GENUS SEBASTODES. 587 



S. serriceps is probably placed a little too high up in the 

 series, as its other characters indicate closer connections 

 with the last group. It is evident that the cranial charac- 

 ters do not furnish a basis for the division of the rock- 

 fishes' of the West Coast into several genera. All the 

 characters that are at all available for purposes of class- 

 ification serve remarkably well for arranging the species 

 in series, but the changes which those characters undergo 

 in the successive species are so perfectly graduated that 

 they cannot be used to break up the genus. Jordan and 

 Gilbert^ first grouped the species in 1883, using the num- 

 ber and degree of development of the cranial ridges and 

 spines as principal characters. Their arrangement not 

 only remains, but is more firmly established, with one or 

 two doubtful exceptions, by the remaining cranial char- 

 acters. 



Connected with this series of cranial characters and 

 their modifications are a number of other characters. Al- 

 though the correlations are not always exact, an arrange- 

 ment of species based on these external characters would 

 differ but little from that given above. 



Ayres long ago pointed out that " the border of the 

 caudal fin changes insensibly in the successive species 

 from the slight emargination of paiLcisfi7iis to the slight 

 rounding of nigrocinctns.^' In paucispinis the anal spines 

 are graduated, but this feature gradually changes in the 

 series until in the rosaceits group the second anal spine is 

 longer than the third. In the group represented hy paii- 

 cispinis and pt'nm'g-er the longest rakers on the anterior 

 limb of the first arch are relatively much longer than in 

 the group represented by rosaceus, etc. The decrease in 

 length is gradual in the series and is quite closely cor- 



^ Jordan & Gilbert: Synopsis of the Fishes of North America, 1883, pp. 

 652-678. 



