SOIL ANALYSIS AND INVESTIGATION 13 



tabulated above indicate that the analyses were made with an 

 exactness that justifies a statement to three decimal places, that 

 is, to three significant figures; and in fact, as was shown, such 

 is necessary if the figures are to have any significance regarding 

 fertilizer applications. It is obvious that the analysis of a finely 

 pulverized definite mineral or rock is less subject to error than 

 a sample of soil sifted through a 2 mm. mesh. Yet the U. S. 

 Geological Survey commonly reports its analytical data to only 

 hundredths of a per cent., that is, to two decimal places. What 

 variation may be expected in duplicate determinations by the 

 same analysts it is difficult to say, for such duplicates are not 

 commonly published. 1 In spite of the widespread view that the 

 chemical analysis of a soil is a statement of great accuracy, it 

 is improbable that as usually determined the potash content is 

 correct to three or even two significant figures; it is also doubt- 

 ful if the phosphoric acid content is correct to even one signifi- 

 cant figure, if the total amount is below o.i per cent, of the soil. 

 That these determinations have a higher accuracy than here 

 stated is not shown by an inspection of the literature including 

 the fairly numerous results reported in the annual Proceedings 

 of the Association of Official Agricultural Chemists. 



It was early felt by some investigators that soil analyses were 

 unsatisfactory for studying the relation of the soil to the food 

 requirements of a crop, and a second method was devised, name- 

 ly, the growing of a crop, and determining the amount of 

 mineral constituents removed from the soil by analyzing the ash 

 of the crop. From the point of view of practical soil manage- 

 ment this procedure involves the serious difficulty of being first 

 obliged to get the crop before determining what must be done to 

 best get it. It apparently has the scientific advantage of direct- 

 ness in determining the mineral needs of the plant from the 

 plant itself. If these needs were constant, the advantage would 



1 See : On the interpretation of mineral analyses, by S. L. Penfield, 

 Amer. Jour. Sci., (4), 10, 33 (1900) ; The analysis of silicate and car- 

 bonate rocks, by W. F. Hillebrand, Bull. No. 305, U. S. Geol. Surv., 1907 ; 

 Manual of the chemical analysis of rocks, by H. S. Washington, 1904, 

 p. 24; Ueber Genauigkeit von Gesteinanalysen, von M. Dittrich, Neues 

 Jahrbuch fur Mineralogie und Palaeontologie, 2, 69 (1903). 



