48 THE AMERICAN CHARACIDAE. 



possible, in the form of the conventional phylogenic tree. In order to better 

 express the relationship of smaller groups of species the genera have been made 

 as small as consistent with facts. The Tetragonopterinae seem to form an 

 interlacing fabric rather than a branching tree. An illustration of the interest 

 and difficulty is given by the series of genera Poecilurichthys, Ctenobrycon, 

 Psellogrammus. The genus Poecilurichthys as here understood, grades in- 

 sensibly into the genus Astyanax, and perhaps should have been retained 

 united with the latter. Be that as it may, P. spilurus is much more closely 

 related to Ctenobrycon hauxwellianus than it is to any species of either Poe- 

 cilurichthys or Astyanax, and yet there is scarcely a technical character by 

 which it can be separated from the remaining species of Poecilurichthys and 

 united with Ctenobrycon. The latter is sharply distinguished by ctenoid scales 

 from all other members of the subfamily. In an another direction, the speci- 

 mens of Ctenobrycon hauxwellianus from Paranagua form a complete bridge so 

 far as the technical characters are concerned, between the genera Ctenobrycon 

 and Psellogrammus. From the parental form, P. spilurus, the Amazonian 

 Ctenobrycon hauxwellianus has become generically distinct. C. hauxwellianus 

 in its turn, in Paranagua, is becoming and, in Paraguay, has developed the char- 

 acter which distinguishes Psellogrammus. 



The difficulties of the systematist could of course be obviated by retaining 

 all of these things in the single old genus Tetragonopterus, but this procedure 

 would neither remove the facts nor explain them. 



I have attempted to express the relationship of the genera by the accom- 

 panying diagram, in which Astyanax occupies the center, and in which it is 

 made to appear that the scaled caudal has appeared once, of which I am not at 

 all certain, and in which the line of the outer ellipse indicates the boundary 

 within which the lateral line is complete. This boundary has been crossed 

 many times and by different radiating lines. 



Fig. 1. — The genera Astyanax, and Moenkhausia, within the central oval, are closely allied and 

 the difference between them is partly bridged by the species M. intermedia. They are considered typical 

 and central for the subfamily. From Moenkhausia have radiated directly Markiana, Gymnocorymbus, 

 Tetragonopterus, and Entomolepis. It is left in doubt whether the genera Hemigrammus and Hyphesso- 

 brycon have diverged independently from Moenkhausia and Astyanax, and then converged till now only 

 the scaled and naked condition of the caudal separate them, or whether they have been derived from 

 one of the above and then developed or lost the caudal scales. The same applies to the genera Brycon- 

 americus and Knodus. The species Astyanax mutator and Hyphessobrycon proteus furnish bridges between 

 Hyphessobrycon and Astyanax. Moenkhausia colinho and Hemigrammus inconslans furnish a similar 

 bridge between Hemigrammus and Moenkhausia. From Astyanax many lines have radiated, some of 

 which have been continued beyond the line of the outer oval, which indicates that their lateral line has 

 become broken. Where a bridge exists between the so-called genera this has been indicated by extending 

 the name of the genus beyond its boundary or by supplying the name of the species which constitutes 

 the bridge in brackets. The origin of two of the radial lines is uncertain, a fact indicated by an interro- 

 gation point. 



