SYSTEMATIC RELATIONS OF FISHES. 585 
In Dipnoi, on the other hand, we have this axis complete, or 
rather with greatly multiplied distal segments, and with or without 
lateral radii. In the Australian Ceratodus Günther finds numerous 
lateral series on both sides of those of the axial row. Hence the 
limb of this order is considered by Owen, the simplest or primary 
type, and this proposition is abundantly confirmed by the beautiful 
researches of Gegenbaur. The foundation laid by this author 
for the history of the genesis of limbs will ever be a landmark 
in the history of modern theories of creation. See his memoir, 
Ueber das Skelet der Gliedmaasen der Wirbelthiere im Allgemei- 
nen, etc., Jenaische Med. Zeitschr., vol. v., p. 397: 
Important as are the characters that patai) the several 
oups indicated by the different types of structure of the limbs 
and pelvis, they do not seem to me to warrant their recognition as 
classes, equivalent to those of the six already pointed out. Taking 
them together there is a greater coherence also in the structure of 
brain and circulatory systems than would be the case with any 
other two of the classes adopted above. The peculiarities of the 
limbs, important as they are, are nearly related in the want of 
specialization of their parts, seen in the Butrachia and other 
classes, the differences consisting rather of number and position 
of similar parts. The pelvis of the Dipnoi might be regarded as 
of primary importance but for its existence in the Holocephali, 
whose limbs again are so near those of the shark. 
It remains, therefore, to adopt the Linnzean and Cuvierian class 
Pisces, and to grant as sub-classes, the groups of Holocephali, 
Selachii and Dipnoi. There remain as sub-classes the groups typ- 
ee by Polypterus on the one hand, and the true fishes on the 
The first has been already distinguished in its external 
Bta by romais Huxley, who again brought light out of ob- 
scurity when he established his *“ third sub-order of ganoids, the 
Crossopterygide.” This division is in my estimation a natural 
one, and to be elevated to arank equivalent to that of each of the 
three above named, being the only part of the original division of 
Ganoids of Miiller entitled to it. Professor Huxley defined it as 
follows: 
“ Dorsal fins, two, or if single, multiplied or very long; the 
pectoral and usually the vertical fins lobate; no branchiostegal 
rays but two principal, with sometimes lateral and medianjugular 
plates situated between the rami of the mandible ; caudal fin diph- 
