y) BULLETIN 1321, U. S. DEPARTMENT OF AGRICULTURE 
ference. The evidence from ear-to-row studies is supported further 
by data from direct comparisons between selected lots of ears differing 
in various characters and from comparisons among unselected ears 
(7). In these comparisons too, however, the differences in general 
were small. : 
Since the above experiments were reviewed the data from the 
——= es 
Nebraska and the New York (Cornell University) Agricultural _ 
Experiment Stations have been brought together (3, 6), and Wallace 
(9) has reported a brief statistical study based on data from the 
Iowa Agricultural Experiment Station. Etheridge (/) has reported 
new data from the Missouri Agricultural Experiment Station in 
which no significant relation was found between ear characters and 
yield. Hayes and Alexander (2) have reported data from the Minne- 
sota Agricultural Experiment Station, on the basis of which “the 
chances are 37 to 1 that selection for the score-card type of ear has 
led to a slight reduction in yield even though the plants were first 
selected on a yield basis from perfect-stand hills’’ (2, p. 17). Finally 
Kyle and Stoneberg (4) have reported that fewer rowed ears tended 
to be more productive than ears with larger numbers of kernel rows, 
both within and between varieties. This was true in some cases in 
spite of the fact that the yield per bearing plant was more for the 
ears with larger numbers of kernel rows. Kyle and Stoneberg also 
-showed that selection within selfed lines tended to reduce the number 
of kernel rows and that self-fertilized strains with fewer kernel 
rows tended to be more desirable from the standpoint of general 
vigor, to be more resistant to corn smut, and to have smaller per- 
centages of plants with heritable deleterious characters than com- 
parable strains with more kernel rows. 
Soil heterogeneity, irregularities in stand, insect damage, and all 
of the other conditions that continually interfere with successful 
field experimentation are related in no way to seed value. Never- 
theless, they determine a large part of the total variation in yield 
from row to row of an ear-row ae and largely obscure the relations 
between yield and factors that are related to seed value. Even 
small correlations may be yabetient under such conditions if they 
are significant statistically. In general, however, the number of 
records has not been large enough to establish the statistical signifi- 
cance of small correlations. Finally, the complex interrelations 
among the ear characters tend to obscure the meaning of the coeffi- 
cients of total correlation that are obtained. ‘Thus, there are very 
obvious physical relations between weight of ear and length or 
circumference of ear, factorial relations between shelling percentage 
and the weight of ear or cob, and presumably there are relations, 
such as that between length and circumference of ears, resulting 
from the inherent vigor of the parent plants and the environment 
in which each grew. These factors must be considered together 
if the whole relation between yield and ear characters is to be analyzed. 
In view of these facts further study of the relation of the charac- 
ters of seed ears to productiveness seems warranted. ‘The problem 
is attacked here through partial and multiple correlation studies on 
data from ear-row plats. 
