466 PROF. ST. GEORGE MIVART ON THE 
impeding the lateral flexure of the body in swimming, because the plane of 
its antero-posterior longitudinal adherence must be at right angles to that 
of the longitudinal adherence of the dorsal fin. 
(2) That generally the pectoral fins join the body at too low a level to enable them 
to abut directly upon the vertebral centra (or their representatives) or upon 
the neural arches, while the solid paraxial skeletal elements are very short in 
Elasmobranchs. 
(3) That they could not be directly connected in a straight line, even obliquely, 
with the skeletal axis without interfering with the body-cavity of that 
region. 
On all these accounts the pectoral fins, and the ventrals also, must (if they are to rest 
on a solid support for their more or less obliquely up-and-down mode of flapping) have 
a narrow connexion with a sustaining structure not directly continuous, in a straight 
line, with the skeletal axis; and these exigencies wonld account for the difference 
existing between the mode of attachment of the pectoral fins generally and the mode of 
attachment of the dorsal fin of Notidanus. 
But it may be said that the radial cartilages of the dorsal fin are really the prolon- 
gations outwards of the axial skeleton. This is the teaching of Gegenbaur, who, as 
we haye seen, considers the dorsal radials to have been originally but productions of 
the neural spines. 
The almost universal absence, however, of concordance or any definite numerical cor- 
respondence between these elements and the subjacent vertebre seems conclusive against 
this view. We have (in Notidanus, Spinax, Acanthias, Pristiophorus, Pristis, and 
Rhynchobates) found a series of forms which agree well with a process of coalescence 
and centripetal extension, but which quite disagree with the opposite view; for, 
according to the latter view, we should have to suppose that the neural spines became 
segmented, that they then enlarged and serially cohered down to their very bases, and 
that subsequently such solid base became absorbed close to the vertebral column, while 
remaining more or less coherent at a greater or less distance from it—a supposition 
which seems to me a very unlikely one. 
It may be objected that in Pristiophorus, Pristis, and Squatina there are elongated 
cartilages preaxial to the dorsal fin, rising up from the axial skeleton, and seeming at 
the same time to be neural spines and serial homologues of the large dorsal fin-plates 
of those genera. But are these parts really neural spines? In Sguatina not only do 
all or part of the basal plates of the dorsal fin seem discontinuous with the subjacent 
skeletal axis, but the neural spine-like cartilages, situated yet more preaxially, also seem 
separate from the axial skeleton on which they rest. Are these structures, then, 
neural spines segmented off, or are they dorsal radials which have as yet imperfectly 
cohered? The study of development can alone solve this problem; but if it should 
turn out that that even these cartilages are centripetal chondrifications, it would fully 
