ON THE PHYLOGENY OF THE CARAPACE, ETC. 805 
respects replaced them by the new mosaic shell, formed by a prolifera- 
tion of the marginals and such other epithecal elements as were 
present “4: epineurals and probably elements in the other keels too. 
{t must for the present be left undecided whether these ‘ other epithecal 
elements ’ were really primitive, connected without interruption with 
the epithecal elements of the Procheloni@, or whether these elements 
had been newly acquired, had reappeared, in the immediate, theco- 
phorous, ancestors of the Atheca (in connection with a strong develop- 
ment of keels in the adult), showing the primitive arrangement that 
was prescribed by the position of the horny shields in these Thecophora. 
The new mosaic shell has obviously the advantage of being lighter, 
more elastic, and more superficial, thus protecting the surface of the 
thick cutis; it seems that a reduced thecal shell could not do this, as 
it would be limited to the deepest layers of the cutis, where it first 
forms in the embryo, and so, with its reduction the skin would have 
been insufficiently protected, especially if the horny scutes had 
been lost very soon too. 
Archelon seems to have already acquired a complete epithecal shell, 
like Dermochelys,*? but it certainly is much more primitive in several 
respects: neural and costal plates have not yet disappeared, the mar- 
ginals are like those of other turtles, and the plastron is much less 
reduced. Archelon, with its largely completed mosaic shell, is an 
interesting intermediate form, connecting the mosaic shell of the 
Leatherback with the thecophorous shell of other turtles. How far 
the shell of the Leatherback is directly connected with that of Archelon 
is difficult to decide, but Archelon shows at least that a transformation 
of the shell as here assumed for Dermochelys was possible, as it was 
largely completed in Archelon. 
It need not be emphasised that the phylogeny of the shell of the 
Leatherback here assumed is hypothetical, and that only new dis- 
coveries and investigations can show whether it or Dollo’s view, that 
the shell of Dermochelys is quite unconnected with the thecophorous 
shell of its ancestors, comes nearest the truth. But I may be permitted 
to point out what seem to me the advantages of the new hypothesis 
over that of Dollo: (1) The view that the marginals take part in the 
formation of the new shell concurs with what is shown by Archelon, 
where they do form part of the mosaic shell. (2) The presence of 
keels in Dermochelys corresponding to the keels in Thecophora points 
to a connection existing between the shells of both types. (8) The 
formation of the new shell by a proliferation of elements already 
present in the thecophorous ancestors seems a less complicated process 
than an entirely new formation. We conclude that: 
(a) The shell of tortoises and turtles is formed by a combination 
of two layers of dermal ossifications, a thecal layer and a more super- 
ficial epithecal layer, the latter generally represented by the marginals 
only. 
41 Jaekel (1914) gives an interpretation of the shell of Dermochelys that has much 
in common with the views of Volker and Menger. 
42 Wieland, 1909, p. 120. 
