SOLITARY CORALS. 207 



stage, the bifurcated outer ends of these same septa constituting the apparent fourth 

 cycle. The further development has not been followed, but it probably continues on 

 the same plan. Thus it appears, as is shown in the diagram above, that at any 

 given stage in the growth of the coral, all the entocoelic septa, after the first cycle, 

 are composite structures, but all the exocoelic septa, forming the apparently last 

 cycle, are the derivatives of the original exocoelic septa which were second in order of 

 appearance. One can scarcely imagine anything more at variance with M. Edwards 

 and Hatmk's law of septal sequence. Dukrdkn has further shown that there is a 

 regular dorsi-ventrality in the mesenterial, and therefore in the septal succession of 

 Siderastrcea radians. New mesenterial pairs appear Hist in the dorsal member of 

 the two exocoeles of each system. This fact may be <>t' assistance in explaining the 

 irregularities in the apparently last cycle of septa which are so common in the 

 Eupsammiidaa and other corals, but, as I shall show, the same sequence does not 

 seem to obtain in Balanophyllia as in Siderastrtea, and it is quite probable that the 

 dorsi-ventral order of the appearance of the mesenteries varies as much in the 

 Kcleractinias as it does in the Malakactiniae. 



The convergence and union of the inner ends of the lower orders of septa is a much 

 more common phenomenon than is generally supposed. Tn the collection of corals 

 forming the subject of this paper, such septal unions occur in Rhodocyathus, Hetero- 

 cyathus, and Paracyathus among the Turbinolidse. They are characteristic of the 

 genera Cycloseris and Diaseris, and, as I have shown elsewhere (6), they are well 

 marked in the anthoblast of Fungia. But they are above all characteristic of the 

 Eupsammiidae, in which the septal arrangement of the adult appears to retain its 

 embryonic character, and it may be inferred from the septal characters of the adults 

 that the septal sequence in the course of development has followed the law of 

 Pourtales and Duerden. It is obvious that, if this inference is correct, all the 

 adult septa of the apparently last cycle will be exocoelic, while those of the apparently 

 penultimate cycle will be entoccelie and contained within the mesenterial pairs of the 

 last formed, and therefore smallest, mesenterial cycle. In the second part of this 

 paper I shall show that this is the case, not only in Hetero'psammia and a species 

 of DendrojyJii/Uia, among the Eupsammiidse, but also in Hetfrocyathus among the 

 Turbinolidse. 



It might seem desirable to amend Milne-Edwards and Haime's method of 

 enumerating the septa, and to adopt a system more consistent with the facts of 

 developmental sequence as now ascertained. But after several attempts to invent a 

 new notation, I have decided to retain the old method. After all, we have only 

 Pourtales' and Dtjerden's accounts to go by, and we are by no means certain that 

 what is true of the species thev describe is true of all Madreporaria. Future researches 

 may bring to light considerable differences in the septal sequences of different groups, 

 and it would be premature to invent a system of notation that might prove to be 

 inapplicable to new cases. I have therefore retained the old system in describing the 



