523 



rapophyses, the hsemapophyses, may severally consist of two or more 

 pieces. This is not all : the like is true even of the centrums. 



" In Heptanchus (Squaliis clnereus] the vertebral centres are feebly aii'J 

 vegetatively marked out by numerous slender rings of hard cartilage in the 

 notochordal capsule, the number of vertebrae being more definitely indicated 

 by the neurapophyses and . parapophyses. ... In the piked dog-iish 

 (Acanthias) and the spotted dog-tish (Scylliurn) the vertebral centres coin- 

 cide in number with the neural arches " (p. 87). 



Is it not strange that the pattern vertebra should be so little ad- 

 hered to, that each of its single typical pieces may be transformed 

 into two or three ? 



But there are still more startling departures from the alleged 

 type. The numerical relations of the elements vary not only in 

 this way, but in the opposite way. A given part may be present 

 not only in greater number than it should be, but also in less. In 

 the tails of homocercal fishes, the centrums " are rendered by cen- 

 tripetal shortening and bony confluence fewer in number than the 

 persistent, neural, and haemal arches of that part " that is, there 

 is only a fraction of a centrum to each vertebra. Nay, even this 

 is not the most heteroclite structure. Paradoxical as it may seem, 

 there are cases in which the same vertebral element is, considered 

 under different aspects, at once in excess and defect. Speaking of 

 the hasmal spine, Professor Owen says : 



" The horizontal extension of this vertebral element is sometimes accom- 

 panied by a median division, or in other words, it is ossitied from two 

 lateral centres ; this is seen in the development of parts of the human 

 sternum ; the same vegetative character is constant in the broader thoracic 

 haiinal spines of birds ; though, sometimes, as e.g., in the struthionidfe, 

 oxfiijication extends from the same lateral centre l".nijt Incite i.e., forwards and 

 backwards, calcifying the connate cartilaginous homologucs of halves of four 

 or fioe Juvmal spinet, before these finally coalesce with their fellows at the 

 median line " (p. 101). 



So that the sternum of the ostrich, which according to the hypo- 

 thesis, should, in its cartilaginous stage, have consisted of four or 

 five transverse pieces, answering to the vertebral segments, and 

 should have been ossified from four or five centres, one to each 

 cartilaginous piece, shows not a trace of this structure; but in- 

 stead, consists of two longitudinal pieces of cartilage, each ossified 

 from one centre, and finally coalescing on the median line. These 

 four or five haemal spines have at the same time doubled their in- 

 dividualities transversely, and entirely lost them longitudinally ! 



There still remains to be considered the test of relative position. 

 It might, be held that, spite of all the foregoing anomalies, if the 

 typical parts of the vertebrae always stood towards each other in 

 the same relations always preserved the same connexions, some- 

 thing like a case would be made out. Doubtless, relative position 



