SOLITARY CORALS. 235 



me to think that Gardiner was in error in describing this form as a stage in the 

 development of the small spiral variety. The mesenterial filaments are relatively 

 small, and the absence of the largest ectodermal nematocysts is explained by the fact 

 that there is no room for bodies of such size. 



The endoderm is remarkable chiefly for the fact that Zooxanthellse are very 

 sparingly distributed in the mesenterial epithelium, and the epithelio-muscular cells 

 are for the most part clearly defined and but little vacuolated. The muscular processes 

 of the cells are remarkably well developed. Scattered through the endoderm, but 

 somewhat sparingly, are minute nematocysts, about - 01 millim. long, containing a 

 loosely and irregularly coiled spiral thread. Each is contained in a transparent cuticle, 

 and there is a flattened nucleus to one side of and outside the nematocyst itself. In 

 the somewhat similar endodermic nematocysts of Flabellum, Gardiner (22) figures 

 the nucleus inside the capsule of the nematocyst. The mesogloea of D. gracilis 

 is relatively thick, no doubt in correlation with the unusually great development of 

 the musculature, but I could find no trace of structure in it. The desmocytes, 

 described by von Heider as calicoblasts. are well developed and form tassel-like 

 groups at the points of attachment of the mesenteries to the theca, but otherwise 

 present the usual features. 



It is evident that the three forms whose anatomy is described in this memoir do not 

 differ in any important points of anatomical or histological structure from the normal 

 Actinian type, which has been shown by many authors to be characteristic of the 

 polyps of Madreporarian corals. There are, indeed, minor characters, both anatomical 

 and histological, which have a certain interest, but none of them can be regarded as 

 having any classificatory value. One's attention is arrested by the presence of both 

 endotentacles and exotentacles in some genera and of endotentacles only in others. 

 But when we see that in the Eupsammiidge, a well-defined family, only endotentacles 

 exist in Heteropsrxmmia while Dendrophyllia possesses both endo- and exotentacles, 

 this character does not appear to be of much value. Moreover, owing to the great 

 difficulty experienced in counting and localising the tentacles in spirit specimens, the 

 information we possess on this point is not altogether trustworthy, and before any 

 attempt is made to use the tentacles as an aid in determining the affinities of different 

 genera of corals, it will be necessary to accumulate a large number of facts based on 

 the study of living or well-preserved expanded polyps. 



But a much more promising field is offered to the future investigator by the study 

 of the relations of the septa to the mesenteries, especially by the developmental 

 sequence of the endosepta and their connections with the exosepta, if present. My 

 chief object in this paper has been to show that it is possible, by a study of the 

 relations of hard and soft parts in the adult corals, to determine whether the septal 

 sequence follows the rule established for Balanophyttia by Pourtales and Siderastrcea 

 by Duerden, and I have given sufficient evidence to show that the peculiar septal 



2 H 2 



