L. J. Curtman and P. Gross 



169 



Chlorides. The results obtained indicate that some other error was 

 responsible for the loss, in addition to the possibiHty of hydrolysis. 

 Careful quahtative tests for both iron and aluminum, in the filtrates 

 after each precipitation, were negative. 



The effect of ignition as a possiblc source of error was next con- 

 sidered. The known stabiHty of AIPO4, even at very high temp., 

 caused attention to be turned to the effect of ignition on FeP04. 

 Nothing could be found in the literature on this subject. The ex- 

 periments in Table 3 were made with sol. of pure FeClg. 



TABLE 3 



Data pertaining to the instahility of FePOi at high temperatures 



The procedure was identical with that employed for the mix- 

 tures (Table 2). The results of these experiments clearly show 

 that the observed losses were due to the instability of the FePO^. 



Fusion of the ignited precipitates with NagCOa, followed by sol. 

 of the melt in HCl and titration of the iron after reduction, showed 

 (Det'ns 2 and 3, Table 3) that all the iron was in the precipitate. 

 Phosphoric acid must therefore be lost, either by hydrolysis or on 

 ignition, or in both processes. 



The results of the above determinations (Tables 2 and 3) re- 

 quire the rejection of the method of Steel on purely theoretical 

 grounds. 



It was thought necessary, however, before passing final judgment 

 on the inaccuracy of Steel's method, to run a number of determina- 

 tions with blood to which definite amounts of aluminum had been 

 added, in order to discover if any compensating errors occurred in 

 the entire process which might cause a correct result to be obtained. 

 The method followed in oxidizing the blood was that given by 

 Steel.''' The blanks were of necessity run by the method of Schmidt 



T Steel : Lac. cit. 



