adjusting yields to their regression on a moving 
AVERAGE, AS A MEANS OF CORRECTING FOR SOIL 
heterogeneity 1 
By Frederick D. Richey 
Agronomist in Charge of Corn Investigations, Office of Cereal Investigations, Bureau of 
Plant Industry, United States Department of Agriculture 
GENERAL CONSIDERATIONS 
The recent tendency has been to eliminate the use of check plats as a 
means of correcting for soil heterogeneity in comparisons of varieties or 
selections under field conditions. Thus, the committee on standardiza¬ 
tion of field experiments of the American Society of Agronomy reports: 3 
“ Check plats .—Adequate replication of varieties or treatments removes 
the necessity of including check plats.” 
The author is in entire accord with this view if the word “check” is 
used in the ordinary sense and “adequate” is defined properly. It is 
felt, however, that replication adequate to remove the advisability of 
having some measure of the variation in soil productivity from point 
to point is impractical under many conditions. Moreover, reducing the 
variability by doubling the number of replications does not necessarily 
make the data obtained as significant as if the same reduction were 
achieved by adjusting the yields to a check occurring in alternate plats. 
In the author’s experiments with com, therefore, the frequency of the 
checks has been increased, when possible, until they occurred in alter¬ 
nate plats and at the same time the number of replications has been 
kept as large as conditions permitted. For any fixed experimental 
facilities the size of the plats, the number of replications, and the fre¬ 
quency of check plats are the elements that determine the number of 
strains that can be compared. If corrections for soil heterogeneity 
could be made without the use of check plats, a material gain in accuracy 
should be possible. 
With this in mind, a planting arrangement was devised for an experi¬ 
ment in which each of the strains to be compared was used as a check in 
one series of replications, thereby constituting that many extra replica¬ 
tions of each strain. The data obtained under this arrangement were 
fairly satisfactory as a whole, but the method was limited in its applica¬ 
tion and there were other objections to it. Some of the concepts on 
which it was based were not fundamental and modification of certain 
phases made it more satisfactory and permitted a wider application. 
As some of the principles involved appear not to have been used in 
connection with varietal and strain comparisons, and appear to promise 
something of value, it seems advisable to present them at this time. 
Before considering the method itself, it is desirable to discuss briefly 
the principles upon which it is based. This can be done more con¬ 
veniently in connection with certain hypothetical experiments, the object 
of which is to determine the relative productiveness of two varieties or 
strains of corn in a given field for a single season. In these experiments 
1 Accepted for publication Nov. 19, 1923. 
* Wiancko, A. T., Arny, A. C., and Salmon, S. C., report of committee on standardization of field experi¬ 
ments. In Jour. Amer. Soc. Agron., v 13, p. 372. 1921. 
Journal of Agricultural Research 
Washington, D. C. 
(79) 
Vol. XXVII, No. 2 
Jan. i2, 1924. 
Key No. G-350 
