ON THE PHYLOGENY OF THE CARAPAOE, ETC. 803 
shields, the lateral row by the marginals and marginal shields, the 
supramarginal row being present in Archelon, Proganochelys, and 
Stegochelys, and indicated by the supramarginal shields of Macro- 
clemmys and Boremys. There is no direct evidence of a costal row 
corresponding to the costal keel of the Leatherback, but the costal 
shields may be taken as evidence of its former presence. On the 
plastron the median row of Dermochelys may be represented by the 
intergular and interhumeral shields present in some forms; the two 
paired rows of plastral ossifications in Dermochelys, probably 
developed as rows of larger horny scales in the young, may be repre- 
sented by the plastral and inframarginal rows of shields in the 
Theocophora. 
In this connection it is important that the keels of the shell of the 
Leatherback are found also in other Testudinata, being rather common in 
very young specimens. We can recognise them on the tail trunk of 
the snapping turtle, Chelydra, as rows of horny scales, partly com- 
bined with small dermal ossifications ; there is one keel more here than 
in the Leatherback, a neurocostal keel lying between the neural- and 
costal-keels, but this row seems to be present also on the neck of the 
young Dermochelys as a row of larger yellow horny scutes like the 
rows that represent the other keels.** Though the keels in the adult 
Dermochelys appear to be rather an adaptation to the swimming habit 
of the animal, they yet are obviously connected with the keels in other 
Testudinata, and cannot well be interpreted as an entirely new 
acquisition, developed after the primitive thecophorous shell and horny 
shields had disappeared. 
We conclude that Hay’s hypothesis of the presence of superficial 
dermal ossifications, arranged in longitudinal rows in the Prochelonia, 
is supported by important evidence. However, we must take into 
account that, on the whole, traces of these epithecal elements, if we 
except the marginals, are very rare in fossil forms, even in the older 
ones, and this makes it improbable that these elements were strongly 
developed, forming continuous zones in the primitive Testudinata. 
There can only have been present small ossifications, loosely attached 
to the underlying thecal elements of the real shell. It seems possible 
that even the epineurals of Toxochelys, Graptemys, and Archelon are 
not really primitive epithecal elements, but newly formed ones; their 
very isolated appearance seems to point rather in this direction. 
I am, on the base of this evidence, inclined to assume that in the 
immediate ancestors of the Testudinata, in the Prochelonia, there 
was present a thecal shell, composed of neurals, costals, the nuchal 
and suprapygals*® dorsally and the plastron ventrally, and rows of 
epithecal dermal ossifications, covered with corresponding horny scutes, 
beginning on the neck and continued over the thecal shelf and the 
tail (fig. 9). These epithecal elements were, however, only feebly 
developed and loosely attached to the thecal shell, with the exception 
of one row (perhaps two or three rows) on each side of the trunk, 
88 Compare Newman, 1906, p. 103, and Volker, 1913, p. 527. 
89 Compare Menger, whose paper brings a discussion of the nature of the nuchal 
and the suprapygals; and Gadow, 1899, p. 219. 
3 F2 
