CEREAL EXPERIMENTS AT DICKINSON, N. DAK. 11 
The plats are prepared for seeding to the cereals by double disking 
and harrowing. ‘The seeding is done with a 6-foot single-disk drill. 
The disks are 6 inches apart, with a covering chain attached to each 
disk. The seed is sown as uniformly as possible at what is considered 
the best rate and covered to a depth of 2 to 3 inches. Before seeding, 
it is treated for smut with a solution containing 1 pound of formal- 
dehyde to 40 gallons of water. The wheat is sown as soon as the 
land can be put in proper tilth in the spring, which is nearly always 
possible by the middle of April. Oats and barley are usually sown 
a week or ten days later. 
When the plants are about 6 inches high, the plats are usually cul- 
tivated across the drill rows with a spring-tooth weeder. This imple- 
ment is very efficient in breaking the crust and is less severe on the 
young plants than the spike-tooth harrow. No further cultivation is 
given. The plats are rogued each year in order to keep the varieties 
as free as possible from mixtures. The harvesting is done with a 
binder. The bundles are shocked and allowed to stand on the fields 
for two to four weeks before they are thrashed. 
CHECK PLATS. 
Check plats for determining the comparative or computed yields 
_ of the varieties have been used since 1908 in the varietal tests of wheat 
and since 1909 in those of oats and barley. These check plats, which 
are distributed through a series, are sown to the same variety and 
given identical treatment. In the series of wheat varieties, two check 
_ plats were sown in 1908 and three in 1909. The results mdicated 
that a greater number of check plats should be used if any value was 
to be attached to the computed yields obtained by their use. In 1910 
the practice of planting every fifth plat to the same variety as a 
eheck was adopted, and since that time this method has been in use 
for all the varietal tests of wheat, oats, and barley. - 
It was believed that the difference in yield of these check plats 
would show the variation of the soil throughout the series and that 
their average yield would equal the average yield of the variety 
used if it had been grown on the entire series. Then, by knowing 
the plat yield of a variety, its average yield per acre in terms of the 
check variety could be computed with reasonable accuracy. From 
the yield of the check plats distributed through the series, the 
rates of increase or decrease were computed for all plats. This dif- 
ference was added to the actual yield of the varieties or subtracted 
from it, as the case might be. A different method, which is con- 
sidered better, was used in 1912 and 1913. After obtaining the actual 
yield of all plats and the average yield of all checks, a graduated 
yield was figured for all plats from the yield of the different checks. 
The actual yield is then divided by the graduated yield and the 
