ELISHA MITCHELL SCIENTIFIC SOCIETY. 49 



50 taken out. (= ^ grm.), precip. with ammonium molybdate. Kept 

 in bath 3 hours at 50° or 60^ then allowed to stand two or three 

 hours in the cold, and filtered— washed with one-half strength am- 

 monium molybdate. Dissolved in ammonia precipitated with mag- 

 nesia mixture and allowed to stand at least five hours in the cold. 

 Filtered and washed thoroughly with ammonia (1.9), dried, ignited 

 and weighed. 



The Atlanta Method was followed as was laid down by the Con- 

 vention at Atlanta and Philadelphia, with the exception of the treat- 

 ment with Magnesium Nitrate for the estimation of total P0O5, which 

 was omitted, being regarded as unnecessary, owing to the absence 

 of organic matter in the samples. The solution vas made by strong 

 HCl and digestion, and the method from that point pursued. In 

 the estimation of the Citrate Insoluble P2O5, it was found impossi- 

 ble to prevent the finer particles of the phosphate from being car 

 ried mechanically through the filter by the washing with water after 

 treatment with Ammonium Citrate solution. This was noticed in 

 every case, and it is very probable that this affected the results of 

 this particular determination. Subsequent experience with the 

 method has found this to hi the case more or less in every instance 

 where an Acid Phosphate is treated, being absent in the Ammoni- 

 ated Fertilizers. 



In both methods the greatest care was observed and the similar 

 determinations were made, not together but separately, as it was 

 thought that this would test the methods and the comparison be- 

 tween the two better than by making the determinations together, 

 for in the latter case if any error had been made in one it would 

 have been followed in the rest, and the treatment in whatever way 

 it was carried out would have made the results identical in all. For 

 this reason each duplicate determination was made at different 

 times, and each determination therefore represented the method 

 per se, and the comparison thus better carried out. 



The Result of the comparison is given in the table following. The 

 figures in each case represent the average of two and sometimes 

 three closely concordant duplicate determinations. W represents 

 the Washington Method, and A the Atlanta : 



