470 
AMERICAN JOURNAL OF BOTANY 
[Vol. io, 
effects on some families than on others. Since there is no difficulty in 
distinguishing true self-fertility from pseudo-fertility, the phenomenon has 
its advantages. By pollinating young buds of old plants under proper 
temperature conditions, one can obtain selfed capsules containing between 
one tenth and one half of the normal complement of seeds, and thus one 
can deal with populations produced by self-fertilization in self-sterile 
strains. 
The most interesting problem of self-sterility is the behavior of self- 
sterile plants when crossed among themselves. Darwin supposed each 
plant to be so specialized in its reproductive organs that, though it could 
not be fertilized by its own pollen, it could be fertilized by the pollen of any 
other plant of the same species. Such a particularized specialization he 
found difficult to explain. Fortunately it is not necessary to explain it, 
for it is not what actually happens. The self-sterile strains are made up of 
groups of plants wherein each individual is cross-sterile with all other plants 
of the same class and cross-fertile with all plants belonging to any other 
class. In other words, if plant A is sterile with plant B and with plant C, 
it may be predicted that plant B will be sterile with plant C; and if plant A 
is fertile with plant X which in turn is sterile with plant Y and plant Z, 
then it may be predicted that plant A will be fertile with plant Y and plant 
Z. I shall not discuss the physiological implications of this peculiar 
phenomenon, but shall describe briefly the facts emerging from the pedigree- 
culture experiments. 
Perhaps the most important genetic question is whether these classes of 
plants, wherein the individuals are all sterile with each other, are classes 
which are analogous to the classes of purples, magentas, reds, pinks, and 
so on, found in the F 2 generation after a white sweet pea is crossed with a 
purple sweet pea. That is to say, it is desirable to know whether one is 
dealing with a case of straight inheritance comparable with other cases of 
inheritance, but where the members of the classes, instead of being dis¬ 
tinguishable by visual methods, can be ticketed and grouped only by the 
criterion of cross-sterility. 
To determine this, a population of plants from a cross between N. 
Forgetiana and N. alata was tested for cross-sterility. Only 2.4 percent of 
cross-sterility was found, a percentage so small that it is not difficult to 
see how easy it was for Darwin to be misled into thinking that every plant 
within a self-sterile species was cross-fertile with every other plant. Though 
computation shows that from 20 to 25 intra-sterile, inter-fertile classes 
would account for such a small percentage of cross-sterility in random 
crosses, it is obvious that with such a slow-going means of testing the 
affinities of each individual a clear analysis of a population containing so 
many groups is impracticable. If the behavior of these self-sterile plants is 
governed by mendelizing factors, however, then it follows that a series of 
self-pollinations or of sister-brother matings for several successive genera- 
