758 PROF. A. DENDY AND Mr/r, W. H. ROW ON 



With the tiansition from the syconoid to the leuconoid type o 

 canal system in this family, and the correlated replacement of th 

 articulate tubar skeleton by irregularly scattered radiates, we ge^ 

 a close approach to the more advanced Leucascidse, such af 

 Leucetta and Pericharax, and we have here one of those cases o 

 convergence which are so frequently met with amongst sponges? 

 but we have already laid sufficient emphasis upon this point. 

 Even in the genus Leucmidra, however, subgastral sagittal tri- 

 radiates are usually present, and when they are absent their 

 absence must be regarded as secondary. 



So far as our experience goes the nucleus of the collared cells 

 is always apical in position in this family. We have been able to 

 determine it in 17 species, as enumerated in an earlier section of 

 this paper. 



The family is a very large one, comprising no less than 23 out 

 of the 51 genera of recent calcareous sponges which we recognise, 

 and containing a great diversity of structural types within it. 

 There are, however, very great difficulties in the way of dividing 

 it into subfamilies, the chief of these being the fact that the 

 possible methods of deriving the various genera from one 

 another within the family are manifold, and it is impossible 

 to determine satisfactoinly which are the true lines upon which 

 evolution has proceeded. We might, for example, place all 

 those genera which have a syconoid canal system and colossal 

 longitudinal oxea in the dermal cortex together in a subfamily 

 Utein?e ; or we might separate the genus Uteopsis from the 

 others, and unite it with Achramor2)ha and Anamixilla in a 

 subfamily characterised by the reduction of the tubar skeleton to 

 a single joint. But neither of these two possible subfamilies 

 would seem to be very sharply defined, and moreover, the cha- 

 racters in question are not confined to members of the Grantiidfe. 

 In short, we feel that in the present state of our knowledge it is 

 impossible to decide which method of grouping would express 

 most correctly the real afi'in.ities of the genera concerned. This 

 is the case with almost all the possible methods of grouping the 

 genera, and we have therefore decided not to attempt to split up 

 the family, but merely to indicate the approximate relationships 

 of the genera, so far as this is possible in a linear series, by the 

 order in which we have arranged them, 



Although it seems probable that the majority of the genera in 

 this family are descended from the genus Sycon, yet it is quite 

 possible that some of them may be descended independently from 

 Sycetta, and thex-efore that the family may be of diphyletic 

 oiigin. 



We have changed the name of the family from Grantidse to 

 Grantiidpe in accordance with the usual practice of systematic 

 zoologists. 



