402 
Journal of Agricultural Research 
Vol. XI, No. 9 
to be about equally reliable as a comparison. In this work the em¬ 
ployment of a check for every other row was said not to be sufficient to 
offset the variability due to soil irregularities; however, by means of io 
replications and alternate check rows the experimental error was reduced 
about 50 per cent. Reduction of the standard deviation was most rapid 
for the first 7 replications. 
Numerous papers by Harris have called attention to the importance 
of the experimental error in variety testing. The more recent one (3) 
gives a method of determining the reliability of a field for experimental 
plot work. If irregularities in the field are large enough to affect areas 
larger than single plots, adjoining plots will be similarly affected. By 
the use of the correlation coefficient the correlation between the yield of 
the small plots and the yield of various groups of adjacent plots is deter¬ 
mined. The more homogeneous the field the lower will be the correlation 
coefficient. 
Surface and Pearl (14) have given a method of correcting plot yield 
which is partially based on Harris’s correlation method. The calculated 
yield of any particular plot is determined by the contingency method. 
If this calculated yield is above or below the mean yield of the field, 
it is supposed that the soil of the plot is above or below the average of 
the field in productivity. The amount that the calculated yield is above 
or below the mean yield is then subtracted or added to the actual yield 
of the plot in question. The reliability of this method was then tested 
by the correlation plan as given by Harris. This test showed a con¬ 
siderable decrease in the heterogeneity of the field for the calculated 
yields as compared with the actual yields. The authors then show how 
this method may be used in variety tests. 
MATERIAL AND METHODS OF WORK 
In 1915 it was decided to make a test of the rod row as a preliminary 
method of determining performance. The method of handling the rows 
is about the same as is used at Cornell University in the Plant Breeding 
Department, although for the purposes of our work some modifications 
have been made. 1 
For our work 18-foot rows are used, spaced at a distance of 1 foot 
apart. At harvest 1 foot is discarded from each end of each row, thus 
leaving 16-foot rows. The use of the same length of row for wheat 
('Triticum spp.), oats (Avena saliva ), and barley (Hordeum vulgare ), 
increases the work of computing the yields per acre. The same length 
of row, however, for the various small grains allows a comparison of 
yield variability. 
The fows were planted at the usual rate per acre for Minnesota, or 
16.8 gm. per row for wheat, 15 gm. for oats, and 18.1 gm. for barley. 
1 Due acknowledgment is made to Dr. H. H. Love, of Cornell University, for descriptions of their row 
plans. 
