CORALS 383 may either be prolonged across the upper surfaces of the tabula to the centre of the corallum, or they may fall short of the centre, and thus leave a larger or smaller area of the tabulae free to view, and conspicuously visible in the floor of the calice. The tabulae may be well developed, approximately horizontal, remote plates, as is usually the case in Zapkrentis and Amplexus, or they may anasto- inoze in various ways, and become so intimately connected with one another as to give rise to a species of vesicular tissue. As regards the affinities of the Rugose corals, doubts have of late years been expressed as to their systematic position and rela tionships. By Professor Louis Agassiz the entire order of the Rugosa was transferred from the Actinozoa to the Hydrozoa, but upon insufficient evidence. It had been observed by Agassiz that the living animal of Millepora was apparently a hydroid, closely allied to Hydractinia, and it was upon the strength of this observa tion alone that the distinguished American naturalist proposed this sweeping change. The following considerations show, however, that this change cannot be accepted. (1) Millepora is not a Rugose coral, but belongs to the so-called Tabulate corals, of which it constitutes a very aberrant member. Even, therefore, were it satisfactorily proved that the genus Millepora is a true Hydrozoon, as to which naturalists are not yet agreed, this would not affect the classification of the Rugosa, which are very distinct in their structure from the Tabulata, and have no affinities with them further than is implied by the fact that tabulae are present in both. (2) It has been shown that the group of the Tabulata itself contains both true Zoantharia and true Alcyonaria, so that the hydroid character of Millepora, if admitted, would not so much as cause the removal of the Tabulata to the Hydrozoa. (3) It has further been shown that tabulae are present in certain forms which are unquestionably Actinozoa, as shown by an examination of the living animal (Podllopora and Lophohelia). It is evident, there fore, that the presence of tabulae in itself should have no weight in determining the systematic position of any given form, unless at the same time the structure of the living animal be known. (4) Apart from the close similarity between the corallum of the Rugosa and that of the recent Zoantharia scleroderinata, the Rugose corals are provided, almost invariably, with stractures which, so far as we know, are absolutely irreconcilable with the belief in their Hydrozoal affinities. They possess, namely, in almost all cases, well developed septa, which, if they do not absolutely imply the existence of mesenteries in the living animal, are, at any rate, wholly unknown as occurring amongst the recent Hydrozoa. "We may, therefore safely accept the conclusion of Verrill, Pourtales, Glaus, Duncan, Milne Edwards, and other distinguished authori ties, that Agassiz has failed to bring forward sufficient evidence in favour of his view that the Rugosa are referable to the Hydrozoa. More lately Dr Lindstrom has endeavoured to show that the Rugosa cannot be placed amongst the Actinozoa, but " must form a class of their own in the great division of the Radiated Animals." The chief grounds upon which this conclusion is reached are " the compact imperforate structure of the calyx and septa (the septa originating from four primary ones), the absence of costae, the strange septal fossae in the bottom of the calyx, the processes resembling rootlets, the transverse floors or tabulae in the interior, which often have a cellular or vesicular structure," and the occasional presence of an operculum. It is obvious, however, that none of the above-mentioned peculiarities are of such fundamental importance as to justify us in overlooking the substantial identity of structure which subsists between the corallum of the Rugosa and that of the Aporose section of the Zoantharia sclerodermata. The wall and septa are often compact amongst the latter ; the septa have a tetrameral arrangement in the unquestionable living Actinozoa, Haplophyllia and Guynia ; the septal fossula is not always recognizable amongst the Rugosa; the presence of rootlets and the nature of the costse are points of secondary importance ; tabulae are present in undoubted corals belonging to the Zoantharia scleroder mata ; and the occasional presence in some abnormal forms of an operculum would no more justify us in removing the Rugosa from the Actinozoa that we should be warranted for the same reason in removing the living Primnoae from the Alcyonaria. Upon the whole, then, there appears to be little danger in accepting the conclusion reached by Professor Verrill, one of the most distinguished of living authorities on the subject, that