472 
AMERICAN JOURNAL OF BOTANY 
[Vol. io, 
their activities before fertilization are concerned they behave as if they were 
all alike. Pollen grains from a single plant may be of variant constitutions 
with reference to their transmission of qualities to the next generation, 
therefore, but they show no selective fertilization. They are controlled in 
their gametophytic activities by the genetic constitution of the mother plant 
on which they are formed. For this reason a genetic experiment conducted 
in such a manner as to have the critical matings made with pollen from a 
single plant will give the results to be expected from chance matings of 
germ cells. But one can easily imagine types of genetic matings where the 
results would be vitiated by not keeping this fact in mind. Suppose, for 
example, that one desired to make a test mating on plant A with pollen 
from plant B. Plant B produced little pollen, however, and additional 
pollen from plant C was used because plant C was assumed to have the 
same constitution as plant B with regard to the particular factors under 
investigation. Unfortunately the pollen tubes from the pollen of plant C 
grow faster than the pollen tubes from the pollen of plant B, and the resulting 
zygotes are all sired by plant C. Naturally, no geneticist in his right mind 
would make a test in this way, but nevertheless it may be well to have in 
mind its possibilities. 
Second, the factors which govern the behavior of self-sterile plants are 
strictly inherited, and are transmitted in accordance with a definite Mende- 
lian mechanism. Class A, for example, is class A wherever found. A 
single class has been identified by the cross-sterility test through three 
successive generations by Dr. Anderson; and the same class has been found 
in collateral families as far removed as fourth cousins. Thus the behavior 
of these factors controlling a peculiar physiological difference is exactly the 
same as that to be found where visually identifiable morphological differences 
are found. 
Third, the genes which govern the behavior of these plants in crosses 
are numerous. About fifteen classes have been proven to be genetically 
distinct from each other by proving each class to be fertile with every other 
class. Eight or ten other classes have proven to be distinct from all other 
classes with which crosses have been made; but since every possibility of 
linking them up with known classes has not been tested, it can not be 
maintained that they must belong to separate groups. Arguing the matter 
as a problem in probabilities, however, it can be said that the chances are 
greatly in favor of there being more than twenty genetically different intra- 
sterile groups of self-sterile plants in Nicotiana Forgetiana and Nicotiana 
alata. 
It is greatly to be regretted that it has been impossible to test out 
thoroughly a good random sample of individuals in these two species. 
When our work on the self-sterility problem was resumed after the war, 
only two packets of N. alata seed and one packet of N. Forgetiana seed 
would grow. The pedigrees of these seeds were such as to make it highly 
