MORb*nor,OGY 01'' TEl.KOSTBAN HEAD SKELETON. 575 



well marked off from the Percesoces and other Telocephali, it 

 descended from the ancestors of these groups. In this they 

 are followed by Kingsley (00), who adds to it the extinct 

 Dercetidfe (Hoploplenrida3) which "show relations towards 

 Belone." This year Boulenger (01, p. 378) remarks on 

 their close affinity with the Lophobranchii, more especially 

 through the Fistularid^e and the fossil Pseudosyngnathus. 



It may at once be stated that there is nothing to support 

 Parker's (68) suggestion of Siluroid affinity for Gasterosteus. 



In order to see liow far these views are supported by the 

 study of the head skeleton, we may now briefly compare that 

 of Gasterosteus with that of Syngnathus, a typical Lopho- 

 branch, and of Fistularia, a typical Hemibranch. McMurrich's 

 paper (83), supplemented and confirmed by uiy own observa- 

 tions, is my authority for details concerning the former. 

 Klein (84, 86) on the cranium, and Rutter (v. Jordan and 

 Evermann) on the branchial skeleton, are the only 

 workers who, so far as my knowledge goes, have dealt with 

 the latter. To attempt to explain the why and the wherefore 

 of Klein's tangle in describing the auditory region would be 

 of no use for »ny present purpose ; I shall therefore give the 

 results of my own examination without reference to his. 



Taking the various parts in the same order as in the former 

 part of this paper, the consideration of the hinder region 

 of the cranium comes first. 



That of Gasterosteus is not so compressed dorsi-ventrally as 

 the other two, but all are alike in the absence of an opistholic 

 and basisphenoid, the even upper surface, the sculpturing of 

 the roofing bones, the simplicity of the post-temporal, the 

 essential shape of the ethmoid, and the great size of the 

 supra-occipital, which separates the parietals widely, and 

 appears to separate the hinder portion of the frontals. In 

 Gasterosteus the exoccipital extends forwards between the 

 pterotics and basioccipital to the pro-otic. In the others the 

 pterotic extends ventrally to the basioccipital, and also part 

 of the way into the large membranous space between this and 

 the pro-otic, thus separating the exoccipital widely, and the 



