^^°l89^o"''] PROCEEDINGS OF THE NATIONAL MUSEUM. 363 



pair of compressed epipharyngeaJs ; hypopharyngea's divergeut and rather 

 compressed. There are three and one-half gills, that is,doiiblebranchiiB 

 on all the arches except the fourth which has a single row of filaments. 

 There is no fissure behind the fourth arch. 



RELATIONSHIPS. 



If we now compare the Cyclopteroidea thus defined with other groups, 

 it is only with those of superfamily Cottoidea that we find many common 

 characters. The two groups agree with each other, and differ among 

 themselves as follows: 



Cyclopteroidea. Cottoidea. 



Third infraorbital boue developed as a "stay," obliquely crossing the cheek and con- 

 uectiDg with the preoperculuiu. 



Myodome suppressed. Myodome more or less developed. 



Post-temporal bifurcate and normally couuected by squamous suture with the 

 cranium, 



Actinosta enlarged and mostly connected with the inner ledge of the proscapula, the 

 hypercoracoid being dislodged upwards and the hypocoracoid downwards on a row 

 with the four actiuosts. 



Ribs sessile on the vertebral centra or h?emapophyses. 



Pharyagobranchials reduced to the enlarged epipharyugeals (homogolous with the 

 third of typical Acanthopterygians). 



Ventrals modified to form a suctorial ; Veutrals normally developed, entirely 



separate, and each composed of a spine 

 and (1-5) articulated rays. 



disk supported by six immovable rays on 

 each side converted into osseous tissue 

 and without articulations (etypically 

 suppressed). 



When it is remembered that all the characters which are common to 

 the Cottoideans and Oyclopteroideans differentiate the Oottoideaus from 

 one or other of those families to which they have been always approxi- 

 mated, the importance of such an aggregate of common characters must 

 be admitted. There can indeed be no doubt that the Cyclopteroideaus 

 are more closely related to the Cottoideans than to any other known 

 fishes. In fact, the former differ from the latter mainly in the sup- 

 pression of the myodome and the modification of the ventral apparatus. 

 It is true that these differences are supplemented by others, but, on 

 the other hand, the characters enumerated as common to the two are 

 also supplemented by others. There is indeed some justification for 

 Professor Putnam's belief that the difference between the Liparidid(c and 

 Cyclopteridware as important as those between either and the CoUidae. 

 I can not, however, admit that the structure of the ventral fins is of as 

 little importance as Professor Putnam urges.* 



In view of the radical modification of the ventral rays and the whole 

 pelvic apparatus, "the discovery of a representative of either family 

 with ventral fins of the ordinary form " is almost inconceivable, and, if 

 discovered, such a form would doubtless be found to have other modi- 



* See antea, p. ofil. 



