OBSERVATIONS ON FISH SCALES. 
141 
indeed to those of such cobitids as Misgurnus and Lepidocephalichthys . They may also be compared 
with the small-scaled catostomids, as Pantosteus, or in Cyprinidse with Oreinus. This general type of 
scale, with radii all around and a sort of latticework pattern, is, I believe, really primitive, with char¬ 
acters not very distantly approaching those of the amphibian Ichthyophis. It appears, therefore, that 
the gymnotids are not “degraded characinids” (Boulenger), but rather specialized pre-characinids. 
The fully developed radial pattern is found in the small-scaled gymnotids, the scale being nearly cir¬ 
cular ( Gymnotus carapo ), or elongated ( Sternopygus macrurus), or transversely oval ( Hypopomus artedi). 
In Eigenmannia virescens more than half the scale is without radii, and if they all disappeared we should 
have a scale not very unlike that of the characinid Serrasalmo. 
In the larger scales of Porotergus and Sternarchus the radii are very fine and the basal ones converge 
to the middle line, curiously simulating the condition found in some Gadidae, as Urophycis. 
Order SCYPHOPHORI. 
For the Mormyridae (including Gymnarchidae) see Smithsonian Miscellaneous Collections, volume 
56, no. 3. In Jordan’s list the Haplochitonidae andGalaxiidae appear as Scyphophori, but this reference 
is opposed to the treatment in the earlier part of the book, and according to Regan (1911) they are 
salmonoids. 
Order HAPLOMI. 
ESOCIDAJ. Pikes. 
Esox (or Lucius) lucius from Toledo, Ohio (Rutter), has oblong scales about 3.5 mm. long and 
slightly over 2.5 wide; nucleus anterior to the middle; no apical radii, but two or three very strong 
basal radii or plicae, ending between the large lobes of the basal margin. The circuli are dense, the apical 
ones distinct and transverse, with no approach whatever to any ctenoid structure. The base of the 
scale is nearly as in Synodus. (PI. xxxv, fig. 20.) 
UMBRIDi®. Mud Minnows. 
Umbra limi (Kirtland) from Milwaukee County, Wis. (Graenicher), has peculiar subquadrate or 
oblong scales, the largest about 3 mm. long, wholly unlike those of Esox. There are no radii, nor is the 
basal margin at all lobed. The very coarse circuli are essentially longitudinal, extending from one end 
of the scale to the other, but the innermost slant toward the middle line, meeting those of the opposite 
side at an extremely acute angle. In the largest scales the apical circuli are not angled in the middle, 
but normally arched, becoming transverse in the middle; the basal circuli in the same scales converge 
to two or three points, producing a transversely zigzag appearance. In Kneria the system of circuli is 
very much the same, though differing in detail. Much more noteworthy and surprising is the great 
resemblance in structure to the scales of Urophycis regius. So far as the scales go, Umbra and Urophycis 
might be imagined to belong to closely allied genera. Dr. Gill (Smiths. Misc. Coll., vol. 45, p. 297) 
has given a rather unsatisfactory figure of an Umbra scale. 
PCECIILIDjE. Killifishes. (PI. xxxv, fig. 21, 22; pi. xxxvi, fig. 23.) 
Scales broad, more or less semicircular, cycloid, with the nucleus apicad of middle and strong basal 
radii. No apical radii; circuli not, or not very, dense. Tilapia (Cichlidae) has the same sort of scale, 
thus quite different from that of Pomacentridae ( Eupomacentrus and Abudefduf examined), which is 
ctenoid, of the ordinary Acanthopterygian type.® Compared with Esox or Lucius the Pceciliid scale 
differs by its broad form (the scale of Lucius is longer than broad), numerous basal radii (two or three in 
a Many Cichlidae have ctenoid scales, and no doubt Tilapia is secondarily, not primitively, cycloid; a parallel development 
to that of the Embioticidae. In fact, it is not quite correct to say that Tilapia is cycloid, since extremely minute marginal teeth 
can be detected with the compound microscope. 
