GEOLOGICAL SURVEY OF THE TERRITORIES. 419 



in the caudals one surface is much more deeply concave than the other, 

 one being funnel-shaped, and the other nearly plane in a few. 



A number of consecutive vertebras are preserved, which represent the 

 posterior portion of the caudal series. One of these is, fortunately, the 

 very extremity, and they demonstrate the tail to have been vertebiated 

 or heterocercal, less extensively than in Amia. On the anterior series of 

 three, the lateral grooves have disappeared from the centra ; the neural 

 canal is very small, and the spines are very massive and curved back- 

 wards, but much less than in the more posterior parts of the column ; 

 they are flattened, wider than deep, and in close contact with each other. 

 The anterior of the three, on the other hand, presents a narrowed edge 

 forward. The hasinapophyses are thin, and suturallyunited by a fiat 

 gomphosis. The terminal series embraces six vertebras, which have a 

 minute or obsolete neural canal, but hasnial canal distinct, but appar- 

 ently interrupted. The hasmal arches are united to the centra by a 

 rather smooth suture. 



The general direction of these vertebras forms a light upward curve. 

 The hasmal spines are fiat and laminar, and their margins in contact ; 

 they decrease in width and length to the end of the series. The neural 

 spine lies obliquely backward, and has a narrowed anterior ridge, but 

 stout shaft. 



An anterior hasinal spine in place exhibits a subglobular base, like an 

 articulation, and its shaft is wider than those posterior to it. The first 

 hasnial spine is a sub-triangular fiat bone, with neck and subglobular ex- 

 tremity, applies very well to a concavity between the anterior pair of pleu- 

 rapophyses, but does not in that position preserve contact with the ante- 

 rior margin of the second spine. One margin of the bone is thin and di- 

 vergent; the other expanded laterally and straight. The latter gives off 

 a transverse prominence like half a globular knob before reaching the ex- 

 tremity. Just within the latter are two large foramina, which are con- 

 nected with the extremity by a groove on each side, which meets in a 

 notch where the thin edge passes into the knob. 



Both sides of the neural and hasinal spines are concealed in this species 

 and in the 18. prognatlius by numerous parallel osseous rods, which are 

 somewhat angulate in section. They lie along the centra of the anterior 

 series of caudal vertebras, but are not to be found on vertebras of any 

 other part of the column. Numerous loose and fragmentary rods of the 

 same character accompany the loose and attached caudal vertebras, and 

 all of them, according to Professor Mudge, belong to the " posterior swim- 

 ming organ" of this animal. There is also a collection of these rods from 

 the anterior region of the body, which Professor Mudge thought occu- 

 pied the position of an interior limb. They do not any of them present 

 a segmentation such as would be exhibited by the cartilaginous radii of 

 caudal and pectoral fins, and their nature might have remained doubtful 

 but for the explanation furnished by the anterior compound ray or spine 

 of the posterior, probably caudal fin. This ray, as in the case of the 

 pectoral spine and first anal rays of some existing sil uroid or loricariid 

 fishes, is composed of a number of parallel rods closely united. Tbese 

 are in their distal portions remarkably and beautifully segmented, of 

 which a very simple form has been figured by Kuer, as existing in the 

 pectoral spine of the siluroid genus, Bangasius. This segmentation 

 becomes more obscure proximally, and finally disappears altogether, 

 leaving the spine and rods homogeneous. This portion of them is quite 

 identical with the rods found in the positions of fins already described, 

 and I therefore regard these as fin radii of the attenuated form presented 

 by cartilaginous rays of most fishes, but ossified sufficiently to destroy 



