423 
species are found, to the nature of the bottom and the consequent 
clearness and purity of the waters, and to the existence and rate of 
current or flow in the waters inhabited by them. In this class of 
divisions, geological distribution merges into ecological relation, the 
distribution of species being no longer by geological areas, but by 
ecological situations. In this sense two species may occupy pre- 
cisely the same territory without ever coming into any effective con- 
tact with each other, because they are differently related to certain 
features of their environment. 
As an explanation of the more general facts of distribution re- 
quires an analysis and interpretation of continental, terrestrial, and 
even cosmic agencies affecting it, so an understanding of what we 
may call the ecological distribution of a species requires a corre- 
sponding analysis of the ecological features of the region. Such an 
analysis can here be carried but a little way, since the ecological data 
borne by our collections are only of a very general type; but such as 
they are, they may, if used with discretion, add definiteness and de- 
tail and some degree of statistical precision to our knowledge of this 
part of the subject. 
My statistics of associate occurrence exhibit in the most inter- 
esting manner the frequent tendency of closely allied species 
inhabiting the same territory to avoid each other’s company and 
thus to evade competition with one another by the choice of different 
haunts and situations within the area of their common habitation. 
In consequence of this tendency, we sometimes find widely unlike 
species more closely and commonly associated in our collections 
than like, the ecological repulsion of each for its similars bringing 
dissimilars together into more or less definite associate groups. 
The sunfishes proper, for example—that is, the Centrarchid@ ex- 
clusive of the black bass—although a homogeneous group of 
species as to form and external structure, are a diverse assemblage 
as to ecological relationships. If we compare the proportionate 
frequency with which the closely similar species of the genus 
Lepomts have been taken together in our collections—in the same 
haul of the net, or from the same situation at the same time—with 
the frequency of associate occurrence of the widely dissimilar 
species of the other genera of the family, we find that the unlike 
species have been taken together much more frequently than the 
like—in a ratio of 14 to 1,—that the species of Lepomis have, 
