shades. Similarly the orange or buff-yellow Rhododendron Javanicum has been split up into various reds; the white having, so to say, eliminated the yellow.

The subsequent effect of crossing with regard to flowers is variety. With this fact florists and horticulturists are familiar for as soon as crossed or hybrid seedlings are raised the varieties of colouring become infinite. It has been observed that the "spots are more persistent than the base-colour of the flower. This fact agrees with the theory advanced that they have, whenever they occur as guides or path-finders, been determined by the insects and then become hereditary as much as the shape of the flower itself; and as that is maintained much more persistently than general colouring, so is that specialized colouring which has been equally due to insect agency.

With regard to the correlations which exist between ▾ colours and insect visitors, Müller especially has observed several. Thus beetles seem to affect yellows, e.g. Thalictrum and Galium verum; wasps and carrion insects, reddishbrowns, such as of Comarum, Epipactis, etc., while the more intelligent bees, etc., delight in purples and blues; and it is thought that their selective agency has determined the survival of such special colours as they prefer. This has been probably the case, but we still want to know what is the immediate cause which induces one colour to change to another.

As high colouring or conspicuousness if the flower be white is due to insects, so pale colouring and inconspicuousness is due to their absence; but what the nature of the stimulus is we cannot tell. It enhances the assimilative powers; for the crossed plants, as Mr. Darwin abundantly proved, are usually larger plants. It usually infuses some of the characters, floral or foliar, of the male parent-but

not always: several experimenters assert that, after every precaution, the offspring exactly resemble the maternal parent. But one rule florists always adopt in order to enhance the colouring is to use the pollen of the bettercoloured plant, the maternal parent being usually the inferior one.

*

As an illustration of the relative effect, of crossing and self-fertilisation respectively on the production of colours, I quote the following passage from Mr. Darwin's work: "The flowers produced by self-fertilised plants of the fourth generation [of Dianthus caryophyllus or Carnation] were as uniform in tint as those of a wild species, being of a pale pink or rose-colour. Analogous cases [occurred] with Mimulus and Ipomoea. On the other hand, the flowers

of plants raised from a cross with the fresh stock which bore dark crimson flowers, varied extremely in colour. . . . The great majority had their petals longitudinally and variously striped with two colours."

Uniformity and paleness of tint are thus correlated with self-fertilisation; and since, whenever the latter process is persevered with, an increase of fertility follows, it is not surprising to find that such tints are usually accompanied by an increased power of seed-bearing. Thus, Mr. Darwin found that, "the proportional number of seeds per capsule produced by the plants [of Dianthus] of crossed origin, to those by the plants of self-fertilised origin, was as 100 : 125." Again, of Antirrhinum majus, the relative self-fertility of red and white varieties was as 98: 20; of Mimulus luteus the same comparison gave the ratio of 100: 147; while palecoloured Pelargoniums are notoriously great seeders.†

* Cross and Self Fertilisation, etc., p. 139.

For further illustrations, see my paper on Self-fertilisation, etc.

CHAPTER XX.

THE EMERGENCE OF THE FLORAL WHORLS.

THEORETICALLY, as already stated, a perfect flower should or might be composed of six whorls, if its parts be not spirally disposed, the perianth, androecium, and gynoecium each consisting of two verticils. The very general rule for their emergence from the axis is centripetal. The subsequent rates of development of the several whorls may vary considerably, so that one part which emerged first, or at least very early, may be late or the last to arrive at maturity.

The calyx or outermost whorl of the perianth when present is nearly always the first to appear, and to grow rapidly to a relatively large size, and thus protects the more rudimentary parts within it; but if it ultimately remains rudimentary itself, or, it may be, is not entirely arrested, then it is the corolla which first emerges, the function of protecting the essential organs being relegated to it. Such is the case with the Composita, Valerianeæ, etc.

The corolla, with rare exception, emerges before the stamens, though it is very generally rapidly passed in development by the latter organs. In Lopezia and Primula, however, the stamens emerge first; and this has led some botanists to regard the petals of the last-named plant as

*

*For references and literature on the structure of Primulaceœ, see Masters's paper, On some Points in the Morphology of the Primulaceœ, Trans. Lin. Soc., 2nd series, BOTANY, vol. i., p. 285.

outgrowths from the stamens. My own observations tend to confirm those of Dr. Masters, that it is an exceptional fact, and not constant. It appeared to him “that in Lysimachia Nummularia the petals did really sometimes (but not always) precede the stamens in their development."

The stamens emerge before the pistil, and if there be two whorls to the andrœcium, it is the sepaline whorl which appears first; though the fully developed stamens sometimes assume a position, as already explained, within the petaline, as in Geraniaceae. Like the corolla and staminal whorls, the carpellary appears all at once, and last of all.

With reference to the emergence of the individual parts of the whorls, it is an almost invariable rule that those of the outermost whorl of the perianth or calyx, if it consist of three or five parts, rise centripetally in succession according to the laws of phyllotaxis. Thus, if the calyx be pentamerous, its parts invariably emerge in quincuncial order, thus constituting a cycle of the type. If it be trimerous, as in Monocotyledons, it is a cycle of the type. If, however, it be tetramerous, then the parts emerge in decussating pairs, as in Tamarix tetrandra, Sparmannia, Philadelphus, and the sepals in the Cruciferæ. This clearly shows that a normally tetramerous calyx is the result of the combination of two pairs of leaves, corresponding to two nodes, the internode between the pairs being suppressed.

*

The parts of the inner whorl of the perianth or petals of the corolla, as also those of each staminal and carpellary whorl, almost invariably emerge simultaneously if the whorls be regular; though pronounced differences may occur in the case of irregular flowers. Similarly, when there is a strong spiral tendency, as in the Ranunculaceae, members may arise

* The lateral sepals, though overlapped by the other pair, are the first to receive their vascular cords from the axis.

successively. If the stamens be very numerous they usually emerge in centripetal order, as in Buttercups; but they may form "centrifugal groups," as in Hypericum; the numerous stamens of Cistus and Helianthemum, as well as of Cactus, Opuntia, and Mesembryanthemum, and the Loaseœ, are also centrifugal in their development. Lastly, if the carpels form a whorl, they, too, emerge simultaneously; but if they be numerous and spirally arranged they emerge and develop in succession.

There are some additional points to be observed. The first is the method of change from tetramerous to pen

If

tamerous in the same plant. Thus in Celastrus scandens, if the flower be tetramerous, the sepals appear in pairs, the antero-posterior first, then the lateral pair afterwards. the flower be pentamerous the sepals arise in succession quincuncially, the numbers 1 and 3 being anterior; numbers 4 and 5 are lateral, and number 2 posterior.

Now, by referring to the diagrams above, it will be seen that this order is in exact agreement with the usual method of passing from opposite to alternate arrangements in the foliage. The correct angular distance or divergence being acquired immediately in the case of the calyx, by shifting the position of the parts so that the divergence of 144° is obtained. In the case of foliage, this is only secured after several internodes (see p. 18).