xm.] ARRANGEMENT OF EXPERIMENTAL PLOTS 367 



receiving the same treatment, arranged about the field 

 as follows : — 



I n this way the experimental errors due to inequalities 

 of soil and other accidentals are greatly reduced. With 

 several stations repeating the same experiment in 

 different districts the effects of soil may also be largely 

 eliminated, and in the course of a single season a result 

 can be obtained which is only subject to the error 

 due to variations of season. The shape of the plots 

 should be long and narrow rather than square, as this 

 tends to equalise, the soil conditions, and their breadth 

 should be chosen to give each exactly the same number 

 of rows of roots or corn, as the case may be, a point 

 which must be closely watched in seeding. When the 

 plots are continued for more than one season, some 

 method must be adopted to mark their position per- 

 manently, but posts at the corners interfere with 

 cultivation and are often in consequence taken up by 

 the labourers. At Rothamsted the method adopted is 

 to set out the plots initially in the field, keeping the 

 boundaries well away from the hedges and trees (the 

 soil on the headlands of a field is generally irregular, 

 and trees have a very wide-reaching effect), and then 

 mark the outside lines of the experimental area by 

 means of stout posts painted white and set in the 

 hedges round the field. These being out of the way 



