500 
GEOLOGY: /. BARRELL 
on the organic features, the successions and relations of fossil faunas or 
floras. On the other hand, it is the work of physical and historical 
geology to res|:ore the ancient environments. The relations of the en- 
vironments to the biotas is a field wherein physical geology and paleon- 
tology meet, to give a better understanding of the underlying physical 
surroundings associated with organic response and progress. It is from 
the standpoint of physical and historical geology, rather than from that 
of paleontology that the present study is made. 
In this subject are two major problems — first as to the environment in 
which fishes developed; second, the changes in the environment and the 
associated organic responses which brought forth amphibians from 
fishes. It is the solution of the second problem which is here especially 
sought,^ but it involves a statement of the evidence in regard to the 
first. 
The question in regard to the origin of fishes is as to whether they 
developed first in the sea and later invaded the land waters, as has been 
generally assumed, or whether their expansion was in the opposite 
direction — from rivers to embayed waters and thence finally to the sea. 
Chamberlin appears to have been the first to have seriously suggested 
that the latter may have been the real direction of their emigration. ^ 
He pointed out that the fishes appeared as fossils only in the embayed 
waters until they had attained such dominance that they spread into 
the truly oceanic realm. Their body form, furthermore, Chamberlin 
notes, is peculiarly adapted to the stemming of currents and suggests an 
initial adaptation to rivers. 
The evidence for this hypothesis of the continental origin of fishes 
has been examined by the writer, by taking up the earliest faunas and 
studying the mode of origin of the associated sediments. The results 
strongly support Chamberlin's position. The exclusively marine habi- 
tats of the lowest remaining chordates, constitute really negative evi- 
dence on the subject, for these forms are far removed from fishes, and 
they leave no fossil record. The original habitats may have been far 
wider and the progressive forms may have lived a freer life than is 
shown by these retrogressive and more or less aberrant relics. The 
positive evidence given by the early fish fossils is more definite and con- 
clusive. A review of this is as follows : 
The earliest known ostracoderm remains were found by Walcott in 
1891 near Canyon City, Colorado, in a horizon belonging to the base 
of the Middle Ordovician, in sandstones which rest upon the Pre-Cam- 
brian.^ These sandstones are marginal marine deposits, holding species 
