510 T. A. STEPHENSON 
line of their own. Since they are in some ways primitive 
we may place them next to the Gonactinids for convenience ; 
but because of their peculiarities they should be kept sufficiently 
apart from those to represent a quite distinct evolutionary 
line. The exact rank of the group Ptychodacteae which 
I propose for their reception will be better discussed in other 
sections (see pp. 540, 552, 554-6. &c.). 
§D. The Corallimorphidae and Discosomidae., 
There has been a growing feeling among those who have 
worked at anemones that there is a good deal of inter-relation 
between them and the corals, and that we can no longer insist 
on a separation of them based on presence or absence of 
a skeleton alone. This feeling has been best expressed by 
Duerden (120) in a study of the Madreporarian relationships 
of certain Stichodactylines. Perhaps in this connexion too 
little attention has been paid to the soft parts of corals. We 
are undoubtedly justified in retaming two groups, Actimiaria 
and Madreporaria ; but the justification is to be found in the 
sum-of-the-characters principle, and not in the presence or 
absence of skeleton merely. The reservation is, that if we 
maintain these two groups we must include in the Madre- 
poraria some forms without skeleton. JI am not familiar 
enough with Madrepores to generalize about them, but am 
relying on the details given in Duerden’s paper—from which 
I gather that there are certain aspects of their soft parts 
which present a fair degree of uniformity through the group. 
With the Actiniaria, as hitherto limited, this is not the case ; 
but if certain forms were removed from among them it would 
be so to a more reasonable extent. There are two families of 
forms, hitherto called anemones, which have all the charac- 
teristics of coral-polyps save a skeleton—in fact which are 
corals but for that one thing. If these two families be removed 
from the Actiniaria and placed under Madreporaria in some 
way, the division into anemones and corals at once becomes 
more intelligible, and various difficulties disappear. The 
families in question are the Corallimorphidae and Disco- 
