so to the biologist, for these mixed and unnatural limitations and separations only form gaps in his argument, which require bridging over.

The Favositida and Thecida, Paleozoic forms, may then be separated, for the purposes of classification, from the Milleporida and Seriatoporidæ, which are almost all post-Paleozoic; but this limitation is not to impede the plain course of the paleontologist, who studies from a biological point of view; nor is it to stand in the way of the assertion, that the break between the Paleozoic and younger Tabulata is almost nil.

The genus Millepora belongs to the Milleporidæ, and the cœnenchyma of its species is very abundant. It is of " a very irregular and spongy structure, rather than tubular" (Ed. & H.). The calices are of very different dimensions on the same corallum. There are no distinct septa, nor is there a columella. The tabulæ are horizontal. These are the diagnostics of the genus according to Milne-Edwards and Jules Hame. A careful examination

of the calices of good specimens determines that the trabecula, of which the cœnenchyma is composed, often projects into them, in the position of septa; but there is nothing like the regular arrangement as seen in Heliopora, or in the Poritida of the Perforata. The cells of the cœnenchyma may occasionally be seen to open into the space above the last tabula.

The absence of septa and this relation of the cœnenchyma to the gastric space are most important. The tubular nature of much of the cœnenchyma is evident, and longitudinal sections of some size prove that the spongy nature of it is by no means constant nor uniform.

In Heliopora, belonging also to the Milleporida, the cœnenchyma is very abundant, and covered here and there with rounded pores arranged more or less regularly and separated by papillose granules. These grains are the extremities of cylindrical "tigelles" which circumscribe the tubules, the calice of which is open at the surface. The calices are circular. The septa are slightly developed, and there are twelve of them. The tabulæ are well developed and horizontal (Ed. & H.). The nature of the conenchyma and the distinct septa distinguish this genus from the last. Both of the extinct species have a papillose or striated structure running over the cœnenchymal surface. In all the species the septa do not project far into the calice; but the amount of projection is not sufficient, as a structural peculiarity, in any case to determine more than a specific distinction. Hence MM. MilneEdwards and Jules Haime when they separate, in their scheme of the Milleporidæ *, Millepora and Heliopora and other genera from Heliolites, Propora, and Lyellia, the particular Palæozoic genera, they can only be permitted to do so on the plea that the plan renders the genera readily distinguishable. The projection or non-projection is not sufficient to determine a generic difference.

Now Heliolites has a beautiful conenchyma, very geometric, and not irregular and spongy; its cellules are placed regularly and symmetrically. In most of the species the septa are distinct, and project far inwards, but in Heliolites Grayi they are almost rudimentary.

The genus Polytremacis links Heliolites and Heliopora together, for its conenchyma is that of the second, and the septa resemble those of the firstnamed genus. Polytremacis is not older than Heliopora in the secondary ages, and the septal distinction which cannot expel Heliolites Grayi from its genus, and which is improperly allowed to distinguish Polytremacis and Heliopora, and these and Heliolites, may well have been produced by variations in a succession of early secondary forms.

* Op. cit. p. 223.

The septal development of Heliolites is exaggerated in Propora, a genus from the Upper Silurian, and which perhaps lasted into the Carboniferous. The costa in this genus are well developed, but the cœnenchymal cells are less geometric than in Heliolites. The structural relations are of the closest, and the generic distinction is not of the usual generic value. Another Upper Silurian genus, Lyellia, represents these symmetrical Milleporida in America. The corallite walls are subcostulate and not so costulate as in Propora. The septa (12) are well developed, as in Heliolites and Propora and Heliopora, and the coenenchyma is perfectly vesicular-spongy, in fact, like Heliopora. Here, then, in the distant and British and Northern European Silurians, there were closely allied forms varying amongst themselves, but more than the secondary types, the variation having some sort of likeness in both instances. It is impossible not to acknowledge the genetic affinities of all these genera except Millepora, of which more will be said, or to hesitate to assert that there has never been a break in the Tabulata, and that the Recent and Palæozoic Heliopora and Heliolites are very closely allied, the one being the descendant of the other*. Axopora is a tertiary genus, and its immense columella, which fills up the corallite inferiorly and leaves but little room in the calice around it, of course prevents the tabulæ from reaching across the axial space. The tabulæ come in contact with but do not perforate the columella, so that this structure grows progressively without any reference to them; they do not form floors upon which a columella is developed. There are no septa, and the cœnenchyma is reticulate in the extreme. No living analogue of this genus exists, and exception may be taken whether it be a true coral. It has no Palæozoic representatives.

Battersbyia is a very remarkable Paleozoic genus, and has been examined by me. MM. Milne-Edwards and Jules Haime § classify it with the Milleporidæ, but apparently only provisionally; but it will be noticed elsewhere. I have associated Battersbyia and Heterophyllia together as a new division of the Aporosa of the Astræidæ, under the name of the Palastræaceæ, which are noticed in the first part of this Report.

