EEPORT ON THE KERATOSA. 83 



As I remarked before, the realisation of this demand necessitates the introduction of 

 subgenera ; and I am the more in favour of this because otherwise I see absolutely no 

 possibility of arranging the Keratosa in a manner not conspicuously artificial, while in the 

 contrary case we should obtain a very natural family of Ceraospongise with the following 

 quite homogeneous genera: — lanthella, Dartvinella, Simiilicella (with two subgenera 

 Ai^lysilla and Dendrilla), SiMngia (embracing forms classed now in the conjectural 

 genera Euspongia, Hipi^o&pongia, Cacospongia, Stelospongos, and Coscinoderma), 

 Phyllospongos (with subgenera Phyllospongia and Carteriospongia, provided that these 

 subgenera are really in a close relationship), Spongelia (with subgenera Dijsidea and 

 Psammoclevia), Psammopemma (supposing that this conjectural genus be not connected 

 with Spongelia by intermediate stages), and Velaria (including forms referred at the 

 present time to Ap>lijsina, Verongia, and Luffaria, which may perhaps all be elevated 

 to the rank of subgenera); and it is clear that each of these genera either admits of a very 

 sharp diagnosis or evidences its right to be regarded as a genus by series of characters 

 concerning the whole of the organisation. If now the question be asked, why instead of 

 following, in the descriptive part of this memoir, the scheme just exposed, I have yet 

 followed the arrangement of Dr. Vosmaer, the answer will be because the above scheme 

 only partly decides the problem of a natural arrangement of horny sponges, and I myself 

 am of opinion that when new systems are proposed they must be well established in all 

 their parts. The reader will remember that in the foregoing three chapters we met with 

 a large number of contradictions issuing from the present mode of classifying the Keratosa. 

 The scheme in question reconciles most of them ; it does not do so, or at least but partly, 

 with respect to what I called circulus vitiosus, characterising the mutual affinities of the 

 genera of the group. This circulus vitiosus is striking as concerning the genera belonging 

 to different families ; it has, however, equally little right to exist as concerning the mutual 

 relationships of the subgenera. It would have been very easy to proclaim the Keratosa 

 as forming but a single family, and to classify according to this the specimens of the 

 Challenger Collection ; but by this proceeding we should not have got rid of difficulties 

 concerning questions of course of a more subordinate nature but nevertheless of vital 

 importance ; this proceeding would not have decided the problems as to whether sub- 

 generic value can be really ascribed to the characters distinguishing the HippospongicB or 

 Coscinodermata, &c. This is the second, and the most difficult part of the task, and 

 without the help of Palseontology and Comparative Physiology it will be scarcely decided. 

 As must be evident from the above discussions, by the term " subgeneric character " I 

 understand a character containing in it a new principle of organisation, the corresponding 

 representatives of the group being connected by intermediate stages. That Euspongia, 

 Cacospongia, Hippospongia, &c., present each in their organisation what may be called a 

 new principle is clear ; but it is by no means evident whether this, their conjecturally 

 main character, be really constant. Should we feel certain that the fossils described as 



