238 A JOURNEY IN BRAZIL. 



zonian species, before taking on its own characteristic 

 features, passes through a stage resembling the perma 

 nent adult condition of the Hemirhamphus. It is interest- 



j ing to find that animals, which have their natural homes so 

 j far from each other that there is no possibility of any ma- 



\ terial connection between them, are yet so linked together by 

 (structural laws, that the development of one species should 

 recall the adult form of another.* The story of .the Acarag^. 



* When I attempted to record my impression of the basin of the Amazons, 

 and characterized it as a fresh-water ocean with an archipelago of islands, I 

 did not mean to limit the comparison to the wide expanse of water and the 

 large number of islands. The resemblance extends much further, and the 

 whole basin may be said to be oceanic also, in the character of its fauna. It is 

 true, we are accustomed to consider the Chromides, the Characines, the Silu- 

 roids, and the Goniodonts, which constitute the chief population of this net 

 work of rivers, as fresh-water fishes ; but in so doing we shut our eyes to their 

 natural affinities, and remember only the medium in which they live. Let any 

 one enter upon a more searching comparison, and he will not fail to perceive 

 that, under the name of Chromides, fishes are united which in their form and 

 general appearance recall several families of the class, only known as inhab 

 itants of the sea. The genus Pterophyllum, for instance, might be placed side 

 by side with the Chaitodonts, without apparently violating its natural affini 

 ties, since even Cuvier considered it as a Platax. The genera Symphysodon 

 and Uaru would not seem very much out of place, by the side of Brama. The 

 genus Geophagus and allied forms recall at once the Sparoids, with which 

 some of them were associated by earlier ichthyologists ; while the genus 

 Crenicichla forms a striking counterpart to the genus Malacanthus. Finally, 

 the genus Acara and their kindred closely resemble the Pomaccntroids. In 

 deed, had not the fresh-water genera Pomotis, Centrarchus, and the like, been 

 erroneously associated with the Percoids, the intimate relations which bind 

 them to the Chromides, and these again to the marine types mentioned above, 

 would long ago have been acknowledged. The genus Monocirrus is a minia 

 ture Toxotcs, with a barbel. Polycentrus, which is also found in the Ama 

 zons, stands nearest to Acara and Heros; it has only a larger number of 

 anal spines. In this connection it ought not to be overlooked that these 

 fishes are not pelagic, like the Scomberoids, but rather archipelagic, if I 

 may use this word to designate fishes dwelling among low islands. If 

 we discard the long-prevailing idea of a close relationship between the 



