of the Fishery Board for Scotland. 



259 



Owing probably to the shortening of these vertebrae, the direct longitu- 

 dinal and cross pieces are absent, and in their place appears generally a 

 series of three or four smaller triangular-shaped cavities, formed by the par- 

 titions running more or less off the direct line and fusing with each other. 



The latero-ventral ridges on each successive vertebra become more pro- 

 minent towards especially their posterior ends, so that on the 24th (im- 

 mediately before the 1st complete haemal arch) they each end in a slight 

 sharp-pointed downwardly and backwardly projecting process — the pos- 

 terior ventral zygapophysis. At the 31st vertebra this sharp process is 

 more prominent, and is overhung, though not touched, by a process 

 (anterior zygapophysis) of the haemal arch. 



About the 3Gth or 37th vertebra, and posteriorly, this articular process 

 or zygapophysis overlaps externally the forward projecting process of the 

 haemal arch. 



In the 1st vertebra these articular processes are largely developed, 

 forming strong sharp-pointed spurs (PI. XVI. fig. 2), and on the 2nd 

 vertebra they are also of considerable size ; behind this and up to the 24th, 

 while gradually increasing, as has been described, they are very minute. 



Beyond the 40th vertebra they appear again to die away, disappearing 

 about the 53rd vertebra. 



The lateral ridges of the centra run along the anterior vertebrae rather 

 below the median line, so as to make the chambers (ventrolateral) con- 

 taining the basal pieces rather narrower than the dorso-lateral (muscle) 

 chambers; but towards the middle of the vertebral column these partitions 

 get shifted into a more directly median lateral position on the vertebrae, 

 and about the 27th vertebra the centrum is divided into two equal longi- 

 tudinal portions. 



Each of the latero-dorsal ridges form, at their posterior ends, a distinct 

 sharp-pointed articular process, projecting upwards and backwards — the 

 posterior darsal zygapophyses (PI. XVI. fig. 4, pz.). These processes are 

 present on all the vertebrae, and they overlap on and behind the 23rd (?) 

 vertebra, corresponding anterior processes of the neural arches — the 

 anterior dorsal zygapophyses. As a fact they are applied to the neural 

 arches throughout, although in a different manner to be described later. 

 The anterior ends of these latero-dorsal ridges terminate, just posterior 

 to the ring forming the end of the vertebra, in a cup-shaped cavity which 

 forms the articulation for the neural arch (fig. 3, a.c). This description 

 applies to the anterior vertebra, where the arch articulates so loosely 

 with this cup-shaped cavity that in preparation it generally falls out. 

 Proceeding backwards, however, it becomes more firmly fixed, and pos- 

 terior to about the 28th vertebra it is so fused to the centrum that 

 the cup is obliterated as such. Exceptions to the above description 

 of the various vertebrae, which are found on the last two or three, will 

 be referred to in describing the skeleton of the caudal fin. In the 

 young herring of 25 mm. to 30 mm. long, these divisions of the vertebrae 

 are not seen, ossification not having proceeded so far, and it is found that 

 they are formed as a rod-like deposit along the vertebrae, where the whole 

 circumference of the external membrane of the notochord has but a slight 

 deposit of bony matter in it, and before its constriction. These divisions 

 then are ossifications partly antecedent to the ossification of the remainder 

 of the centrum. 



The articulation of the various, vertebrae, one with another, behind the 

 2nd is simple. The ring forming the ends of each amphicoelus vertebra 

 simply butts against its neighbour, each being held against the other by 

 ligament, the articular processes forming but a slight additional security 

 against transverse movement, and that not in every part of the column. 



