black matter continuous with that which covers the plates, and which seems to be the remains of the ruptured film or dermal envelope. In fact, in one or two places where the plates are more approximated than usual, the continuity of the connecting film is quite obvious.
Now comes the question, Are the whole of these plates and rods the component parts of an endo-skeleton, or do they partially belong to an exo-skeleton ?
At first sight one is disposed to look upon these plates and rods as the remains of an endo-skeleton in a more or less disturbed condition ; but on a closer examination a certain difficulty presents itself. There is an element or two too much, that are not found in the usual ichthyic skeleton. The sigmoidal plates and their rods (fig. l, j, j', i) are additional parts that are very perplexing ; had they not existed, there would have been no difficulty in the matter. The lozenge-shaped areas and the lateral plates in that ease might have been assumed at once to be a partially ossified vertebral column with the neural and haemal spines attached, and the hour-glass-shaped plates the fin-supports. But these redundant sigmoidal plates and rods must likewise be taken into account. What are they ?
We have to express our indebtedness to Prof. Huxley for an answer to this question.
All the Pycnodonts have supplementary spinous processes, which extend from the vertebral spines to the interspinous processes or fin-supports. The lateral rods, then, in Dorypterus would appear to be the homologues of these supplementary spinous processes. Such being the case, the only remaining anomalous feature to be explained is the sigmoidal plates. These, there can be little doubt, are dermal, and consequently do not belong to the endo-skeleton ; they are parts of an exo-skeleton, and therefore belong to the same category as the ventral plates, which are assuredly dermal. But this is not so clear with regard to the peculiar compound columns with which the latter are connected, though it would seem evident that the lower portion, which is continuous with the plates, should be so considered. The upper and larger portion, however, of the column, which is composed of numerous elongated rods articulated lengthwise with each other and with the upper prolongation of the plates, probably belongs to the endo-skeleton. These peculiar columns, in their compound structure and twisted appearance, closely resemble similar columns, which are considered to be ribs, in Pycnodus rhombus, Ag., and in which they are apparently joined to the vertebrae. In Dorypterus, too, the compound columns seem to have their upper extremities united to the lateral plates or neural spines. It would therefore seem probable that in this genus the upper portions at least of the columns are likewise ribs.
The great abdominal rods have a strong resemblance to the bone similarly situated in such fishes as Vomer, Zeus, and, according to Agassiz's restoration, also Platysomus : but in these it seems to be an enlarged development of an interhaemal spinous process ; while in Dorypterus, in which there appear to be two such rods, they seem
