DENDY— HOMOSOLEROPHORA AND ASTROTETRAXONIDA 243 



Aurora providentice and A. rowi. In addition I propose to include in this genus three 

 reduced or 'epipolasid' species, viz. Coppatias (Rhahdastrella) distinctus Thiele [1900], 

 Diastra sterrastrcBa Row [1911] and Aurora cribriporosa n. sp. 



Aurora glohostdlata remains the type of the genus, and I therefore call special 

 attention to the fact mentioned later on (p. 247), that it probably -does not possess tricho- 

 dragmata*, as described by Sollas. 



The genus is of great phylogenetic interest as representing almost certainly the 

 starting point for the evolution of the Geodiidae, as well as for other reasons. To work 

 out this problem as fully as it deserves would require more time and material than I at 

 present have at my disposal, but the following considerations appea,r to me to leave little 

 doubt as to the results which such an investigation would yield. 



The Geodiidse are, as is well known, distinguished by the possession of a very 

 peculiarly modified spheraster, known as the sterraster, and these spicules form a dense 

 cortical crust. In the more typical forms of Aurora, such as A. glohostellata and A. provi- 

 dentice, the characteristic microsclere is a large spheraster with conical rays, very like that 

 of Donatia or Chondrilla, and these also are arranged in a cortical layer. Now in certain 

 species of Aurora, such as A. memhranacea (Hentschel), A. sterrastrcea (Row) and A. rowi, 

 n. sp., the typical spheraster is either associated with or replaced by a peculiarly modified 

 spheraster resembling a sterraster, which I propose to call a sterrospheraster (Plate 46, 

 fig. 4 c). What appears to be this type of spicule was indeed actually described as a 

 sterraster by Hentschel [1909] in his Isops (Aurora) memhranacea. The same type of 

 spicule was described by Row [1911] in his Diastra sterrastrcea, which may be looked 

 upon as an Aurora with reduced spiculation. Row pointed out the resemblance which it 

 bears to a sterraster. He also noted the absence of a " hilum," and it is possible that this 

 may prove to be a distinctive feature of the sterrospheraster, though I think it hardly 

 likely. 



Row gave some account of the development of the sterrospheraster in Diastra, and 

 I have been able to work it out somewhat more fully in the case of Aurora roivi [vide 

 infrd and Plate 46, fig. 4). It certainly resembles pretty closely the development of a 

 typical geodiid sterraster, but at the same time it passes through a stage in which it is 

 a fairly typical spheraster, identical, in fact, with the spheraster of Aurora aurora, which 

 has no sterrospherasters. There can be no doubt that the typical spheraster of Aurora, the 

 sterrospheraster, and the sterraster of Geodia, are all closely related spicule-forms, the 

 sterrospheraster being in some respects intermediate between the other two. At the same 

 time we must not forget that a "sterrospira," practically, indistinguishable in the adult 

 condition from the sterraster of the Geodiidce, has arisen independently in the spii^astrellid 

 genus Placospongia, as shown by Vosmaer and Vernhout [1902]. 



The sterrospheraster, however, actually occurs in certain undoubted Geodiidee, as, for 



example, Geodia carteri Sollas [1888, p. 247], associated with true sterrasters. Carter, 



who originally described that species from the south coast of Australia under the name 



Geodia canaliculata, Sdt., and figured the spicules [1883 b], regarded the sterrospheraster 



* Sollas [1888] uses the term " orthodragma," but the spicules in question are identical with the 

 spicules described by Ridley and Dendy [1887] as " trichodragmata." 



31—2 



