136 EBPOiiT — 1871. 



ordinarily occupied by a series of tabulra or vesicular endotheca, and tbc 

 eudotheca often occupies the greater part of the corallum. The septal 

 laminfc, although generally very incomplete, arc never perforated or ' pou- 

 trellaire ; ' finally, their lateral faces are not furnished with synapticulse, and 

 are only rarely granular. 



" The individual corallites increase by gemmation, and never by fissiparity. 

 The buds are generally calicular, and this form of gemmation may continue 

 in the same individual. In some cases the gemmation is lateral." 



The originators of the " Rugosa" divide them into four families : — 



1. Stauridffi. 3. Cyathophyllidas. 



2. Cyathosonidge. 4. Cystiphyllidse, 



In criticising this classification some definite plan must be adopted, which 

 should refer to the philosophy of the classification of the Aporosa and Per- 

 forata. In fact the scheme of generic subdivision and differentiation adopted 

 in the Neozoic corals can be made to apply to those of the Palceozoic age. 

 Thus an essential distinction is made amongst the Neozoic corals by the 

 simple or compound nature of the corallum. Simple Caryophyllinse constitute 

 a series of genera, and the compound forms are separated as Ccenocyathi. 

 Now in the Paltcozoic genus CijafJiojjJu/Ihun, MM. Milne-Edwards and Jules 

 Haime admit, in direct opposition to the Neozoic scheme, both simple and 

 compound forms. This, I think, is an error, but only an error of classifica- 

 tion, for there can be no reasonable doubt of the intimate genealogical 

 relation of the simple and compound genera of Ct/athophyllum, 



Families *. 



1. SiATJEiDiE. — Genera: Siauria, Hohcystis, Polyccelia, Metrio][jliyllum, 

 Conosmilia. 



Of these Ilolocystis is a Lower Greensand form, and Conosmilia is Austra- 

 lian and Tertiary. 



MM. Milne-Edwards and Jules Haime place the Stauridse first in their 

 list of families ; but it would have made the classification more simple if the 

 second family took their place ; and I propose to change the order of arrange- 

 ment, but proceed at present in the recognized method. 



There is a well-developed wall in the Stauridaj ; the septa are continuous 

 from the top to the bottom of the calico, and are eminently quaternary in 

 their arrangement. The endotheca assumes the vesicular structure between 

 the septa, and then crosses over in the form of horizontal tabulps. The 

 Staiiridaj approach the CyathophyUidaj more than the Cjathoxonidae ; and, 

 indeed, the only essential distinction between the first two families is in the 

 truly lamellar state of the septa in the first instance, and in the incomplete 

 condition of them in the second. Nevertheless it should constitute a family 

 distinction. 



Two of the Stauridian genera are compound, and three arc simple forms. 



Stauria, which as yet has not been found in British strata, has neither colu- 

 mella nor costa?, whilst Hohcystis has both of these structures. There is no 

 reason why the last-named genus should not be the lineal descendant of the 

 former. Both were probably shallow-water forms in the neighbourhood of 

 reefs. 



The simple forms Conosmilia and Poh/ccelia are closely allied, and the 

 presence of the first in the Australian Tertiaries, and of the other in the Euro- 



* See Hist. Nat. cles Coralliaires, vol. iii. p. 325 c( scq. (Milne-Edwards aud Jules 

 Haime). 



