37 GENUS SEBASTODES. 605 



of the parietals is of no significance. A more thorough 

 study has, however, convinced us that the species with 

 united parietals are related, and that the relationships 

 pointed out by Jordan & Gilbert are at fault. 



The value placed on such a cranial character as the 

 union or nonunion of the parietals need not be defended 

 here. It may only be mentioned that in mystinus, which 

 for other reasons we considered the hub to which the other 

 groups proposed here are related as spokes, the parietals 

 are united in 8 out of 10 specimens. The variation of 

 this character in mystinus but confirmed our view that it 

 is the radiating point. 



Leaving the parietals, the next prominent characters 

 are the development or nondevelopment of certain cranial 

 spines and ridges. These spines are found in all stages, 

 from minute points to comparatively huge spines. The 

 variation in size for this reason, if there were no other 

 objections, cannot be utilized for determining generic re- 

 lationship. The spines are very regularly arranged, and 

 in any given species certain ones are always present. 

 (Individual variations should of course be expected in 

 this character, as in every other, if a sufficient number of 

 specimens are examined.) The constancy of the presence 

 of certain spines in a given species warrants the use of 

 the presence or absence of these spines in the different 

 species in determining their true relationship. This rela- 

 tionship is usually borne out by a number of subsidiary 

 characters. Considering the constancy of the spines, 

 reinforced by subsidiary characters, we have divided the 

 species usually united under the generic name Sebastodes 

 as follows : 



a. Parietals meeting above the supra-occipital. 



b. Jaws equal; bead narrow above; high and prominent cranial ridges 

 ending in spines; preocular, supraocular, tympanic and parietals 

 present. Scales usually very strongly ctenoid; accessory scales 



