SPONGES 109 



three mother-cells, each of which divides into a basal and an apical 

 formative cell, but the unpaired ray at first greatly outstrips the other 

 two in its growth. 



Classification. The earliest general classification of the Calcarea was 

 that of Haeckel [7], who divided them by characters of the canal 

 system into Ascons, Sycons, and Leucons. Each of these groups was 

 further classified into seven genera, each genus being characterised by a 

 skeleton made up of one of the seven possible combinations of the three 

 types of spicules. 



The threefold division proposed by Haeckel has generally been super- 

 seded by the binary classification of Polejaetf [18], who divided the entire 

 group into Homocoela, with the gastral layer continuous, and Heterocoela, 

 with the gastral layer discontinuous. The former group comprises 

 Haeckel's Ascons, the latter his two remaining groups. 



There can be little doubt that Polejaeff's two groups do not represent a 

 natural classification of the group, but only two tjrades of structure. His 

 classification is, in short, a horizontal cleavage of the phylogenetic tree, 

 not a vertical one. It is highly probable that the Heterocoela are a 

 polyphyletic group, derived from more than one stock of Homo- 

 coela. 



Amongst the Homocoela we have two very sharply defined families ; 

 on the one hand, the Clathrinidae with reticulate form, equiangular 

 triradiates, collar cells with basal nucleus, and parenchymula larva 

 (Ascetta line) ; on the other hand, the Leucosoleniidae with erect form, alate 

 triradiates, collar cells with apical nucleus, and amphiblastula larva (Ascyssa 

 line). The divergence between the two families of Ascons indicates the 

 deepest phylogenetic cleft in calcareous sponges. While the majority of 

 the Heterocoela approach the Leucosoleniidae, a few forms (e.g. Heteropegma) 

 certainly find their nearest allies among Clathrinidae. Hence a truly 

 natural classification of the Calcarea must proceed along these lines. 

 Nevertheless?, any such classification, though to be looked for in the 

 future, seems to us premature and inconvenient at present. The 

 Heterocoela have not yet been studied in detail from this point of view, 

 and their phylogenetic connections are not yet sufficiently unravelled. 

 We cannot therefore adopt here for practical purposes the division of 

 Calcarea proposed by Bidder (1898) into the two groups Calcaronea 

 (Calcarea on the Ascyssa line) and Calcinea (Calcarea on the Ascetta line). 

 We retain for the present the two groups of Polejaeff, not as natural 

 orders, but as two grades of structure, indicating a frankly artificial 

 classification. 



Rauff has recently proposed to divide the Calcarea into two divisions 

 Dialytina, with spicules separate, and Lithonina, with spicules united into 

 a continuous framework (Petrostroma). This classification is obviously 

 unsuitable for the entire group, but may be usefully employed within 

 the limits of Pharetronidae, where we retain it. 



As regards families, we adopt in the main the grouping proposed by 

 Dendy, but we are unable, in the first place, to retain his so-called 

 heterocoelous family Leucascidae. The true position of the forms included 

 in this family is amongst the Clathrinidae. In the second place, we retain 



