442 



confining my speculations for the present to these two groups. From 

 such a hypothetical ancestral form the OlyntJms, and consequently all 

 existing Calcarea, would be derived by the more or less complete res- 

 triction of the dermal layer to the outer or dermal side of the gastral 

 layer, as above pointed out and the acquisition of the calcareous skeleton, 

 while the Hexactinellids as known to us at the present day, on the other 

 hand, would be derived from it by folding of the gastral layer to form 

 the chambers, and by the acquisition of the characteristic skeleton of 

 siliceous triaxon spicules. 



In the evolution of the chamber-system of HexactinelHds we may 

 assume, therefore, three phylogenetic stages, which may be termed con- 

 veniently J., B, and C (fig. 1). In stage Ä we should have the collar-cells 

 as a continuous sheet of epithelium, not folded in any way, as in most 

 Ascons. In stage B this epithelium would be commencing to increase in 

 superficial extent by becoming thrown into folds, leading to a condition 

 similar to that found in some Ascons, as for examjDle Ascandra falcata. 

 In stage C, represented by present-day Hexactinellids, the folds of stage i? 

 have become distinct flagellated chambers. Stages A and B are both hy- 

 pothetical, but the condition found in Hyalonematidae approaches closely 

 to stage i?, since the chambers grouped round each excurrent canal are 

 continuous with one another at their excurrent apertures, the gastral 

 epithelium passing without interruption from chamber to chamber. 



The origin of the triaxon skeleton must now be considered. In his 

 masterly monograph of the Hexactinellida collected by the Challenger 

 expedition [10], Schulze has pointed out how exactly the type of spicule 

 regarded as most primitive in this group, namely the regular hexactine, 

 is adapted to form the support of a simple Hexactinellid sponge. While 

 in such a form one of the six rays is directed radially between the 

 chambers, the four rays at right angles to this lie tangentially in the 

 body-wall between the chambers. This theory gives a perfectly feasible 

 and natural explanation of the origin of the triaxon sj^icules, on the 

 sujDposition that the skeleton was acquired after the gastral 

 layer had become folded to form the flagellated chambers. 

 It is obvious, however, that the hypothesis requires modification, at 

 least, if we suppose the skeleton to have been already present during 

 the earlier stages of evolution when the gastral layer was still in the 

 condition of a continuous sheet of collar-cells. It was certainly during 

 a stage corresponding to my stage A, that the skeleton first appeared 

 in the Calcarea, as proved by the existence of the Homocoela, and it is 

 my belief that the same is true for the Hexactinellida also. If this were 

 the case, it is at once evident that the primitive form of spicule in the 

 latter group is not the hexactine, as Schulze 's theory assumes, since 



