62 THE VOYAGE OF H.M.S. CHALLENGER. 



the elongation in the transverse axis brings about an arrangement of the tentacles in two 

 subparallel rows of three each. If one were not acquainted with the arrangement of the 

 mesenteries, the tentacles might be described as consisting of an anterior, a middle, and 

 a posterior pair. In Antipathella the tentacles are close together, and younger zooids 

 still show them somewhat radiately arranged. In Parantipathes, however, the difference 

 between the length of the zooid in the transverse and sagittal axes is very great, so that 

 the tentacles, still arranged in rows, become considerably isolated. In this case there is a 

 faint indication of a division of the zooid into three lobes, each bearing two tentacles. 

 This is brought about by a slight depression in the peristome on each side of the stomo- 

 dseum. This depression crosses the transverse axis at right angles, and pushes down 

 the transverse mesenteries before it for a short distance. In Parantipathes larix the 

 ccelenteron is thus imperfectly divided into three lobes, the central containing the 

 stomodseum and all the mesenteries, the lateral lobes only the distal portions of the 

 transverse mesenteries. The reproductive organs are confined to those sections of the 

 transverse mesenteries situated in the lateral lobes of the ccelenteron. From this type 

 the dimorphism in Schizopathes is easily derived by the formation of a mesoglceal 

 partition in each depression of the peristome which, passing down to the base of the zooid, 

 divides it into three individuals, a central gastrozooid and two lateral gonozooids. In the 

 Parantipathes type the ccelentera of the various zooids on a pinnule are in communica- 

 tion by means of a basal prolongation of each. The hex-tentaculate individuals are 

 imperfectly separated from one another by mesoglceal partitions which do not reach the 

 base. In Schizopathes similar mesoglceal partitions separate the dimorphic individuals 

 from each other, as well as one triplet from another. In Schizopathes there are, typically, 

 no prolongations of the ccelentera in the direction of the branch, because all are closely 

 packed, as in Pteropathes amongst the Antipathinse. In Bathypathes, on the other 

 hand, the dimorphic individuals are always separated from each other by a considerable 

 interval, and an interzooidal communication is kept up by lateral prolongations of their 

 ccelentera. Thus the Schizopathinse appear to be directly derived from the Antipathinge. 

 The transition is brought about, first by an elongation of a zooid along the axis of a 

 branch, and secondly by a division of such an elongate zooid into three individuals by 

 the formation of two vertical mesoglceal partitions, one on each side of the stomodseum. 

 Parantipathes larix forms an interesting link between the two subfamilies. 



Colony Formation. 



The production of a colony from the primary oozooicl has not been observed, 

 but it is possible to gain a general idea of the process from an examination of the 

 blastozooids of an existing colony. New zooids are added by a form of budding in all 

 essential features similar to that which obtains in Gorgonidse. The process is probably 



