4 BULLETIN 1209, U. S. DEPARTMENT OF AGRICULTURE. 
11 caused prompt germination, and a good stand resulted. The plat 
thinned to the final stand of two plants per hill, conditions 
permitting, early in June. There was no rain from May 15 until 
July 1 . just prior to blossoming, but from that time to the end of the 
season there was too much rather than too little moisture. The 
weed Beeds which failed to germinate during the May-June drought 
when the corn was being cultivated developed rapidly following the 
July rains and after it was too late to control them. This not only 
reduced the yield as a whole but was an additional cause of irregulari- 
ties in the yields. Heavy storms in late August and September 
broke off some of the plants, which also interfered with the results. 
To represent the generations and parent strains in each case, 24 of 
the hand-pollinated ears grown in one of the 1020 plats were shelled 
and mixed thoroughly. Each composite lot so obtained was grown 
in a single row between one-row check plats. The rows were 10 hills 
long. A single comparison of all the seed classes with the necessary] 
cheek plats constituted a unit which was replicated ten times. A 
different class, however, was used as the check in each replication or 
series. Whatley was the check in the first series, St. Charles White 
in series 2, and so on, each of the 10 seed classes in the experiment 
being used as the check in one series. Replications 1 to 5, inclusive, 
were grown end to end, and series 6 to 10 were adjacent to numbers 
1 to 5. In addition to the seed grown at Burdette in 1920, the 
original cross, No. 201, F. (1914 seed), was included in each series. 
Because of lack of seed this strain was not used as a check. The 
order of planting is sIiowti in Table 1. 
This method possessed several advantages. The use of the small 
sections permitted 22 replications of each seed class (except No. 201, 
F.), which were distributed at right angles across the experimental 
plat. Moreover, the odd-numbered rows constituted controls on 
the fertility of the different series, whereas heterogeneity within a 
single series was shown by the even-numbered rows. There were, 
therefore, alternate check plats throughout the experiment without 
any increase being required in the area. In addition to these advan- 
tages, each class competed with every other class, so that although 
competition existed there should be no distortion of its effects. 
EXPERIMENTAL DATA. 
Counts were made of the number of plants per row prior to harvest- 
ing. The data on suckers and barren plants are not presented, as 
they are practically negligible and of no significance except as they 
vary with a few of the extremes in productiveness of the soil. As 
each row was husked the ears Were weighed. The product of five 
rows of each seed class was retained and the number of ears counted. 
Approximately 50 pounds of ears from this sample were shelled, the 
cobs weighed, and the grain determined by difference. Samples of 
this shelled grain were mailed to Washington in moisture-proof cans 
and the moisture content determined. 3 
e determinations were made in the research laboratory of urain Investigations, Bureau of Markets 
anii<r"i> Estimates, through the courtesy of Dr. D. A. Coleman, in charge. 
