22 THE VOYAGE OF H.M.S. CHALLENGEE. 



contrast in thickness between the branches and branchlets as is figured by Morison, and 

 as Pallas refers to this figure, it is to be presumed that his type specimen agreed with it. 



The species Arachnopaihes panicidata, Duchassaing and Michelotti, and Arachno- 

 patlies columnaris, Duchassaing, referred to the genus Arachnopaihes by the authors, 

 appear to me to have no place there. The former species, judging from the figure, is a 

 form allied to Antipathes gracilis, Gray, and is one of the fan-like species, in which the 

 branches are less confluent than in typical RMpidipath.es. Arachnopaihes columnaris, 

 of which Pourtales has given us a photograph, has a similar corallum and polyp (?) to 

 Antipathes larix, Esper, and has been provisionally referred to Parantipathes, n. gen. 

 Fusions occur occasionally between the branches, but it must be remembered that in this 

 type the stem is simple, and therefore the typical Arachnopathes form, brought about 

 chiefly by fusion between branchlets belonging to adjacent branches, cannot occur. The 

 reticulum to which Duchassaing refers forms a tube for a parasitic worm, and its presence 

 is therefore neither generic nor specific, but depends on the presence of the parasite. 



In the systematic portion of this Report I have temporarily retained the name 

 Arachnopatlies as generic, in order to link together the three species Arachnopathes 

 ericoides, Arachnopathes clathrata, and Arachnopathes aculeata, until more detailed 

 information is obtained regarding them. 



Rhipidipathes. — Milne-Edwards gives the following definition of this genus : — 

 " Polypier sclerobasique dont les branches s'^talent sur un meme plan en forme d'evantail 

 et se soudent entre elles aux points de contact, de facon a constituer un re'seau." At the 

 time only two species had been described which were considered referable to this genus, 

 viz. : — Antipathes fiabellum, Pallas, and Antipathes reticulata, Esper. Gray in a 

 paper published about the same time (40), and other authors more recently, have 

 described quite a number of species which possess in a more or less marked degree the 

 characters referred to. One of the Challenger species {Aphanipathes cancellata) shows a 

 closer and more regular network than Antipathes fiabellum, Pallas. Starting with this 

 species as the one in which the network is most complete, one may trace this character 

 through a number of forms in which it is less and less marked until finally the 

 original feature has entirely disappeared. Such a series might include A. cancellata, 

 A. fiabellum, A. hypnoides, A. reticulata, A. gracilis, A. paniculata, D. and M. (non 

 Esper), and A. tristis. To begin with, the reticulum is formed by bridges of 

 sclerenchyma which pass across from branch to branch, giving a more or less 

 rectangular network, the sides of the meshes being subequal in thickness. In other 

 forms a similar result is obtained by fusion between pinnules of adjacent branches 

 and a general confluence of the stouter portions of the corallum. In A. hypnoides 

 one begins to find the terminal pinnate branchlets free, and not showing the fusions 

 so abundant in other parts. From this type onwards in the series there is a 

 gradual increase in the size of the terminal fronds in which fusions do not occur, until in 



