Feb. 9, 1924 
Effect of Fertilizers on Stem Rust of Wheat 
37i 
prior to the maturity of the plants could not be expected. The average 
percentages of stem rust for the three replications are fairly uniform and 
show no consistent correlations with the fertilizers. A few more plat 
replications undoubtedly would have made the results just as conclusive 
as those for 1918. 
In 1922, observations were made on a field of winter wheat at the 
Crookston substation in which several strips across the field had been 
fertilized with acid phosphate. No difference in growth or maturity 
of the plants could be observed and there was a perfectly uniform stem 
rust infection of 10 per cent on all the plats. Observations were made 
also on plats of Mindum, C. I. No. 5296, a durum wheat, in a four-year 
rotation series with different rates of manuring. The results are given 
in Table XVII. The amount of lodging was directly proportional to the 
amount of manure applied, but the degree of stem rust infection was 
uniform in all of the plats, at least as late as about two weeks before 
harvest. 
Table XVII. —The effect of barnyard manure on degree of lodging and percentages of 
stem rust on Mindum in IQ22 in the fertilizer plats at Crookston , Minn. 
Amount of manure. 
Percentage 
of plants 
lodged. 
Percentage 
of stem 
rust. 
None. 
O 
IO 
4 tons. 
Trace. 
10 
8 tons....;.. 
5 
25 
35 
60 
IO 
16 tons. 
IO 
16 tons (in 2 applications). 
IO 
32 tons.. 
IO 
SIGNIFICANCE OF RESULTS 
The average percentage of rust in all plats receiving nitrogen compared 
with the average percentage in those receiving none gives a fair indica¬ 
tion of the resultant differences from the two treatments. The averages, 
however, may or may not be a true expression of the differences between 
the two series of plats. If the individual percentages entering into the 
average are relatively small, it becomes a fairly accurate expression, 
but, as these increase, the average becomes less and less a reliable expres¬ 
sion of the true condition. In considering such a variable as the effect of 
fertilizers, comparison of two series in an experiment comprising a small 
number of plats is most availably presented by the method given by 
Student in Biometrika (34). By this method the resulting differences 
between the treatments are expressed in odds. 
In the following discussion of results, the odds have been calculated 
mainly for the rust percentages and yield, these being the two chief 
characters under consideration in this paper. The odds are calculated 
from the results on two adjacent series of plats. The series of four plats 
receiving 250 pounds of sodium nitrate were adjacent to a series of four 
receiving none. These two series of homologous plats are used in all of 
the experiments with commercial fertilizers. They are shown in Table I 
as plats 3, 10,17, and 24 for the nitrogen series; and plats 4, 11,18, and 25 
for the adjacent non-nitrogen series. In the case of manured plats in the 
years 1918 and 1920 (Table XVII) the odds are calculated by comparing 
