HISTORICAL AND INTRODUCTORY 29 



The nature of the subject necessitates a further departure 

 from the usual procedure. In purely laboratory investigations 

 it is customary to adopt the Baconian method in which factors 

 are studied one at a time, all others being kept constant 

 except the particular one under investigation. In dealing 

 with soils in natural conditions, however, it is impossible to 

 proceed in this way : climatic factors will not be kept constant, 

 and however careful the effort to ensure equality of conditions 

 there is always the probability, and sometimes the certainty, 

 that the variable factor under investigation is interacting with 

 climatic factors and exerting indirect effects which modify or 

 even obscure the direct effects it is desired to study. 



Of recent years statisticians have devised methods for 

 dealing with cases where several factors are varying simul- 

 taneously. The data obtained by the various workers at 

 Rothamsted are therefore examined by a statistician who 

 endeavours to disentangle the effects of various factors and 

 to state a number of probable relationships which can then 

 be investigated in the laboratory by the ordinary single factor 

 method. 



The modern methods as applied at Rothamsted include 

 three distinct processes : 



1. Observations or experiments in the field by a group of 

 specialists working independently, but with full cognisance of 

 each other's results. 



2. Examination of the data by modern statistical methods 

 so as to ascertain the probable effects of the known factors 

 and to indicate where known factors are insufficient to account 

 for the results, and where, therefore, new factors must be 

 sought. 



3. Laboratory studies by the specialist staff of the relation- 

 ships indicated by the statistical examination, these being 

 reduced to single factor problems. 



