494 T. A. STEPHENSON 
anemones in a fairly detailed way; the residue of forms is 
much larger, and there will not be space available in Part II 
for as much detail. I have not set apart a section of the 
paper as a criticism of the classification I wish to modify, as 
it has economized space to let objections emerge here and there 
in connexion with the individual changes suggested. Part I 
tried to clear the ground and discuss the method of attack, 
so that the arguments there given need not be repeated, and 
so that the general principle and method suggested there might 
be taken for granted in Part II. I should like to record here 
that in these papers on Classification there will be found points 
in contradiction to certain remarks in earlier papers— Terra 
Nova’ and ‘ Actiniaria collected off Ireland’—but the point 
of view is bound to become modified in some particulars as 
further experience opens new vistas. That the view-point 
should remain immovably fixed in the light of developing 
knowledge would more need apology than that it should march 
with necessity. Work on Part IT has served only to strengthen 
and confirm the plan suggested in Part I of this paper. 
Definitions to be given in Part III are based as far as possible 
on anatomically-described species, leaving the more doubtful. 
forms to fit themselves in as knowledge of them increases. 
Consequently lists of species given include rather the better- 
known forms on which the definition is founded, than exhaus- 
tive enumerations. Even to identify an anemone from an old 
figure or description is very risky ; to be sure of an old species 
one must obtain and re-describe the type-specimens if such 
exist. If there are none, it is guess-work—cf. Pax (75), p. 309, 
and others. 
One result of working through all the Actinian genera 
(supported by a personal anatomical study of a large number 
of them) is the recurrence of impressions connected with the 
difficulty of species-identification of some of them from 
preserved material—and the unfruitfulness of the pursuit. 
It would seem that family and genus are fairly easily tracked 
down when once a certain number of data are gained, and that 
these are intelligible quantities. But when it becomes a matter 
