1040 
Journal of Agricultural Research 
Vol. V, No. 23 
theless, where for various reasons it is impossible to make a large number 
(10 to 20) of repetitions, the factor of soil heterogeneity still enters into 
the average yield. One or two exceptionally high or exceptionally low 
yields will unduly influence the average where the number of repetitions 
is only four or five. 
In a series of papers Harris (3, 4, 5, 6) has called attention to various 
phases of the experimental error in field tests. In his most recent paper 
on this subject Harris (6) has proposed a method of measuring the hete¬ 
rogeneity of the soil of a field. The principle employed by Harris is 
stated thus (432-433) : 
If the irregularities in the experimental field are so large as to influence the yield 
of areas larger than single plots, they will tend to bring about a similarity of adjoin¬ 
ing plots, some groups tending to yield higher than the average, others lower. 
This tendency to grouping of the high- and low-yielding plots is evi¬ 
dent in most field experiments. It is clearly shown in the diagrams pub¬ 
lished by Montgomery (10). 
The measure which Harris proposes for this heterogeneity (or homo¬ 
geneity) of a field is the correlation between the yield of the ultimate 
small plots and the yield of various groups of contiguous plots. The 
more nearly this correlation approaches zero the more homogeneous the 
field. The more differentiated a given field is in regard to good and 
poor soil, the greater will be the value of the correlation coefficient. 
This method of measuring the heterogeneity of a field is dependent 
somewhat upon the size of the ultimate plots and also upon the method 
of grouping. It does, however, mark a distinct advance in our method 
of dealing with small plot experiments. 
While Harris's method provides a measure of the substratum hetero¬ 
geneity in a given field, it does not provide any means of obtaining a cor¬ 
rective term for individual plots. While in field experiments it is of 
importance to know the amount of heterogeneity in the field as a whole, 
it is usually of much more importance to obtain some correction to 
apply to individual plots which will in some measure even up the 
differences in soil conditions. 
The present paper is the result of an attempt to obtain such a correc¬ 
tive term. It is realized that the method proposed is far from ideal. 
It is believed, however, that it marks a step in this direction, and it is 
hoped that it may lead to further study of this important question. 
The usual method of taking account of soil heterogeneity is the use of 
check plots. However, in very many cases this method has been far from 
satisfactory. It is not at all difficult to find examples in the literature 
of variety tests in which the amount of variation in the check plots is 
nearly or quite as great as the variation in the other varieties. 1 If check 
1 Davenport, Eugene, and Fraser, W. J. Experiments with wheat, 1888-1895, Experiments with oats, 
1888-1895. Ill. Agr. Exp, Sta. Bui. 41, p. 147-160. 1896. 
Noll,C. F. Tests of varieties of wheat. Penn. Agr. Exp. Sta. Bui. 125, p. 43-56. 1913. 
