The number of septa and the habit of growth of the two genera separate them very widely; and the propriety of connecting the last-named one with the Milleporidae must be considered.

There are four genera in the family of the Seriatoporida :-Seriatopora, Dendropora, Rhabdopora, Trachypora.

The family is characterized by the continual growth of the lower parts of the corallites and the rarity of tabulæ.

Seriatopora is a recent genus, and therefore those associated with it must be carefully examined.

Dendropora, Michelin, is clearly too closely allied to Rhabdopora to be separated generically.

Rhabdopora, formed for the Dendropora megastoma, M'Coy, by MM. MilneEdwards and Jules Haime, has only one species, the diagnosis of which is as follows::

Rhabdopora megastoma, M'Coy, sp.-The corallum is branching. Branches four-sided, starting from the stem at an angle of 70°, and very equal. Cœnenchyma granulated or subechinulated and obscurely striated. Calices in vertical series on each face of the branches. Septa (teeth) 12 in number and subequal.

It is impossible to separate this from Seriatopora, for the four-sided suture of the branches is only a specific (if that) distinction.

Trachypora appears to be an Alcyonarian.

The distinction between Pocillopora and Seriatopora is not generic, and therefore these genera and Dendropora (for Dendropora and Rhabdopora are equal, and the first name is the oldest) are absorbed in one. Oken's name Acropora (1815) may be used as the generic term :- -ACROPORA (Seriatopora, Lamarck; Pocillopora, Lamarck; Dendropora, Michelin; Rhabdopora, Ed. & Haime).

All the species of the absorbed genera should take the generic name of Acropora, and the family becomes that of the Acroporinæ. Thus the sharp distinction between the recent and Paleozoic forms is partly smoothed down, and the old Dendropora and Rhabdopora were doubtless the ancestral forms of the recent Acropora. Conites cannot be associated with the family.

The family of the Thecidea is characterized by well-formed septa, which are prolonged throughout the visceral chamber, well-developed tabulæ, which grow like dissepiments upon the sides of the septa, and these last do not spring from the upper surface of the tabulæ, as in some Tabulata. The walls are solid, compact, and united.

The corals contained in the family are all Silurian forms, so far as is known at present.

Thecia, Ed. & Haime. It is a most remarkable fact that this genus, the species of which have no true wall, but a dense cœnenchyma between septal prolongations or costa, should here give the family name. Thecia Swinderniana, Goldfuss, sp., has been called Agaricia, Porites, Astreopora, and Palæopora by different authors, so that its classificatory position may well be a matter of doubt. It is not in the least allied to Columnariæ, which has solid walls, and which fulfils all the characteristics of the Thecida.

In Thecia, Ed. & H., there is a long visceral cavity surrounded by a dense tissue, as in Millepora, through which the septa, or rather the costa, run. What is the structure of Plasmopora and Propora but that of Thecia slightly modified. The genus clearly must be associated with them amongst the Milleporida.

Columnaria is a fine form; the great septa (12 to 18) and tabulæ, with the compact walls, distinguish it at once. Col, alveolata is a Lower Silurian form, C. Gothlandica is Upper Silurian. It is a most important genus, and its affinities will be noticed.

The Favositidæ have a massive corallum without coenenchyma, septa, and perforate walls; that is, there are openings which permit the visceral cavity of one corallite to communicate with that of another in several places. The following genera are included by MM. Milne-Edwards and Jules Haime :Favosites, Emmonsia, Michelinia, Roemeria, Koninckia, Alveolites.

Favosites is the typical genus. In some species the mural foramina are scanty in number, in others numerous; and they are even in relation with the angles of the wall, especially in F. alveolaris.

The earliest species of the genus are Lower Silurian, for instance :-F. Gothlandica, F. multipora, F. aspera, F. Forbesi (which ranges through to the Upper Silurian), and F. fibrosa (having the same vertical range, and is found as a Devonian fossil).

F. Hisingeri has the same range as F. fibrosa. F. cristata and F. cervicornis are the same, and the range is from the Upper Silurian of England to the Devonian of Russia.

The species which are Devonian, and do not range above or below, are :— F. Goldfussi, F. basaltica, F. polymorpha, F. alveolaris, F. pediculata, F. Tchihatcheff, and F. mammillaris. The only known Carboniferous Favosites is F. parasitica, and it is a degenerate form.

F. Gothlandica has rounded processes encircling the mural pores, and the projections formed upon one fit against those of the neighbouring corallite. F. multipora has three vertical series of pores, and its walls are almost as perforate as some Alveopora.

The tabulæ are almost universally horizontal in the Favosites, but some are wavy in their course; and the septa are a series of vertical spines which vary in size according to the cycle, and are often referable to three cycles in six systems. In some there is a faint columellary swelling on the tabulæ.

