RHAMPHOGNATHUS y 



genus, the gill-rakers are greatly reduced, the body is rather robust, the 

 mouth large, the ventrals inserted forward ; the dorsal fins are separated, 

 however, and the structure of the mouth is peculiar. In Rheocles, which 

 also inhabits the fresh waters of Madagascar, the dorsal fins are closely 

 approximated, the caudal is emarginate and the structure of the mouth 

 is normal, the body robust; both Bedotia and Rheocles thus show 

 relationships with Nannatherina. In the Melanot^niin^ of the fresh 

 waters of "Australia, New Guinea, and neighboring islands, the body is 

 typically robust, the ventrals inserted well forward, the dorsals closely 

 approximated, and the first dorsal spine usually pungent. In one genus, 

 Centratherina, the spines are all pungent; in Rhadinocentrus all 

 are flexible, as also in Telmatherina and Pseudomugil, genera which 

 show an approach toward the more typical members of the subfamily 

 Atherinin.e. The large group of genera comprising the Atherin- 

 OPSiN.E, and confined wholly to the Americas, are doubtless derived from 

 the Atherinin.e. Each of these groups of the Atherinid.e, however, 

 is specialized in its own way: Nannatherina and Bedotia have the 

 caudal rounded, and the premaxillaries little protractile ; Bedotia has the 

 mouth peculiar in structure ; Rheocles has few and very short gill- 

 rakers ; most of the Melanot-eniin.!!; have the anal fin greatly elongate 

 and the belly unusually short; the Atherinin.^ have the anus usually 

 inserted forward of its normal position ; the Atherinopsin^e have the 

 gape arched and the premaxillaries widely dilated posteriorly. Of all 

 the groups however, the Atherinopsin.e are the least perch-like, and 

 for that reason may be placed at the end of the series. 



Despite the wide range of variation exhibited by them, the numer- 

 ous genera of the Atherinid^ (Nannatherina perhaps excepted) 

 form a compact and obviously natural group. The few fossils known 

 appear to belong in or near Atherina or in one case perhaps approach- 

 ing Melanot^nia. 



Of the fossil Atherinid,^, not much is known. The type is not 

 ancient and may be as already suggested a comparatively modern oflf- 

 shoot from ancestors of the Apogonid^, or Ambassid.e. 



The European fossils from the Upper Eocene and Miocene referred 

 to Atherina seem to belong to that genus or to Hepsetia which as 

 fossil could hardly be distinguished. 



A second genus, Rhamphognathus Agassiz, has the jaws produced, 

 the snout ending in an acute point. In this form, as in Atherina, 

 the premaxillary border is straight, not curved as in Labidesthes and 

 American forms generally. Rhamphognathus paralepoides Agassiz 

 is from the Upper Eocene, at Monte Bolca, Verona. 



