486 
Journal of Agricultural Research 
■Vol. XIV, No. II 
method of study and presentation of results 
Tables I to XII, inclusive, present in a separate table for each field 
Station the yields of each crop each year on plot B (not subsoiled) and 
on plot E (subsoiled). The average yield of each plot for the entire 
period of years is also presented. The principal comparisons have been 
made and the conclusions of the effects of subsoiling drawn from the 
annual and average yields. But in order to obtain a uniform expression 
of difference on which to calculate the probable error of the results, and 
through which to compare the relative effect upon different crops and 
at the several stations, it has seemed necessary to express the differences 
in some form of ratio or percentage. 
A thorough study of the data has been made by means of percentages 
calculated on four different bases. The ratios of the average yields have 
been computed for each crop at each station by dividing the average 
yield of plot E by the average yield of plot B. This single calculation 
does not afford either an expression of the results each year or opportu¬ 
nity for the determination of the probable error based on annual differ¬ 
ences. 
In each experimental field there are a large number of plots of each 
crop grown each year by various methods. The difference in yield 
between B and E each year has been calculated as a percentage of the 
average yield of all plots of the same crop in the field for that year. 
Objection has been made to the use of this method for this particular 
study, for the reason that with some crops the yields of these methods 
are below the general average and with other crops they are above it. 
This results in some cases in a disproportionate valuation of the entity 
under study. The average departure of the results as calculated by this 
method from those obtained by the first method is so high as to make its 
use for the present study unsatisfactory. 
The difference in yield between plots B and E each year has been 
calculated as a percentage of the yield of B, which may be considered as 
the control plot, or the one giving the yield that it is sought to increase 
on E by means of subsoiling. For the results in hand this method gives 
undue weight to comparatively small differences in yield as the yield of 
the plot selected as the base of comparison approaches zero. There are 
so many of these cases that the results of comparison by this method are 
not satisfactory when compared with actual differences in average yield. 
This objectionable weighting of small differences in yield is largely 
overcome and the results smoothed by using the arithmetic mean of the 
yield of the two plots under study as the base on which to calculate the 
difference between the two as a percentage. This method tends, however, 
to reduce the percentage when it is above and to increase it when it is 
below ioo. In over half the comparisons the ratios depart from ioo 
by less than 10, and within this range the distortion is not great. The 
average results are further made up of varying combinations of increases 
