flowers (Ibid. page 230.). Nevertheless the variety continues to be
reproduced from seed, because in addition to the double and infertile
flowers, the seeds always produce a certain number of single, fertile
blossoms, and these are used to reproduce the double variety. These single
and fertile plants correspond "to the males and females of an ant-colony,
the infertile plants, which are regularly produced in large numbers, to the
neuter workers of the colony."

This illustration is entirely apt, the only difference between the two
cases consisting in the fact that the variation in the flower is not a
useful, but a disadvantageous one, which can only be preserved by
artificial selection on the part of the gardener, while the transformations
that have taken place parallel with the sterility of the ants are useful,
since they procure for the colony an advantage in the struggle for
existence, and they are therefore preserved by natural selection. Even the
sterility itself in this case is not disadvantageous, since the fertility
of the true females has at the same time considerably increased. We may
therefore regard the sterile forms of ants, which have gradually been
adapted in several directions to varying functions, AS A CERTAIN PROOF that
selection really takes place in the germ-cells of the fathers and mothers
of the workers, and that SPECIAL COMPLEXES OF PRIMORDIA (IDS) are present
in the workers and in the males and females, and these complexes contain
the primordia of the individual parts (DETERMINANTS). But since all living
entities vary, the determinants must also vary, now in a favourable, now in
an unfavourable direction. If a female produces eggs, which contain
favourably varying determinants in the worker-ids, then these eggs will
give rise to workers modified in the favourable direction, and if this
happens with many females, the colony concerned will contain a better kind
of worker than other colonies.