nor yet can they be dismissed as representing another line of development, for they possess
some of the most highly specialized characters of the ^Solids, such as the hepatic system
(both), cerata (Janidse), and cnidosacs (Pleurophyllidiidge). With these characters they
combine others which are either unknown or uncommon among the Oladohepatica. To
represent adequately their affinities we require, not a linear arrangement, but a figure in
three dimensions,1 and if we wish to provide for them a plausible phylogeny, we must, I
think, assume the existence of ancestors possessing combinations of characters now unknown
but not improbable. There is no reason, for instance, why cnidosacs and the clado-
hepatic system which seems a necessary condition for them may not once have existed in
animals shaped like Doris or Tritonia. It is easy, too, to find a reason for the disappearance
of such forms, for cnidosacs are useless without a particular diet, and large numbers of
Nudibranchs live on sponges and other animals which supply no cnidge.
More or less parallel to the AEolids are a number of small families, some more, some
less specialized, such as Fionidge, Dotonidas, Heroidge, Myrrhenidge. As with the Dorididge
among the Holohepatica, so here we may conclude that the HSolids are the most successful
of Nature’s experiments in this branch of molluscan construction, but that other variations
have a sufficient number of good points to maintain their existence. Hero is reminiscent
of Dendronotus, but Fiona and Myrrhine (and Doto in some points though not all) foreshadow
further developments and point in the direction of the Ascoglossa. Yet I find it hard to
think that these latter are derived from the ./Eolids simply by a process of specialization, for
the nervous system shows a different and apparently more primitive arrangement. But
it seems impossible to dissociate Hermaea from the HDolids, and I should suppose that here,
as in the Pleurophyllidiidge and Janidge, we meet with a collection of affinities which
cannot be represented in a linear arrangement.
The Nudibranchiata may be imagined as filling an area divided down the-middle by
a line ab which separates the Holo- and Oladohepatica. A—3 stand for the various types
of structure, those nearest ab being the least specialized and those in the corners the most
E
G A B D
F
G
H \
b
specialized. Various special characters will be found common not only to A and G and to
B and D, where there may be direct descent, or to A and B, which are near the dividing
line, but also to 0 and D, E and A, E and H, and to many other pairs composed of two
members whose structure is in other respects very divergent. Such combinations of
characters are illustrated by the Pleurophyllidiidae and other families already cited, but
many other instances on a smaller scale occur. Thus the Holohepatica have the hermaphrodite
gland as a rule in the form of a layer spread over the liver, but in Bathydoris,
AUoiodoris (otherwise an ordinary cryptobranchiate Dorid), and Trevelyana (a Polycerid),
it is collected in one or more solid masses, much as in Scyllzea. The verge is armed with
1 As suggested by Sedgwick. See the diagram on p. 410 of his ‘ Students’ Text-book of
Zoology,’ vol. i.
a straight spine (not a number of little spines) in a few Cryptobranchiatge (e. g. Kentrodoris),
in some HSolids (e. g. Amphorina), and some Ascoglossa (e. g. Limapontia), but not in the
intervening forms. The tentacles are grooved in Tritoniidge, some Dorids (Platydons and
Doris), and some Ascoglossa (Gyerce, Phyllobranchus). The kidney is a simple and
unramified sac in Tritonia and Doridopsis. The nervous system is fused into a mass in
which the ganglia are indistinguishable in the Tethymelibidse (which are cladohepatic), in
the Doridopsidge and Phyllidiidge, in Hexabranchus, and in a few ordinary Cryptobranchiate
(e. g. Asteronotus). Some of these characters may be due to convergence caused by simplification
or by concentration, but it will be observed that they do not depend on feeding
habits or on the tendency towards external symmetry. Remarkable coincidences apparently
due to these causes are seen in the ingluvies buccalis found in specialized Holohepatica
(Pseudodorididge, Goniodorididge and Corambidse) and in Ascoglossa (Phyllobranchidas,
Placobranchus, Tlmridilla), and the terminal position of the vent found in similarly
specialized but very different forms such as Fryeria, Gorambe, and a few Ascoglossa (Alderia,
Tlmridilla, Limapontia depressa).
Nor do these instances of one or more remarkable common characters appearing
to connect animals which are not nearly related in the rest of their organization occur
only between Nudibranchs of different groups; they occur also between Nudibranchs and
Tectibranchs, that is they extend to the whole group of the Opisthobranchiata. As
already mentioned the Lophocercidge and the Pleurobranchidge are the families which
approximate most closely to the Nudibranchiata, and the resemblances presented by the
first named have been discussed. The Pleurobranchidse are strikingly similar to the
Dorids in many points, but these resemblances are divided among different genera and no
one Pleurobranchid unites them or forms a connecting link. Plenrobranchaea is sometimes
credited with this role because it has no shell, but in its less concentrated nervous system,
its single genital orifice, its teeth, and even its external appearance, it resembles the
Dorids less than does Oscanius, which, however, has an internal shell. The chief
resemblances between Pleurobranchids and Dorids are the absence of the shell (Pleuro-
brancliae), the presence of spicules in the integuments, the general shape, the nervous
system and sense organs, the presence of a blood-gland, the reno-pericardial tube, the
radula, and the two spermatothecas.1 The similarity in the nervous system is particularly
striking. There are three pairs of large ganglia concentrated round and above
the oesophagus, as well as smaller buccal andgastro-cesophageal ganglia; the commissures
are generally short; the visceral ganglia tend to disappear from the visceral commissure;
the osphradium is replaced by tentacular ganglia, and the eyes are buried in the integuments.
On the other hand the stomach of the Pleurobranchidge is in two divisions, the
salivary glands are different from those of the Dorids, and the blood-gland lies on the
heart, not on the anterior part of the aorta. I t is also remarkable that the Pleurobranchids
'have more points in common with the Dorids than with Tritonia, which has a different
shape, no spicules, no blood-gland, one spermatotheca, a median tooth, and different jaws.2
1 But in some Pleurobranchids one of them is said to be imperfectly developed or absent.
2 On the other hand, Tritonia has an oral veil and a lateral anus. The remarkably unspecialized
character of Tritonia is shown by its affinities to both Holo- and Oladohepatica and also by the
slightly asymmetrical disposition of its heart, which is evidently a vestige of the arrangement when
there was a lateral ctenidium.