Feb. a8,19x6 
Correcting for Soil Heterogeneity 
1045 
Now, if we calculate the correlation between the observed yield of the 
ultimate plots and the observed yield of the combination plots it is 
found that 
r= 4-0.358i0.039 
This shows a fairly large coefficient of correlation, indicating a rela¬ 
tively large heterogeneity in the soil of this field. 
If we calculate the correlation between the corrected yields of the 
ultimate plots and the corrected yields of the combination plots it is 
found that 
r= +0.111 ±0.045 
This coefficient is less than three times its probable error and is hardly 
to be regarded as significantly greater than o. In any case it indicates 
that this method of correcting the yields has practically, if not quite, 
eliminated the influence of differences in soil of different plots. 
2,699 
2,616 
2,806 
2,703 
2,826 
2,600 
2,758 
2,894 
2,587 
2,759 
2,798 
2,684 
2,996 
2,953 
2, 766 
2,943 
3,007 
2,658 
2,915 
2, 766 
2,872 
2,975 
2,887 
2 , 8 ix 
2,650 
2,707 
2,666 
2,665 
2,924 
2,464 
2,625 
2,994 
2,354 
2,844 
2,895 
2,672 
3,157 
3,056 
2,824 
3,300 
3,112 
2 , 9 tl 
3,206 
2, 862 
3,067 
3,090 
2,987 
2,826 
2,488 
2, 594 
2,617 
2,642 
2,802 
2,563 
2,854 
2,869 
2, 708 
2,692 
2,774 
2,641 
2,805 
2,928 
2,600 
3,088 
2,982 
2,829 
2,958 
2,743 
2,938 
3,029 
2,863 
2,889 
2,305 
2,414 
2,614 
2,505 
2,608 
2,620 
2,637 
2,670 
2,690 
2,471 
2,583 
2,611 
2,498 
2,725 
2,496 
3,106 
2,776 
3,053 
2,734 
2,553 
2,904 
2, 737 
2,664 
2, 796 
2,119 
2,444 
2,398 
2,835 
2,640 
2,928 
2,993 
2,703 
3,013 
2,840 
2,614 
2,949 
2,812 
2,759 
2, 776 
2,627 
2,809 
2,541 
2,411 
2,584 
2,540 
2,6ll 
2,697 
2,637 
2,564 
2,345 
2,942 
2,901 
2,533 
3,091 
2,637 
2,593 
2,767 
2,624 
2,508 
2,839 
2,880 
2,647 
2,956 
2.549 
2,694 
2,578 
1,953 
2,479 
2,197 
2,278 
2,587 
2,414 
2, 716 
2,421 
3,018 
2,696 
2,615 
2, 804 
2,897 
2,677 
2,943 
2,532 
2,589 
2,666 
2,652 
2,732 
2,643 
2,550 
2? 782 
2,491 
2,367 
2,559 
2 , 53 i 
2,636 
2,671 
2,688 
Fra. 3 —Diagram showing the observed, corrected, and calculated yield (in grams) of Montgomery’s * 
wheat plots in groups of four, taken from figure 2. 
Similar coefficients have been calculated for other fields with corre¬ 
sponding results. 
It will next be of interest to test this method in the case of an actual 
variety test. This has been done in the case of all of our own variety 
test fields. The results will be published in another place in connection 
with a discussion of some pure-line oat varieties. In order to furnish 
an example of the use of this method in a variety test, the results of our 
1915 test of oat varieties are given below. 
Figure 4 represents a diagram of the 19*5 plots of oats at the Highmoor 
Farm (Monmouth, Me.). In the upper left-hand comer of each square is 
the plot number as it occurs in our records. Immediately below this is the 
name of the variety. In the case of the pure-line varieties these are 
22535°—16 - i 
