page 66, that certain varieties of a closely allied plant, the
Convolvulus tricolor, cannot be kept pure unless grown at a distance
from all other varieties.) It is, therefore, remarkable that the plants
raised by me from flowers which were, in all probability,
self-fertilised for the first time after many generations of crossing,
should have been so markedly inferior in height to the intercrossed
plants as they were, namely, as 76 to 100. As the plants which were
self-fertilised in each succeeding generation necessarily became much
more closely interbred in the later than in the earlier generations, it
might have been expected that the difference in height between them and
the crossed plants would have gone on increasing; but, so far is this
from being the case, that the difference between the two sets of plants
in the seventh, eighth, and ninth generations taken together is less
than in the first and second generations together. When, however, we
remember that the self-fertilised and crossed plants are all descended
from the same mother-plant, that many of the crossed plants in each
generation were related, often closely related, and that all were
exposed to the same conditions, which, as we shall hereafter find, is a
very important circumstance, it is not at all surprising that the
difference between them should have somewhat decreased in the later
generations. It is, on the contrary, an astonishing fact, that the
crossed plants should have been victorious, even to a slight degree,
over the self-fertilised plants of the later generations.

The much greater constitutional vigour of the crossed than of the
self-fertilised plants, was proved on five occasions in various ways;
namely, by exposing them, while young, to a low temperature or to a
sudden change of temperature, or by growing them, under very
unfavourable conditions, in competition with full-grown plants of other
kinds.