712 PROF. A. DENDY AND MR. R. W. H. ROW ON 



were expovmded by one of us some twenty years ago [Dendy 

 1891 A, 1893 A], as the following review of our present position 

 will show. 



The canal system, including the form of the flagellate chambers, 

 is, we are convinced, of comparatively little taxonomic value in the 

 higher Calcarea. In the lower forms it necessarily determines the 

 arrangement of the skeleton, which must lie in the walls of the 

 ascon tubes, however these may be arranged, and in the Sycet- 

 tidpe the arrangement of the radial tubes has undoubtedly been 

 the determining factor in the development of the articulate tubar 

 skeleton. With the appearance of a definite dermal cortex, how- 

 ever, the arrangement of the skeleton begins to vary more or less 

 independently of the canal system, so that with an identical 

 canal system we find such different types of skeleton as that 

 of the Grantiidse, the Heteropiidse and the Amphoriscidse. In 

 each of these families, while the type of skeleton remains fairty 

 constant, the canal system ranges from syconoid to leuconoid, or 

 at least sylleibid. The syconoid type is again met with on a 

 totally difi'erent line of descent in the homocoel genus Dendya, 

 and also in Leucaltis, and again gives rise to a leuconoid type 

 both in the Leucascidaj and the Leucaltidse. 



Our view that it is the canal system rather than the skeleton 

 that has repeatedly undergone convergent evolution is strongty 

 supported by the distribution of the different types of spicules 

 and of the two types of collared cells. The form of the spicule, 

 however, must be used with great caution as a guide to genetic 

 relationships, for it is largely a question of adaptation. The tri- 

 radiate is undoubtedly the fundamental spicule form in the group, 

 but one might almost say that it tends to become quadiiradiate 

 on the slightest provocation. Thus we almost invariably find 

 quadriradiates in the gastral cortex, whose inwardly directed 

 apical rays are undoubtedly of great value as a protection against 

 enemies, such as small crustaceans, approaching through the 

 osculum. Then, again, the ordinary triradiates of the dermal 

 cortex not infrequently develop a more or less conspicuous, 

 centripetally directed, apical ray; and this latter tendency appears 

 to have led, in the case of the Amphoriscidse, to a constant 

 skeletal character which foi-ms the most chai-acteiistic feature of 

 the group. Similarly with regard to the distinction between 

 equiangulai- and sagittal trii-adiates, we find that the latter can 

 always be developed, when the situation in the sponge demands 

 this form, by the bending back of the oral rays during growth. 

 This nearly always takes place, for example, in the oscular collar, 

 where there is no room for the oral rays to extend forward at the 

 usual angle. 



Whether or not there is a fundamental difference between an 

 equiangular triradiate, however its rays may be bent, and an 

 alate one in which the primitive oral angle is really different 

 from the paired angles, and in which there is a correspond- 

 ing difference in the position of the crystalline optic axis, as 



