42 U.S. NATIONAL MUSEUM BULLETIN 284 



ening of the ascending processes of the premaxillaries and a carnivorous 

 habit. In contrast, the Salariini have some portion of the wings of the 

 lateral ethmoids anterior to the median ethmoid (the median ethmoid 

 forms the posterior border of a depression bounded laterally by the 

 lateral ethmoids) and the median ethmoid is not visible (fig. 3, pi. 10) 

 when the skull is viewed dorsally. This condition reaches the extreme 

 in some species of Ecsenius (pi. 10), wherein the lateral ethmoids 

 almost encapsulate the median ethmoid and the dorsal end of the 

 latter may be forced back into the orbital region. The condition in 

 Ecsenius appears to be the result of extreme shortening of the para- 

 sphenoid, with an associated ventroposteriorly curving of the lateral 

 ethmoids to compensate for the shortening. In Andamia, the only 

 exception in the Salariini examined, the median ethmoid is the most 

 anterior skull bone and is obvious when the skull is viewed dorsally 

 (pi. 11). The condition in Andamia appears to have developed as a 

 result of a depression of the frontals in the orbital region. 



The relative size of the median ethmoid is much smaller in the 

 Salariini than in the other blenniids, and this reduction is associated 

 with the extreme weakening of the ascending processes of the pre- 

 maxillaries and the peculiar nature of the teeth, which in turn are 

 associated with a primarily herbivorous, grazing habit. 



In the Blenniinae, as in many perciforms, there is a sphere of 

 cartilage (rostral cartilage) attached to the posterior edges of the 

 distal ends of the ascending processes of the premaxillaries. This 

 cartilage was not seen in the Nemophidinae. The purpose of this 

 cartilage is not clear, but in blenniids it seems to serve as a cushion 

 between the ascending processes and the median ethmoid. In some 

 species of a few genera, an endochondral ossification develops in 

 the rostral cartilage. Such an ossification has not been described 

 for other perciform fishes, but for the cyprinoids Harrington (1955) 

 proposed the name "kinethmoid" for a bone in relatively the same 

 position as the blenniid rostral ossification (since previous names 

 for the cyprinoid kinethmoid, particularly "rostral," have been 

 applied to other bones, their usage might result in confusion). The 

 kinethmoid of the cyprinoids functions in the mechanism for pro- 

 trusion of the premaxillaries, but since the premaxillaries of blenniids 

 are not protrusible, the function must be different. Although it is 

 not at all certain that the cyprinoid kinethmoid is homologous with 

 the rostral ossification of blenniids, I am provisionally appropriating 

 the name "kinethmoid" for use in the blenniids. Harrington used 

 the name to denote a movable bone in the ethmoid region, and, 

 in the sense of a movably attached bone, the name can be applied 

 also to blenniids. 



