CLASSIFICATION OF ACTINIARIA 499 
THENARIA: Sagartiidae, Paractidae, Boloceridae, Acti- 
niidae, Bunodidae, Aliciidae, Phyllactidae, Dendro- 
meliidae, Minyadidae. 
4. STICHODACTYLINAE: Discosomidae, Stoichac- 
tidae, Heteranthidae, Homostichanthidae, Aureli- 
anidae, Actinodendridae, Phymanthidae, Thalas- 
sianthidae. 
This is, of course, the list as it stands without taking any account 
of the present paper, even Part I of it. The work of Part I 
was chiefly devoted to a revision of the Sagartiidae and 
Paractidae, taking those names in the old sense as used on this 
page. 
3. Discussion OF CHARACTERS TO BE USED IN 
CLASSIFICATION. 
The characters already discussed in Part I, pp. 456-68, will 
of course be used here again, where they come in, but a few 
others remain to be mentioned. 
In the families under discussion now, there are no mesogloeal 
sphincters save in Halcampa, but it has to be decided 
how far the character of the endodermal sphincter is to be 
trusted as a family feature. All grades of it exist, from very 
weak diffuse or very weak circumscribed to very strong 
circumscribed, through various degrees of diffuseness and cir- 
cumscribed diffuseness (cf. Text-figs. 11 and 12). It may 
be quite absent. In some families the range is not more than 
from absent to weak diffuse. But in other cases there are 
so many grades that one can draw no line of demarcation 
anywhere ; and it must be admitted that the form and grade 
of development of the sphincter cannot be used as a family 
character except where it is fairly stable. The same thing 
really applies to mesogloeal sphincters, but here it has been 
less noticed because no one happens to have suggested an 
artificial distinction between diffuse and circumscribed meso- 
gloeal sphincters. 
It has long ago been realized that presence or absence of 
