272 6 



and third vertebrae possess an elongated part, directed forwards (PI. 1, fig. 10); that 

 of the second vertebra projects a little (PI. I, fig. 9, 10*) and is connected by a strong 

 ligament with the posterior end of the postteniporal. 



On examining the lateral aspect of the coalesced vertebrae (PI. I, fig. 10, PI. II, 

 fig. 10) we find the sutures partiallj' repeating the features characteristic of the fol- 

 lowing free vertebrae: in the small bend (a) looking forwards we recognize the part 

 lettered in the same way on the following free vertebrae (PI. II, fig. 10) and on the 

 front end of the first of the fused vertebrae; further, the process (b) of the latter, 

 which is connected with the skull, will be found to correspond with the process (b) 

 on the free vertebrae, that is to say, it must be regarded as an articular process, 

 not as a transverse one, and the first vertebra has thus only developed the posterior 

 part of the transverse process (t). 



The nerve-openings observable in side view (n') are in a position about corre- 

 sponding to that of the ventral ones (n, Pi. I, fig. 9). 



All the following, free, abdominal vertebrae possess double transverse 

 processes (PI. II, fig. 10), i.e. there is an anterior and a posterior process on each 

 side (or it may be put in this way: there is one transverse process of the same 

 length as the whole vertebra but divided through a deep incision into an anterior 

 and a posterior part). A posterior process on one vertebra joins with its hind 

 margin the front margin of the anterior process on the following vertebra etc.; the 

 two adjoining processes, belonging to two different vertebrae, are almost of the 

 same length on the foremost vertebrae; but about from the 17th vertebra the 

 posterior element (i. e. the anterior pi-ocess of the I7lh vertebra) grows a little 

 longer than its fellow (the posterior process of No. 16), and on the 20th — 24th vertebrae 

 the difference is still more marked. On the foremost vertebrae these "double pro- 

 cesses" are nearly horizontal and directed straight outwards but farther back they 

 gradually pass into a position directed downwards and a little backwai'ds, still, on 

 the 24th vertebra (sometimes on the 25th) the long anterior transverse process does 

 not unite with its fellow from the opposite side; but on the next the anterior pro- 

 cesses join from opposite sides and form an inferior arch, terminating in a long, 

 backwards directed, lower spine. This vertebra (No. 25 or 26) I therefore count as 

 the first caudal vertebra. On the preceding, the last abdominal, vertebra the 

 posterior transverse process is already much shortened, on the first caudal it is 

 very small, and on the following it gradually vanishes. 



The spinous processes originate from the posterior part of the neural 

 arches. On most of the abdominal vertebrae they are only low, and laterally com- 

 pressed; from about the 17th they rise somewhat and from the 20 — 21st they grow 

 long and slender as in most bony fishes. The anterior part of the neural arch is 

 shaped like a rather large, rounded process, which, together with its fellow from 

 the opposite side, embraces the base of the preceding spinous process; this part is 

 larger on the foremost vertebrae, and diminishes gradually backwards. Below this 

 part is found the proper articulation, formed by a triangular pit into which fits a 

 small process from the preceding neural arch. 



