midrib elongated, and its apex discoloured, or a little distended. If we compare this with the pistil of a single cherry, the margins of the leaf with the ventral suture, the elongated midrib with the style, the discoloured distended apex with the stigma, they will be found to correspond exactly.

In this case there is an indisputable identity of origin and nature between the ovary and the blade of a leaf; between the little suture that occupies one angle of the carpel of a cherry, and the line of union of the two edges of the leaf; and between the elongated midrib, with its distended apex, and the style and stigma. There can be no doubt that the plan of all carpels is the same; so that the ovary is the blade of a leaf, the style an elongated midrib, and the stigma the denuded, secreting, humid apex of the latter.

Such being the origin of the carpel, its two edges will correspond, one to the midrib, the other to the united margins of the leaf. These edges often appear in the carpel like two sutures, of which that which corresponds to the midrib is called the dorsal, that which corresponds to the united margins is named the ventral, suture.

It is at some part of the ventral suture that is formed the placenta, which is a copious developement of cellular substance, out of which the ovules or young seeds arise. It, the placenta, originates in both margins of the carpellary leaf: but, as they are generally in a state of cohesion, there appears to be but one placenta; nevertheless, if, as sometimes happens, the margins of the carpellary leaf do not unite, there will be two obvious placenta to each carpel. Now, as the stigma is the termination of the dorsal suture, it occupies the same. position as that suture with regard to the two placentæ; consequently the normal position of the two placenta of a single carpel will, if they are separate, be right and left of the stigma. This is a fact important to bear in mind.

Pistils consisting of but one carpel are simple; of several, are compound. If the carpels of a compound pistil are distinct entirely or in part, they are apocarpous, as in Caltha; if they are completely united into an undivided body, as in Pyrus, they are syncarpous. That syncarpous pistils are really made up of a number of united carpels is easily shown, as

Goethe has well remarked, in the genus Nigella, in which N. orientalis has the carpels partially united, while N. damascena has them completely so; in the latter case, however, the styles are distinct. They and the stigmas are all consolidated in a single body, when the pistil acquires its most complete state of complication, as in the Tulip; which is, however, if carefully examined, nothing but an obvious modification of such a pistil as that of Nigella damascena.

This important conclusion is deducible from the foregoing considerations: viz., that, as the carpels are modified leaves, they are necessarily subject to the same laws of arrangement, and to no others, as leaves developed around a common axis upon one or several planes. For no axiom appears more incontestable in botany, than that all modifications of a given organ are controlled essentially in the same way, and by the same influences, as the organ itself in an unmodified state: and hence every theory of the structure of fruit which is not reducible to that which would be applicable to the structure of whorls of leaves is vicious of necessity. I shall proceed to demonstrate the perfect accordance of the carpellary theory of structure in every point with these principles.

The placenta usually arises from the two margins, either distinct or combined, of a leaf folded inwards. When a leaf is folded inwards, its margins will point towards the stem or axis around which it is developed; and in a whorl of leaves such inflected margins would all be collected round a common centre; or, if the axis were imaginary, in consequence of the whorl being terminal, would be placed next each other, in a circle of which the back of the leaves would represent the circumference. Therefore the placenta will always be turned towards the axis, or will actually meet there, forming a common centre; and, which is a very important consequence of this law, if one carpel only, with its single placenta, be formed in a flower, the true centre of that flower will be indicated by the side of the carpel occupied by the placenta. Proofs of this may be found in every blossom: but particularly in such as, habitually having but one carpel, occasionally form two, as the Wistaria sinensis, Alchemilla arvensis, Cerasus acida, &c.; in these the

second carpel, when added, does not arise by the side of the first, but opposite to it, the face of its placenta being in front of that of the habitual carpel. A fourth proof of this uniform direction of the placenta towards the axis is afforded by those pistils in which a great number of carpels is developed in several rows, as in the Strawberry and the Ranunculus : in all these the placenta will be, without exception, found directed towards the axis, and consequently towards the

123

124

back of every row, except the inner. For example, in the following diagram (123.), let o be the axis, bb placentæ, c c the backs of carpels; the placentæ, bb, of the inner row will be next the centre. o; the placenta, b b, of the second row will be next the backs, cc, of the first row; and so on.

