SPONGES 145 



flagellated layer. In the families Poeciloscleridae and Axinellidae there is 

 no such ring of special flagellated cells, and the whole flagellated layer 

 is pigmented, while the exposed portion of the inner mass is unpigmented. 

 This may be compared to the way in which the families Clathrinidae and 

 Leucosoleniidae, amongst Ascons, are characterised by the possession of 

 parenchymula and amphiblastula larvae respectively. 



Classification. The subdivision of the class Demospongiae is a 

 matter of great difficulty, and one upon which little agreement is 

 to be found amongst the authorities ; not because the mutual affini- 

 ties of the various forms comprised in this group are not clear, but 

 on account of the very frequent occurrence of convergent evolution 

 and parallel adaptations. The characters which can most con- 

 veniently be used for denning and delimiting systematic groups, 

 and above all, the characters of the skeleton, have not always a uni- 

 form origin, and therefore do not indicate natural relationships. 

 It may, indeed, be said that at present, at any rate, it is not 

 possible to construct a system which shall be at once strictly logical 

 and perfectly natural. The most obvious and simple classification 

 is into four grades, characterised respectively (1) by the possession 

 of tetraxon spicules, (2) by monaxon spicules, without tetraxons, 

 (3) by a horny skeleton, without siliceous spicules, and (4) by the 

 absence of a skeleton of any kind. If these four groups are to have 

 any pretence to being natural, however, it is absolutely necessary 

 to overstep in every case the limits imposed by rigidly logical 

 definitions. Thus in the first sub-class, Tetraxonida, it is necessary 

 to include such forms as Placospongidae and Chondrosidae which lack 

 tetraxon spicules and sometimes even spicules of any kind, but 

 whose affinities with the other families of the sub-class are indicated 

 by a number of secondary characters. In the Monaxonida we have 

 three sub-orders which are less closely allied to one another than to 

 forms outside the group, and the same must be said of the two 

 orders of Keratosa. The climax is reached, however, when we come 

 to the so-called Myxospongiae, forms devoid of a skeleton. In the 

 first place, we have to remove Chondrosia, which, as has been said, 

 is undoubtedly a degenerate Tetraxonid. Of those that remain, 

 Oscarella is certainly a very close ally of Plakina, among the Tetrax- 

 onida, while Hexadella, and probably also Halisarca, seem to have 

 close affinities with the Dendroceratina amongst the horny sponges. 

 So long, however, as it is by no means certain, in the case of these 

 forms, whether their lack of a skeleton is due to degeneration, or 

 represents, as seems more probable, a primitive feature, and until 

 there is more evidence bearing upon this point, the genera in ques- 

 tion, in spite of their divergent affinities, may well be left as a sub- 

 class together, as representing, perhaps, a more primitive grade 

 of organisation than any other Demospongiae. It is inevitable 

 that any system at present proposed should be more or less of a 



