is still more marked in that the number of ovaries is reduced to six, that is to say, two whorls of three each, accompanied by nine stamens, similarly divisible into three rows. In all these very early forms (as in their analogues the buttercups) the main point to notice is this, that there is as yet no regular definiteness in the numerical relations of the parts. They tend to run, it is true, in rows of three; but often these rows are so numerous and so confused that nature loses count, so to speak, and it is only in their higher and more developed members that we begin to arrive at any distinct symmetry, such as that of the flowering rush. Even here, the symmetry is far from being so perfect as in the later lilies. There are, however, a few very special members of the alisma family in which the approach to the true lilies is even greater. These are well represented in England by our own common arrowgrasses—inconspicuous little green flowers, with three calyx-pieces, three petals, six stamens, and either six or three ovaries. Here, too, the ovaries are at first united into a single pistil (as it is technically called), though they afterward separate as they ripen into three or six distinct little capsules. One of our British kinds, the marsh arrowgrass, has almost reached the lily stage of development; for it has three calyx-pieces, three petals, six stamens, and three ovaries, exactly like the true lilies; but it falls short of their full type in the fact that its pistil divides when ripe into separate capsules, whereas the pistil of the lilies always remains united to the very end; and this minute difference suffices, in the eyes of systematic botanists, to make it an alisma rather than a lily. In reality, it ought to be regarded as a benevolent neutral—a surviving intermediate link between the two larger classes.
The specialization which makes the true lilies thus depends upon two points. In the first place, all the parts are regularly symmetrical, except that there are two rows of stamens to each one of the other organs: the common formula being three calyx-pieces, three petals, six stamens, and three ovaries. In the second place, the three ovaries are completely combined together into a single three-celled pistil. The advantage which the lilies thus gain is obvious enough. Then bright petals, usually larger and more attractive than those of the alismas, allure a sufficient number of insects to enable them to dispense with the numerous stamens and ovaries of their primitive ancestors. Moreover, this diminution in number is accompanied by an increase in effectiveness and specialization: for the lilies have only three sensitive surfaces to their pistil, combined on a single stalk; and the honey is usually so placed at its base that the insect can not fail to brush off pollen at every visit against all three surfaces at once. Again, while the number of ovaries has been lessened, the number of seeds in each has been generally increased, which also marks a step in advance, since it allows many seeds to be impregnated by a single act of pollination. The result of all these improvements, carried further by