Exactly the same procedure occurs in Sparmannia and Philadelphus, which are tetramerous, as compared with Tilia and Deutzia respectively, which are pentamerous (see p. 18).

The next point to be noticed is the alteration in the order of emergence which takes place in irregular flowers. The rule seems to be that those parts of the flowers which assume a greater prominence in the mature state, or have some special function beyond the rest, emerge and develop before the others. Thus in Leguminosa and Labiata, where there is a prominent "landing-place" for insects, the petals issue successively in an antero-posterior order. The carina of papilionaceous flowers composed of two petals appears first, then the alæ together, and finally the vexillum. In Reseda, the sepals, petals, and stamens issue in a posteroanterior manner; but while the sepals finally attain to much the same dimensions, the petals remain more or less atrophied as they emerge towards the anterior side. Then the stamens appear in the same order upon a cellular ring, which, later on, grows out into the unilateral disk between the petals and

stamens.

In a few regular flowers the simultaneity is also wanting: thus in Adoxa the sepals of the tetramerous terminal flower emerge in pairs, and the four petals simultaneously; but in the lateral flowers the posterior sepals issue before the anterior; and of the five petals the posterior one emerges first, the two lateral secondly, and the two anterior ones last of all. These modifications are continued in the order of flowering. Thus the terminal flower expands first, and "all at once." Of the lower lateral flowers the two upper posterior sepals open out first, then the posterior stamens mature and shed their pollen. The anthers dehisce in succession from the lateral stamens, and lastly from the anterior

ones. The lower sepals do not separate until after the upper stamens have shed their pollen.*

Though we are not in a position yet to account for all such deviations from general rules, yet I think in such cases as the Leguminosa and Labiata, and probably all irregular flowers, that the rationale may with great probability be assumed to be the stimulus given from without to meet the extra strain which certain petals or stamens or both have to sustain while supporting the weight of an insect when visiting them. To meet this demand an extra supply of nutriment is sent to the parts which thus require it; and, in fact, I believe the final result has thus been actually brought about by the effort of the plant itself, so that it has developed parts in accordance with its requirements in a manner parallel with that which has obtained in the animal kingdom.

In the case of Adoxa I would regard the above-mentioned orders of development as a result of unequal distribution of nutriment in order of time. Thus the apical flower receives its nutriment first and develops first; then the other flowers which are placed laterally subsequently. And this order of supply has affected the parts of the latter flowers in the same way, so that they develop from above downwards, or in a postero-anterior manner. It may be compared to a three-flowered cyme, of which the central flower expands first, and the two lower ones afterwards.

A feature must here be noticed, though I do not think much stress need be laid upon it, which botanists have called “obdiplostemony." If a flower have one whorl of stamens of the same number as the petals it is isostemonous; of two, diplostemonous; and if the stamens of the outer whorl be opposite or

* For a note on Adoxa, see my paper On the Origin of Floral Estivations, Trans. Lin. Soc., 2nd series, BOTANY, vol. i., p. 194.

† Sachs' Text-Book, 2nd edition, p. 601.

superposed to the petals, and therefore antipetalous, then the above term is used for the rule is that the calycine whorl should be outermost and emerge first; then the petaline, which usually takes a position higher up on the axil; and, in at least most of the genera and orders where obdiplostemony has been noticed in the completely developed flower, it is simply due to the petaline whorl of filaments being, so to say, thrust outside the level of the calycine whorl by the protruding buttress-like bases of the carpels, as in Geranium pratense. This is still more the case in Oxalis, where, as in Geranium, the sepaline stamens become the taller set, the petaline the shorter; and the position of the former being more internal than usual, apparently in consequence of the appendages which grow on the outer side of the filaments.*

Again, the order of emergence may be the same as usual, namely the sepaline stamens first, then the petaline; but the position of the latter, instead of being within as is the rule, may be apparently on exactly the same plane as the sepaline, as in Heaths. Since, however, they do not emerge simultaneously, but one set is intercalated between the other, or even outside of it (Fig. 51), this order of appearance is, to my mind, a sufficient proof that they do not really belong to the calycine Fig. 51. — whorl.

8$

- Diagram of emergence of petaline stamens of Peganum, outside the sepaline (after Payer).

There is no greater difficulty in understanding this, than in seeing that a compression of the internodes of opposite and verticillate leaves has taken place when double the usual number are present in a whorl. Thus privet has sometimes four leaves at one node, forming a quaternary whorl, and all on the same plane; and

* According to Frank, in Oxalideœ and Geraniaceœ, it is the antipetalous stamens which are developed first. See above, p. 150.

this will remind the reader that, since floral whorls are based upon phyllotaxis, ten stamens could not possibly form a cycle; and although the eight stamens of a Heath might do so, there is nothing in the leaf arrangement of that genus to suggest their being a whorl of the 3 type.

Since the petaline cords are usually united to the staminal ones, the fact that the petaline stamens get sometimes, as it were, "dragged outwards," offers really no great difficulty; but is, so to say, a mere accident brought about by the adaptations of the flower to insect agency.

Indeed, to interpret these irregularities in the emergence, one must look to the final condition to see if there are any ultimate results in correlation with them. In Oxalis we get heterostylism with its corresponding different lengths of the filaments, and the necessary adjustments of the latter; since there are at least two sets in each flower, for insects to readily secure the pollen. In Heaths all the anthers are arranged in a ring round the style, pressing their cells against it, and so closely approximated, that when a bee dislocates one by pushing the lever-like auricle to one side, she dislocates the whole, and so receives a shower of pollen.

These final arrangements, therefore, are suggestive of the reason why the points of emergence of the stamens occur just where they do.

In the case of Hypericum, where the stamens emerge centrifugally, from a definite number of original papillæ, three or five as the case may be, the stigmas extend outwards; so that, if they have not been pollinated by insects, they can come in contact with the latest formed or the outermost anthers.

CHAPTER XXI,

THE DEVELOPMENT OF THE FLORAL WHORLS.

THE order in which the several whorls of flowers emerge from the axis is, as stated above, almost invariable; but the rates of development are very various, and important sexual and other differences follow as the results. For flowers with conspicuous corollas or other structures attractive to insects, the prevailing order of progression subsequent to emergence is first the calyx, secondly the stamens, and, if there be two series, the whorl superposed to the sepals grows first, afterwards the whorl superposed to the petals; then follows the pistil to a point approaching maturity, when the corolla, just before expansion, grows very rapidly to its full size; and finally the stigmas mature. The anthers have also grown long before the filaments, which at last elongate very rapidly. The usual result on maturity is various degrees of protandry, coupled with conspicuousness or attractiveness to insects. As a few of the examples I have examined may be mentioned Ranunculus acris, Cardamine pratensis, Stellaria Holostea, Lychnis dioica (male), Malva moschata, Geranium (larger flowered sp.), Pelargonium, Tropaeolum, Epilobium hirsutum, Enothera biennis, Ipomea, Veronica Chamadrys, etc. In fact, this order of growth and development prevails generally with flowers having conspicuous corollas.

The interpretation appears to be as follows. In such