i'2 THE VOYAGE OF H.M.S. CHALLENGES. 



whereas the number of tentacles and mesenteries is constant in the former order, it is 

 most variable in the latter. In this respect the Antipatharia agree with other Zoantharia, 

 at least to this extent that there is a variation in the number of mesenteries present in 

 the various species already examined. This distinction has necessarily led to a difference 

 in the characters selected as being of generic value in the two great sections of the 

 Anthozoa. I am thus inclined to think that the method adopted by E. Hertwig in the 

 classification of the Actiniaria will be likely to yield the most reliable results if applied to 

 the Antipatharia also. The characters available for a classification of the Antipatharia 

 are more limited than might be at first supposed. The sclerenchyma is apparently 

 always chitinous, and is more or less spinose in all the species described, excepting 

 Savaglia lamarcki, which constitutes the only known species of the Savagliidse. The mode 

 of branching has been generally admitted to give no characters of generic value, indeed, so 

 far as can be made out at present, this feature is sometimes not even of specific value. 

 For instance, in Antipathes picea and Antipathes tanacetum, described by Pourtales, the 

 two forms are said to be precisely similar in the mode of branching, the species being con- 

 sidered distinct on account of differences in the spines. The value of the mode of branching 

 for generic purposes has already been partly discussed when considering the value of the 

 genera proposed by Milne-Edwards. More need not be added at present. Further investi- 

 gation may, however, show that some of the genera here defined (e.g., Aphanipathes) may 

 bear subdivision, and that in this case distinct types of branching may yield characters of 

 value. With regard to the form, size, and mode of distribution of the spines, it may be 

 stated that three distinct types, viz., cylindrical, triangular, and knobbed, have been 

 observed ; their value as an aid to classification does not at present seem clear. Un- 

 doubtedly, as Pourtales has already pointed out, the form of the polyp frequently bears a 

 definite relation to the form of the spines, but this is not invariably the case, and the three 

 types are closely linked together by intermediate forms. The cylindrical type may become 

 compressed and shortened, whilst in the other direction Antipathes Jilix, Pourtales, 

 lorms a link between those forms having simple cylindrical spines and others, such as 

 Aphanipathes pedata and Aphanipathes cancellata, in which the knobbed feature is most 

 marked. We have thus to fall back on the structure of the zooids and ccenenchyma to 

 supply the chief characters, and these are precisely the features which have hitherto 

 received the least attention. 



The characters to be considered of ordinal value depend to a great extent on whether 

 the Savagliidse are to be included in the Antipatharia. Undoubtedly Savaglia 

 lamarcki has little in common with the Antipathidse beyond the possession of a con- 

 tinuous and branched horny sclerenchyma and a non-spiculate ccenenchyma. Its zooid 

 has the typical Actinian structure, and the system of canals in the ccenenchyma are, so 

 far as is known at present, without parallel in the Antipathidse. A horny sclerenchyma, 

 such as that of Savaglia, is by no means confined to the Antipatharia. The colonial 



