848 
Journal of Agricultural Research 
Vol. XXIV, No. 10 
DISCUSSION OF RESULTS. 
True-breeding hybrids have been found in other genera than Pisum, 
notably in Oenothera , and have been interpreted by Muller (10) as due 
to the action of balanced lethal factors. In self-fertilized Gradus plants 
there is no evidence of zygotic lethals since under proper growing con¬ 
ditions all the ovules in a pod develop. Although an occasional sterile 
rogue is met with, all ovules generally form viable seed. In* over 60 
instances crosses between Gradus and Gradus rogue gave an average of 
4.5 seeds per pod out of a possible 7 or 9. About 4 seeds per pollination 
is the number generally obtained from artificial crosses of Gradus with 
other varieties. However, it has been customary to pollinate two or 
three stigmas with the pollen of one flower, the result being one well 
filled pod and one or two partially filled, as a consequence of deficient 
pollination. 
A microscopical examination of the pollen of Gradus and Gradus 
rogue showed well-developed pollen grains in both forms. Their pollen 
is about 100 per cent perfect. There was no morphological evidence 
of defective pollen which might be interpreted as due to the action of 
gametic lethals. 
It is, therefore, apparently out of the. question to explain the anoma¬ 
lous behavior of the rogues in crosses with Gradus types as due to the 
presence of zygotic or gametic lethals. 
As previously stated, the nonappearance of Gradus segregates in the 
F2 generation of the cross Gradus with Gradus rogue is explained by 
Bateson and Pellew (i) as probably due to somatic segregation of the 
type and rogue “elements” in the hybrid. The same authors have 
not suggested, however, a mechanism by which somatic segregation is 
accomplished. Such segregation would occur if somatic nondisjunction 
took place early in the development of the F^ plants, so that of the two 
daughter cells formed, one would receive both of the chromosomes 
derived from one member of the heterozygous chromosome pair present 
in the parent cell, and the other daughter cell would receive the two 
chromosomes derived from the second member of the heterozygous pair. 
As a result, both cells would retain the characteristic number of somatic 
chromosomes and at the same time be homozygous for either the type 
or rogue factors. 
Besides assuming the phenomenon of somatic nondisjunction, the 
further assumption must be made that the daughter cells homozygous 
for the rogue factors increase more rapidly than cells containing the type 
factors; and consequently the F^ plant becomes more roguelike as it 
matures. At sporogenesis only rogue tissue would take part in gamete 
formation; therefore type plants would not appear among the F2 progeny. 
Although somatic nondisjunction offers a hypothesis to account for 
somatic segregation it does not explain satisfactorily the failure of the 
type factors or “ elements ” to be present in the gametes of the F^ hybrids. 
Certainly accepted instances of somatic nondisjunction are rare, the 
best examples being the g3mandromorphs in Drosophila described by 
Morgan and Bridges (9) and attributed to the effect of the dropping out 
of the sex chromosome at an early division of the fertilized egg. The 
demonstration of the loss of a chromosome was possible from a knowledge 
of the factors linked with sex. As there are no factors known which are 
linked to the type allelomorphs of the rogue factors, the loss, by non¬ 
disjunction in the F^ hybrid, of the chromosome carrying the type factor 
can not be detected. In addition, in the F^ plants nondisjunctional 
