REVISION OF THE FISH FAMILY LIPARIDAE 15 



spines as there are unsegmented rays in the Hving species. If so we 

 may be deahng with the retention of a primitive condition of the 

 dorsal fin and the unsegmented rays may be homologous to or repre- 

 sent the spinous dorsal of the hypothetical ancestor. There are, 

 however, a number of serious objections to considering the unseg- 

 mented rays of the dorsal as primitive spines. These objections 

 are as follows: 



1. If the unsegmented rays represent spines we have the unusual 

 condition of the notch dividing the spinous dorsal. In no other 

 group in the Loricati do we find that, when only one notch is present, 

 it divides the spinous dorsal. We have seen, however, that in some 

 of the Cottidae two notches may be present and that the anterior 

 one divides the spinous dorsal. The loss of the second notch would 

 bring about the condition we are discussing the possibility of in the 

 Liparidae. It is possible that this is what has occurred in the latter 

 famUy. The presence of the notch in the middle of the unsegmented 

 rays, while casting some doubt upon the theory that these rays 

 represent spines, is not fatal to it. 



2. In the Cyclopteridae, a family which is generally considered to 

 be very closely related to and but slightly differing from the Liparidae, 

 the dorsal notch separates the spinous and soft dorsal. The number 

 of spines is small, eight or less, and about equal to the number of 

 undivided rays in the Liparidae. Considering the close relationships 

 between the Cylopteridae and Liparidae (some ichthyologists recog- 

 nize but one family) and the apparent stability of the dorsal notch 

 in the Loricati (if present in other families it divides the spinous and 

 soft dorsal), we should expect to find the dorsal notch bearing the 

 same relation to the spines and soft rays in these two families. The 

 agreement in the number of spines in front of the notch in both 

 families favors this view. If the spines in front of the notch in the 

 Liparidae are homologous to the spinous dorsal of the Cyclopteridae, 

 we can reasonably assume that the rays behind the notch in the 

 Liparidae represent the soft dorsal of the Cyclopteridae. Therefore 

 we can not consider the unsegmented rays behind the notch as repre- 

 senting ancestral spines, but should look upon them as having been 

 modified independently in regard to segmentation. 



3. The most important evidence contradicting the assumption 

 that all the unsegmented rays represent a primitive spinous condition 

 is obtained from a study of the anal fin. We find that the number 

 of unsegmented rays in the anal fin varies from 1 to 12 or 16. None 

 of the families of the Loricati have more than three anal spines. 

 These spines are absent from the Cottidae and Cyclopteridae. We 

 certainly can not homologize all the unsegmented anal rays with 

 spines, but are forced to explain the loss of segmentation of these rays 



