128 
bulletin of the bureau of fisheries. 
of the Anguilla type, the structure, with its upper and lower layers, being much the same; but the 
upper layer exhibits more or less quadrangular spaces, due to the crossing of the radial and circular 
systems. The specialization of detachable oval plates does not exist, but it is possible to see how it 
might arise from a type resembling Synaphobranchus . 
Although I have found nothing among the primitive fishes which seemed to lead toward the eels, 
there is a distinct resemblance, I believe not altogether superficial, to be observed in the scales of the 
caeciliid amphibians. I am greatly indebted to Dr. Boulenger for material of Ichthyophis glutinosus 
(see text fig. 7). Here the elongate quadrangular spaces are arranged like bricks in a wall, but in con- 
centric circles, while the center of the scale is reticulate and minutely tuberculate, quite closely simu- 
lating the condition in the Lepidosirenidae. Thus in a remote and general way, Ichthyophis scales 
connect tliose of the eels with those of primitive fishes, suggesting that the eel type is really a very 
ancient one, not derived from any of the groups of fishes nearly associated with it in the classification. 
In this connection the extinct Archencheli, considered primitive eels with isospondylous traits, 
become very important. I have never seen Urenchelys, but Dr. A. S. Woodward says of it, “scales 
rudimentary”; and under U. avus A. S. Woodward, he says of one specimen, “there are no traces of 
scales,” but of another, “the brownish impression of the soft parts of the fish is curiously mottled, as if 
there were indications of small, rudimentary quadrangular scales.” Looking in the opposite direction, 
we find an unexpected revival of the ichthyophoid type of scale, at least as regards the quadrang;ular 
areas, in the gadid Brosmius brosme, the cusk. The fine radii are broken so tliat, as in Ichthyophis, the 
bricklike spaces are largely laid in alternate fashion, not one directly above the other. In the pol- 
locks, Pollachius and other genera, however, the radii are essentially continuous, and one can hardly 
doubt that the Brosmius scale was derived from a Pollachius-\\ke. type. In this coimection it is worth 
while to note that Sphyrcena picuda has a scale pattern practically identical with that of Pollachius. 
All things considered, it must be admitted that the eel type of scale could have been derived from 
something of the Sphyrocna or gadid pattern, and hence it is not necessarily primitive. The resemblances 
noted are, however, interesting, and may not be without some significance. 
According to Boulenger, the scales of the Murasnolepididse resemble those of the Anguillidae. 
Order OSTARIOPHYSI. 
Suborder Heterognathi. 
CHARACID^ (CHARACINID^). Characins. 
The scales of the African species have been discussed in Smithsonian Miscellaneous Oallections, 
volume 56, no. i. Those of the South American genera form the subject of a lengthy paper to be 
published by the Carnegie Museum. This also includes the Er5rthrinidae. 
Suborder Eventognathi. 
CYPRINID.®. Minnows and carps. 
The scales of this family have been discussed in numerous papers. I give here a list of the genera 
which I have examined, with references to literature and some additional information and figures. 
(a) AMERICAN GENERA. 
It is doubtful whether any genera of Cyprinidas are common to America and the Old World, 
except, of course, those introduced by man. 
Abramis auctt. Amer., the sh.meT=Noiemigonus. 
Acrocheilus Agassiz, the chisel-mouth. Proc. Biol. Soc. Wash., xxii, p. 210 (subfamily Chondros- 
tominse). 
