Methods of Field Manurial Trials. 221 



down are of such general application that it has been thought 

 desirable to reproduce a summary of his article. These prin- 

 ciples are based largely on his own experience, and much of 

 the following information is accordingly an account of the 

 practice he himself follows in carrying out experiments. 



Variations in the yield of differently manured plots of a field 

 can only be accepted as attributable to the action of the manure 

 when it has been definitely ascertained that the only difference 

 in the plots consists in the manures applied. In the conduct of 

 experiments, a field should be chosen which is uniform in all 

 respects. It will be readily understood that absolute uniformity 

 in the conditions is practically unattainable, and an examina- 

 tion should therefore be made to determine how far the field 

 corresponds to such a requirement. Analysis of the soil, although 

 useful, is not sufficient ; and a more satisfactory judgment can 

 be formed by exact and repeated observations on the develop- 

 ment of the plants. The best method of doing this is to conduct 

 a preliminary trial during one season, no manure, or at least the 

 same manure throughout the field, being used for this purpose ; 

 although here again it must be remembered that very different 

 results may be shown in different seasons. 



Although it is not possible to eliminate entirely the various 

 irregularities in a field, whether due to the nature of the soil or 

 to the operations performed on it, much can be done to minimise 

 the errors due to this source. One method consists in giving 

 the same manures to several plots and totalling or averaging 

 the results. And here it may be remarked that the reliability 

 of the deductions to be drawn from an experiment is propor- 

 tional to the accuracy of the standard with which it is com- 

 pared ; hence four or five control plots should be employed, so 

 as to ensure as much accuracy as possible in the standard used 

 for comparison. Another means of rendering the errors less 

 sensible is to use the poorest land, and to manure as heavily as 

 practicable (within limits and, of course, consistently with the 

 question to be elucidated), so as to allow the manures full scope 

 for their action : nitrogenous manures, for instance, will yield 

 little result if the land already contains a considerable supply of 

 that element. 



It is often considered that the plots should be as large as 



rll 



