SPECIAL FARM TOPICS 205 



himself an experimenter. Experience brings him certain knowl- 

 edge. Many experiments which farmers attempt, however, are 

 either valueless or actually misleading, because of failure to ob- 

 serve some of the essential conditions of successful experimenta- 

 tion, for investigation in agriculture by experiment is a business 

 by itself, entirely distinct from ordinary farming, and many a good 

 farmer will overlook points of vital importance to the success of 

 an experiment until his attention is called to them. 



A matter of first importance in preparation for field experi- 

 ments is that the soil shall be as nearly uniform as possible in char- 

 acter; to know whether the soil is of sedentary drift or alluvial 

 origin. 



A sendentary soil is liable to be more uniform than a drift or 

 alluvium, because in the case of transported soils there is generally 

 a more or less uneven deposit of materials, an excess of gravel and 

 coarse sand appearing in one spot and of silt and finer particles in 

 another. A heavy sheet of drift may sometimes become weathered 

 into practically the same condition as a sedentary soil, while it 

 would seem that some of the great loess deposits would offer espe- 

 cially good conditions to the field experimenter, the loess being the 

 fine grained, silty soils found in some of the Western States, and 

 whose origin is apparently due to the blowing of the dry surface 

 dust into banks and hillocks, sometimes many feet in depth. 



Experiment Plots. Farmers generally have the idea that ex- 

 periment plots should be made as large as possible, an idea natur- 

 ally following their observation of the inequalities of most soils ; but 

 the practical difficulty in the way of using large plots is the fact 

 that for a comparative experiment the soil must be as nearly ab- 

 solutely uniform as possible, and it is extremely difficult to find 

 large areas having sufficient uniformity. In almost all cases it 

 will be found better to use a large number of small plots than a 

 small number of large plots, since by multiplying the plots the 

 variations of the soil can be more evenly distributed. A field of 

 ten acres, for example, in which it is desired to make ten compari- 

 sons, will yield results of far greater value if cut into 100 plots 

 containing one-tenth acre each, giving ten plots, distributed over 

 the field, to each particular comparison, than if only ten plots are 

 employed. 



It is highly important that the plowing and fitting of the land 

 for a comparative experiment be as uniform as possible in every 

 respect. A difference of a few weeks in date of plowing may cause 

 as great a difference in yield of crop as will be produced by differ- 

 ence in fertilizing, and far greater than is usually observed between 

 varieties. For this reason the plowing should, as a rule, be done 

 across the plots, and when it becomes necessary to plow the plots 

 separately, in order to ridge them, the work should be pushed for- 

 ward as expeditiously as possible. 



Seed. The selection of seed is a very important matter in 

 field experimentation. Defective seed means an imperfect stand, 

 and an imperfect stand means an undesirable and inconclusive 



