1 64 SPONGES 



an evolutionary series in which the siliceous material of the triaxon 

 spicule has become replaced by spongiii, giving rise to horny spicules, 

 combined with a dendritic horny skeleton, as in Darwinella. Further steps 

 in this direction would lead to loss of the liorny spicules, as seen in the other 

 Dendroceratina ; while finally, the Halisarcidae are supposed to arise by 

 complete suppression of the entire horny skeleton. However interesting 

 and suggestive this theory may be, it cannot be said, in the present state 

 of our knowledge, to be more than a guess. It is not clear how the 

 alleged Hexactinellid ancestor acquired its dendritic horny skeleton, and 

 it is just as easy to take the Halisarcidae as the starting-point of the series, 

 to derive the dendritic skeleton from an upgrowth of a basal plate of 

 spongin, and to regard the horny spicules, present in a single genus, as 

 originating liy discontinuous secretion of spongin in a form already 

 provided with a dendritic skeleton. 



While, in tlie Myxospongida, the athnities of the Halisarcidae 

 are uncertain and in need of further elucidation, the Oscarellidae, on 

 the other hand, seem to stand very near the Plakinidae. Oscardla 

 scarcely differs from the simpler species of Plakina except in the 

 absence of any skeleton, and there is no evidence whatever that it 

 is degenerate in this respect. Oscardla may be regarded therefore as 

 in many respects the most primitive Demosponge, representing more 

 than any other the simple Khagon ancestor of the group. Since, 

 on the other hand, it seems to have little affinity with any 

 Hexactinellids, it points to the siliceous skeleton having been 

 independently evolved in the ancestors of the Triaxonia and 

 Tetraxonia respectively. 



The Hexactinellids, unlike the Demospongiae, are a compact and 

 homogeneous group of very uniform structure, presenting no special 

 phylogenetic difficulties. The starting-point is a simple Rhagon- 

 like form, as described above, from which all the known types are 

 easily derived. 



The i)liylogen)' of the Calcarea has alrcad}'' been briefly dis- 

 cussed, and it has been seen that the two families of the Honiocoela 

 represent the two main divergent branches of the genealogical tree. 

 It is not possible at present, however, to trace these branches upwards 

 through the grade Heterocoela, until the latter have been further 

 studied from this point of view. 



LlTKRATUItE OF THE PORIFEKA. 



The following list of rcfereuces comprises, in the first place, those memoirs oi 

 monographs which deal with some sponge question or group in a comprehensivi' 

 manner, and contain exliaustive references to the literature ; and, in the second 

 place, works of very recent date, wliich supplement and extend the larger 

 treatises. Many important memoirs are, therefore, not cited separately, as they 

 are to be found (juoted by all authors ; such as, for instance, in morphology, 

 histology, and development, tlu' chissical memoirs of Carter, Lieherkidiii. 



