f life or growth they
are primarily caused. We shall see in the two following chapters that
various agencies, such as an abundant supply of food, exposure to a
different climate, increased use or disuse of parts, &c., prolonged during
several generations, certainly modify either the whole organisation or
certain organs. This direct action of changed conditions perhaps comes into
play much more frequently than can be proved, and it is at least clear that
in all cases of {268} bud-variation the action cannot have been through the
reproductive system.

With respect to the part which the reproductive system takes in causing
variability, we have seen in the eighteenth chapter that even slight
changes in the conditions of life have a remarkable power in causing a
greater or less degree of sterility. Hence it seems not improbable that
being generated though a system so easily affected should themselves be
affected, or should fail to inherit, or inherit in excess, characters
proper to their parents. We know that certain groups of organic beings,
but with exceptions in each group, have their reproductive systems much
more easily affected by changed conditions than other groups; for
instance, carnivorous birds more readily than carnivorous mammals, and
parrots more readily than pigeons; and this fact harmonizes with the
apparently capricious manner and degree in which various groups of
animals and plants vary under domestication.

Koelreuter[649] was struck with the parallelism between the excessive
variability of hybrids when crossed and recrossed in various
ways,--these hybrids having their reproductive powers more or less
affected,--and the variability of anciently cultivated plants. Max
Wichura[650] has gone one step farther, and shows that with many of our
highly cultivated plants, such as the hyacinth, tulip, auricula,
snapdragon, potato, cabbage, &c., which there is no reason to believe
have been hybridized, the anthers contain many irregular pollen-grains,
in the same state as in hybrids. He finds also in certain wild forms,
the same coincidence between the state of the pollen and a high degree
of variability, as in many species of Rubus; but in _R. caesius_ and
_idaeus_, which are not highly variable species, the pollen is sound.
It is also notorious that many cultivated plants, such as the banana,
pine-apple, brea