16 THE DISCOBOLI. 



to connect them each with the other and with the members of the large 

 group containing the Cottoids, Scorpaenoids, and allies, the cheek-boned 

 fishes {Pareiostichtht/es), all of which possess a common character in the 

 prolongation of one of the suborbitals backward toward the preoperculum. 

 It was not, however, until Putnam, 1874, began the study of the group, 

 that the full significance of this cheek bone was developed and the real 

 affinities with the Cottoids determined. The publication of the outcome 

 of his investigations, " Notes on Liparis and Cyclopterus," has brought 

 about a more accurate understanding of the relations of the group as a 

 whole, and of its position in the system. 



In the present study various comparisons have been made for the pur- 

 pose of determining the degrees of relationship existing between the 

 several families of the Discoboles, and also between them and allied 

 groups. There is nothing in the results obtained that declares with any 

 emphasis against the conclusion that the closest affinities are with the 

 Cottoids. Within the group it is found that the nearest approach is made 

 by the Cyclopteridae, and that the Liparididse have diverged more from 

 the progenitors common to these disk-bearers and the Cottidoe. Excepting, 

 in cases, the shape and the disk, the agreement in general features with 

 the latter is rather close. This is shown by comparison in the fins and 

 their attachments, particularly in the connection of carpals and pubes with 

 the claviculars, in the elements of the skulls (Plates IX. and X.), especially 

 in the suborbital stay and the opercles, in the branchihyals, in the teeth, 

 in the gills, in the stomach, in the caeca, and in the urogenital apparatus. 

 The general correspondence is very noticeable in regard to the glossohyal 

 and basibranchial elements, which are either suppressed in part or greatly 

 reduced in size. The apparent difference in shape and the softness of the 

 bones notwithstanding, it is Cyclopterus that is to be placed nearest to 

 Cottus. Without going far into details, reasons for this will be found by 

 comparing the opercular bones and the expansion and connection of the 

 suborbital process (Plate X. Figs. 28, 30, 32, 33, 19"). In the Liparididce 

 the other opercular bones are more reduced, while the preoperculum is 

 more expanded in the upper limb, and the suborbital process is more 

 slender and crosses the preoperculum to its hinder edge. Better agree- 

 ment also exists between the brain of Cyclopterus and that of Cottus; 

 in both the brain is shortened, and cerebellum and hemispheres are small 

 as compared with the optic lobes. Liparis has a more elongate brain, and 



