100 GEODIA MESOTRIAENA. 



afferent anil efferent oanals proper are cribriporal, some of the pore-groups being 

 the beghuiings of the afferent, the others the termini of the efferent system. 

 Some of the radial choanosomal canals, as stated above, are surrounded by thick 

 mantles of tissue free from flagellate chambers, others are not so surrounded. 

 I think that the former are efferent, the latter afferent canals. Since in G. 

 mcsutriaena the walls of the tubes leading down from the large depressions on 

 the outer surface have, as in G. yniilleri, a cortex, continuous and virtually iden- 

 tical with the cortex of the outer, exposed parts, I do not doubt that the tubes 

 themselves are in the former, as in the latter, produced by a plicature and local 

 fusion of the growing sponge in such a manner that their lumina are in reality 

 outside the sponge and the tubes themselves not to be consiflered as canals 

 I^rojier. As the afferent pores seem to predominate on the outer, exposed sur- 

 ^face, and as efferent pores only seem to occur in the walls of the tubes, I con- 

 sider the lumina of the latter as annexes of the efferent canal-system, that is, 

 as praeoscular cavities. 



The specimen of var. megana, the best preserved of the three, is, in its 

 histological stntciure on the whole similar to G. viiilleri, but also exhibits some 

 peculiarities. There is no accumulation of stainable cells (nuclei) at the surface. 

 The dermal membrane is traversed paratangentially by slender spindle-shaped 

 elements, drawn out at each end to a fine thread. The central swelling meas- 

 ures 1.5-2 n, each terminal thread 0.3 fi in thickness. These elements consist 

 of a somewhat granular substance, the granulation being coarser and more 

 distinct in the spindle-shaped thickening than elsewhere. Lower down in the 

 outer layer of the cortex similar fibres, not situated paratangentially but ar- 

 ranged irregularly, occur. The fibres joining the sterrasters stain only very 

 slightly with azure but deeply with iron-haematoxylin. In the proximal part 

 of the cortex paratangential threads are often observed. These are not homol- 

 ogous to the threads in the distal part of the outer cortical layer but appear 

 as the connecting fibres, stretched out between the most proximal of the sterr- 

 asters. Since these sterrasters at the proximal limit of the sterraster-layer are 

 usually much farther apart than those above, the fibres connecting them are 

 often of considerable length. The differentiated contractile tissue, forming a 

 ring, 300-400 // broad, round the proximal part of the chonal canal, the chone, 

 is brown in colour. I assume that all the chones observed are considerably 

 contracted. The greater, outer part of the chone consists of a tissue composed 

 of circular fibres (Plate 22, fig. 10c) and scattered euasters. This tissue does not 

 extend right down to the chonal canal, a layer of massive or radially elongated 



