E. MARSHALLI. — TRABECULE. 147 



rouuded by three or four adjacent cells, there were sometimes 

 found roundish pores, Avhich should allow the entrance of water 

 into the chamber. In E. mavshalli I have seen no trace what- 

 ever of this kind of connection between the cells. It is jirobably 

 a sort of an accessory arrangement which is not of nuivi^rsal 

 occurrence among the Hexactinellida. 



The trabecules. — These are in general fine and thread- 

 like, in places band-like. The disposition of their branching and 

 anastomosing strongly reminds one of the irregular web woven 

 by certain spiders {tr., PL IV, fig. 22 ; PL V, figs. 36, 43 ; see 

 also PL VIII, figs. 29, 30). The meshes are of quite indefinite 

 shape and size. Generally speaking, the trabeculfe stand for 

 the mesenchyme of other sponges, but never and nowhere do they 

 form a bulky mass or a compact layer of any considerable 

 thickness'", though in places they may be expanded into film-like 

 membranes. Such expansions occur here and there in the deeper 

 parts, at points where three or more trabecule join together, 

 aud are in appearance somewhat alike the nodal confluence of 

 the filamentous pseudopodia of certain Rhizopods. 



The same membranous development of the trabecuhe, but on 

 a far greater scale, is seen in certain definite positions, especially 

 on the surfaces delimiting the sponge from the exterior. Here 

 belong the dermalf, the gastral aud the canalar membranes, as 



*The diagi-auiraatic iigiire, recently given by Delage and IIerouakd in the ' Zoologie 

 Concrete' (T. II, PI. 8, fig. 4), representing the rchition of parts iu the waU of EuijkclcUa, 

 is fitted to give an altogether erroneous idea of the stnictiu-e, in that the choanosome is 

 shown as a thick folded layer of compact ' meioderme ' inclosing the chambers, while botli 

 tlie ectosnme iiud the endosome are given likewise as thick, minutely perforated layers con- 

 nected witii the ch<ianosome by solid pillars. Their other diagrams on PI. 11, relating to 

 the Hexactinellida, are miicli better. But tliesc, as also the matters embodied iu the text, 

 do not call for special comment, being entirely based on the representations of F. K. Schulze. 



t For the fact that this layer in H. vMnhaUl hardly deserves to be called a membrane, 

 on account of the general thread-like ilevelopmcnl of its beams, see ante, \i. 12u. 



