position of the placenta with regard to the margins of the carpel is reducible to no certain rule, but depends upon specific organization. Consequently we shall no longer be unable to account for the unusual situation of the placenta opposite the stigma, in Papaver (as M. Kunth has lately noticed), in Parnassia, or elsewhere.

We ought not indeed to be surprised at coming to this result; for if the ovules are, as botanists generally believe them to be, a modification of buds, then the uncertainty in the position of the placentary lines will only be conformable to the uncertainty in the origin of buds from leaves. If in Bryophyllum, Malaxis paludosa, and most other cases, they usually spring from the edge of the leaf, they also arise from its surface in ferns; and in the famous case of the Ornithogalum leaf mentioned by Turpin, they were found issuing indiscriminately from all parts of its face.

When two leaves are developed upon a stem, they are always opposite, and never side by side. As carpels are modified leaves, they necessarily obey this law; and, consequently, when a pair of carpels forms a bilocular ovarium, the separation of the two cells is directly across the axis of the flower.

The partitions in ovaries, that are formed by the united sides of cohering carpels, and which separate the inside into cells, are called dissepiments or septa. It is important to bear in mind, not only that such is really their origin, but that they cannot possibly have any other origin, in order to form an exact idea of the structure of pistils. Now, as each dissepiment is thus formed of two united sides, it necessarily consists of two plates, which are, in the ovary state, often so completely united, that their double origin is undiscoverable, but which frequently separate in the ripe pericarp. This happens in Rhododendron, Euphorbia, Pentstemon, and a multitude of other plants. The consideration of this circumstance leads to certain laws which cannot be subject to exception, but which are of great importance; the principal of which are these:

1. All dissepiments are vertical and never horizontal.- For if a b, in fig. 128., represents the side of one carpel, and c d, that

128

f

a

с

b

d

g

h

of another, the dissepiment a c bd, formed by this union, will have precisely the same direction as that of the carpels, and can never acquire any other; and the same would be true of the sides e ƒ and g h, if they formed themselves into dissepiments by

uniting with other carpels: consequently a partition in any cell in the direction of ik could not be a dissepiment, but would be of a different nature.

A

d

129

C

margin of the carpel.

B

2. They are uniformly equal in number to the carpels out of which the pistillum is formed.-Suppose the triangle A B C represented a transverse section of an ovary formed by the union of three carpels 0, 0, 0;, then d, e, f would be the dissepiments, and could not be either more or less in number.

3. They proceed directly from the placenta, when that part originates in the As the placenta is the margin of the carpellary leaf, and as the dissepiment is the side of the carpellary leaf, it is evident that in such a case a dissepiment cannot exist apart from the placenta. Hence, when any partition exists in an ovary and is not connected with the placenta, if marginal, it follows that such a partition is not a dissepiment, however much it may otherwise resemble one.

A

130

m

d

B

4. They are alternate with placenta formed by the cohesion of the margins of the same carpel, and opposite to placenta formed by the cohesion of the contiguous margins of different carpels.-Let the triangle A B C represent a transverse section of a three-celled ovary of which d, e, f are the dissepiments: the dissepiments d and e will alternate with the placenta m, g, both belonging to the carpel A; but the dissepiment d will be opposite the placentæ

C

m, l, formed by the cohesion of the contiguous margins of the carpels A and B.

5. A single carpel can have no true dissepiment.

131

6. The dissepiment will alternate with the stigma: — for the stigma is the extremity of the midrib of the carpellary leaf, or of the dorsal suture of the carpel; and the sides of either of these (which form dissepiments) will be right and left of the stigma, or in the same position with regard to the latter organ as the sides of the lamina of a leaf to its apex. Let the triangle a b c represent a transverse section of a three-celled ovary, of which d, e, fare the dissepiments. The stigmas would occupy a position equal to that of the spaces s, s, s, and would consequently be alternate with d, e, f, the dissepiments: they could not possibly be placed opposite d, e, f, upon any principle of structure with which we are acquainted.

a

This law proves that neither the membrane which separates the two cells of a cruciferous siliqua, nor the vertical plate that divides the ovary of Astragalus into two equal portions, are dissepiments; both are expansions of the placenta, or of some other part, in different degrees.

