OPISTHOBEANCRIATA FROM THE ABKOLHOS ISLANDS. 541 
Sub-order NUDIBRANCHIATA Cuvier, 1817. 
Nudibranclis are marine hermaphrodite Opisthobranchiata without 
ctenidium and osphradiam, and in the adult state without a shell. The larva, 
however, has a shell and operculum. The visceral mass is not marked off 
from the foot; the body shows complete or appro.ximate external symmetry 
and usually bears plumes or other appendages which assist respiration. The 
nervous system is concentrated in a collar behind the buccal bulb, and the 
chief ganglia are placed on the dorsal surface of the oesophagus, being often 
partially united and sometimes fused into a single mass. The vas deferens 
is always an internal tube, never an external groove. The teeth, when used, 
are discharged from the body and not retained in a sac. Among the common, 
but not universal, characteristics of the Nudibranchiata are also the 
following : — (1) The dorsal tentacles or rhinophores are often laminated and 
retractile, features not recorded in any other group. (2) The kidney is 
rarely compact, but usually a system of ramified tabes. (3j The genitalia are 
often extremelv complicated, both in their essential plan and also owing to the 
presence of accessory glands and armatures. Besides this, the various sub- 
divisions show remarkable peculiarities of their own, such as the ramification 
of the digestive organs, the reduction of the teeth to a single row, and the 
presence of nematocysts. 
The classification of the Nudibranchiata here adopted is essentially that of 
Eliot (40) with the following four main differences : — (1) The Ascoglossa have 
been removed and treated separately for reasons noted in the introduction, 
Eliot's family-names Dorididse Cryptobranchiatae and Dorididae Phanero- 
branchiatae, being binomial, cannot strictly be used, as the family must have a 
single name derived from that of its oldest and typical genus. Moreover, both 
these terms include a number of groups of forms, which, for the present at unj 
rate, would appear to be more conveniently handled separately. It must be 
confessed that the classification of both these assemblages stands in need of 
further revision, a task outside the scope of the present work. (2) For the 
reasons just given the group Dorididae Cryptobranchiatae has been sub-divided 
into smaller families, whose limits correspond with the similarly named sub-, 
families as given by Bergh in his 'System der Nudibranchiaten Gastero- 
poden " in 1892. In certain cases the laws of priority have necessitated a 
change of name. (3) Eliot's four divisions of the Dorididse PhanerobrancHatse 
have been regarded as inde]:)endent families. (4) Investigation of the synonyms 
of various species has revealed the fact that certain names are incorrectly 
applied, as was pointed out previously. These have been rectified by the 
alteration of the names in the cases that appear to admit of no doubt, and such 
a change sometimes involves both the names of the genus and the family, even 
though the limits of these groups have not been altered. 
The sub-order is divided into two Tribes (A. Holohepatica and B. Clado- 
hepatica), a separation dependent in the main, as the terms imply, upon 
