observations on fish scaees. 
149 
Holocentriis xantherythrus Jordan & Evermann. Hawaii. 
Holocentrus microstoma Gunther. Samoa. 
The species of Holocejitrus are so uniform in their characters that I can not construct a satisfactory 
key. There are, however, minor differences; e. g., the apical teeth of H. xantherythrm are much coarser 
than those of H. laticeps, and the basal radial folds are more prominent in H. diadema and laticeps than 
in the others. 
Finding that H. laticeps was not recorded from the Hawaiian Islands, I asked Dr. Garman to look 
at the specimens (M. C. Z., 3440). This he kindly did, and reports that the identification is correct. 
Sauvage, in his work on the fishes of Madagascar, has figured scales of Myripristis seychellensis, 
M. pralinus, M. borbonicus, Holocentrus (or Holocentrum) spiniferus, H. diadema, H. leo and Holotrackys 
lima. These show various peculiarities, but ail confirm the essential uniformity of the holocentrid t}rpe, 
and indicate its general affinity with that of the Berycidae. 
There is an evident resemblance between the holocentrid scale and that of Aphredoderus. 
POLYMIXIID^. Barbudos. 
From the United States National Museum I have scales of Polymixia japonicaSteinAacim&r, from 
Japan (fish 7X inches long). They are verj* broad, with the same general shape and apical spines as 
Holocentrus , but the basal circuli are very much less dense, and the basal radial folds are very well 
marked, throwing the basal margin into numerous strong imdulations. The laterobasal comers are less 
prominent than in the holocentrids, and the subapical region is without spines. Thus the BeryCoidea, 
as regards their scales, form a sufficiently compact group, quite isolated from the groups standing on 
either side of it in the system. 
Suborder Percomorphi. Perciform fishes. 
Superfamily SCOMBROIDEA. Mackerel-like fishes. 
Broadly speaking, tire families Scombridse, Carangidse, Cheilodipteridse, Stromateidse, Hemirham- 
phidse, Scombresocidse, Belonidae, and Exocoetidae maybe ranged together (and apart from the percoid 
series) on their scales. Atherinid® show some resemblances also. The fishes are of course variously 
diverse in other characters. 
SCOMBRID^. Mackerels. 
I have figured the scales of Scomber in Smithsonian Miscellaneous Collections, volume 56, no. i. 
They are small and thin, broader than long, with the circuli transverse. In the common mackerel. 
Scomber scombrus (Einnseus), from Woods Hole, the apical margin is more or less, but very irregularly, 
dentate, without the formation of definite stmctmres like those on genuinely ctenoid scales. At these 
teeth the circuli are bent upward and angled, and this irregular waviness and angulation is seen here and 
there in the subapical field, indicating no doubt the teeth of an earlier stage. The fusiform area repre- 
senting the nucleus is below the middle of the scale, and on each side of it the circuli are acutely angled. 
The circuli above the nucleus are somewhat denser than those below. The angulation of the lateral 
circuli in 5 . scombrus is like that of the basal ones in Macrurus. 
In the bonito, Sarda sarda (Bloch), and the frigate mackerel, Auxis thazard (Eacepede), the scales 
are of the same general type, but still smaller and more degenerate, often practically without sculpture. 
Those of Auxis are often pointed laterally. 
CARANGIDdJ. Cavallas, etc. (PI. xxxvi, fig. 27.) 
The jack, Caranx hippos (Einnseus), has circular scales about 2^2 mm. diameter, with central nucleus 
and fine concentric circuli, not unlike some salmonid. There are no radii or marginal teeth. The inner 
circuli are variously angled laterally, and in the young the scales are transversely oval, and look just 
like those of Scomber. The carangid and scombrid scales are therefore very closely allied, although when 
adult appearing different. 
