CHAPTRR XIII 



THE CONDUCT OK EXFEKIMKNTS WITH FERTILISERS 



Magnitude of Experimental Error involved in Field Experiments 

 —Choice of Land for Field Flxperiments— Size and Shape of 

 l>lols_Machines for sowing Fertilisers— Should Farmers con- 

 duct Experiments upon their own Land ? 



The value of any fertiliser, new or old, on any par- 

 ticular soil can onl)- be settled by experiment ; for 

 though it is now possible to a large extent to recognise 

 types of soil by their analysis and predict their 

 behaviour, because the main outlines of the principles of 

 the manuring are understood, yet unknown factors will 

 often intervene and upset expectations. The proper 

 conduct of field experiments is therefore a matter of 

 considerable moment, and it is of particular importance 

 that the degree of accuracy which may be expected 

 from a series of such trials should be realised before any 

 scheme of experimentation is embarked upon. One 

 often sees experiments so designed that the differences 

 between the plots are likely to be less than the 

 experimental error ; still more often one sees conclu- 

 sions drawn from differences between the yields of the 

 plots that are smaller than the experimental error. Nor 

 must it be supposed that by any amount of care the 

 experimental error can be got r'd of; there are various 

 ways by which it may be diminished, but in some form 