A careful examination of the species proves that the earliest known forms are as highly developed as the Devonian, but that the species parasitica is dwarfed.

Emmonsia has imperfect tabulæ. The tabulæ are vesicular at the sides, or dissepimental, and they communicate more or less with each other. Roemeria has infundibuliform tabulæ, and the species is Devonian. Koninckia is an Upper Cretaceous form; it has thin and nearly horizontal tabulæ, thin walls very much perforated, and six series of large spiny septa. Michelinia has irregular and vesicular (dissepimental) tabulæ, and simple striæ for septa (Devonian and Carboniferous). The alliance of Michelinia, Romeria, and Emmonsia is very evident. Mr. Kent has written a most interesting description of Favositipora (Kent), Ann. & Mag. Nat. Hist. 1870, vol. vi. p. 384, which unites the Favositinæ and the Favositidæ.

Alveolites offers the same objection to being united to Favosites that Cœnites does to Pocillopora; in fact Alveolites is a Conites with perforated walls, and it is proposed to deal with both genera by disassociating them from their recognized families.

Syringopora I propose uniting with the Favositidæ, as it has tubular connexions between the visceral centres of the corallites, which are foreshadowed in F. Gothlandica.

After this analysis of the Tabulata, it is necessary to state the opinions of Prof. Agassiz respecting their Hydrozoan characteristics.

Prof. Agassiz (senior) writes as follows in the American Journal of Science and Arts,' 2nd series, vol. xxvi. p. 140, November 1858 :

"The animals of Millepora are Hydroid Acalephs and not polyps;" that is to say, they are Hydrozoa and not Actinozoa. The résumé of several letters to Dana is given at the same place. "I have seen," writes Agassiz, "in the Tortugas something very unexpected. Millepora is not an actinoid polyp but a genuine Hydroid, closely allied to Hydractinia. This seems to carry the whole group of Favositidæ over to the Acalephs, and displays a beautiful array of this class from the Silurian to this day."

Dana adds a note to this statement. "The drawings of Professor Agassiz which have been sent us for examination are so obviously Hydractinian in most of their characters that no one can question the relation. With regard to the reference of all the Favositidae (a group including Favosites, Fenestella, Pocillopora, &c., as well as the minuter Millepora, Chatetes, &c.) to the Acaleph class, direct evidence is not yet complete, as the animal of the Pocillopora has not been figured by any author on zoophytes. From the specimens of the species of this genus which I procured in the Pacific, I never obtained a clear view of the polyps, and hence made no figure. The brief description on page 523 of my Report may be reasonably doubted until confirmed by new researches. The much larger cells in Pocillopora, Favosites, and Fenestella than in Millepora, and the frequently distinct rays in these cells, are the characters I had mentioned to Prof. Agassiz as suggesting a doubt as to their being Acalephs, and to this what follows-above relates."

Agassiz observes, in a subsequent letter, after observing that the Sideroporæ obviously are polyps, "There are two types of radiating lamella which are not homologous. In true polyps (excluding Favositidæ as Hydroids) the lamellæ extend from the outer body-wall inward along the whole height of that wall, and the transverse partitions reach only from one lamella to the other, so that there is no continuity between them, while the radiating lamellæ are continuous from top to bottom in each cell. In Milleporida the partitions are transverse and continuous across the cells; so are they in Pocillopora and in all Tabulata and Rugosa; while the radiating lamella, where they exist, as in Pocillopora and many other Favositidæ, rise from these horizontal floors, and do not extend through the transverse partitions; indeed they are limited within the spaces of two successive floors, or to the upper surface of the last. A careful comparison of the corallum of Millepora and Pocillopora with that of Hydractinia has satisfied me that these radiating partitions of the Favositidæ, far from being productions of the body-wall, are foot-secretions, to be compared to the axis of the Gordonia corallum &c., and their sceming radiating lamella to the vertical groove or keel upon the surface of the latter, which, reduced to a horizontal projection, would also make the impression of radiating lamellæ in the foot of the polyp. If this be so, you see at once that apparent radiating lamella of the Favositida do no longer indicate an affinity with the true polyps, but simply a peculiar mode of growth of the corallum; and of these we have already several types, that of Actinoids, that of Alcyonoids, that of Bryozoa, that of Millepora, and other corallines, to which we now add that of Hydroids. Considering the subject in this light, is there any further objection to uniting all the Favositide with the Hydroids? Sideropora and Alveopora being of course removed from the Favositidæ. It is a point of great importance in a geological point of view, and for years I have been anticipating some such result, as you may see by comparing my remarks in the American Journal,' May 1854, p. 315. If all the Tabulata and Rugosa are Hydroids, as I believe them to be, the class of

Acalephs is no longer an exception to the simultaneous appearance of all the types of Radiata in the lowest fossiliferous formations, and the peculiar characters which these old Hydroid corals present appears in a new and very instructive aspect."

