70 THE ANTHOZOA 



borne by the latter, the cavities of the two being still in free com- 

 munication below. As a rule, this communication is eventually 

 cut off by a secondary deposit of calcareous tissue, and then the 

 two zooids are united only by their confluent edge-zones. But as 

 growth proceeds this union also is broken, and the zooids in the 

 older parts of the colony are isolated, and have no organic connec- 

 tion with one another (see Fig. XXX, 4). 



The classification here adopted is based upon Martin Duncan's 

 revision of the Madreporaria (79), with the modifications intro- 

 duced by Quelch (86). It cannot be pretended that it is a natural 

 or a satisfactory classification, yet it is the best which can be 

 off"ered in the present state of our knowledge. Other systems 

 have been proposed, but they have not stood the test of criticism, 

 and have been ephemeral. Milne-Edwards and Haime divided 

 the Scleractineae into five sections — Aporosa, Perforata, Rugosa, 

 Tabulata, and Tubulosa. The two last named have long since 

 been broken up and their families distributed, some among the 

 Alcyonaria, other's among the Aporosa. The Rugosa, also termed 

 the Tetracoralla, held their ground for a long time ; but it has _ , 

 been shown that the structure of the skeleton of the rugose corals ■! 

 does not differ from that of recent corals, and the tetrameral 

 symmetry, which so many of them exhibit, is to be considered of 

 less importance, since it has been shown that a hcxameral symmetry 

 is by no means characteristic of recent corals. Moreover, tlie 

 tetrameral symmetry is an inconstant feature in Rugosa. The 

 discovery of Mosclcya latistellata, a reef coral from Wednesday 

 Island, Torres Straits, leaves no doubt as to the close relationship 

 of the Astraeidae to the Cyathophyllidae. ' Moseleya is a compound 

 coral with polygonal calicles, a thin epithcca, a rudimentary theca, 

 and the cavity of the calicle is filled up nearly to the margin by 

 tabulae separated by an abundant dissepimental cndotheca. The 

 septa in adult calicles are numerous and give no indication of 

 a hexameral arrangement, but in young calicles a tetrameral 

 symmetry is distinctly visible, owing to the cruciate arrangement 

 of four larger septa. Moseleya shows decided affinities on the 

 one hand to a typical Astraeid, such as Frionastnica ; on the other 

 hand to a Cyathophyllid, such as Ciiathophiilhtm reijuim, and it cannot 

 be doubted that the Cyathophyllidae and the forms allied to them 

 can no longer be classified apart as Rugosa, but must be placed 

 along with or close to the Astraeidae. 



There is some doubt as to the distinctness of the sections 

 Aporosa and Perforata of M. Edwards and Ilaime. The anatomical 

 features on which the division is based have been referred to above, 

 but there are corals ranked among the Aporosa in which the theca 

 is perforated by a few canals, and amongst the Perforata there is 

 every grade between trabeculate and spongiose theca and septa 



