Florida Agricultural Experiment Station


ing to emergence might have been lengthened by disease, low
temperature, or lack of moisture, but none of those factors
was operating. It cannot, therefore, be denied that every F,
seed had a considerable rest period and it is indicated that the
F1 rest period is slightly above the mean of parents. While
these tests are limited in number and precision, the very ex-
istence of F1 rest period is entirely adequate to disprove dom-
inance of short rest period in sufficient degree to recover the
parental phenotype in 50 percent of F4 seeds.
 It remains to examine in some detail the behavior of rest
period in parent-offspring relations and in other respects. First
it is necessary to recognize the positive bias in mean rest period
which must be obtained from unequal expression of experi-
mental errors or random variation of any kind because the
range of variation extends to zero. It is noted that the range
of variation in the pure strain of Table 2 is sufficiently removed
from the limit to avoid most of the bias., The situation in Table
1 is quite different where the modal type of each strain is zero.
It is difficult to avoid the conclusion that mean rest period of
each strain in Table 1 is zero except for the bias of unequal
expression of experimental errors, i. e. random variations of
seed condition. If the present theory of the ranges of seed
condition and genotype extending beyond zero rest period is
correct a similar bias due to unequal reflection in rest period
will be similarly obtained. In both cases the bias is increased
as the position of the sample lies farther to the left or by an
increase of variance from genetic or non-genetic sources.
 Other significant features of rest period behavior may be sum-
marized:
 1. Less dormant parent strains in Table 1 all have zero as
the modal type but differ in means and variances. More dor-
mant parent strains in Table 2 differ significantly in mean rest
periods as judged by their standard errors, Table 2b.
 2. Differences between parent strains are generally confirmed
in their hybrids, Table 6. Comparisons of 1 x 14 with 13 x 14,
3 x 8 with 13 x 8, and 13 x 14 with 13 x 8 are more reliable,
being based on much larger numbers.
 3. Mean rest period of the F1, Table 6, is approximately equal
to the mean of parents. Later hybrids average not only con-
siderably less than the parental means but also considerably
less than one-half of the greater parents. This in spite of the
considerable positive bias which must occur in each hybrid mean