SEEDLING STRUCTURE IN THE LEGUMINOSE. 115 
comparatively small seeds *. "Ihe small class of “shrubs or small trees ” 
shows a distinet upward tendency in the size of seed. There is a great gap in 
size between this class, however, and the true trees. The most noticeable 
feature of this correlation is the uniformity of each class. Out of 78 
(uncultivated) herbaceous species only 11 have seeds more than 0°03 g. in 
weight; the 40 perennials only having one of these, this and seven of the 
other ten annuals being species of Lupinus. Also only 14 of the 74 shrubs 
had seeds weighing more than 0°03 g. In the two other classes of habit the 
statistics are quite different. Of the 16 “shrubs or small trees? 10 had 
seeds weighing more than 0'03 g., 8 being more than 01 g. Of the 40 
species of trees no fewer than 34 surpassed 0°03 g., and 22 of these weighed 
more than 0'1 g. 
Moreover, within single genera indications of the same kind are not 
wanting. The genera Acacia and Bauhinia, for instance, comprise shrubs, 
shrubs or small trees, and trees. In Acacia 14 species of shrubs give an 
average weight of 00163, 12 species of the other two classes give 00525. In 
Bauhinia the average for five shrubs is 0:162, for five species of trees and 
“shrubs or small trees” the average is 0:308. 
It thus appears that the production of large seeds is definitely correlated 
with the tree-habit. What may be the nature of this correlation it is hard to 
say. To produce a large seed may be a symptom of the greater vigour of 
growth that distinguishes trees from shrubs. (It is difficult to institute a 
comparison in this respect between woody and herbaceous plants.) Or, on 
the other hand, the large seed may be rendered necessary by the advisability 
of producing a large seedling which will give the tree a good start in life. 
It must be borne in mind that on the whole trees seem to produce, relatively 
to bulk, much fewer seeds than herbaceous plants in the same period of time: 
what is lost in number is balanced by the superior equipment with reserves 
of each individual embryo. It appears that a tree devotes itself to the pro- 
duetion of few offspring, each well furnished with the means of support; 
while the more opportunist herb produces an enormous number of offspring, 
among which there must be a very high mortality. 
We have seen that the characters of tetrarchy and a low transition are 
correlated with massiveness of the seedling; and also that these same 
characters are largely correlated with the tree-habit. It is now clear that 
the reason why both of these correlations can exist simultaneously is that, 
whatever may be its significance, there is a deeper correlation between the 
tree-habit and the production of large seeds and seedlings. 
* The large seeds of the Phaseolew and certain Lupinus spp. are noteworthy exceptions. 
If we omit Lupinus hirsutus, for instance, the average for annual herbs is reduced to 0:0293, 
