372 ON THE ANATOMY OF A SUPPOSED XEW SPECIES 



do not appear to be especially massed in any position, and near the base of the 

 polj'ps are very distant from one another. 



The layer however appears to be complete, and to separate the structureless 

 membrane everywhere from the coraUum except at the attachments of the mesenteries, 

 and dividing walls of the coenosarcal canals outside the calices. Three structureless 

 membranes join one another in these positions (Fig. 20), and at their junction 

 a number of small bands are given off, which broaden out at their ends, and 

 are directly attached to the corallum. In section these bands are seen to be 

 striated, and the calicoblastic ectoderm between them is especially well developed, 

 almost completely filling up their interspaces. In some polyps, especially at the bases 

 of the mesenteries, the striations appear to run for some distance into the structureless 

 membrane and to be due to fibres, but generally it appears quite structureless. 



In oblique sections the bands have the appearance, represented by Sclater (25), 

 in Stephanotrochus, and appear to have no connection with the structureless membrane. 

 Sclater believed these blocks to be the calicoblasts, but Bourne (4), and subsequently 

 Fowler (10), pointed out their real nature. In Coenopsammia they are short and have 

 rather broad bases of attachment to the structureless membrane of the mesenteries, while 

 at the attachments of the dividing walls of the coenosarcal canals they are generally very 

 long, much narrower, where they leave the structureless membrane, broadening out very 

 greatly at their ends. The spaces between them, in the latter position, are crowded with 

 nuclei, and they often present in thick transverse sections (10 — 12/i) almost exactly the 

 same appearance as represented by von Heider (13), for the calicoblasts in the same 

 position in DendrophyUia, a closely allied genus (12. Taf xxxi. Figs. 8, 9 and 11). 

 I have too very carefully examined the calicoblasts, together with their attaching bands 

 from the same position in undecalcified and partially decalcified preparations of 

 Coenopsammia and other Madreporaria, and I can confidently state that neither in them 

 nor in the calicoblasts are there any crj'stals of any sort. In decalcified preparations of 

 Pocillopora (5 species), Seriatopora, Prionastraea, Madrepora and C'oenopsaiuinia, I have 

 found no trace of any organic tissue or remains in the corallum, other than that due 

 to Clione or boring algae, and there does not seem to be any room for doubting von 

 Koch's conclusion (18), that the corallum lies completely outside the animal, and is the 

 result of secretion by the calicoblastic ectoderm, the elements of which retain their cnvn 

 organic existence. 



Endoderm. (end. Figs. 5, 20 — 22). I found it impossible in the hardened and 

 preserved specimens to tease out separate cells from the endoderm, and in sections no 

 cell outlines can be seen. In the coenosarcal canals and generally in the contracted polyp 

 above the tentacles and over the corallum, the endoderm consists of a ragged much 

 vacuolated epithelium of a cubical facies with large round granular nuclei with distinct 

 membranes together with a few scattered mucous cells (Fig. 5). Over the muscles and 

 between the attachments of the mesenteries to the external body wall it has a very 

 elongated columnar facies with small oval or round, deeply and homogeneously staining 

 nuclei with a few of the larger granular nuclei (Fig. 21). The structureless membrane 

 is irregularly drawn out under the endoderm into processes, to which fibres from the 

 endoderm appear directly to be attached, one or more corresponding to each nucleus. 



