98 MISSOURI BOTANICAL GARDEN. 
conditions are different. It is, therefore, only with the 
explicit statement that final conclusions must await the 
comparative treatment of wider series of data, that I indi- 
cate some of the points of more general interest. 
2. On comparative grounds, one must assume that the 
anomalous seed habit of C. longifolium, and other similar 
species, is not primitive but probably recently acquired 
from an ancestral form producing a large number of ovules 
per fruit. Several peculiarities of the inflorescence and 
fruits may be referred to the large size of the seed. We 
note that :— 
(a) Variation in the number of flowers produced or in 
the number of fruits matured per inflorescence is not greater 
than that generally found in inflorescences. The peculiari- 
ties of the seeds apparently have not produced any effect 
on variation in the inflorescence. 
(b) The variation in number of seeds per fruit, measured 
by range, standard deviation or coefficient of variation is 
very great. The distribution is also very skew. These con- 
ditions are probably directly due to the existence of a large 
number of ovules in each ovary (an ancestral character- 
istic?) of which only a part can, because of the great size 
of the seeds, be developed to maturity. 
(c) The distribution of seed weight is very skew and the 
variability very high. This is probably to be attributed to 
the limitation imposed upon the tendency of a large num- 
ber of ovules to develop into seeds by the inadequacy of 
plastic materials for all. 
3. There is a moderately close positive correlation of 
the order r = .35, between the absolute numbers of flowers 
formed and fruits developing per inflorescence. The correla- 
tion between the number of flowers per inflorescence and 
the deviation of the number of fruits developing from the 
probable, on the assumption of proportionate fertility 
throughout, is negative in sign and of about the same order 
of magnitude. The larger inflorescences are, therefore, less 
