may in some species support two or more acrogene growths, and in just these cases also the acrogene part may be small or wanting, arguing that the basal expansion might have been the origin of some strictly perigene species, the acrogene part being wholly suppressed. The laminate form may have again become massive. Finally, an acrogene growth may be round, compressed, flattened, frond-shaped, or bifoliate. In short, the series between zoarial forms is very complete, and genetic relationship between the most extreme forms may be presumed. The lines of evolutionary development have never been traced, however, and they evidently cross or parallel in a confusing manner; hence this character is of taxonomic use in species chiefly.

The different zoarial forms result not from changed shape and size of the component cells, as one can readily observe in similar zoaria of extremely different cells, but, as seen firstly in variations of a species, it results from increase of cells, the region of greatest cell increase being that of greatest zoarial growth, and inversely. The change of zoarial form, nevertheless, must be made to explain the change of cell as seen in different regions in the same species. Thus, in the acrogene growth, where the cells are turned from the axis of rapid growth to the peripheral slow growth region, it changes markedly, becoming thick walled, closer tabulated, etc. (Pl. A, Fig. 6, c). Noticing that the cell apertures do not spring apart under any circumstances, and presuming that this is because the respective zooids were bound together by a cortex, it can be understood how the same cell can be different in two parts, and why there are certain differences in cells. Thus, the tip of an acrogene zoarium, like a hemispheric or massive zoarium, has the cells subparallel, lengthening as new cells develop, so that the surface or circumference widens and the radius or cell-length increases proportionately. But as a cell turns into the peripheral region it comes into a zone of more restricted circumference and radial lengthening; hence the cell is shorter, thicker-walled, and closer-tabulated. Moreover, the cell-increase lessens to some degree, which further restricts the circumference and radius with added effect. In the

latter way the apex of branches often retarded building thickwalled and close-tabulated cells, and sometimes the basal expansions did likewise; and it is evident then that a kind of zoarial maturity exists, simulating if not derived from the "peripheral " stage. Further, specimens in which a mature stage has developed at the apex may renew axial growth and cell characters, a second maturity following, by which it would seem that environmental causes have had to do with the time of maturity.

In a specimen of Eridotrypa mutabilis Ulr. at hand, a fortuitous branch has arisen from the mature region instead of the axial, and consequently a cell can be traced as axial, peripheral, axial, and peripheral again. Injured or broken zoarial parts are commonly renewed by thin (axial) cell growth, even in the peripheral region, a stolon-like, expansion first overspreading the dead area. In perigene growth also the young cells arising at the margin are at first more prostrate than later, giving rise to a basal stolonal region, and in thin, laminar, or encrusting forms it is very distinct. In these cases the stolonal has been said to be probably the homologue of the axial immature region of dendroid zoaria. The stolonal and immature regions in these may coincide, but it is wrong to assume that they do in other and all cases. The stolonal part need not be considered as coördinate with the developmental peripheral and axial regions. It seems, in fact, unnecessary to attach any genetic significance to the stolonal region further than that it is incidental to perigene growth.

In a given species one specimen larger than the other may be so either from more vigorous growth, as seen in larger axial or immature region, or from longer continued growth, as seen in thicker peripheral or mature region.

Cell increase is, as a rule, intermural, i. e., young cells begin as small, round pits in the wall at the angle between three or more older cells. The reproduction or budding of zooids doubtless took place in the cortex above the zoarial skeleton, and only later the young zooid comes to build a cell, which, however, from its initial, has its own wall, i. e., wall-half. The young cell

becomes triangular, quadrangular, etc., in proportion as it grows large enough to neighbor on three or more cells. Young cells have shallower calycals and are generally closer-tabulated, but proportionate to their size, as compared to mature ones. The number of young cells in proportion to large ones affects the cell pattern at the surface. This is especially notable when the simplest case of rapid, direct, continuous expansion of young cells is contrasted with that where "mesopores" are numerous (Plate A, Figs. I and 10).

Mesopores are said to be present if the young cells when about half-grown in diameter, retard or cease their expansion, but, of course, continuing length growth. If the mesopore stage of cell is short, few mesopores, if long, many mesopores, are present. They may outnumber the autocells so greatly that a small proportion only could become autocells. They may be more numerous in the peripheral, mature, or retarded cell growth region. Again, as in Prasopora simulatrix many mesopores remain such while other newer ones develop to autocells. Yet, apparently any mesopore may finally become autocell. They are, however, something more than retarded young cells, in case like Stellipora (Constellaria), where the angular mesopores are rather larger instead of smaller than the rounded autocells.

