1914 ] N. Annandale : Sponges of Lake Baikal. 141 



structure, the terminal or subterminal spines of the gemmule- 

 spicule form a regular crown or rotule at both extremities ; in 

 Pectispongilla they fuse together at both ends of the spicule to 

 form smooth-edged disks ; in Tubella one of the rotules begins to 

 disappear and a trumpet- shaped spicule is the result, while in 

 Parmula this rotule has vanished together with the greater part of 

 the shaft of the spicule, which takes the form of a flat plate 

 (representing the other rotule) with a spine (all that is left of the 

 shaft) projecting upwards from its centre 



The evolutionary development of the free microscleres or 

 flesh-spicules of the Spongillidae is much less striking than that of 

 the gemmule-spicules and need not be considered here. In several 

 genera and many species these free microscleres disappear alto- 

 gether; whereas this is the case with the gemmule-spicules only in 

 a few degenerate forms. 



The skeleton of the Spongillidae differs in no essential feature 

 from that of the Haploscleridae ; but the chitinoid sheath of the 

 spicule- fibres, if it exists at all, is never so stout as it is in some 

 Haploscleridae, notably in those of the subfamily Chalininae, and 

 the lattice-like network of single spicules characteristic of Reniera 

 among the Haploscleridae is never found in its full development. 



The free-swimming larva of the Spongillidae has a very 

 characteristic structure, consisting of a hollow, bladder-like body, 

 entirely covered externally with homogeneous cilia and invariabl}^ 

 without a pigment spot. 



The most characteristic feature of the SpongilHdae has, 

 however, as yet been mentioned only incidentally, viz. the 

 elaboration of the gemmule. 



Gemmules are produced by many Haploscleridae, but consist 

 merely of masses of cells stored with food-material and enclosed 

 in a simple chitinoid case without specialized spicules or a pneuma- 

 tic covering. In the Spongillidae on the other hand both these 

 structures are commonl}^ associated with the gemmule ; in the 

 subfamily Spongillinae the one critical character of most of the 

 genera is the form of the microscleres with which the gemmule is 

 armed, a foraminal tubule or cup (or at any rate a very definite 

 depression in the covering at which the contents of the gemmules 

 may escape on germination) is found in all but a few cases, while 

 in most instances there is a special coat of chitinoid substance 

 containing air-spaces of one kind or another. In the subfamily 

 Potamolepidinae, in which microscleres of all kinds are absent, the 

 gemmule, if it exists at all, is of a much simplified nature and 

 resembles in many respects that of the Haploscleridae : that this 

 is the result of convergence rather than of genetic relationship 

 is proved by the very close structural resemblance between cer- 

 tain Potamolepidinae and certain Spongillinae not of a primitive 

 type. 



Hexactinellida, free microscleres very similar superficially to the gemmule-spicules 

 of Ephydatia have been produced in totally different lines of evolution. 



