ON THE PROBABLE ERROR OF SAMPLING IN 

 SOIL SURVEYS. 



By G. W. ROBINSON, B.A., 

 Adinser in Agricultural Chemistry, 



AND \V. E. LLOYD, B.Sc, 

 Research Scholar, University College of North Wales, Bangor. 



The following paper records experiments made to determine the 

 magnitude of the error involved in the sampling of a soil for survey 

 purposes. While it is not held that the results obtained are true for all 

 localities, they serve to indicate the order of magnitude of the various 

 errors. 



The ordinary method of sampling a soil for survey purposes is, 

 first to select a field uniform in itself and representative of the soil type 

 which is being examined. Borings are then taken in various parts of 

 the field. The borings from the top soil are united to form the soil 

 sample. Similarly those from the subsoil are united to form the sub- 

 soil sample. The number of borings recommended in text-books is 

 from five up to ten or twelve according to the size of the field sampled. 



Now the probable error affecting the analysis of a single boring is 

 a function of two probable errors, namely, (1) the laboratory error, that 

 is the error of analytical determination (which itself includes an error 

 of sampling from the laboratory sample), and (2) the field error, which 

 is the error due to the normal variation in the composition of the soil 

 from point to point in the field. 



Now if the probable error of a determination on one boring be P, 

 the probable laboratory error be p^, and the probable field error be p^, 



P, pi and P2 will have different values according to the soil constituent 

 which is determined. There will thus be a different set of P, p^ and p.^ 



