EXPERIMENT STATION BUiLLETINS. 629 



2. That the amount dissolved decreases with the alkaliuity of tlie 

 solution. 



3. That, roughly n-2o% of the precipitated tricalciuni i)hosphate is 

 rated as available. 



4. That the amount of this soluble in citrate solution also decreases 

 with the alkalinity of the citrate solution. 



5. Only about "^2-4% of the original rock phosphate is soluble in 

 citrate solution. 



P 

 G. Decreasing the ratio —r '' I and adding calcium sulphate in- 



citrate 



creases the amount of phosphoric acid removed. 



The results with solutions B and C as well as the two series G-I and 

 J-L whicli differ only in the kind of tricalciuni phosphate used verify 

 Gladdings statement that, judged by its solubility in ammonium citrate 

 solutions, precipitated tricalciuni phosphate is much more available than 

 the natural rock. Hence if the action of ammonium citrate is a criterion 

 of the action of natural forces we are warranted in accepting a certain 

 increased amount of precipitated tricalciuni phosphate as available. 



It is perfectly apparent from the data presented above as well as that 

 appearing in reports by others that ammonium citrate solution is under 

 no circumstances a dififereiitial solvent for any two forms of calcium 

 phosphate. From results given in Table VI an approximate determina- 

 tion of dicalcium phosphate in the presence of rock phosphate can be 

 obtained with an acid citrate solution, but since at least 10% of the 

 phosphoric acid of tlie rock is removed the results cannot be considered 

 as accurate beyond that figure. The question at issue is then reduced 

 to — What conditions give results most nearly approaching those ob- 

 tained in soil? This question cannot be answerecl except by means of 

 vegetation experiments. It is evident that insofar as the composition 

 and reaction of the solutions used are concerned the difference in the 

 results obtained are relatively small, generally not exceeding 1%. 

 This, of course, is vrell within the range of experimental error of vege- 

 tation experiments and it may be safely assumed that any solution 

 selected which approximates in composition those usually considered, 

 will give results Avhich will agiee as wxll as any others with figures 

 obtained by vegetation tests. In other words, assuming the validity of 

 the camparison between the solvent powers of approximately neutral 

 amm.onium citrate solution and the soil solution, any citrate solution 

 having a composition within the limits of those used in the experiment 

 reported above, may be selected as the reagent for evaluating com- 

 mercial i^hosphates. 



While any of these solutions could, from the point of view of the 

 practical value of the phosphate, be used, it is necessary for purposes of 

 comparing the analj'tical results obtained by different chemists and 

 accurately fixing the market value of various goods that one solution 

 be chosen and consistently used. Its selection however cannot be made 

 on the basis of the practical value of the results obtained with it but 

 must depend upon such considerations as the accuracy and convenience 

 with wdiich it can be prepared, constancy of comi)osition, etc. 



In the original description of the metliod and in official procedures 

 based upon it, a neutral solution is prescribed. Owing to the un- 