The Favositidæ are divided by MM. Milne-Edwards and Jules Haime into the following subfamilies: Favositina, Chatetina, Halysitinæ, Pocilloporinæ. All are presumed to present the following family characteristics :-" The corallum is formed essentially of the lamellar walls of the corallites, and possesses hardly any or no cœnenchyma. The visceral chambers are divided

by tabulæ, which are numerous and well developed."

The subfamilies without any conenchyma, and those whose corallites form a massive corallum, are the Favositinæ and the Chatetinæ, and the genera whose corallites are not united on all sides the Halysitinæ. The Pocilloporinæ constitute the cœnenchymal subfamily. One of the great difficulties of the zoophytologist appears strongly enough whilst investigating these Tabulata, for the question constantly arises, and can only be answered very unsatisfactorily, are such and such forms really Actinozoa? are they not Polyzoa, Hydrozoa, or of some class which has become extinct, and which has no modern representatives?

Some genera are characterized by the absence of septa. Thus Chatetes has long basaltiform corallites, numerous tabula which do not correspond in their plane throughout the corallum, no septa, and the reproduction is fissiparous.

* See Huxley's Address, Geol. Soc. 1870.

+ Pal. Soc. Tertiary Corals, 3rd Series, P. M. Duncan, pl. vii. figs. 11-15.
Op. cit. p. 244.

Phil. Trans. 1867.

Keyserling considered the genus to belong to the Alcyonaria amongst the Actinozoa; but MM. Milne-Edwards and Jules Haime, considering the great analogy between Chatetes and Favosites, and particularly with Beaumontia, "où la présence de cloisons n'est pas contestable "*, determined its position to be amongst the true Tabulata.

The same authors now recognize the necessity of separating Chatetes from Monticulipora, and assert that the members of the last-named genus increase by gemmation.

The genus Dania differs from Chaetetes in having the tabulæ on regular planes which traverse the whole corallum. This peculiarity is hardly of generic value.

Stellipora (Hall) is not generically different from Monticulipora, and the truth of this assertion can be estimated by comparing the diagnosis of the genera given by MM. Milne-Edwards and Jules Haime +.

The differentiation of Dekayia (Ed. & H.)‡ and of Labechia is also unsatisfactory, and their more or less mammillated cœnenchyma ranges them together by the side of Stellipora as subgenera of Monticulipora.

Now Jules Haime, when investigating the Oolitic Polyzoa, classified forms without septa and with tabulæ, like Chatetes or Monticulipora, as Polyzoa, and the beautiful Stellipore were especially included.

Now the question arises, are there any recent Polyzoa whose soft parts have been examined that have tabulæ? From our knowledge of the recent Polyzoa, it is unsafe to answer this in the affirmative. There is a freshwater species which is said to have tabulæ, but the assertion requires confirmation. The classification, then, of these forms amongst the Polyzoa must be deferred, and I propose to decide against it now.

Beaumontia, the genus noticed above, is distinguished by MM. MilneEdwards and Jules Haime § as follows:-" This genus is distinguished from all other Chætetinæ by the formation of its tabulæ, which are irregular or vesicular, and it thus resembles Michelenia, belonging to Favositinæ." The presence of septa belonging to three cycles is asserted by the same authors, and this fact must remove the genus quite out of the neighbourhood of septaless forms.

The genera of the Chatetina were formerly Chatetes, Monticulipora, Dania, Stellipora, Dekayia, Beaumontia, and Labechia. It has been shown that Stellipora, Dekayia, and Labechia are subgenera of Monticulipora, that Dania cannot be separated from Chatetes, and that Beaumontia has no correct affinity with the others, and that it belongs to another family. The genera should stand thus:

This result is interesting because it eliminates Beaumontia and makes a compact series, the affinities of which are not Polyzoan, but which may be Alcyonarian or Hydrozoan.

The long tabular or basaltiform corallites of Chatetes and its allied forms,

*Op. cit. pp. 271.

Ibid. p. 283.

+ Op. cit. vol. iii. pp. 272, 281.

§ Op. cit. vol. iii. p. 282.

and their more or less horizontal and perfect tabulæ, recall the Tubiporina amongst the order of the Alcyonaria.

The Alcyonaria are Actinozoa which are separated by MM. Milne-Edwards and Jules Haime from the Zoantharia on account of the pinnate structure of the tentacles, and from these important organs being invariably eight in

number.

The zoantharian tentacles, on the contrary, are simple or irregularly ramified, and increase in number with age.

The Alcyonaria are divided into the families of the Alcyonidæ, the Gorgonidæ, and the Pennatulidæ.

The first two families have an adherent corallum, and the last consists of free forms.

The Alcyonida have no hard central axis, but this characterizes the Gorgonidæ.