All partitions whose position is at variance with the foregoing laws are spurious. Such spurious dissepiments are caused by many circumstances, the chief of which are the following:-they are caused by expansions of the placenta, as in Cruciferæ, when they form a partition stretching from one side to the other of the fruit; or they are mere dilatations of the lining of the pericarp, as in Cathartocarpus Fistula, in which they are horizontal; or they are internal expansions of the dorsal or ventral suture, as in Amelanchier, Astragalus, and Thespesia, in which they are distinguishable from the dissepiments by not bearing the placenta, and by being opposite the stigma, or by projecting beyond the placentæ; or, finally, they are caused by the sides of the ovary projecting into the cavity, uniting and forming many supernumerary cells, as in Diplophractum.

Such is the structure of an ovary in its most common state; certain deviations from it remain to be explained. We have seen that when carpels become syncarpous, they form a pistil, the ovary of which has as many cells and dissepiments as there are carpels employed in its construction. But sometimes the united sides of the carpels do not project so far into the cavity of the ovary as to meet in the axis; and then the result is an ovary, which, although composed of many carpels, is nevertheless one-celled (fig. 132.). In such case the dissepiments project a short distance only beyond the inner lining, or paries, of the ovary, and, bearing on their edges the placenta, the latter are said to be parietal. In other plants, such as Corydalis, Viola, and Orchis, the carpels are not folded together at all, but are spread open and united by their edges (fig. 133.): in that case the placenta do not project at all into the cavity of the ovary, but are still more strictly parietal than the last.

Another class of anomalies, of a still more remarkable character, is that in which there are no dissepiments, while the placentæ form a distinct mass in the centre of the ovary, as in Lychnis; this is what is called a free central placenta (fig. 134.). But, if we examine these plants at a very early period of their formation, long before the flowers expand, the explanation of the anomaly is sometimes not difficult. It will be found that such plants as Alsinaceæ have, at that time, their dissepiments meeting in the centre, and forming there a fungous placenta; but subsequently the shell of the ovary grows more rapidly than the dissepiments, and breaks away from them; while the excessive growth of the placenta afterwards destroys almost all trace of them: their previous presence is only to be detected by lines upon

the shell of the ovary, or by the separation of the mass of ovules into distinct parcels upon the placenta.

It seems to me difficult to explain the usual nature of the pistil and its parts more simply or in a more satisfactory manner than this; but Schleiden altogether objects to that part which attributes the placenta to the developement of ovules upon the edge of a carpel, or from a carpel at all. He maintains that the formation of the ovule in Taxus, where it terminates a branch, and is naked, and where the leaves are arranged in the customary spiral direction, even to the extreme summit, and where no one leaf implies in the slightest degree an adaptation to the female part more than another,— is incompatible with this theory; and he also adverts to the difficulty of explaining by it such a structure as that of Armeria, in which five carpels surround a single ovule, rising from the bottom of a cell upon a cord, which curves downwards at its apex, and thus suspends the ovule free in the centre of the cavity; he therefore supposes the ovule, and consequently the placenta, to be in all cases a production of the axis. As the opinions of Dr. Schleiden are in my mind always deserving of great attention, I extract a rather long passage from his paper on this subject.

We find in many

Although we cannot doubt that in plants possessing a free central placenta, or in those where, as in the Polygonacea, Taxus, Juglans, Myrica, the placenta cannot be supposed to exist as a separate organ, the nucleus of the ovule is only the summit of the axis, yet the question suggests itself as to how the parietal placenta is to be understood; and I do not consider the explanation to be very difficult. Araceae that the axis is expanded at its summit into a kind of disc, upon which is a number of buds or ovules, arranged like the flowers in the capitulum of Compositæ and other families. We next observe these discs expanded into lobed processes, and adherent to the edges of the carpellary leaves in all parietal or pseudocentral placenta; such a modification of the axis as this is what occurs in Dorstenia. The parietal placenta may be explained equally well, and perhaps with greater simplicity, as a mere ramification of the axis. It will