at the base of the buccal apparatus. In the opinion of the writers mentioned, the group
consists (a) of Lobiger and Lophocercus, forming a small family of Tectibranchs known as
Lophocercidae, and provided with a shell, an internal vas deferens, and a laminated gill
somewhat different from the ordinary ctenidium, and (b) of four families here classed as
Nudibranchs, viz. the Hermaeidae, Phyllobranchidae, Elysiidae (including Placob'anchus),
and Limapontiidae. These four families undoubtedly answer to the definition of Nudibranchs
given above,1 and in discussing their systematic position it will be most convenient
to consider first the general classification of the sub-order and then to inquire whether
there is sufficient ground for excluding them from it.
Although the name Ascoglossa, strictly speaking, denotes the whole group, including
both Nudibranchs and Tectibranchs which certain authors wish to unite, I frequently use
it in these pages as a convenient designation for the four families of Nudibranchs (Her-
maeidae, Phyllobranchidse, Elysiidae including Placobranchus, and Limapontiidge) which
possess the ascoglossan radula.
Several primary divisions of the Nudibranchiata have been suggested, but the simplest
plan is to make two tribes, the Holo- and Cladohepatica, those with entire and those
with branched livers. This appears to me preferable to Pelseneer’s division into four tribes,
Tritonioidea, Doridioidea, iEolidioidea, and Elysioidea,2 for these tribes are not equally
important and distinct. Thus the JEolidioidea and Tritonioidea offer a greater contrast
to the Dorids than to one another, and a distinction drawn between them may separate
inconveniently forms which are closely related. For instance, Notasolidia has the appearance
and most of the characters of an JEolid, but the liver is contained partly in the body-
cavity, so that according to Pelseneer’s system we must place it amoqg the Tritonids and
cannot show its close relationship *to the Solids. Even Tergipes hardly falls within the
iEolidioidea if Pelseneer’s definition (“ the whole of the liver is contained in the integuments
and tegumentary papillge ”) is taken literally, for the digestive ducts in its body-
cavity are lined with coloured liver cells. On the other hand, the Pleurophyllidiidae
cannot be classed simply as a family of the JEolidioidea, for their differences from that
group are as great as their resemblances to it. The Elysioidea have many peculiarities in
common but are distinctly cladohepatic, and it cannot, in my opinion, be maintained that
the Hermgeidge differ from the Solids as much as both of them differ from the Holo-
hepatica. But the iEolids are connected with the Holohepatica by a long and continuous
series of links, whereas the relatively small gap between them and the Hermgeidge is
less completely bridged.
The Holohepatica are Nudibranchs in which the liver forms a compact mass, neither
branched nor divided. This feature is usually (though with a few exceptions to each
point) accompanied by the following characters: There is almost complete external
symmetry, the vent being usually in the medio-dorsal line and surrounded by a circle of
branchial plumes. Above the central nervous system is a double or single blood-gland.
1 Bergh adds to the above definitions of the Nudibranchiata the following: Lingua fortis, dentibus
imi-, tri-, vel pluriseriatis a/pice linguse paullaiim caducis et eliminatis. But these words are clearly
added simply in order to exclude all Ascoglossa, and apart from their radula the four families come
under his definition.
8 Also called by him Tritoniomorpha, Doridomorpha, Eolidomorpha, Elysiomorpha.
The radula is usually of moderate or considerable width, and is not known to be reduced
to a single tooth in any genus.1 The labial cuticle often bears an armature of various kinds,
but true mandibles are rare. The hermaphrodite gland is (with rare exceptions) not a
separate mass but a layer spread over the liver. As a rule there are two receptacula
seminis, and the genital ducts are triaulic. This is indicated externally by the presence
of three openings, but the essence of the diaulic and triaulic arrangements consists not
in the number of orifices but in the undivided or bifurcate configuration of the female
branch of the genitalia. After the bifurcation the tubes may reunite before reaching the
orifice.
The Cladohepatica are Nudibranchs in which the liver is branched or at least
divided. In the majority of families it is contained wholly or mainly in the dorsal papilla,
into which its ramifications extend. Such papillas containing liver branches are called
cerata, and often bear cnidosacs. The vent is usually (but not always) on the right side.
The gills assume various forms, but never that of a branchial rosette. There is no blood-
gland. The hermaphrodite gland is an independent organ, and not attached to the liver.
Mandibles are usually present, and the radula is frequently reduced to a single row of teeth.
But the median tooth never disappears, and, in the uniseriate radula, is the only one that
remains. The. Holohepatica are (with the exception of the Polyceridae) solidly built
sluggish animals, usually of oval shape, and rarely bearing any processes larger than
tubercles and branchias, but the Cladohepatica show great variety of form, and are often
extremely elegant and active in their movements. They usually bear dorsal appendages,
which may be very large.
The following table represents the classification of the Holohepatica here proposed.
T ribe I. H olohepatica.
Family 1" Tritoniidae (Tritonia, Marionia, Atthila, Tritoniella, Tritoniopsis, etc.).
2. Doridoxidae (Doridoxa).
3. Bathydorididae (Bothydoris).
4. Hexabranchidae (Hexdbranchus).
5. Dorididae Cryptobranchiatae (about fifty genera).
6. Dorididae Phanerobranchiatae with four sub-families.
a. Notodorididae (Notodoris, JEgires, etc.).
f3. Polyceridae (about twenty genera).
y. Pseudorididae (!Acanthodoris, Adalaria, Lamellidoris, etc.).
8. G-oniodorididae (Goniodoris, Ancula, Idalia, etc.).
7. Doridopsidae (Doridopsis, Doriopsilla).
8. Phyllidiidas (Phyllidia, Fryeria, Phyllidiopsis} etc.).
9. Corambidae (Corarribe, etc.).
Tritonia is to some extent a connecting link between the two tribes. Bergh classes
it among the Cladohepatica while admitting its affinities to the Holohepatica, but it seems
more logical to refer it to the Holohepatica (since it has a compact and undivided liver)
while recognizing its affinities to some of the Cladohepatica. How real those affinities are
is shown by the fact that its near relation Marionm (which must be regarded as a
i The narrowest known radula among.the Holohepatica is in Lrepama 1.0.1.