there can no longer be any reasonable doubt but that the coralla of the great majority of the Rugosa were made by " true polypes essentially similar to those of the existing corals." Indeed, if any great change in our classification is to be made, it would seem rather to be in the direction of more closely approximating the Rugusa to the Zoantharia sclerodermata. The tetra meral arrangement of the septa is by no means always very conspicuous amongst the Rugosa, and it seems not unlikely that too high a classificatory value has been attached to it. Leaving this out of account, the affinities of the typical Rugosa with the Aporosa are very close, and it may be doubted if it would not be proper to establish a more intimate union between these groups. At the same time, it must not be overlooked, as pointed out by Mosely, that the Rugosa have certain affinities with the Alcyonaria, and especially with singular recent Heliopora. Great stress, however, can hardly be laid upon the existence of paired opercular structures in Goniophyllum and in the existing Alcyonarian genus Primnoa, since Goniophyllum, if a Rugose coral at all, must be referred to the Cystiphyllidce, a family in which all the more characteristic features of the Rugose organization have disappeared. The divisions of the Rugosa recognized and founded by Milne- Edwards andJulesHaime, and subsequently generally adopted, are the following : Fam. I. STAURID^:. The wall is well developed ; the septa lamellar, extending without interruption from the bottom to the top of the visceral chamber, and showing a conspicuous quaternary arrangement. The interseptal loculi are crossed by endothecal dissepiments, and there is a central tabulate area. Of the five genera which form the family, Stauria has a compound corallum, astraeiform in shape, and increasing by calicular gemmation ; there is no columella ; and the four primitive septa form a conspicuous cross. Holocystis is also composite and astrseiform, the corallites being united by well developed costae, and a styliform columella being present. Of the simple genera, the Permian Polyccelia and the Tertiary Conosmilia are closely allied to one another ; but ihe third, Metriophyllum, is so aberrant that it may require to be placed in a separate family, as its septa are arranged in four distinct groups, separated from one another by four distinct fossulae. Fam. II. CYATHAXONIB^;. In this family the corallum is simple, with a deep calice ; the septa are well developed, and extend from the bottom of the visceral chamber to the floor of the calice ; the interseptal loculi are completely open, and there are neither dissepiments nor tabulae. Though exhibiting a quaternary arrangement, the four primaiy septa are not conspicuously developed above the others. This family makes an exceedingly close approach to the Aporose group of the Turbinolidce, from which it is separated by the tetrameral arrangement of the septa. Of the genera of the family, Cyathaxonia is Palaeozoic, whilst Haplophyllia and Guynia are recent. No Secondary or Tertiary forms are as yet known. Fam. III. CYATHOPHYLLID.K. In this family the corallum is simple or compound ; and the septa are always more or less inter rupted, and do not extend as complete lamellae from the bottom to the top of the visceral chamber, being more or less imperfect either internally or externally. The four primitive septa are not pre-eminently developed, so as to give rise to a conspicuous cross. The interseptal loculi are generally more or less interrupted by the development of dissepiments, and tabulae are invariably present. The family of the Cyathophyllidce is divided by Milne-Edwards and Haime into the two tribes of the Zaphrentince and the Cyatko- phyllince. a. Zaphrentince. In this tribe the corallum is simple and free ; a well-developed septal fossula is present, which may be formed by a tubular inflection of the tabulae on one side, or may be replaced by a cristiform process. The septa usually fall short of the centre of the visceral chamber, and, from the presence of a septal fossula, are invariably more or less irregular in their arrangement. _ The tabulae are complete, and pass from one side of the visceral chamber to the other ; while dissepiments are poorly developed, and there is usually nothing of the nature of a columella. The Zaphrentince are exclusively Palaeozoic, and the principal genera of the family are Zaphrentis, Amplexus, Mcnophyllum, Loj>ho- phyllum,Anisophyllum, Baryphyllum, Hadrophyllum, Microcyclus, Combophyllum, Trochophyllum, Aulacophyllum, Uallia, and Streptelasma. b. Cyatho2)hyllincc.Io. this family the septa do not exhibit the irregularity which is so conspicuous in the Zaphrentince, but are
more or less regularly radiate in their arrangement. In some