XIII.] CHOICE OF LAND FOR FIELD PILOTS 365 



one or more outcrops should be avoided ; trial holes 

 should be sunk to see that the depths of soil and subsoil 

 are fairly uniform ; thin soils on the chalk or limestone 

 should be avoided, because of the very irregular surface 

 of the underlying rock. Naturally, sharp slopes should be 

 avoided ; if there is any gradient, the plots should be laid 

 out to run parallel to one another up and down the slope, 

 so that each plot shares both the higher and the lower 

 levels. Other points will suggest themselves ; speaking 

 generally, the opinion of an intelligent farmer well 

 acquainted with the land is the most valuable guide. 

 It has been suggested to weigh up a number of areas 

 when the field is in ordinary crop, but, as indicated above, 

 the normal variations are so great that several years of 

 such trials would be required to arrive at any exact 

 conclusion. The condition of the land is equally im- 

 portant , land in high condition should be avoided, since 

 for some years at least the effect of the manures would 

 be swamped and all the plots would give very similar 

 results. On the other hand, bad land is not desirable, if 

 the object is to illustrate the action of fertilisers and not 

 to work out a method of dealing with that particular 

 class of land ; good land in poor condition after two or 

 more white straw crops is the best. Care should also 

 be taken to ascertain that the field has not been cropped 

 or manured irregularly for the five or six years previous 

 to the trials ; it is astonishing how long the disturbing 

 effect of farmyard manure or a leguminous crop, or 

 folding sheep on a portion only of a field, will 

 persist and become manifest under experimental con- 

 ditions. 



The size of experimental plots is a matter on which 

 there are considerable differences of opinion ; on the 

 one hand, large plots smooth out the small irregularities 

 due to minor differences of soil and drainage, insect 



