ON THE BRITISH FOSSIL CORALS. 125 



were not Actiuozoa, but Hydrozoa, aud lately he has reasserted this state- 

 ment. If MUlepora is one of the Hydrozoa, those tabulate forms which 

 resemble it in structure, such as Heliolites, must reasonably be asso- 

 ciated with it in classification. The importance, then, of determining this 

 point is very great, and unfortunately it is accompanied by many difficulties. 

 Before proceeding to criticise Agassiz's remarks, it is necessary to examine 

 the nature of the structures of the genera associated with MUlepora, or, in 

 fact, to review the classification of the Tabulata, and to note their affinities 

 with the other sections. Milne-Edwards and Jules Haime divide the Tabulata 

 into four families : — MiUeporidEe, Seriatoporidae, Favositidae, Thecidas. 



The principle upon which this classification is founded is philosophical and 

 natural to a certain degree. The first two families have more or less coenen- 

 chyma between the coraUites, and the last two have little or none, the co- 

 raUites being soldered together by their walls. 



The genus Pocillopora unites the two divisions, for it belongs to the Favo- 

 sitidae, and yet has a compact coenenchyma on the surface of the corallum. 



The classificatory value of the presence of coenenchyma in the whole of the 

 Madreporaria may be estimated by examining the scheme of MM. MUne- 

 Edwards aud Jules Haime. 



When treating of the Madreporidae (Hist. Nat. des Corall. vol. iii. p. 91), 

 they subdivide them into Eupsamminae without an independent coenenchyma, 

 Madreporince and Turbinarinae with a very abundant coenenchyma. 



The Poritidae they subdivide into the Poritinae without coenenchyma, and 

 the Montiporinae with an abundance of that structure in the spongy or 

 alveolar form. 



The EiiphyUiaceae (Ed. & H. op. cit. pp. 184 & 197) have such genera 

 as BarysmiUa and Dichocosnia, associated with Dendrogyra, Gyrosmilia, Pa- 

 chyr/yra, Rhipidogyra, which have or have not much cosnenchyma. 



The StylinaceaD are divided into independent, " empatees," aud agglomerate. 

 The independent genera have no coenenchyma; the " empatees" possess it in 

 the extreme so as to merit the term peritheca. 



The agglomerate have an excess of exotheca, but some genera are admitted 

 which are united by their walls, and are therefore without exotheca or coe- 

 nenchyma. Thiis PhyJIocosnia has an exotheca quite ccenenchymatous, and 

 Astroca'iiia has none. The corallites of Elasmoccenia have large mural ex- 

 pansions, and those of Aphccenia are soldered by their walls. Heteroccenia 

 and Pentaccenia present the same anomalies. 



The Astraeinae present such genera as ApJirastrcea and Septastrcea, the one 

 with and the other without extramural tissue, and Heliastrcea and Solenastrcea 

 with and Isastrcea without the same structure. 



It is then evident that the presence or absence of coenenchyma had difierent 

 significations in the estimation of the distinguished French zoophytologists. 



It is e%'ident that the structure of the corallites of Isastraeae and their defi- 

 ciency in coenenchyma in comparison with the Heliastraeae and Solenastraeae 

 cannot be of any very great organic significance ; for the coraUites of Heli- 

 astraeae occasionally grow so close together as to produce absorption of the 

 exotheca and costas, and the same occurs in the Astroccenife. The presence 

 of exotheca, peritheca, aud coenenchyma (for they are grades of a particular 

 structure) depends very much upon the habits of the corallum, and the notion 

 of teleology can hardly be separated from the consideration of this presence 

 a7id absence. Certainly to separate great groups by the presence or absence 

 of coenenchyma is not natural. It may be very useful to the classificatory 

 student, because the limitation qf forms is the prevaUing want; but it is not 



