Page:Quarterly Journal of the Geological Society of London, vol. 26.djvu/873

 to be developed in connexion with the posterior pair of ventral plates ; nevertheless they are probably internal bones.

On the whole, then, it seems pretty clear that these peculiar plates and rods in Dorypterus are the component parts of both an endo- and an exo-skeleton. The lozenge-shaped plates or areas are the vertebral centres, or rather the remains of them ; the lateral plates are the neural and haemal spines, and the rods in connexion with them are the supplementary spinous processes ; while the hour- glass-shaped plates are the interspinal processes or fin-supports, and the greater portion of the compound thoracic columns are apparently ribs. All these, as well as the great abdominal rods, are component parts of the internal bony skeleton.

The sigmoidal and ventral or thoracic plates, together with the upward prolongation of the latter, forming the basal portion of the compound thoracic columns, seem to be developments of the skin, and consequently belong to an exo-skeleton. The plates, too, that apparently correspond on the shoulder to the sigmoidal plates of the posterior portion of the body, are most likely also dermal ; and if so, the ridges in their transverse centre which form a line extending from the nape to some distance down the body are probably true mucus-tubes, which they so much resemble.

There is not much more to say with respect to this matter. We may observe, however, that in Dorypterus the ganoid type is modified ; and in this interesting form we see, for the first time in the geological series, a true thoracic, or rather a true jugular fish of the Linnean classification ; for the ventral fins are considerably in advance of the pectorals, and, indeed, they are placed as far forward as it is possible for them to be. Some change in the body- scales might therefore be expected, though the fins have the usual structure of those of the group to which this fish belongs; and the tail is decidedly heterocercal, with the lobes, which are deeply forked, of nearly equal length. And in the upper lobe there is a double row of the usual rhomboidal scales, resembling in their form and mode of articulation those on the tail of Acipenser Sturio, Linn. No other scales are observed on any part of the body of either of the four specimens, unless, indeed, the ventral and sigmoidal plates be so considered. We have, then, in Dorypterus a ganoid fish not only deprived of body-scales, but without plates of any kind except those just alluded to.

Having now given the result of a careful examination of all the four specimens at our disposal, collating and rigorously estimating the facts exhibited by each, we shall now give separate descriptions of the specimens themselves, marking emphatically the points for which each is distinguished.

The first specimen (Pl. XLIII. fig. 1) that we shall notice is from the cabinet of our friend Edward Wood, Esq., Richmond, Yorkshire. It is 4 inches long, including the tail, and is about 1-3/4 inch deep at the deepest part. The anterior portion is well preserved. The head is one-third of the entire length of the fish, exclusive of the tail- lobes, and is one-third higher than long ; it is arched in front, being