326 Journal of Agricultural Research voi. xii, No. 6 



Quadruplicate portions of these solutions were analyzed without further 

 treatment. Large aliquots from the same stock bottles were also passed 

 through new and old Pasteur-Chamberland filters, and some portions 

 were treated with G. Elf carbon black. These treatments were included 

 to check the possible effect of absorption of elements by the filter candles, 

 as well as the effect of the carbon black which was used to decolorize 

 soil extracts containing organic matter. 



Table III gives the results of the analyses of the untreated and of the 

 filtered solutions. It will be seen that the determinations of ammonia 

 and nitrate are satisfactory throughout. The methods used are essentially 

 those almost universally applied in sanitary water analysis (z) and with 

 the proper precautions are not believed to be open to criticism. The re- 

 sults for potassium, calcium, sulphate, and phosphate are seen to be 

 extremely inaccurate in the lowest concentration. Of these the calcium 

 is by far the most questionable. The error in the determination of this 

 element is so large that the results for calcium obtained in 191 5 have con- 

 sequently been discarded. Even with the potash and phosphate the 

 percentage of error is very large in some determinations. In the higher 

 range of concentrations the results are much closer throughout, and if it 

 were possible to confine all the work done to such solutions there would 

 perhaps be no objection to the methods employed. In many solutions 

 smaller concentrations of one or two elements are frequently encountered. 

 In such a case the error may be 50 per cent or more, either plus or minus. 

 Such a variation can readily obscure any significant change in concentra- 

 tion. It became necessary to obtain a more reliable procedure for potas- 

 sium, calcium, and phosphate, all of which are extremely important 

 elements in any study of soluble plant nutrients. 



