43 
primarily geographical, and the second is essentially biological. Both 
methods are useful and the product of both is necessary to a full knowl- 
edge of our complex subject; but of the two, the biological method seems 
to me much the more useful and significant. = 
The most general division of the environment of fresh-water fishes 
into habitats distinguishes large rivers, small rivers, creeks, upland lakes, 
lowland lakes, marshes, and stagnant ponds. Streams are still further 
divisible into those with rocky bottoms, bottoms of sand, and bottoms of 
mud; into those with a swift or with a sluggish current; and into those 
with clear or with turbid waters. Even parts of streams are distinguish- 
able into different habitats—the rippled reaches of rocky streams differ ma- 
terially from the deep, still pools between. In the larger rivers, like the 
Illinois, we may sometimes distinguish between the opposite margins, 
where, as at Havana, one has a muddy bank and the other a sandy one. 
In lakes there are notable differences between the marginal shoals and 
the deep interior waters, between sandy bars and mud flats; between 
open waters and those filled with weeds; and in the weedy parts, be- 
tween those in which reeds, rushes, and other coarse, rooting plants are 
present and those in which the plants are mostly submerged. These pre- 
sent their characteristic differences in the fishes which resort to them—- 
differences clearly discernible, however, only by the use of quantitative 
methods, which give us the relative numbers of each species found in 
each situation over a sufficient length of time and variety of external 
conditions to make us sure that we are getting fair and stable averages. 
The first thoroughly practical work of this kind that I know of was 
done at Havana, Illinois, under my direction, in 1898-99, by Wallace 
Craig, later a doctor of philosophy of the University of Chicago, but at 
that time a temporary assistant on the Illinois Natural History Survey 
and also a graduate student in the University of Illinois. He began in 
August, 1898, a detailed study of the local distribution and the move- 
rnents of fishes, with a view to making out preferences of situation or 
choices of environment of the various species of fish under varying con- 
ditions and at different times of the year. Using identical apparatus by 
uniform methods at regular intervals in the waters of the locality, it was 
possible to get totals and averages by a comparison of which the strik- 
ing features of the different situations were made manifest when the 
statistical tests were compared with each other. By this method it was 
shown that the Illinois River gave us different data of frequency for 
the different species on the two sides of the stream, one of which was 
