566 
Journal of Agricultural Research 
Vol. XI, No. ii 
Since the reduction in damage per larva is more than outweighed by 
the increased number of larvae introduced into the ear, the practical con¬ 
clusion is that husk leaves are to be avoided in breeding worm-resistant 
strains. 
The regression of damage on husk leaves is 0.98; that is, an increase 
of one grade in the husk leaves is attended by an increased damage of 
nearly one grade or about 10 per cent. This relation is not the result of 
the slight negative correlations between husk leaves and prolongation or 
length of husks, since the partial correlation between husk leaves and 
damage for constant prolongation and length of husks is higher than the 
direct correlations. 
NUMBER OP LAYERS AND NUMBER OF HUSKS 
Number of layers and number of husks are a closely associated pair of 
characters, the correlation between them being 0.68. 
As a protective character number of layers appears to have the advan¬ 
tage. The direct correlations between number of husks and each of the 
three measurements of damage are negative, but too low to be of signifi¬ 
cance; and since the correlations of number of layers with the measures 
of injury are in every case higher, the correlation between number of 
husks and damage is doubtless largely a secondary relation, owing to the 
relatively close relation between husks and layers. The partial correla¬ 
tions between number of husks and the measures of injury for constant 
number of layers are, in fact, all positive instead of negative. 
On the other hand, a large number of layers appears to be a protective 
character second only to prolongation in importance. The direct corre¬ 
lation between number of layers and damage is —0.52. Its operation 
must be largely independent of prolongation, for the partial correlation 
between layers and damage for constant prolongation is —0.51 ±0.09. 
With days to silking constant, the correlation is reduced to —0.45 and 
for both days to silking and prolongation constant the correlation is 
— 0.40. 
Since the records show that only 4.5 per cent of the larvae gained 
access to ears by penetrating the husks, it is difficult to explain the 
correlation between number of layers and damage as a result of any 
direct protection. A large number of layers, which, of course, means 
wide husks, might bring about a closer wrapping of the husks and thus 
to some extent impede the progress of the larvae. It seems more prob¬ 
able, however, that the true relation is that suggested as a partial explana¬ 
tion for the relation between prolongation and damage and that number 
of layers is positively correlated with some protective character not 
considered in these experiments. 
Since the correlation between layers and number of larvae is closer than 
that between layers and damage per larva, a large number of layers 
