CEREAL INVESTIGATIONS AT THE NEPHI SUBSTATION. 1B 
a=yield of nearest check plat. 
b=average yield of all check plats. 
x=actual yield of nearest variety. 
y=computed yield of nearest variety. 
y=x+(b—a). 
In 1908, 7 check plats of Koffoid wheat (C. I. No. 2997) were used. 
This number was found to be too few, because they had to be scattered 
so widely that they did not represent the true value of soil variations. 
This was the case also in 1909, when 8 checks of the same variety 
were used. In 1910, 23 checks of Turkey wheat (C. I. No. 2998) 
were used and more accurate computations were made. Seldom 
were there more than 7 varietal plats between any 2 checks. In 
1911 and 1912 about the same number of checks was used as in 1910. 
The check-plat system described above did not prove as satisfactory 
as was hoped. It afforded a means of gaining a better idea of the 
relative yielding power of the several varieties than would be possible 
without the use of any system designed for that purpose. But the 
checking system used at Nephi was lacking in accuracy. The yields 
of the various check plats were influenced by so many uncontrollable 
factors (soil variation, winterkilling, weeds, etc.) that to assume that 
any one of the other varieties would be affected to a like degree would 
be more a theory than a determination. Though the check yields 
were fairly uniform in many instances, occasionaliy there would be a 
variation as great as 100 per cent. More than this, the variety 
growing on a plat next to a low-yielding check plat sometimes would 
yield very high. In the latter case the computed yield of the variety 
would necessarily fall far below its actual yield. In other cases the 
reverse condition would exist and the computed yield of the variety 
would greatly exceed its actual yield. Such instances show the inac- 
curacy of the checking system used. 
It is believed that a replication of plats devoted to a test of any 
one variety would be far more dependable as a means of showing 
the relative yielding power of that variety than would the check- 
plat system. Where a replication of plats 1s practiced, the number 
of varieties tested might necessarily be reduced. Then, too, more 
work in connection with the varietal test would be necessitated, but 
the accuracy of the results obtained would compensate for the extra 
jabor involved. 
The computed yields of the winter varieties grown at Nephi during 
the years 1908 to 1911, inclusive, are reported in connection with 
the absolute yields, but only for the purpose of comparison. 
Collection of Data. 
Throughout the entire season the behavior of the crops on the plats 
in the varietal test was observed closely. In the spring, after it was 
reasonably certain that no more plants would appear, the number 
