Cockerell — The Scales of the Blennioid Fishes. 91 



former ctenoid features. There is indeed a certain resemblance in 

 structure between the scales of Genypterus and Anarrhichas, but it is 

 probably not fundamental. Considering the scale-structure, together with 

 the osteological characters mentioned by Regan, it seems that theOphidii- 



formes should stand against the other two divisions combined. Thus the 

 Blennioidea will include the Blenniiformes and Cliniformes, while the 

 Ophidiiformes will go in a distinct suhorder or superf amily Brotuloidea. 



Mr. Regan expresses the opinion (in litt. ) that "the Gadoids seem to 

 have sprung from some more primitive group [than the Percoids], such 

 as generalized Iniomi (Aulopidse) or Berycoids (near Polymixiidse)." It 

 is recognized that the Brotuloidea are quite distinct from the Gadoids, 

 but may they not represent still another independent branch, arising 

 from fishes more specialized than the immediate ancestors of the Gadoids, 

 yet without the ctenoid scales of the typical Percoids? It should be added 

 that the scales of Polymixia are ctenoid, and are as different from those of 

 the Gadoids as possible. I do not know the scales of Aulopidae, but those 

 of the Synodontidse (Trachinocephalus and Synodus) show no Gadoid 

 features. 



Postscript.— Since the above was written I have received through the 

 kindness of Dr. Edwin C. Starks scales of Dactylagnus mundus (Car- 

 men I., Gulf of California ; Albatross collections), a representative of the 

 Dactyloscopidee. This family is placed by Regan in the Cliniformes, but 

 has been referred by other authors to quite different groups. The scales 

 are about 2 mm. long and 2% broad ; nucleus considerably basad of the 

 middle; radii basal, very numerous (over 40), extending to the sides of 

 the base, only the middle ones reaching the nuclear area ; basal margin 

 not or hardly crenulated; basal circuli very numerous but irregular, cut 

 by the radii; nuclear area, and the region basad of its level, finely orna- 

 mented with a pattern curiously simulating a finger-mark, composed of 

 variously broken and anastomosing circuli; in the nuclear field producing 

 irregular reticulations, but apically largely resolved into dots, the trend 

 of the circuli mainly longitudinal, but spreading laterally, pointing at an 

 angle of about 45° to the margin ; apical margin of scale very thin, easily 

 frayed or broken, finely granular. The lateral line canals are forked 

 basally, forming a Y with very short arms and a long stem. By the 

 position of the nucleus, this is very different from Labrisomus, but the 

 general structural features are not dissimilar, and so far as it goes, the 

 scale of Dactylagnus confirms the position of the fish in the Blennioidea. 

 The Dactyloscopida? have been placed near the Uranoscopidse by Gill. 

 Dr. Starks sends scales of Uranoscopus scaber which he collected at 

 Naples, Italy. The very small (l-l^ mm. long) scales are covered by 

 skin and scarcely at all imbricated. They are elongated, with the shape 

 of an oyster shell, the apex obtusely pointed, the base broadly rounded. 

 The nucleus is very near the apex, and from its vicinity run about ten 

 long, partly broken, basal radii. The circuli are coarse and dense, not 

 only interrupted by the radii, hut considerably broken up in the apical 

 field. This scale is entirely diverse from that of Dactylagnus, and does 

 not suggest that of any Blennioid. 



