Owing to the unequal rates at which the reduction of the
various parts might occur, some amount of irregularity in the flower
might arise, and on a second development towards insect
cross-fertilisation this irregularity, if useful, might be increased by
variation and selection.

The rapidity and comparative certainty with which such changes as are
here supposed do really take place, are well shown by the great
differences in floral structure, as regards the mode of fertilisation,
in allied genera and species, and even in some cases in varieties of the
same species. Thus in the Ranunculaceae we find the conspicuous part of
the flower to be the petals in Ranunculus, the sepals in Helleborus,
Anemone, etc., and the stamens in most species of Thalictrum. In all
these we have a simple regular flower, but in Aquilegia it is made
complex by the spurred petals, and in Delphinium and Aconitum it becomes
quite irregular. In the more simple class self-fertilisation occurs
freely, but it is prevented in the more complex flowers by the stamens
maturing before the pistil. In the Caprifoliaceae we have small and
regular greenish flowers, as in the moschatel (Adoxa); more conspicuous
regular open flowers without honey, as in the elder (Sambucus); and
tubular flowers increasing in length and irregularity, till in some,
like our common honeysuckle, they are adapted for fertilisation by moths
only, with abundant honey and delicious perfume to attract them. In the
Scrophulariaceae we find open, almost regular flowers, as Veronica and
Verbascum, fertilised by flies and bees, but also self-fertilised;
Scrophularia adapted in form and colour to be fertilised by wasps; and
the more complex and irregular flowers of Linaria, Rhinanthus,
Melampyrum, Pedicularis, etc., mostly adapted to be fertilised by bees.

In the genera Geranium, Polygonum, Veronica, and several others there is
a gradation of forms from large and bright to small and obscure coloured
flowers, and in every case the former are adapted for insect
fertilisation, often exclusively, while in the latter self-fertilisation
constantly occurs. In the yellow rattle (Rhinanthus Crista-galli) there
are two forms (which have been named _major_ and _minor_), the larger
and more conspicuous adapted to insect fertilisation only, the smaller
capable of self-fertilisation; and two similar forms exist in the
eyebright (Euphrasia officinalis). In both these cases there are special
modifications i