_Delphinium_, as shown by Braun,[376] the stamens and carpels are
members of a continuous spiral series, and in the double balsam an extra
corolline whorl is produced, without the suppression of the stamens, in
the following manner: the ordinary stamens are replaced by petals, the
carpels by stamens, while an additional whorl of carpels is produced at
the summit of the axis. In this instance, therefore, the doubling is
distinctly referrible to an absolute increase in the number of whorls,
and not to chorisis.[377]

On the other hand, it must be admitted that there are many cases which
are not to be explained in any other way than that suggested by the
French botanists before alluded to. Probably, the main difficulty in the
way of accepting the doctrine of chorisis is the unfortunate selection
of the word used to designate the process; this naturally suggests a
splitting of an organ already perfectly formed into two or more
portions, either in the same plane as the original organs, "parallel
chorisis;" or at right angles to it "collateral chorisis." Indeed,
before so much attention had been paid to the way in which the floral
organs are developed, it was thought that an actual splitting and
dilamination did really take place; Dunal and Moquin both assert as
much. The truth would rather seem to be that, in the so-called parallel
chorisis at least, the process is one of hypertrophy and
over-development rather than of splitting. The adventitious petal or
scale is an excrescence or an outgrowth from the primary organ, and
formed subsequently to it.

In the case of "compound stamens" the original stamens are first
developed each from its own cellular "mamelon," or growing point; and,
after a time, other secondary growing points emerge from the primary
one, and in this way the stamens are increased in number, without
reference, necessarily, to the so-called law of alternation. Outgrowths
from leaves, multiplying the laminar surface, are alluded to under the
head of hypertrophy, and it is probable that some of the cases of
duplication of the flower, or of the formation of adventitious segments
outside the ordinary corolla as alluded to in succeeding paragraphs (see
Pleiotaxy of the corolla), are due to a similar process.[378]

The formation of parts in unwonted numbers may be merely a reversion to
what is supposed to have been the original form, and in this way there
may be a restoration of parts that are usually undeveloped