Now the Cornularinæ, Telestinæ, and Alcyoninæ, subfamilies of the Alcynidæ, are clearly allied to the Tubiporine by their soft structures; but the hard external structures of this subfamily are only faintly shown in the spiculate scoriaceous conditions of the external tegument of Nephthya, Spoggodes, and Paralcyonium. The polypes of Nephthya and Paralcyonium enter their spiculate and dense external covering when they contract; but the hard structures of Spoggodes celosia, Lesson, are very slightly developed.

Tubipora, Lamarck.

TUBIPORA (pars), Linnæus.

The genus has been examined by MM. Milne-Edwards and Jules Haime with their usual care and acumen.

The specimens of Tubipora are so common that the descriptions of these authors concerning the hard parts of the corallum can readily be followed.

The corallites are formed principally by a tabular wall, the tissue of which is calcareous and readily fractured. There are no septa, but there are rudimentary tabulæ, which cut off the visceral cavity into more or less perfect stages. The corallites are cylindrical, and usually attain an equal height; but they do not touch each other, for they are united by a peritheca, which is only seen here and there in distinct floors. The budding takes place from the connecting peritheca, which is therefore a true cœnenchyma, and not like that of Solenastræa. Were the corallites in contact the appearance of Chatetes would be presented; so that the presence of the cœnenchyma is the differentiating structure. It is only of generic value, and thus there is a very strong reason for associating the Chatetinæ and all the other fossils with long tubular structures, no septa, and tabulæ with the Alcyonidæ in the subfamily Tubiporine and near the genus Tubipora. These remarks are subject, of course, to the consideration whether the views of Agassiz already noticed are correct.

Reuss's genus Stylophyllum (Gosau Chalk) cannot be associated with the Alcyonidæ, for the species has septa. The corallites are united by their walls without there being a conenchyma, and the walls are imperforate. The junction of the corallites takes place by means of an epitheca.

The junction may occur at any part of the corallite.

The resemblance of Stylophyllum to some of the Halysitinæ (Ed. & H.)* necessitates an examination of their structural peculiarities.

Op. cit. vol. iii. p. 286.

1871.

K

MM. Milne-Edwards and Jules Haime differentiate the Halysitinæ as follows:

"The corallum is compound, but its corallites unite imperfectly, and constitute lamellar expansions or long fasciculi; they are either free on two sides, or are united together by 'expansions murales.'"

The septa are small, but usually very distinct; finally the walls are well developed and aporose.

The genera are:- Halysites, Fischer; Syringopora, Goldfuss; Thecostegites, Ed. & H. (Harmodites, Michelin); Conostegites, Ed. & H.; Fletcheria, Ed. & H.

Halysites. The species are invariably formed by corallites which are joined on two sides, and which in transverse outline resemble links of a chain. The epitheca is very strong, and unites the corallites perfectly where they are in contact from the base to the calice. Septa 12. Tabulæ horizontal and well developed. (Silurian.)

Thecostegites. The corallites have septa, horizontal tabulæ, and an exotheca unites them, and it is more or less tabular in structure, and exists in stages like the Tubipora. In T. parvula the cœnenchyma is nearly compact. (Devonian.)

Conostegites. There are numerous septal striæ, which mark also the smooth and convex surfaces of the tabulæ. The tabulæ are more or less infundibuliform, and the epitheca unites the corallites here and there. Syringopora. The corallum is fasciculate; the corallites are cylindrical and very long, parallel, and free laterally, except where horizontal tubes connect them. The walls are well developed, and clothed with a strong epitheca; septa exist. The tabulæ are infundibuliform.

Fletcheria. The corallum is fasciculate; the corallites are cylindrical, close, and long. The epitheca is complete; septa exist. Tabulæ horizontal and well developed. No intercorallite tubes or expansions of epitheca. Gemmation calicular.

It is evident that some of these genera are very slightly allied; for instance, Syringopora and Fletcheria, and both of them and Halysites.

Halysites, with its stout epitheca and simple tabulæ with non-tubular joints, is a very definite form.

Thecostegites should belong to the Milleporida.

Conostegites, with infundibuliform tabulæ, is related to Halysites as Michelinia is to Favosites.

Fletcheria is altogether aberrant.

The Halysitinæ comprehend, according to this analysis, Halysites, Fischer; Stylophyllum, Reuss; Conostegites, Ed. & H.

The genera Syringopora and Fletcheria will be considered further on. The subfamily of the Pocilloporinæ contains the genera Pocillopora and Coenites.

Pocillopora has septa (and my specimens show 12), which, even in fossil specimens, mark the top of the tabulæ. There is a columellary swelling on its tabulæ. The cœnenchyma is very stout and thick in old portions of the corallum, less so where growth has just ceased, and the cœnenchyma barely exists where the corallites or calices are developing. It is cellular at first, and then fills up with calcite and other coral salts.

Fossil forms have been described by Reuss and myself from the Cainozoic formations.

Canites resembles Pocillopora in a certain density of its conenchyma, but differs in only having three tooth-like septa, like the genus Alveolites.