Vol. 6, 1920 AGRICULTURE: LIPMAN AND LINHART 
685 
portant findings of our studies, so that our colleagues may be apprised 
of what we consider to be an investigation of a fundamental nature, not 
merely on account of the importance of the subject itself, but because of 
its important bearings on agricultural research. 
It seems that the originators of both the Ohio and the Pennsylvania 
experiments saw the value of replicating check plots, while not recognizing 
the value of such replication for the treated plots. Only in the case of 
two complete fertilizer plots in the Ohio experiments, was the fertilizer 
treatment the same. In all the others every treatment was different from 
that of every other. This situation renders it difficult to subject their 
data to statistical treatment in the most desirable way. Fortunately, 
however, the long series of years during which the plots were studied gives, 
in a certain sense, a replication of plots of each kind. Our procedure has 
been, therefore, to group the plots in different ways including the following : 
1. All the check plots for each year. 
2. One check plot for all the years. 
3. All the plots treated with "one-element" fertilizers in one year. 
4. Every "one-element" fertilizer plot for all the years. 
5. The same double study for the "two-element fertilizer plots." 
6. The same for the "three-element" fertilizer plots. 
In addition to these groupings, others were made in which a given plot 
was studied with each crop in the rotation, so as to allow of only three to 
eight "yield" data for each plot. It may be said in passing that the ro- 
tation study included in these experiments only serves to complicate a 
situation which was already sufficiently difficult. We shall show in de- 
tail in the complete study which is to be published, that these experiments 
are utterly inadequate for solution of the questions involved and that all 
such fertilizer experiments are so fallacious as that they do not justify the 
use of the large amounts of money that are constantly being lavished 
upon them. 
For the sake of brevity, only a few of our findings, from the study of the 
Ohio data alone, are summarized here. 
1. The "one-element" fertilizer treatments show no significant increases 
in yield over the yields obtained on the untreated check plots. All state- 
ments to the contrary which have heretofore been made are erroneous 
and the data on which they are based were obtained by a misleading method 
of procedure and evaluation. 
2. The "two-element" fertilizer plots give definite increases in yield 
over the untreated control plots. Whether or not the increases are profit- 
yielding, however, is a question which will be discussed later. 
3. The "three-element" fertilizer plots give definite increases in yield 
over both the untreated control plots and the "two-element" fertilizer 
plots. Here again the question whether the increases are profit-yielding 
still remains to be determined. 
