Interrelationships of the Madroporidae. 131 



and Astraopora — can be usefully united in a second sub- 

 family ; so that, for the future, the Madreporidaj will consist, 

 so far as we at present know, ot" two subfamilies — the Madre- 

 porinff", conijirising three genera, and the Montiporinai, 

 comprising two genera. 



The strongest argument in favour of this classification lies 

 in the fact that the five genera can be deduced from a 

 common ancestral form. In describing this form we are, for 

 obvious reasons, confined to a consideration of its parent polyp, 

 and not of its colony. Every colony starts from a parent 

 polyp, and, indeed, receives its chief characteristic from the 

 structure, growth, and method of budding of this individual, 

 directly developed from the attached larva. Hence it is enough 

 if we can trace any group of colony formations back to a 

 common ancestral parent polyp. 



Reference to the analyses already given in this and in the 

 earlier papers on Turhinaria and Astrceopora shows that this 

 common parent polyp possessed the following leading charac- 

 teristics: — (1) a porous wall, with laminate radial structures; 

 (2) a well-developed saucer-shaped epitheca ; (3) the habit 

 of very early budding while the parent polyp Avas still very 

 small ; (4) the production of true buds, starting from the 

 smallest beginnings out of the sides of the polyp, and forming 

 their skeletons, at least in the first stages, upon and with 

 some slight modification of the radial symmetry of the porous 

 wall of the parent polyp*. 



From such a form we may deduce the genera under dis- 

 cussion along the following lines of specialization : — 



Madre2)ora. — The skeleton of the parent polyp grew in 

 height, and consequently somewhat in size, shooting upwards 

 in a tall cone with thickening base (fig. 4 a) . The buds grew out 

 in tiers from its sides, remaining comparatively small. The 

 radial structures persist as laminae, and those septa of the 

 buds would be largest which could start at once upon, and in 

 the same plane with, one of the radial laminate structures 

 (costee) of the parent; hence the "directive" septa of the 

 buds are typically radially symmetrical with those of the 

 parent. The epitheca is left behind. 



Turhinaria. — K ring of buds shoots up round and from 

 the sides of the parent polyp, together forming a cup, the wall 

 of each bud rising up as a distinct cone above the level of the 

 fusion of their walls to form the common coeneuchym.a (fig. 46), 



* For Miss Ogilvie's alternative derivation of the Madreporidse see 

 Phil. Trans, vol. clxxxvii., IfcUU. This has been criticized by me in the 

 Geological Mag. vol. iv. Ib97, p, 170. 



