i. No stomach plates ; no branchial tufts.
2. Cerata branched. Dendronotidae.
3. Cerata not branched. Lomanotidse.
ii. Stomach plates present. Branchial tufts presept except in Tethys.
4. Two pairs of large cerata or a single pair of lateral wings. Scyllaeidm.
5. Several pairs of branched cerata; branched appendages near mouth;
male genitalia armed. Bornellidse.
6. Large asolidiform animals with several pairs of large cerata and a hood
or funnel round the mouth. No radula. Tethymelibidas.
Hancockia seems intermediate between Lomanotus and Doto, but may perhaps be
included in one family with the former in virtue of the manner in which its cerata arise
from the mantle-margin, the processes on its oral veil, and its perfoliate rhinophores.
But the radula is 1.1.1.
There are a few forms which seem intermediate between the Dendronotoidea and
JEolidioidea. They have most of the characters of the latter and are seolidiform in
appearance, but they possess some features of the Dendronotoidea which prevent us from
classifying them with the true Eolids, and none of them have cnidosacs. At the same
time they show so little resemblance to one another that we cannot unite them in a single
family. The Notaeolidiidas, represented by very large Antarctic nudibranchs, resemble the
true iEolids externally in all particulars, including the presence of cnidosacs. But they
have a hepatic mass in the body-cavity, as well as hepatic diverticula in the cerata and
integuments, and the radula is 5.1.5. The Dotonidaa are small animals, bearing tuber-
culate cerata which resemble minute fir-cones. In structure they agree with the iEolids
in most respects, but the rhinophores are retractile into sheaths, and though the radula is
generally uniseriate, yet in Dotilla there are several small lateral teeth. These points
connect them with the Dendronotoidea, but in others they seem more specialized than the
/Enlids. The ganglionic commissures are very short; the anal papilla is anterior and
dorsal; the jaws are reduced and very thin. The Heroidae, represented by the genus
Hero only, have branched umbellate cerata, one pair of which is in front of the rhinophores.
They cannot be placed with the Dendronotoidea, for the liver does not form a mass in the
body-cavity, but is arranged in two longitudinal canals. In essential structure the
Grlaucidas approach the true Eolids more nearly than the families mentioned, but they are
specially modified for a pelagic life. The body is produced into three lateral lobes, and
the cerata which grow out of these are directed not vertically but horizontally, so that
they help the animal to keep its balance, if not to swim.
The Janidae (Antiopella, Janus, Proctonotus) offer more complicated and perplexing
affinities. The shape is aeolidiform, but there is a crest between the rhinophores; the
cerata (and consequently the hepatic ducts) pass round the front of the head. Besides
sending branched prolongations into the cerata the liver forms a net-work in the integuments.
The anal papilla is medio-dorsal and set far back. The radula consists of hamate
teeth with a formula of about 40.1.40.1
Thus, like the Pleurophyllidiidae, the Janidae combine some qf the characters of the
1 According to Alder and Hancock (letterpress to Family 3, pi. 43) the genitalia show analogies
to those of Doris.
Dorididae (a broad hamate radula and medio-dorsal anus) with a hepatic system resembling
that of the Eolididge though not identical with it. It is interesting to find that another
genus (Madrella) which resembles the Janidae in the arrangement of its liver-system and
cerata has a triseriate radula and lateral anus. But it has rhinophores which are more
like those of Tritonia than of any cladohepatic Nudibranch.
The true iEolids are almost as numerous as the Dorids, and, as they are often small
and transparent, it is likely that a great many more remain to be described. The head
bears tentacles (except in Embletonia) and rhinophores, which maybe simple or perfoliate,
but never have sheaths. On either side of the back are arranged cerata containing diverticula
of the liver. They may be of various shapes, cylindrical, inflated, flattened, nodulous,
etc., but are never branched. They usually bear cnidosacs in the tips. The foot is usually
of the same breadth as the body, and the anterior corners are frequently produced into
tentacular processes. "Well developed jaws are always present. The radula is very
narrow. It sometimes contains three teeth in a row; more frequently it consists of a
single line of teeth. There is no compact or flocculent mass of liver in the body-cavity,
but the core of hepatic substance in the cerata may extend more or less into the ducts,
which finally enter the stomach by three openings. The male genitalia are often provided
with spines, hooks, or accessory glands.
The iEolids are even more than the Dorids a very homogeneous group, in which it is
difficult to find characters of taxinomic value. As such have been utilized the number
and shape of the teeth, the shape of the jaws, the surface of the rhinophores (smooth,
perfoliate, etc.), the conformation of the foot (round or angled), the shape and arrangement
of the cerata, the armature and annexes of the genitalia. But these characters
seem to be combined in all possible ways and none of them are absolutely satisfactory as
a basis of classification. Thus the smooth or perfoliate surface of the rhinophores seems
an important feature, but if we put together all the iEolids with perfoliate rhinophores
the result is not a natural group; it simply shows that iEolids of all kinds may have
perfoliate rhinophores. Further, the perfoliation may be perfect or rudimentary, and
when it is rudimentary it is hard to distinguish it from a slight wrinkling often found in
life (not merely in preserved specimens) in Goryphella, JEolidiella, and other genera, which
are considered to have simple rhinophores. The least objectionable basis for classification
is, I think, the number of teeth,1 and I would divide the great bulk of Eolids into
Triseriatas and Uniseriatas. This system is not without inconvenience, for it separates
allied genera, such as Flabellina and Pterseolidia, which differ chiefly in the radula. But
the character selected has this advantage, that the uniseriate radula is the more specialized
of the two, and the triseriate radula points in the direction of the Dendronotoidea.
Agreeably to this we find that, although we cannot say that all Triseriatse have a primitive
character, they at least include the form Ghlamylla, which seems an annectent type such as
1 I leave aside the question of classification based on the structure of the genitalia. In the last
chapter ,1 have shown that the genitalia of JEolictia papillosa do not conform to the diagnosis
generally given for the family, and it is quite possible that the different genera of Eolids may
exhibit considerable variation in these organs. But much difficult anatomical investigation is
necessary before such variations can be tabulated or compared, and for present practical purposes
this basis of classification is useless.