372 
Journal of Agricultural Research voi. xxvii, No. 6 
the results from the manured plats with those from the plats which re¬ 
ceived no manure. 
From a general inspection of the data obtained in 1915 (Table III), 
one can readily determine that the difference in rust percentages on the 
two series of plats are not great enough to be considered significant. 
This fact is further demonstrated by the calculated odds of only 4.1:1 
for Haynes Bluestem in favor of the nitrogen plats. Mathematicians 
have shown that in order to be significantly different the odds in favor 
of one treatment over another ought to be 30:1, which is approximately 
3.2 times the probable error. With Iumillo, the odds for rust percentages 
are 7.4:1 in favor of the plats not receiving nitrogen. The differences 
obtained on these plats in this experiment, therefore, are not significant, 
and the only possible conclusion that can be drawn is that the differ¬ 
ences in rust percentage caused by the two treatments are negligible. 
In the case of yield of grain of Haynes Bluestem, the odds are 42.5:1 in 
favor of the series not receiving nitrogen, indicating a significant differ¬ 
ence for this character. The average acre yield for all plats receiving 
nitrogen is 12.6 bushels, and for all others it is 23 bushels. 
In 1916 (Table VI) the calculated odds for stem rust on Haynes 
Bluestem are 9.2 : 1 in favor of the series not receiving applications of 
nitrogen, and therefore are not significant. With the hybrid, the odds 
are 55.5 : 1 in favor of the plats not receiving applications of nitrogen. 
These odds indicate a significant difference. The average rust per¬ 
centage on the hybrid, however, for all plats receiving nitrogen and for 
those not receiving nitrogen are 47.9 and 43.3 per cent, respectively. 
The calculation of the odds for these two series appears to give a sig¬ 
nificant difference, but when considered with the average rust percentages 
for the experiment as a whole, the results are not so convincing. In 
the case of yield of grain of Haynes Bluestem, the odds are 28.9 : 1 in 
favor of the series not receiving nitrogen, indicating a fairly significant 
difference for this character. The average acre yield for all plats re¬ 
ceiving nitrogen is 0.77 of a bushel and for all others it is 2.9 bushels. 
In the experiment at University Farm in 1917 (Table IX) the cal¬ 
culated odds for rust on Marquis are 7.5 : 1 in favor of the series receiving 
applications of nitrogen. For the hybrid, the odds are only 2 : 1 in 
favor of the series not receiving applications of nitrogen. It is evident 
that the differences in rust percentages on either variety in the two 
series are not large enough to be significant. In the case of yield of 
grain of Marquis, the odds are only 11.6 : 1 in favor of the series not 
receiving nitrogen, and therefore are not significant. The average acre 
yields for all nitrogen plats is 20.7 bushels and for all others it is 29.9 
bushels. 
On the plats at the Quinn farm in 1917 (Table XI), the calculated 
odds for rust on Marquis are 6.8 : 1 in favor of the series receiving 
nitrogen. For the hybrid, the odds are 8.2 : 1 in favor of the series 
receiving nitrogen. These odds are too small to be considered significant 
for either variety. In the case of yield of grain of Marquis the odds 
are 67 : 1 in favor of the series receiving nitrogen, a very significant 
difference. The average acre yield for all nitrogen plats is 33.5 bushels 
and for all other plats is 25.8 bushels. This experiment was made on a 
rather light soil and the yields were increased by applications of nitrogen, 
contrasting strongly with the results of the preceding experiments which 
were on a rather heavy soil on which nitrogen decreased yields. 
