29 
application of the test of texture or form of scales would 
remove true Ganoids from the group, and introduce alien 
genera having no real afl&nity with the order. Amia, with 
its cycloid scales, the two species of Foli/odon, with their 
nearly naked skin, would go in spite of the ganoid character 
of their heart, intestine and optic nerves. Balistes, Lori- 
caria, and a number of plectognathous fishes would enter the 
order, entitled to do so solely by their osseous scutes. The 
zoologist would not tolerate for a day the consequences of 
any principle of division which involves such anomalies as 
these. 
Miiller has pointed out how destitute of value is any 
classification of teleostean fishes based upon scale characters. 
Well-defined and natural families contain both cycloid and 
ctenoid genera. Genera can be named with both cycloid 
and ctenoid species. Atherina (the Sand Smelt), and Mugil 
(the common Mullet), both belong to one family, the 
Mugilidce, yet the first is cycloid and the second ctenoid. 
The families Percidce, Clupeidce, GoUdce, and many others 
include both forms. In short, scale-characters, like characters 
taken from the caudal fin, tell us nothing unless the affinities 
of the fish in question are beforehand nearly determined. 
In 1844 appeared the celebrated Memoir of Miiller, 
" TJeber den Bau und die Grenzen der Ganoiden,"* to which 
we have already referred as a work which marks the era of 
sound general principles of definition and classification of 
Ganoid Fishes. 
Miiller gives first a historical resume of the work done 
by his predecessors in this field of inquiry, Eafinesque, De 
Blainville, Cuvier, and Agassiz. The views of the last only 
of these receive a full discussion, for he alone had recognized 
the ties which connect the Ganoids together. Briefly, but 
with irresistible weight of argument, Miiller disposes of the 
* Abhandlungen der Akademie der Wissenschaften zu Berlin, 1844. 
