110 THE SPONGES. 



however, strengthens the genus, which may be a useful one. Topsent 

 characterizes it as having but one sort of streptaster, and that a spiraster, 

 which definition excludes my form. Lendenfeld (1903) retains Ndhea 

 Sollas as a subgenus of Pachastrclla, a genus which, as conceived by Len- 

 denfeld, is heterogeneous, including as it does forms with eurypylous 

 {Poecillastra Sollas) and forms with aphodal [Pachastrella dbyssi Schm., 

 Sollas, 1888, Plate XI.) chambers. According to Lendenf eld's classification 

 my species, Foecillastra tricornis, owing to the degeneration of the tetraxon 

 rhabdome, would be separated from the above-mentioned " Challenger " 

 species of Poecillastra, and would be classed along with the certainly very 

 different Netliea nana Carter. I cannot but regard this part of Lendenfeld's 

 classification as artificial. 



In the absence of tetraxons from the interior, Poecillastra crihraria agrees 

 with Characella Sollas. But this is a negative point of resemblance, which 

 cannot be used to exclude the species from Poecillastra, since the tetraxons 

 are rare in one of the type species of the latter genus. If Characella Sollas 

 is maintained at all, it must be based, as Topsent (1902) points out, on some 

 other and positive characters. In Topsent's classification, the microscleres 

 especially are made use of for this purpose, and Characella is re-defined as 

 having microxeas and amphiasters, which effectually excludes my species. 



In re-defining Poecillastra, Topsent (1902, p. 10) has altered the original 

 diagnosis of Sollas so as to make the genus include forms in wliich the 

 calthrops are absent from the interior: " Triaenes inegalement developpes 

 suivant les especes, souvent rares, localises a la peripheric ou epars en outre 

 a Vinterieur ou ils simulent des calthropses." I would adopt this emen- 

 dation. The existence of Poecillastra tricornis makes it necessary to add 

 to the definition the following clause : An actine of the tetraxon may de- 

 generate, the spicule becoming characteristically in some species a triod. 

 This qualification is really implied in Sollas's definition, since he includes 

 under Poecillastra, P. [Paclmstrella) amygdaloides Carter, the tetraxon of 



which is a triod. Topsent's definition includes a clause which must be 



modified : " Des microxes epineiix par tout le corps." The microxeas are 

 " roughened or minutely spined or smooth " in one of the type species, 

 " roughened " in another, and apparently smooth in the others, since a 

 spinous surface is not mentioned. 



