quite as functionless as are the cilia on the leaf margin of our common barberry. These cilia we saw good reason for believing to be rudimentary spines. The supposition that the stamen appendages are degenerate nectar glands would seem to be scarcely less probable, in spite of our inability to find as full a series of intermediate stages. For it should be remembered that the time
| Fig. 14.—Berberis aquifolium. Stamen showing appendages (A). |
| Fig. 15.—Lindera benzoin. Stamen showing nectar glands (N). |
| Fig. 16.—Berberis vulgaris. Petal showing nectar glands (N). |
which has elapsed since the development of the floral peculiarities here considered is surely much greater than in the case of the foliar modifications, and consequently it would be strange indeed if the intermediate stages had not disappeared.
Although, in regard to the evolution of the floral organs, there is so much less opportunity than with the vegetative system to test the validity of our conjectures, yet it may not be entirely profitless to follow such clews as are available, and endeavor to reconstruct hypothetically the main features of those more primitive flowers from which the present barberry type was derived.
A multitude of stamens and pistils is generally recognized as characteristic of primitive flowers;[1] hence, we shall probably be not far wrong in considering the remote ancestor of the barberries and their kin to be in this regard very like a marsh marigold (Caltha), although doubtless less conspicuous, and with the parts more definitely arranged. As it is a very general characteristic of berberidaceous flowers that the parts are in whorls of three, we should expect this to be the case with the common ancestor. Accordingly, we arrive at a generalized type of flower, the structure of which may be expressed diagrammatically as in
- ↑ In the Lardizabalacæ, an exotic group which some botanists consider to be a subfamily of Berberidaceæ, the pistils are from three to nine in number.