The more distinct the mesopore development the more subordinate they appear to become. The autocells are angular from contact with each other or young cells, but rounded when crowding mesopores, the latter alone remaining angular. Also the mesopores become shallow, the tabulæ developing close to the wall margin forming "closed" mesopores, or they even filled solid with superimposed tabulæ; or the tabulæ overlap the walls, forming "vesiculose" structure. Autocells arise, displacing several mesopores at once, "coenenchymal gemmation," in some species with vesiculose or even regular mesopores.

Monticules, as Nicholson pointed out, are rapid cell increase areas. In simplest cases they appear at the surface as mere small elevations or more or less elevated clusters of slightly larger sized cells, among which young cells are seen except

rarely in a growth retardation stage. A few young cells or mesopores are present, or many mesopores, or again aggregated mesopores between the cells are found in some species, or exclusively mesopores form maculæ or large aggregates in other species, with or without surrounding major-sized autocells. The elevated clusters are the typical monticules. The extreme degree of elevation is expressed by the name monticule. Maculose ones or "macula" are nearer plane or slightly depressed. Exceptionally a typical monticule may extend like a ramulet, in which case it is probably to be considered as such, a fortuitous acrogene growth having sprung from the area of one monticule as it might also have from that of several.

The monticules are distributed on the zoarial surface more or less regularly, new monticules arising from the widening intervals, and never, apparently, one monticule from another. The function of the zooids that built the major autocells is unknown if it was different from that of others. The only discernible peculiarity of the monticules to which a special function could be assigned is their more rapid reproduction or cell increase than the interspaces. In this respect they resemble the acrogene growths or the axial region, and when very prominent they might suggest an origin as retarded branches, but no unquestionable transitional forms to these are known to me. They differ from branches in their size and in relation to the zoarium as a whole; for, if monticules produce the more new cells, their interspaces produce the less, the zoarium being unchanged; while, on the contrary, growth of branches comprises essentially the zoarial growth as well as the maximum increase of cells.

The surface pattern, as shown, is very diverse. The average size of cells in a species is quite constant, but in different species differs several diameters. The form of aperture, quadrangular, polygonal, to rounded; the relative size and numbers. of young cells and mesopores; the monticules and maculæ of varied style; all these form conspicuous essential characters by which species can be recognized. The calycals, too, are deep

or shallow, relatively, in different species. The autocell walls, which are thin and then more or less thickened, give respectively polygonal or rounded calycals, since the thickening is greatest at the cell angles. A beveled wall edge or impressed rim around the cell aperture gives it in many a saucer-shape, especially when the walls are very thick (Plate A, Fig. 11).

Acanthopores or warts may be present on the walls, usually at cell angles. A lunarium, when present, gives the cells another peculiarity. The lunarial wall and the acanthopores are, however, mural modifications, and will be explained in that connection later.

Thin section is quite necessary to bring out the wall's structure. The walls are dense. If thin they may show no differentiation; yet, if a specimen split, the cleavage may pass longitudinally and so as to leave part of the wall attached to each stone core, which fact has been taken to indicate that the wall is always structurally double! Presumably the wall is built double always, i. e., the increment on the margin is continued within every calycal, the wall being thus double with a median, third part, which is, however, not known to be double. A thickened wall tends to show greater differentiation, both in structure and composition, than a thin one. The median wall may appear either distinct or not, and correspondingly the striping parallel to the wall's surface when seen, showing the laminæ of growth, is either interrupted by the median wall or more or less distinctly crosses continuously from one side wall to the

other.

To explain the structural aspect of acanthopores in thin sections, the following analysis may serve. A distinctly double wall shows the median wall as a line in transverse section (Fig. I, a) and when the wall edge is scalloped (Fig. b) the median line appears interrupted in section (Fig. c), corrresponding to the angles. A rounded wart (Fig. d) would produce a similar effect (Fig. e); or, if very distinct, as in Fig. f. When once begun, the wart may have its own growth, so to say, independent of the wall thickness (Fig. g), becoming so large as to appear