EE VISION OF THE FISH FAMILY LIPARIDAE 13 



The number of dorsal rays varies from about 28 or 30 in some 

 species of Liparis to 65 or more in species of Paraliparis. An exam- 

 ination of the table on page 35 indicates that the number of dorsal 

 rays is typically increased in the species inhabiting the deep water. 

 The number of rays varies in Liparis from 28 to 48, in Careproctus 

 from 39 to 61, and in Paraliparis from 48 to 66. 



The anterior dorsal rays are primitively connected with the fin 

 membrane, and the tips project above as in the pectoral fin. In 

 many of the deep-water species, especially in Careproctus, the anterior 

 rays are buried in tissue beneath the skm and free from it. Such rays 

 are weak, and their tips frequently extend, undulating, backward to 

 the succeeding ray. 



All the dorsal rays are simple; that is, unbranched. A varying 

 number of the anterior rays are also unsegmented. In the species 

 examined the number of unsegmented rays varies from 6 to 18, and 

 appears to be fairly constant for each species. The species of the 

 more highly modified genera do not appear to have more unsegmented 

 rays than the species of Liparis. (See following chart.) Some of 

 the anterior rays in some of the species, especially those of Liparis^ 

 in addition to being unsegmented, are undivided, the two shafts or 

 lateral halves of the ordinary ray being united to form a single rod. 



The relation between the dorsal notch and the unsegmented and 

 undivided rays is of some interest. The number of unsegmented rays 

 usually, if not always, exceeds the number of rays in front of the 

 notch. In some species only one or two of the rays behind the notch 

 are unsegmented, while in other species there are as many or more un- 

 segmented rays behind as in front of the notch. The number of 

 undivided rays is always less than the number of unsegmented rays. 

 It appears that the notch, in some species at least, marks the division 

 between the undivided and divided rays, the rays in front of the 

 notch consisting of a single rod and therefore spinelike. 



In all the species in which the dorsal notch is present the rays in 

 front of it appear to be unsegmented and undivided. We will now 

 describe the condition of the anterior rays in those species in which 

 the notch is absent. As previously mentioned, the anterior rays are 

 always unsegmented. The question of their being undivided in 

 those species in which the notch is absent has been investigated in 

 but a few species. In Liparis tunicatus and tanaTcae the anterior 

 dorsal rays are undivided and spinelike. This may represent the 

 condition of these rays in all the species of this genus. In Care- 

 proctus and Paraliparis, more degenerate genera, some of the species 

 have all the dorsal rays divided. 



The significance of the dorsal notch and the unsegmented and 

 undivided rays may now be considered. Does the dorsal notch 

 mark the division between the ancestral spinous and soft dorsal? 

 If so what is the significance of the unsegmented rays behind the 



