16 BULLETIN 150, UNITED STATES NATIONAL MUSEUM 



as a newly acquired condition. If we are forced to explain the lack 

 of segmentation of the anterior anal rays in this manner, it would be 

 illogical to explain the condition of the unsegmented dorsal rays in 

 any other manner without the support of very strong evidence. 

 Such evidence is lacking except as regards the rays in front of the 

 dorsal notch. 



The dorsal rays in front of and possibly forming the notch appear 

 to be homologous to spines. They differ from the unsegmented 

 rays behind the notch in being undivided and rodlike. In the Cyclop- 

 teridae and Cottidae the notch separates the spines from the soft 

 rays. We should expect to find the same condition in the Liparidae. 



We may now discuss the fate of the undivided rays or spines in 

 those species in which the notch is absent. We have seen that in 

 the Cyclopteridae the absence of the dorsal notch indicates that the 

 spines have been lost. The same is true of the Agonidae. In these 

 two families the notch disappeai-s only with loss of the spines. But 

 in the Liparidae the notch may disappear and the spines be retained. 

 This becomes very evident when we examine species like Liparis 

 dennyi, in which the notch is faintly indicated or absent. In the 

 species of Liparis, in which the notch is absent, it appears that the 

 anterior rays remain undivided and spinelike. The absence of the 

 notch in this genus does not indicate that the spines have been lost. 



The fate of the undivided rays in Careprodus and Paraliparis 

 is not so easily solved. In these genera, with a few exceptions, 

 the dorsal notch is absent and the anterior rays are divided.* The 

 absence of the notch in these species, judging from what has occurred 

 in Liparis, does not indicate that the spines have been lost. There 

 are two other criteria that we may use in determining whether the 

 spines have been lost or become divided. These are the position of 

 the origin of the dorsal and the number of rays. 



The position of the origin of the dorsal in Careprodus and Para- 

 liparis favors the view that the spines have not been lost. In the 

 species of Cyclopteridae and Agonidae, in which the spinous dorsal 

 is absent, the origin of the dorsal is proportionally farther back on 

 the body. If the spinous dorsal has been lost in Careprodus and 

 Paraliparis we should expect to find the origin of the dorsal farther 

 back than in the species of Liparis, in which we know the spinous 

 dorsal is present. The base of the spinous dorsal in Liparis is shorter 

 than in the Cyclopteridae, but if this fin disappeared there would be 



* We assume here that in the ancestry of these genera a dorsal notch and spines were present. In two of 

 the linown species of Careprocttis a dorsal notch appears to be faintly indicated. The condition of the rays 

 in these species has not been examined. We are justified in assuming, I think, that in the ancestry of these 

 genera a dorsal notch and spines were at one time present. 



