SIL.ICIOUS AND HORNY SPONGES WILSON. 305 



occur in the development of the sterraster of Geodia and its allies. 

 He therefore separates Erylus, designating its sterraster as an aspi- 

 daster, from the other Geodiidae, creating for the genus a new- 

 family, the Erylidae. The remaining genera are left, " for the 

 present at least, in the Geodidae." 



Dendy (1916, p. 256) accepts Lendenf eld's conception of the genus 

 and the family Erylidae. This does not seem to me advisable, for 

 the following reasons: 



Erylus is (has been) characterized primarily by its complex of 

 adult spicules. Accepting this idea of the genus, we find that of the 

 25 species there are only eight (four in Lendenfeld, 1910&, two 

 in Dendy, 1916, together with E. formosus and cylindrigerus, see 

 Sollas 1888 (pp. 213, 240)), in which the sterraster is known to pass 

 through an aspidaster stage. It is by no means safe to assume that 

 the others pass through this stage, for in the species described in 

 this report, E. oomutus, they do not.. On the contrary, in this 

 species they develop in the usual way described for sterrasters 

 (Sollas, 1888, p. lxiv). Hence if we follow logically Lendenfeld's 

 proposition we disrupt Erylus — namely the group of species, charac- 

 terized primarily by a certain spicule-complex, and must assign 

 some species, as E. cornutus, to another genus and indeed family. I 

 take it, no one would wish to do this. Without minimizing in the 

 least the interest and value of Lendenfeld's new facts, it does not 

 seem to me that they are of such a kind as should influence the 

 definition of genera or families. 



Retaining Erylus in the older sense, we have to regard, with Sollas, 

 the sterrasters as variable not only in facial outline but in thickness: 

 in some species, very thin; in other species, comparatively thick. 

 When the sterraster is very thin, its mode of growth leads through 

 an aspidaster stage; but such spicules are not radically different 

 from other sterrasters, and detailed study will probably result in the 

 discovery of intermediate modes of development. 



This interpretation of the aspidaster as representing only an ex- 

 treme in a graduated series of morphogenetic methods, leading up 

 from that practiced in Geodia, receives support from the recorded 

 details. Thus Sollas (1888, p. 213) who regards the narrow, elong- 

 ated, flattened, sterraster of E. formosus as only a modification of the 

 spherical type, describes for this species a developmental stage made 

 up of a center and trichite-]ike rays, already of unequal length in 

 different regions (pi. 28, fig. 30), such that the shape of the spicule is 

 thereby determined. This stage which is clearly only a modification 

 of the corresponding one, with equal rays, in the formation of a 

 spherical sterraster, is followed just before the completion of the 

 spicule by a smooth stage (pi. 28, fig. 12), on which " the small spines 

 which granulate the surface " of the adult subsequently appear. 

 S1709— 25 1 