If the order of developement of

leaves were exactly followed in that of the stamens and carpels, it would happen that the latter. would be invariably alternate with the inner row of stamens; for, if a a (fig. 124.) are the stations of five stamens, bb would be the situations of the carpels: this relative position is therefore considered the normal one, and is in fact that which usually exists in perfectly regular flowers; but as all the parts of a flower, in consequence of the non-developement of some parts, or the excessive developement of others, are subject to deviations, either real or apparent, from what is considered their normal state, it frequently happens that the carpels either bear no apparent relation to the stamens or are opposite to them. In papilionaceous plants, for example, where only one carpel is present, it is difficult to say that it bears any exact relation to the stamens, although it is probable that its position is really normal with regard to them; and so also in rosaceous plants, with numerous carpels, no exact relation

can be proved to exist between the latter and the stamens, unless it may be said to be indicated by those genera, such as Spiræa, in which the carpels are reduced to five; and, finally, in such plants as Delphinium, in which the carpels are three, while the floral envelopes and male system are divided upon a quinary plan, it is manifest that no alternation can exist between the stamens and carpels.

As the sepals and petals most commonly consist each of a single whorl of parts, so the pistil is more frequently composed of one whorl of carpels than of more. There are, however, certain families in which several whorls are produced one within the other, as in Fragaria, Ranunculus, Magnolia, Anona, and the like. In these cases it mostly happens that the carpels are either entirely separate or nearly so; but it sometimes is found that syncarpous pistils are habitually produced with more than one whorl of carpels, and consequently of cells, as Nicotiana multivalvis, and some varieties of the genus Citrus. In such instances the placenta of the outer series will necessarily be applied to the backs of the inner series, as has been just demonstrated.

This mutual relation of the different rows of carpels is sometimes observed when the receptacle from which they arise is either convex or concave: in the former state the outer series will obviously be lowermost, and in the latter uppermost; a circumstance that leads to no intricacy of structure when the carpels are distinct, but which may cause an exceedingly anomalous structure in syncarpous pistils, especially when accompanied by other unusual modifications.

There can be no doubt that the true nature of the composition of the pomegranate is to be explained upon this

principle. In order to make these considerations more clear, let figs. 125, 126, and 127. represent-fig. 125. a convex receptacle, with distinct carpels; fig. 126. a concave one, with the same; and fig. 127. a concave one, with the carpels consolidated. In these, a a are the outer row of carpels, bb the next, and d d the central row. The relative position of these, as the receptacle is convex or concave, will now be apparent.

I have stated that the placenta, however simple it may appear to be, is usually produced by the union of two united margins of a carpellary leaf: it is, therefore, essentially double; and, accordingly, we find that in polyspermous ovaries the ovules are almost always arranged in two rows, as in the Pea and Bean, the Quince, the Pæony, &c.

But there are exceptions to this rule. In Taylor's Magazine for Nov. 1837, I have shown that in Orobanchacea the placenta undoubtedly arise from the face of the carpel. That their capsule consists of two carpels standing right and left of the axis of inflorescence, and with the margins not inflected in the form of dissepiments, is incontestable. Yet in Orobanche and Phelypæa the capsule has the placentæ placed equidistant in pairs upon the face of each valve or carpel, and considerably within the margin. In Epiphegus each carpel has two intramarginal placenta, which diverge from the base upwards, and terminate before reaching the apex. In Lathræa there is to each valve but one placenta, which may be regarded as two confluent ones occupying the very face of the dorsal suture of the carpel. And finally in Æginetia indica, and I believe in Æginetia abbreviata also, the placenta is in like manner confined to the axis of the valve, occupying the same position upon the carpels as in Lathræa, but broken up into a number of parallel plates of unequal depth, over the whole surface of which multitudes of minute seeds are distributed. If we connect these facts, about which there can be no sort of question, with the well known placentation of Flacourtiaceæ, Nymphæaceæ, and Butomaceæ, we shall find that they invalidate the general carpological rule, that the placentæ belong to the ventral suture of a carpel, and consequently alternate with the dorsal; and we shall have to admit that the