[*o the principle involved in the firsl of these groups I would offer no objections, the 



inthiniaria undoubtedly standing in closer phylogenetic relationship to the II' actiniaria than 

 to any of the other groups, and just as undoubtedly the Madreporaria ai ly related to 



the Actiniaria, an idea sug by Hertwig and abundantly confirmed by recent studies oi 



the soft parts and development of the coralligenous forms (see especiallj Duerden, 1902 With 



,r>l to the third group, however, I am entirely in disaccord. In the first place 1 cannot 



elationship that the arrangement predicates for the Ceriantharia and Antipatharia. 



Indications of affinity van Beneden finds in the presence in both groups ol a longitudinal 



idermal musculature in the column wall, in the weak development of the mesenterial mus- 

 culature and in the similarity in the arrangement of the principal mesenteries. The first two 

 characters do not seem to me to be entitled to great importance, occurring and disappearing 

 .is they do in various groups of Anthozoa; the third is of significance. But is there similarity in 

 this Van Beneden finds it by recognizing but three protocnernes in the Ceriantharia 



and in this 1 believe he is mistaken. I can sec no reasons for regarding the fourth couple ol 

 m. , that appear in Cerianthan larvae as other than protocnernes, and in later pages ol 



this Report I shall present further evidence that is afforded by the development in favour of 

 this view. 1 hope to show that the development <>f what I shall term the deuterocnemes in the 



riantharia begins with the fifth couple of mesenteries and that, consequently, the members 

 of that group take origin from an octamerous condition. Whether the Antipatharia really start 

 trom an hexamerous condition in which there are only three couples ol protocnernes or not 



.not be determined with certainty at present; we must wait tor further light on the «piestion, 

 until. perhaps, a study of the embryonic development ^\ some member of the group may offer 

 data for a final conclusion. In the mean time it seems that any such close alliance of Ceriantharia 

 and Antipatharia as van Beni den has proposed is inadvisable. 



In the association of the Scyphomedusae with the Ceriantharia, etc, I believe van Beneden 

 has also fallen into error. Certainly these medusa are more nearly related to the Anthozoa than 

 ;■■ the I Ivdromedtisa-, but instead of forming a suli^roup of a subgroup of the Anthozoa the\ 

 should be of equal value with that group. Their relationship traces back only to remote ancestors, 

 to a stage before the differentiation of the Alcyonaria, Ceriantharia, etc, and to make the 

 phomedusse equivalent to these is an imperfect expression of their phylogenetic atlmities. 

 Finally, with regard to the Rugosae it seems probable that van Beneden himself, with 

 the results obtained by DüERDEN (1905) now at his disposal, would materially alter his views 

 as to their affmities and recognizc their close relationship to the Zoanthactiniaria. The evidence 

 which Duerden has presented as to their primary hexactinian characters seems wel) founded, 

 and while I am not inclined to recognize tbr them so close an affinity to the Zoanthinaria 

 he suggests, nor \ et to follow Carlgren in including them among the Madreporaria. yet I 



lieve there is reason for associating them with both these groups to the same extent that 

 in associates the Zoanthids and Hexactinians. 

 I would accordingly do away with van Beneden's group Scyphactiniaria and frame a 

 much more resembling that proposed by Carlgren (1908), though differing trom 

 this in tlv n assigned to the Rugosae. Carlgren's classification is as follows: 



