12 The Evolution of Fruits.
to the fruit as a guide to their affinities. It is impossible
here to do more than indicate my results, which have far
exceeded my original hopes.
A very slight knowledge of plants convinces one of the
great natural importance of the distinction, first pointed out
by the botanical glory of Essex, John Ray, between Mono-
cotyledons and Dicotyledons. Adopting this primary division,
and considering Mr. Bentham's Nudiflorae as the lowest
among Monocotyledons, we find that we have many mono-
carpellary fruits in very rudimentary flower-types, followed
by a ring of apocarpous follicles, generally six in number, in
the flowering-rush (Butomus), and by a spiral of more
numerous ones in the water-plantain (Alisma); but that
syncarpy is early originated (even in the Lemanea), the
fruit becoming capsular, and that succulence also originates
early (in the Araceae). We constantly find in tracing the
evolution of fruits that Nature arrives at the same result,
such as the capsule or nuculane, by various routes; hence the
practical uselessness of an artificial grouping of fruits.
Among Mr. Bentham's Coronariae a syncarpous fruit of
three carpels asserts itself as the typical monocotyledonous
fruit; succulence re-appears, as in Asparagus and Ruscus;
and we get the previously-mentioned reduction in the cocoa-
nut and date, accompanied in the former with that develop-
ment of woody texture in the pericarp, which proves an
effectual protection against the action of sea-water, if not
against crabs or monkeys.
The Glumiflorae I consider merely as reduced from the last-
mentioned type, their one-seeded caryopsis illustrating a
general principle which should have been before stated, that
when numerous small flowers form a crowded inflorescence
(as also in Composite and Umbelliferae) we have a reduction in
the number of carpels answering to a reduced need of them.
Among Dicotyledons my examination led me to recognise
four great groups, for which I select the names Thalamiflorae,
Disciflorae, Peryginae, and Bicarpellatae. Of these the second
seems undoubtedly a branch-phylum from the first, and the
fourth probably from the third. Among the Thalamiflorae