182 GENETICS IN RELATION TO AGRICULTURE 
two varieties of tobacco of the species Nicotiana longiflora. The smaller 
of these two varieties has a tube length of about 40 mm., whereas the 
contrasted variety bears flowers the tube length of which is over twice 
as great, namely about 93 mm. The two varieties had been self-fertil- 
ized for a number of generations preceding hybridization, and since it 
can be demonstrated that continuous self-fertilization tends to reduce a 
variety to a homozygous condition, it is fair to conclude that the parents 
represented varieties homozygous for nearly, if not quite, all their fac- 
tors. We are not surprised, there- 
fore, to find that they display only 
a slight variability in flower size. 
This slight variability is to be con- 
sidered merely an evidence of the 
influence of external conditions and 
of inherent variability in character 
expression and not of internal 
heterozygosity, for there is a limit 
below which it is apparently im- 
possible to force the reduction in 
variability of any given character. 
In this case the accompanying table 
which has been reproduced from 
East in its entirety will serve as the 
material for the following discussion. 
It will be seen in Table XX XIII 
that when the two varieties were 
crossed the Fy distribution occupied 
Fic. 87.—Average flowers of two ® position midway between the two 
varieties of Nicotiana longiflora with an parents. The number of plants 
average flower of the F2 from a cross be- 
tween them in the middle. (After East.) grown was somewhat larger than that 
for the parents, consequently the 
range covered by the F; distribution is slightly greater, but calcula- 
tions of the coefficient of variability show that the variability of 
the Fy is only slightly and not significantly greater than that of 
the smaller flowered parent. When we look at the F, from such a 
cross, we find that although it, like the F,, occupies an intermediate 
position, the range has been doubled and this in spite of the fact that the 
population contained only a few more individuals than that of the F. 
This increased variability is borne out by calculations of the coefficients of 
variability which are over twice as great for fF. as for F;. That the in- 
creased variability in 2 is the result of genetic segregation of some sort 
is shown by the distributions of 7’; families. They are strikingly different 
from each other in their position on the range, and in the variability which 
they display, as is shown clearly in the table. 
Digitized by Microsoft® 
