THE CYPRINODONTS. 9 



but reliance is not to be placed entirely on them. The same may be said of 

 the anal fin, or other features, when taken alone. By means of the anal the 

 family might readily be subdivided into three groiips : a first, in which the 

 anal is not modified on the male, a second, in which it forms a clasper with- 

 out an urogenital tube, and a third, in which a tube passes to the extremity 

 of the fin ; but in the first we should have Haplochilinos with conical teeth 

 and Cyprinodontinte with compressed tricuspidate, and in the second Gam- 

 busiinse, with conical, and Pceciliinas with compressed teeth, while in the 

 third would be placed the conical toothed Anablepinte with the compressed 

 tricuspidate toothed Jenynsiinas, both of which are still more widely separated 

 by the eyes and the ventral fins. To depend on the teeth alone would bring 

 together the Cyprinodontinoe, Jenynsiinse, and some PoeciliinoB, by the com- 

 pressed and fixed dentition ; by the conical shapes, Haplochilinse, Gam- 

 busiinas, and Anablepinae would be included ; and by the oar-shaped and 

 movable, only a part of the Poeciliinse would be placed. The unnatural 

 nature of such arrangements is sufficiently evident. 



Five or six branchiostegal rays are most common ; in a few types the 

 number is smaller. 



With the variety in the food, the gill rakers vary in the different genera 

 from short and tubercular to elongate. 



The greatest departure from the average vertebra is seen in Anableps 

 where each side of each segment of the column over the body chamber has 

 an elongate triangular process, grooved on the upper side, to the end of 

 which, the rib is attached. Great breadth of back with shortness and greater 

 strength of rib is secured in this genus by means of these processes. At the 

 base of the tail behind the terminal vertebra the processes forming the sup- 

 port of the central portion of the caudal are broad and fan-like posteriorly 

 throughout the family ; in some instances the hindmost pair anchylose, and 

 form a single broad expanse, in which the original lines of separation are 

 hardly visible. In number the vertebrae vary considerably. Anableps has 

 more than fifty ; certain species of Cyprinodon have hardly half as many. 

 A peculiar modification of several of the vertebrae is to be noticed on males 

 of some species, in which the anal fin is modified and carried forward ; an 

 inferior process from the centra of two or more of the vertebra? over tlie 

 hinder portion of the body cavity is sent down to furnish support for the 

 base of the transformed fin, Plate VIII. In Poecilia there are two of these 



stays ; in Gambusia there are two in one species, and three, with more or 



2 



