441 



over to some extent the change from the larval to the adult condition, 

 so far as the relative position of the layers composing the body is 

 concerned. 



I pass to the consideration of another structural feature, in which 

 all existing Hexactinellid sponges are less archaic than the Ascons among 

 the Calcarea. In the latter we find the only instance known to occur 

 among sponges of the gastral being a continuous, uninterrupted layer, 

 not folded to form distinct flagellated chambers. In the Hexactinellids, 

 on the other hand, even in the youngest specimens that have been 

 examined, the gastral layer occurs invariably in the form of thimble- 

 shaped chambers, forming the middle or chamber-layer of the body- 

 wall, and an Ascon-like condition is nowhere found in recent members 



A. B. C. 



Fig. 1. Diagi'ams to represent the hypothetical evolution of the structure of Hexac- 

 tinellida, as shown by vertical sections of the body-wall. The thick black line repre- 

 sents the gastral layer, the finer lines the trabecular system of the dermal layer. The 

 outer surface is towards the right hand, the inner surface towards the left, in each 

 illustration. For further explanations, see the text. 



of this group. Nevertheless, all the facts and analogies of sponge- 

 morphology lead irresistibly to the conclusion that the homocoele con- 

 dition is the more primitive, and must have preceded the heterocoele 

 condition with distinct flagellated chambers. We may therefore assume 

 for the Hexactinellids also an ancestral type of structure in which the 

 collar-cells were arranged as a continuous layer, as in the present-day 

 Ascons, but here suspended in the midst of the body-wall, with the 

 trabeculae of the dermal layer situated both internal and external to it. 

 (.1, fig. 1). I think it extremely probable that such a type of structure 

 was the ancestral condition both in the Hexactinellids and the Calcarea, 



