GREGORY: FISH SKULLS 219 



curved jaws and fine cardiform teeth arranged in many rows on the premaxillse and den- 

 taries. These premaxillas are movably hinged on the mesethmoids and vomer. The 

 maxillae are produced downward below the level of the dentaries and serve as levers for 

 depressing the premaxillse. The skull agrees in most essentials with that of Fundulus. 



Another acanthopterygian resemblance, noted by Hubbs, is to be seen in the form and 

 arrangement of the branchiostegal rays: "The MicYocyprini (Poeciliidse and Amblyopsidae) 

 were long confused with the Haplomi, but have recently been shown to have a more 

 advanced organization. The structure of the branchiostegal rays in the two groups con- 

 firms this view; those of the Haplomi are quite like those of the Isospondyli, whereas those 

 of the Microcyprini are similar to those of the Acanthopteri. In the Poeciliidae there are 

 six, or fewer, branchiostegals, which are folded up behind the opercular and above its 

 lower margin. The upper four saber-shaped rays are attached to the outer surface of 

 both the ceratohyal and epihyal, postero-superior to the prominent angle of the hyoid arch; 

 the lower rays arise from the inner face of the ceratohyal." (Hubbs, 1919, p. 67.) 



In his very thorough investigation of the osteology of the order Haplomi, Starks (1904fl, 

 pp. 258-261) records the many technical characters which distinguish the "Superfamily 

 Poeciloidea " from the Esocoidea, or true Haplomi, as well as from the Amblyopsoidea. The 

 ethmoid in the Poeciloidea is represented by a nearly circular scale of bone not unlike that 

 seen in the Synentognathi (see also Starks, 1926a, pp. 206-211). There are no "proeth- 

 moids." 



The palatopterygoid arch is toothless; "posttemporal is attached to the epiotic without 

 the intervention of a ligament; supraclavicle is a very small scale of bone, scarcely sufficient 

 to separate the posttemporal from the clavicle . . ." (Starks). Starks also shows that the 

 three subfamilies of the Poeciliidse differ widely from each other in skull characters (1904a, 

 pp. 259-261). 



The lower pharyngeal bones in Fundulus similis are "joined to each other by a deeply 

 dentate and interlocked suture, together forming a triangular bone with concave sides, 

 covered with teeth similar to those above." In Fundulus heteroclitus the opposite pharyn- 

 geals were elongate, attached only at their anterior ends and diverging posteriorly. In 

 other species intermediate conditions were observed (Starks, 1904a, p. 259). Similarly in 

 Fundulus similis the three superior pharyngeals on each side are "joined (not ankylosed) 

 to form an ovate plate," covered with molar-like to conical teeth; in Poecilia the elongate 

 pharyngeal teeth, both upper and lower, are bristle-like. In Amblyopsis the upper pharyn- 

 geals are separate and bear normal teeth and the lower pharyngeals are but loosely in 

 contact in the mid-line. Hence in these various Microcyprini we may observe the steps 

 by which tooth-bearing pharyngeals approaching the synentognath type are being evolved. 



In conclusion, the Microcyprini seem to have paralleled the true Acanthopterygii in 

 a number of characters, including the predominance of the premaxillse, the exclusion of the 

 maxillse from the gape, the loss of the orbitosphenoid (Regan, 1909a, p. 78; 1911c/, p. 321), 

 the loss of the mesocoracoid arch, the form and arrangement of the branchiostegals 

 (Hubbs). On the other hand, they have lagged behind the Acanthopterygii in retaining 

 solt-rayed fins, abdominal ventrals, unarmed opercles, etc. 



According to Starks (1904(2, p. 254), Umbra (of the true Haplomi) is certainly nearer 

 to the family Poeciliidse than it is to any family in the order Iniomi, while Tate Regan 

 says (1910a, p. 7) that "the relationship of the Cyprinodontidas to the Esocidae is generally 



