SOLITARY CORALS. 225 



(V06 millim. in length, and the central thread stains intensely blue in picro-indigo- 

 carmine. They are clearly of the same type as, but somewhat larger than, the 

 mesenterial nematocysts described by Gardiner (21) in Ccenopsammia. 



As regards the layer of calicoblasts and the desmocytes, they are of the usual 

 character, and I have nothing to add to what I have published concerning these 

 structures in a previous paper (7). 



The En do derm. As is usual in corals, the highly vacuolated endoderm cells 

 are so badly preserved that nothing very definite can be said of their structure. 

 They differ, however, in different regions of the body. The endoderm cells covering 

 the muscle-banners and the extrathecal continuations of the mesenteries, and also 

 those of the tentacular endoderm, are very long and columnar, and are crowded with 

 Zooxanthellse. As a rule there is a similar modification just within the mesenterial 

 filaments, this modification being most pronounced in the primary mesenteries. 

 Elsewhere the endoderm consists of a rather low columnar or cubical epithelium, and 

 Zooxanthellae are more scarce or, in some places, absent. Gardiner (21) lays great 

 stress on the absence of glandular elements in the endoderm, but my observations 

 do not support his conclusions. Glandular elements, it is true, are few or altogether 

 absent, not only in the elongated endoderm above referred to, but in the whole or 

 the greater part of the endodermic investment of the mesenteries. But in Hetero- 

 cyathus the endoderm of the axial gastrovascular cavity, that is to say the tissue 

 investing the pali and columellar upgrowths, is invested by a moderately thick 

 cubical endoderm in which there are few Zooxanthellse, but numerous ovoid or bean- 

 shaped cells containing clear refractive granules which do not stain with any of the 

 aniline dyes used, nor with hematoxylin. These cells are shown in Plate IV., fig. 19. 

 It is, of course, possible that they may be modified nematocysts. Similar cells are 

 found, though not so abundantly, in the endoderm wherever it is opposite a layer of 

 calicoblast, and this would seem to suggest that the function of the cells in question is 

 to elaborate material which is passed through the thin mesoglceal lamina to the layer 

 of calicoblasts and converted into calcareous tissue by the agency of the latter. This 

 view has a certain probability, because the calicoblasts form so thin a layer, and are 

 themselves so retrograde in structure that it is difficult to believe that they are the 

 only agents in the active growth of the corallum. On the other hand, I have 

 observed fragments of copepods and diatoms in the intermesenterial chambers, at 

 some distance from the filaments, and as these have evidently been or are being 

 digested, it is equally possible that these glandular-looking cells in the endoderm may 

 secrete a digestive fluid. At all events the facts do not warrant so sweeping an 

 assertion as that of Gardiner, that the endoderm is excretory but not glandular, < ti- 

 the conclusions as to the homology of the Anthozoan layers that he has founded 

 on it. 



In some, but not in all my series of sections, the endoderm, in addition to the 

 glandular elements described above, contained a number of large amoeboid cells of 



2 g 



