr. upon 0.0512 gr. of phosphoric oxide, 

 or — |— 1 .37 per cent. ; and in experiments (13) and (15) in which 12 

 cm. 3 of the precipitant were used, while about 2.1 cm. 3 were theoret- 

 ically required, the error upon 0.0512 gr. of phosphoric oxide was 

 -f 0.0017 gr., or +3.31 per cent. 



Experiments (21) to (2G) inclusive show that when the process was 

 varied so as to precipitate the phosphate solution, either containing no 

 free ammonia or a very small quantity, by a magnesic sulphate mixture 

 from which the ammonia had been expelled, the error of the result 

 varied from -{-0.0061 gr. to — |— 0.01 1 7 gr. upon 0.1875 gr. of phos- 

 phoric oxide, or from —(—3.25 to -{-6.24 per cent. ; and from -(-0.0077 gr. 

 to -{-0.0106 gr. upon 0.2142 gr. of phosphoric oxide, or from -^j— 3.59 to 

 -f-4.94 per cent., — the greatest errors appearing naturally in the 

 experiments made with hot solutions containing no free ammonia, in 

 which cases the hydio-magnesic phosphate first thrown down was 

 partially changed into tri-magnesic phosphate by the action of 

 the heat, and therefore not converted into ammonio-magnesic phos- 

 phate upon the subsequent addition of ammonia. 



Experiments (12), (14), (16), and (17) to (20) inclusive indicate 

 that tolerably good results may be obtained by draining the first pre- 

 cipitate, dissolving in hydrochloric acid and reprecipitating with am- 

 monia ; the first three experiments showing incidentally that the 

 addition of small amounts of citric acid does not prevent precipitation 

 to any very marked degree in presence of a considerable excess — 

 in this case about 10 cm. 3 to 0.5 gr. of citric acid — of magnesic 

 sulphate mixture. Thus in experiments (12), (14), and (16) the error 

 was 0; and in experiments (17) to (20) the error varied from 

 -{-0.0006 gr. to -{-0.0015 gr. on 0.1863 gr. of phosphoric oxide, or 

 from -{-0.32 to -{-0.80 per cent., the mean error being about -4-0.0006 

 gr., or -{-0.32 per cent. 



It appears from these experiments that the method of precipitating 

 by magnesic sulphate mixture without the solution and second pre- 

 cipitation ought to give an accurate determination of phosphoric acid, 

 providing only that the amount of the precipitant used were just 

 sufficient to complete the precipitation. But this condition renders 

 the process practically useless in ordinary cases ; and inasmuch as in 



