member of the same family) shows a division of the liver, one small portion being separate
from the main mass, and has a stomach armed with horny plates. Besides this, the
appearance of Tritonia recalls such forms as Dendronotus and Lomanotus; also it has no
blood-gland, only one spermatotheca and a lateral anus, all features characteristic
of the Cladohepatica. But, on the other hand, besides the undivided liver, it has the
structure of the hermaphrodite gland (a layer on the liver) characteristic of Dorids, and
its branchias are simply respiratory organs, not cerata. Also its buccal parts are very like
those of Bathydoris, which, though an abnormal form, has no cladohepatic affinities.
Most Tritoniidas have gills, but in Tritonidoxa and Tntoniella they are absent. Other
gill-less forms are Doridoeides and Doridoxa. The first of these combines the characters
of the Holo- and Cladohepatica, for it has the appearance of a Dorid (except that it has
no branchiae) and three genital openings, but its liver is ramified. Doridoxa (found in the
northern Atlantic, and perhaps occurring off our northern coasts) has the shape of Doris,
a blood-gland, and two spermatothecas, but there are no branchiae and the anus is lateral.
It appears to me probable that these genera, most of which combine characters that are
not found united in ordinary forms, are survivals of some of the earliest types of
Nudibranchs in which no new pattern of gill was developed to replace the ctenidium. In
any case Doridoxa is a connecting link between the Tritoniidae and Dorididge, and the
connection is further illustrated by Bathydoris,* which in general shape and many
anatomical features resembles Doris but has the buccal parts of Tritonia. I t is remarkable
that it also has a hermaphrodite gland entirely separate from the liver; that is to
say the connecting link presents an anomalous feature not found in either of the types
which it connects, and suggestive of a different group.
The most important subdivisions of the Holohepatica-are Dorididge Phanerobranchiatae
and Dorididae Cryptobranchiatae. Both have gills in the form of plumes set in a complete or
incomplete circle round the anal papilla, and may be called collectively Anthobranchiata.
In the Phanerobranchiatae the plumes, though sensitive and capable of contracting when
touched, cannot be withdrawn below the level of the back into a permanent pocket2; in the
Cryptobranchiat® they can be retracted into such a pocket, which is often closed by
teeth or tubercles, so that the gills entirely disappear.8 Sometimes when the gills are
large they can be withdrawn into the pocket only partially or with difficulty.
The phanerobranchiate Dorids exhibit great differences in both appearance and
structure. In nearly all the sub-divisions there is a strong tendency to differentiate the
1 Bathydoris vngolfiana is found in the north Atlantic.
2 Phanerobranchiate gills usually shrink together when touched, and a hollow may be formed
temporarily a t their base and subsequently disappear. The characteristic of the Cryptobranchiatee is
a permanent pocket. Only rarely is it doubtful to which class a form should be referred.
8 Bergh includes in the Doridid® Cryptobranchiatae Bathydoris and Eexabranchus, two genera
which have no branchial pocket. If they must be classed in one group or the other they certainly
belong to the Phanerobranchiate, but I think it is better to consider each of them as the representative
of an independent family, Bathydoris, as already mentioned, has affinities to Tritonia. Eexa-
bra/nchus is a tropical genus, which in many respects agrees with the typical Dorids, but its mouth
is armed with strong labial plates, it has separate, non-retractile branchiae, and is quasi-pelagic in its
habits, being frequently found swimming at some distance from the land. The statement (repeated
in the most modem manuals) that its branchiae are retractile into separate cavities is not correct.
teeth of the radula and narrow its dimensions. With very few exceptions the blood-
gland is single (usually double in the Cryptobranchiatae) and the verge is armed with
spines. The class may be divided into four families: (1) Polyceridae; (2) Notodorididae;
(3) Pseudodorididae; (4) Goniodorididge. The Polyceridae form a fairly numerous group
and comprise many brightly coloured and graceful animals. They are usually elongate
and often bear dorsal appendages. The analogy of the Cladohepatica suggests that there
is some mechanical connection between these features, and that an elongate body can bear
appendages better than a flat and oval one. It is not difficult to connect this group with
those that we have already mentioned. Such an animal as Polycera resembles Tritonia in
nearly all external features, except that the branchiae, instead of being arranged along the
mantle margin, are collected round the anal papilla. The mantle-margin is narrow and
sometimes do,es not exist at all at the sides of the body, but over the head it is developed
into an oral veil which often bears simple or branched processes (as in Tritonia). The
rhinophores, which are nearly always perfoliate and retractile, are often provided with
raised sheaths. The oral tentacles are generally small. Mandibular plates (hardly
amounting to true jaws) are usually present. The tendency towards a narrow radula
with differentiated teeth is very marked. In the Holohepatica this narrowing is always
effected by the degeneration or loss of the outer teeth and the enlargement of the first
few laterals or one of them, so that the radula reduced to its lowest terms (Drepania)
becomes 1.0.1. In the Cladohepatica on the other hand, the reduction takes place by
dispensing with the laterals and throwing all the work on the middle tooth, so that the
minimum formula is 0.1.0. In Trevelyana the differentiation' is slight, but the first
tooth is often much larger than the others. In the other genera1 some of the inner teeth
are large and hamate, whereas the outer ones are reduced to flat plates. In Euplocamus
and Plocamopherus we find radulse like 12 + xi. 0. xi + 12, where .the Roman figures
represent hooked teeth and the Arabic ones plates. But in a large number of common
genera there are only two hooked teeth, of which the second is the larger, and the radula
varies from 12 + ii. 0. ii + 12 in Triopa to 2 + ii. 0. ii + 2 in Polycerella. Central teeth
are rare, and when present (Issa, Triopha, Nembrotha) are, as a rule, imperfectly developed.
The Notodorididge are a very small group comprising the three genera Notodoris,
Triopella and JEgires, of which the last two inhabit the northern Atlantic. They are
hard and stiff: the branchige are protected by special lobes or valves; the rhinophores
are retractile and (except in Triopella) not perfoliate. There are no jaws in Notodoris,
but in JEgires there is a single upper jaw, much as in the Pulmonata. The radula
consists of simply hamate teeth, all alike and without any central tooth. This form of
radula and the branchial valves approximate the Notodorididge to the Cryptobranchiatge,
for the effect of the valves in protecting the branchige is much the same as that of a pocket.
I propose the name Pseudodorididge for some genera with non-retractile branchige, on
account of their resemblance to true Dorids, which is so great that in the letterpress
accompanying Alder and Hancock’s plates they are called Doris. They are flattish
animals covered with small tubercles or papillge.. The parts round the mouth are dilated
into a semi-circular veil on which tentacular appendages are hardly visible. The
branchiae, which may be pinnate or tri-pinnate, are contractile, but have no pocket.
1 Except the very anomalous KaUnga.