A. Agassiz includes the Tabulata amongst the Hydrozoa. He notices "that the absence of radiating partitions in the Tabulata seems to show without much doubt that their true place is among the Hydroids." It is true that Prof. Agassiz has not observed the Medusa-buds on the specimens he has figured, yet the Hydroid character of the animal and their similarity to Halocharis-like Hydroids is very striking (Havard Catalogue, 1865, p. 219).

Prof. Alexander Agassiz informs me that his father still holds these opinions, and that new researches have satisfied him about the correctness of the drawings which have been lately reproduced. "Millepora is not an actinoid polyp, but a genuine Hydroid, closely allied to Hydractinia."

This very strong expression of opinion is founded upon the appearance presented by the polyps of Millepora alcicornis, the drawing of which has been reproduced by A. Agassiz. Now the distinction between the Actinozoa and the Hydrozoa is well marked; in the first the generative apparatus is included in the gastric and perigastric cavities, and in the last the digestive and generative organs are perfectly apart. Every variety of tentacular and disk apparatus may exist in either, but the external development of the gemmules, ova, and embryonic forms must be recognized before any Colenterate animal can be associated with the Hydrozoa.

Here is the point at which Agassiz fails. His researches are only suggestive, until the generative organs are recognized on the protruded polypes of Millepora, and until the mesenterico-ovarian layers are proved not to exist within the calices. The external resemblance of the Millepore polypes to the sterile Hydractinia is evident.

The remarks upon Favositidæ, Sideroporæ, and other genera, made by Agassiz in consequence of the assumption that Millepora is Hydrozoan, are of doubtful value; and I must refer back to my analysis of the Tabulata to show how a confused classification between both classes imperils research. Sideropora is not a tabulate form even. A careful examination of Columnaria satisfies me that Agassiz's description of the lamellæ fails in that genus; and inasmuch as the wavy lines of Gorgonia and Corallium are connected with the water system of the species, they can have no possible relation with the radiate amellæ or groovings of the Milleporan calices. The homologues of the grooves are the depressions and irregular interstriated portions on top of the cœnenchyma between the calices in the Tabulata.

The perforate walls and the septa of the true Favositidæ seem to remove them from the range of the remarks of Agassiz, which may well deserve attention, so far as Millepora is concerned, for it is a genus with marked distinctions from all other corals.

It is not reasonable to include the Rugosa, because some of them have no tabulæ, and others have them so much like dissepiments, or associated with dissepiments, that we are impressed with the unimportance of the differentiations established by the presence of horizontal tabulæ.

It is most important that the minute structure of the Milleporida should be thoroughly investigated, and any report on the Paleozoic corals must be very incomplete without a detailed description of its study.

With conenchyma

....

Section TABULATA.

Families.

Milleporida. Coenenchyma cellular.
Acroporida. Coenenchyma compact.
Favositide. Walls perforated.

Without conenchyma.. Halysitide. Walls imperforate.

MILLEPORIDE

ACROPORIDE

FAVOSITIDE

HALYSITIDE

ALVEOLITIDE

Incerta sedis

Chatetes.

Alveolitida. Septa tridentate.

Lyellia.

Thecostegites.

Axopora.

Acropora, Seriatopora, Pocillopora, Dendropora, Rhab-
dopora.

Favosites, Koninckia, Favositipora, genus nov. (Kent).
Michelinia, Roemeria, Emmonsia.

Syringopora.

Aulopora.
Halysites.

Stylophyllum.

IV. The Rugosa.-MM. Milne-Edwards and Jules Haime observe (op. cit. vol. iii. p. 323), "that this division comprehends simple and compound corals, and that the septal apparatus never forms six distinct systems, and appears to be derived from four primitive elements. Sometimes this disposition is shown by the great development of four principal septa, or by the existence of four depressions which occupy the bottom of the calice and take on a cross-like look. In other instances there is observed only one of these depressions or excavations, or one large septum interferes with the regularly radiate and star-shape of the septal arrangement. Finally, thero are instances where no traces of distinct groups or systems of septa can be recognized, and where the septa are represented by numerous striæ arising on the upper surface of the tabulæ or dissepiments near the calicular margin." They continue as follows:-"The corallites are always perfectly distinct amongst themselves, and are never united by independent coenenchyma. The walls are in general very slightly developed. The visceral chamber is

*Millepora is a most aberrant genus if it is one of the Madreporaria Tabulata. I have not yet satisfied myself about the Hydroidean characteristics of its soft parts; but an examination of the cœnenchyma of a series of species throws great doubt upon the Madreporarian affinities.

The relation of Heliopora to Heliolites is of the closest.