BY E. F. HALLMAXN. 635 



and Dendy(33) to receive also the species named by them T. 

 cervicornis, and was defined by them tlius: "Sponge ramose, with 

 a dense central axis of spiculo-fibre; megasclera styH and (or) 

 oxea, and (in some species) cladostrong3'la. Microscleres present 

 in the form of trichodragmata." More recently Dendy(8) has 

 amplified the definition so as to embrace in the genus all Axi- 

 nellidie in which the microscleres are trichodragmata and the 

 skeleton is more or less plumose. A similar disposition to employ 

 the genus in a wider sense than that in which it had been under- 

 stood by Ridley and Dendy, had previously been shown by 

 Topsent(46), when he assigned to it provisionally, under the name 

 ThrinciCophoraiV) spissa, a species of massive habit and halichon- 

 droid skeletal structure, with oxea alone as megascleres. This 

 species was also admitted in the genus by Dendy; but as the 

 result of a second investigation of it Topsent(53) has found that 

 the microscleres include toxa (in addition to trichodragmata), 

 thereby definitely establishing the correct position of the species 

 to be in the genus Gellius as defined by Lundbeck(30). The 

 known species that propei4y admit of inclusion in Thrinacojihora 

 as defined by Dendy, are ten in number*, — comprising, in addi- 

 tion to those already assigned to the genus, Axindla padina 

 Topsent(47), Easpailia [Syringella) rhaphidophora Hentschel(15), 

 and the species originally described by Whitelegge as Spongo- 

 sorites variabilis. The great diversity of spiculation and skeletal 

 structure exhibited by these species renders it obvious that they 

 do not constitute a natural genus; and the only justification for 

 their association together in a single genus would be the impos- 

 sibility of separating them into simpler and apparently more 

 homogeneous groups susceptible of precise and adequate defini- 

 tion. It is easy, however, to subdivide them into at least four 

 such groups, distinguished by differences sufficiently great to be 

 regarded as generic. I px'oj^ose, therefore, to restrict the name 

 I'hrinacopJiora to the species with special dermal megascleres, 



* Since this was written, I have discovered the existence of trichodrag- 

 mata in Whitelegge's Ciocalypta incnf<tans(56), which, therefore, consti- 

 tutes an eleventh species of this kind. 



