of Nudibranchs can be variations of a single type. As mentioned in the last chapter, I
think the family ought logically to be classed on the holohepatic side because its liver is
not divided. But Marionia, indistinguishable externally from Tritonia, leans more to the
other side, for it has a slight division of the liver, and, as in several of the Cladohepatica,
the stomach is armed with plates. As might be expected there are several other
annectent forms about the dividing line, mostly scarce and not well known. Doridoeides
has the shape of a Dorid but a ramified liver, and Dirona in many characters resembles
Dendronotus or the Solids, but its liver is solid, although three-lobed.
It is easier to explain the Dorididaa as variations of one type than the Cladohepatica,
but the number of plausible explanations is embarrassing and means of course that the
facts which serve as sign-posts are few. Bathydoris and Doridoxa connect the Dorids with
Tritonia, but opposite views have been held as to the relationship of the Crypto- and
Phanerobranchiatm. Bergh considered that the latter are derived from the former, and
this order of development is supported by the case of Doridopsis rubra mentioned above
and by the specialized buccal parts1 of the Phanerobranchiatm. But Pelseneer derives
these latter from the Cryptobranchiat*, and, as far as the structure of the branchim is
concerned, this seems the probable order of evolution in most cases, though not without
exception. He selects Goniodotis as the connecting link, but this seems to me improbable
since this genus is in some respects highly specialized. I would rather suppose that some
extinct form with differentiated teeth and in external characters resembling Trevehjana gave
rise to Chromodoms, or that the Notodorididae, which have simply hamate teeth and valves
to protect their branchim, gave rise to some form like Miamira. But of the vast majority
of existing Dorids it may be said that the Cryptobranchiatae are more specialized in their
gills and the Phanerobranchiatse in their buccal parts, so that the-one cannot be derived
from the other, though both may be regarded as variations of a type like Bathydwis.
The suctorial Doridopsidas may be explained as Cryptobranchiatse with a modified
buccal apparatus, in, some respects analogous to that of the Pseudo--and Goniodorididm
but much more highly specialized. But they also resemble the Phyllidiidm (especially
Phyllidiopsis) which have the same suctorial mouth and branchiae in the form of lamellae
under the mantle-margin. The Corambidae have a similar arrangement of the branchiae
(especially analogous to the disposition found in Fryeria) but mouth-parts like those of the
Pseudodorididae. An analogous combination of infra-pallial branchiae with a simply hamate
radula has not been found, but an approximation to such an arrangement occurs in Miamira,
which has normal cryptobranchiate gills, but also a mantle-margin produced into lobes
bearing lamellae on the lower surface. If this arrangement occurred in a Doridopsis,
Phyllidiopsis might be evolved by suppression of the dorsal branchial plumes and extension
of the lateral lamellae.2 But this form of respiratory apparatus involves a considerable
change in the whole economy and it is also found in the cladohepatic Plmrophyllidia. It
might therefore be reasonably argued that it must date from an early stage in the history
of the Nudibranchiata when important characters could be modified or re-combined more
1 In the great majority of genera we find'either (1) differentiated teeth or (2) an ingluvies
buccalis accompanied by a much reduced radula which usually exhibits two kinds of teeth.
3 In Phyllidiopsis berghii there is a small circular cavity round the anal papilla which looks like
the remains of a branchial pocket.
freely than now, and in support of this it might be urged that the simple renal organ of
Doridopsis resembles that of Tritonia. In this case the resemblance between Doridopsis
and the Phyllidiidæ would be due mainly to convergence.1
Of all thèse families there can be no doubt that the Cryptobranchiatae are the most
numerous and thriving. They number about three hundred out of a total of some five
hundred known species of Holohepatica. This can only mean that the type of alimentary
and respiratory organs which they all possess is eminently suited to animals of their habits.
It may be observed that the branchial rosette of the Holohepatica seems to require and
nearly always to receive some kind of special protection. The defence may take the form
of (a) a pocket,. (b) a valve closing over the branchiae, (c) dorsal appendages near the
branchiae which keep off dangerous objects. When none of these are present the
necessary protection may be secured by making the branchiae very small with the greater
part of the stem attached to the dorsal surface, or, on the other hand, by having the main
stem of each plume very broad and strong (e. g. in Nembrotha) so that when it contracts
it acts as a cover to the smaller pinnules which it bears. But of all these devices the
branchial pocket is the commonest and most effective.
The Æolids form in the Cladohepatica an extensive and homogeneous group comparable
to the Dorids. Being for the most part small and pellucid animals they are
probably even more numerous than they appear to be, and will continue to furnish new
species to the investigator. They can be derived from an ancestor like Tritonia, and a
series of living forms, if not exactly connecting links; mark stages analogous to the line of
development traversed. In the clumsy Scyllsea and more elegant Bornella, the structure
of the Tritoniidæ (especially of Marionia) is still preserved in essentials. There are
dorsal papillæ, but they are not wholly analogous to the cerata of Æolids. Branchial
tufts are scattered over them, and they do not always contain diverticula of the liver,
which consists mainly of two or three portions in the body-cavity. Dendronotus marks
another step in the direction of the Æolids. The radula becomes narrower, there are no
accessory branchiæ on the cerata, but the delicately ramified cerata act themselves as
branchiæ. The liver still mainly consists of three portions in the body-cavity, and is
sometimes ramified in the cerata and sometimes not, but in all cases is less solid and more
flocculent than in the Tritoniidæ. Lomanotus is a parallel genus. I t is farther from the
Æolids than Dendronotus, inasmuch as it has a wide radula ; nearer, inasmuch as its cerata
are not branched but simple columns connected by a membrane. An antarctic genus
Notæolidia effects the transition from these forms to Æolids like Gklamylla and Goryphella.
One or two families diverge somewhat from this line of development, though not indicating
another origin. The Phylliroidæ are specially modified for pelagic life. Tethys and Melibe
are in some points of structure intermediate between Dendronotus and the Æolids, but
their feeding habits have modified their alimentary- organs, both internal and external.
More difficult to explain than these are the Pleurophyllidiidæ and Janidæ. I do not
see how it is possible to describe them as either ancestors or descendants of the Æolids,
1 But side by side with differences there are many resemblances in anatomical details. In both
families (1) the nervous system is so concentrated that the ganglia are hardly distinguishable ; (2)
the male genitalia are armed with spines; (3) there are folds on the wall of the pericardium—the
so-called pericardial gill.