VIII CALCAREA ASTROSCLERIDAE I 95 



united by their surfaces to the total exclusion of soft parts 

 (Fig. 88). Each element consists of crystalline fibres radially 

 disposed around a few central granules, and terminating peri- 

 pherally in contact with the fibres of adjacent elements. Young 

 polyhedra are to be found free in the soft parts at the surface. 

 Tlie chambers are exceptionally minute, especially for a calcareous 

 sponge, comparing with those of other .sponges as follows : — 



Astrosdem chambers, 10/xX S/j, to ISfxX 11/u,. 

 Smallest chambers in Silicea, 15//, x 18/t to 'I-i/xx '31 fx. 

 Smallest chambers in Calcarea, 60/i x 40/i,. 



In its outward form Astrosdera resembles certain Pharetronids. 

 The minute dimensions of the ciliated chambers relegate Astro- 

 sdera to the Micromastictora, and the fortunate fact that the 

 calcium carbonate of its skeleton possesses the mineral characters 

 not of calcite, but of aragonite, renders it less difticult to conceive 

 that its relations may be rather with the non-calcareous than 

 the calcareous sponges. 



BEANCH II. MICEOMASTICTOEA 



All sponges which do not possess calcareous skeletons are 

 characterised by choanocytes, which, when compared with those 

 of Calcarea, are conspicuous for their smaller size. The great 

 majority (Silicispongiae) of the non- calcareous sponges either 

 secrete siliceous skeletons or are connected with siliceous sponges 

 by a nicely graded series of forms. The small remainder are 

 entirely askeletal. All these non-calcareous sponges are included, 

 under the title ]\Iicromastictora, in a natural group, opposed to 

 the Megasmastictora as of equal value. 



The subdivision of the Micromastictora is a matter of some 

 difhculty. The Hexactinellida alone are a well circumscribed 

 group. After their separation there remains, besides the askeletal 

 genera, an assemblage of forms, the Demospongiae, whicli fall 

 into two main tribes. These betray their relationship by series of 

 intermediate types, but a clue is wanting which shall determine 

 decisively the direction in which the series are to be read. The 

 askeletal genera are the crnx of the systematist. It is perhaps 

 safest, while recoe;nisinn: that manv of them bear a likeness of 



