William J. Gies 193 



the figures given by Smith and Hawk, in Table i (page 187). Com- 

 menting on these results, Smith and Hawk State : " In the case of 

 Det'ns 38, 41 and 42, the amounts of ferric phosphate alone, calculated 

 f rom the permanganate titration values, are greater than the respective 

 amounts of ferric phosphate plus aluminum phosphate, as obtained 

 gravimetrically. In Det'n 37 the total weight of ferric phosphate and 

 aluminum phosphate is but very little (1.3 mg.) larger than the cal- 

 culated weight of the ferric phosphate alone. These findings indicate 

 a failure of complete precipitation of either the iron or the aluminum, 

 or both. The exact reason for this failure was not determined." 



These anomalous results of Smith and Hawk are in accord with 

 some of our findings. Thus (page 170) in our Table 4 (ox) the weight 

 of the FeP04, calculated from the " iron by titration " (463.9), exceeds 

 the weight of the sum of the AIPO^ and FePO^ found, giving a nega- 

 tive value for the weight of AIPO4. These results, considered in con- 

 junction with the values given in Tables 2 and 3 (pages 168-9), point 

 not to a failure to secure complete precipitation of the iron and alumi- 

 ntim or both, but rather to a decomposition of the ferric phosphate 

 either in washing or in igniting, or in both, as the most probable cause 

 of these irregularities. 



After receiving the foregoing comment by Curtman and Gross, 

 I requested Dr. Howe to give me, for publication here, the " details 

 as to the manner in which he standardized his iron sol." His reply 

 is appended. 



Addendum by Paul E. Howe. Approximately 2 gm. of iron 

 (piano wire) were dissolved in hydrochloric acid and this sol. diluted 

 to 2000 cc. with dist. water. Portions of this sol., approximately 10 

 gm., were weighed out and the iron determined according to the Zim- 

 merman-Rheinhardt method. The average of five determinations of 

 the iron sol. was 0.931 mg. for each gm. of sol. ; the greatest deviation 

 from the mean was o.oi mg. 



The facts in the foregoing discussion do not require revision of 

 the general conclusion, stated in the opening paragraph of this 

 paper, that the Schmidt-Hoagland process is apparently somewhat 

 more accurate, and more convenient of execution, than any other 

 of the available methods for the quantitative determination of 



