x11 INTRODUCTION. 
to truly thoracic ventral fins (with the pelvic bones directly attached to the clavicles) 
in the series Ophiocephalus, Anabas, Osphromenus and Tetragonurus, Centrolophus, 
Stromateus respectively, and in order to attain precise diagnoses it becomes necessary 
to regard each of these two groups as of primary, é. é. sub-ordinal, rank, and to neglect 
the character of the attachment of the pelvic bones in defining them. 
The first and second of these sub-orders have the names Percesoces and Labyrinthici 
respectively. For the third I propose the name Zenopharyngii, whilst the fourth, 
comprising fishes with the skeleton in great part cartilaginous, may be termed 
Malacichthyes. Of these groups only the Percesoces is represented in the fresh waters 
of Mexico and Central America. 
The sub-order Acanthopterygii, as understood by Boulenger, is defined by a single 
character, viz., the direct attachment of the pelvic bones to the pectoral arch. When 
the systematic portion of the present work was commenced, more than two years ago, 
the sub-order was accepted by me; but I now think that the fact has to be 
recognized that the pelvic bones have become directly attached to the clavicles in 
several groups and that a large group defined by this character alone is unnatural. 
It has been mentioned above that each of two natural groups, Zenopharyngii and 
__ Labyrinthici, includes forms with the pelvic bones remote from the clavicles and 
others in which they have acquired a direct attachment to the clavicles. In a third 
group, the Berycoids, the two conditions are also found, Polymixia exemplifying the 
first and Z'rachichthys the second. It seems pretty certain that the Cyttide and 
the Percoid fishes have evolved independently from the Berycoids, and we cannot 
lay much stress on the fact that the pelvic bones are directly attached to the clavicles 
in the two first-named groups as indicating any special relationship between them. 
Enough has been said to show that the Acanthopterygii should be split up into 
several sub-orders; two of these are represented in the fresh waters of Mexico and 
Central America, viz. Percomorphi and Heterosomata, the latter including the single 
family Pleuronectide. 
4. GEOGRAPHICAL DISTRIBUTION. 
In discussing the distribution of fresh-water fishes we need not take notice of 
marine fishes (Centropomus, Pomadasis, Mugil, &c.) which ascend rivers for con- 
siderable distances, either in search of food er for breeding-purposes, nor need we 
consider fishes which descend the rivers to breed in the sea (e.g. Anguilla). 
Fresh-water species of marine genera (e.g. Chirostoma humboldtianum) and fresh- 
water genera of marine families (e.g. Aenatherina) have little importance. True 
