14 BULLETIN 150, UNITED STATES NATIONAL MUSEUM 



notch? In those species in which the notch is absent and all the 

 rays are divided have the primitive spines, such as are present in 

 front of the notch in Liparis, been lost or have they become modified 

 to resemble the remaining unsegmented rays? 



We can best approach the solution of the questions involved by an 

 examination of the ancestral or related families. In the Cyclop- 

 teridae the dorsal notch appears to separate the spines from the 

 soft rays. In some of the species the spinous dorsal is absent. The 

 notch is always present when the spines are present. When the 

 spines are absent it is very evident from the position of the first 

 dorsal ray and the number of rays that the spines have been lost 

 and not transformed into rays. In such forms the anterior dorsal 

 ray bears the same relative position in regard to the first anal ray 

 as the first ray of the soft dorsal bears to the first anal ray in the 

 species in which the spinous dorsal is present. Also the number 

 of dorsal rays in those species in which the spinous dorsal is absent 

 is the same as the number of rays in the soft dorsal of those species 

 in which the spinous dorsal is present. It also bears the same pro- 

 portion to the number of anal rays. The number of anal rays re- 

 mains fairly constant for both types of species. The dorsal notch 

 in these fishes disappears, not as the result of the lengthening or 

 shortening of certain rays, but as the result of their disappearance. 

 The number of spines is small, usually eight or less. In the Agonidae 

 also we find some species with the spinous dorsal absent. In the 

 Cottidae the spinous dorsal is always present. In Psychrolutes, 

 however, it is weak and buried in a ridge of skin. The number of 

 spines varies from 6 to 18. A notch marks the change from the 

 spinous to the soft dorsal. In some of the genera, as in Ohitonotus, 

 Hemilepidotus , Blepsias, and Hemitripterus , a second or anterior 

 notch divides the spinous dorsal. In the Scorpaenidae the number 

 of dorsal spines varies from 8 to 16. In the Hexagrammidae there 

 are as many as 26 dorsal spines. In this family the notch is some- 

 times absent, as in Pleurogrammiis . 



From the above review of the condition of the dorsal fin in the 

 principal families of the Loricati we are forced to consider the pos- 

 sibility that in the Liparidae all the unsegmented rays are ho- 

 mologous to spines such as are found in the Scorpaenidae. At least 

 the large number of unsegmented rays, 15 or more, is not fatal to 

 any such assumption. We have seen that species of both the Cot- 

 tidae and Scorpaenidae may have as many as 16 or 18 and the Hexa- 

 grammidae as many as 26 spines. These families are less modified 

 than the Liparidae and we may assume that a large number of spines 

 is a primitive condition for the Loricati. It therefore seems prob- 

 ably that in the ancestry of the Liparidae there have been as many 



