OF ARTS AND SCIENCES. 67 



in distinct excess. To portions (11 6) to (120) inclusive were added 

 20 cm. 3 of magnesic chloride mixture, and to portions (121) and (122) 

 20 cm. 8 of magnesic sulphate mixture. The first precipitates of (1 1 9), 

 (120), (121), and (122) were filtered off on paper, dissolved in hydro- 

 chloric acid and thrown down again with ammonia, collected on 

 asbestus (as the first precipitates of the other portions were), washed 

 and ignited as usual. Portions (123) and (124) were precipitated 

 with 1.5 cm. 3 of magnesic chloride mixture, 1.5 cm. 3 being added 

 after the precipitate had settled, together with ammonia in strong 

 excess. Portions (125) to (129) inclusive were precipitated with 

 20 cm. 3 of magnesic chloride mixture from which the ammonia had 

 been expelled, and ammonia was added in strong excess after the 

 settling of the precipitate. 



XIII 



XVI 



Mg,PjO r . P,0 5 found. P,0 3 required. Error. Percent. 



f (116) 0.0441 gr. 0.0282 gr. 0.0270 gr. +00012 gr. -4-4.44 



XII. J 17 ) 0-0452 gr. 0.0289 gr. 0.0270 gr. +0.0019 gr. +7.04 



((118) 0.0458 gr. 0.0293 gr. 0.0270 gr. +0.0023 gr. +8.52 



(119) 0.0423 gr. 0.0271 gr. 0.0270 gr. +0.0001 gr. +0.37 



(120) 0.0425 gr. 0.0272 gr. 0.0270 gr. +0.0002 gr. +0.74 



(121) 0.0423 gr. 0.0271 gr. 0.0270 gr. +0.0001 gr. +0.37 



(122) 0.0427 gr. 0.0273 gr. 0.0270 gr. +0.0003 gr. +1.11 



(123) 0.0423 gr. 0.0270 gr. 0.0269 gr. +0.0001 gr. +0.37 



(124) 0.0424 gr. 0.0271 gr. 0.0269 gr. +0.0002 gr. +0.74 

 1 (125) 0.0424 gr. 0.0271 gr. 0.0269 gr. +0.0002 gr. +0.74 

 [ (126) 0.0425 gr. 0.0272 gr. 0.0269 gr. +0.0003 gr. +1.11 



f (127) 0.3168 gr. 0.2026 gr. 0.2017 gr. +0.0009 gr. +0.44 



IX. \ (128) 0.3173 gr. 0.2030 gr. 0.2017 gr. +0.0013 gr. +0.64 



[ (129) 0.3180 gr. 0.2033 gr. 0.2015 gr. +0.0016 gr. +0.79 



Experiments (1) to (10) inclusive, (11), (13), and (15) demonstrate 

 that the method of estimating the phosphoric oxide of alkaline phos- 

 phates by precipitation with magnesic sulphate mixture and washing 

 until the chlorine reaction vanishes from the filtrate gives results far 

 too high when the amount of the precipitant used is in any consider- 

 able excess of the amount theoretically required. Thus, in experi- 

 ments (1) to (6), in each of which 20 cm. 3 of magnesic sulphate 

 mixture were used, the amount required by theory being about 5.2 

 cm. 3 , the error varies from +0.0033 gr. to +0.0060 gr. on 0.1298 gr. 

 of phosphoric oxide, or from +2.54 to +3.39 per cent. ; in experi- 



