490 WILLIAM K. GREGORY 



These characters and the many similarities to the Salmonidae 

 incline Jordan to regard the Galaxiidae and Haplochitonidae 

 as Isospondyls. But they lack the mesocoracoid, and Boulenger 

 has consequently placed them with the Haplomi. In order to 

 express their wide differences from the typical Haplomi we set 

 them apart from the Esocoidea in a coordinate superfamily 

 Galaxoidea. Boulenger shows that the present range of the 

 group in Southern Africa, Australia, New Zealand, and South 

 America may be accounted for by the fact that the genus 

 Galaxias is not confined to fresh waters but occurs also in the sea. 



2. The Dalliidse or Alaska Black Fishes. These peculiar 

 forms, in which the coracoids are coalesced and cartilaginous 

 and the actinosts are represented by a longitudinally divided 

 and distally fringed cartilaginous plate, were set apart by Gill 

 ■as the order Xenomi; but their close relationship to the true 

 Haplomi has been demonstrated by Starks. 1 We may provision- 

 ally assign them to a separate superfamily, the Dalloidea. 



3. The Stephanoberycidae (Crowned Beryces). "This [abys- 

 sal] family has hitherto been placed near the Berycidae, among 

 the Acanthopterygii, but there are no spinous rays in the dorsal 

 and anal fins, and the ventrals formed of one simple and four 

 or five branched rays are abdominal" (Boulenger.) The air 

 bladder has a wide duct (Boulenger). Mr. Tate Regan (see p. 498) 

 regards this family as possibly related to the ancestors of 

 the Anacanthini or Cods. Gill suggests that the Stephanobsry- 

 cidae may be degraded berycoids in which the ventral fins have 

 lost their normal connection with the clavicle. 



4. The Percopsidae. The Sand-rollers inhabit the Great 

 Lakes, the rivers and streams of the northern Mississippi valley 

 ■and of Canada, and the Columbia River. This family (repre- 

 sented in the present fauna by only two genera, each with a single 

 species) is of extraordinary interest, since, according to Jordan 

 and Evermann, it is apparently derived directly from "the 

 extinct transitional forms through which the Haplomi and 

 Acanthopterygii have descended from allies of the Isospondyli. 

 The group shows the remarkable combination of true fin spines, 



1 "The Osteology of Dallia pectorahs." Zool. Jahrb., Bd. XXI, Heft 3 

 19.04, pp. 249—262. 



