prevailed, the earlier forms may long remain unaltered or altered only in a
slight degree, and we are thus sometimes enabled to recover the connecting
links. This is the case in Persia and India with the tumbler and carrier,
which there differ but slightly from the rock-pigeon in the {220}
proportions of their beaks. So again in Java, the fantail sometimes has
only fourteen caudal feathers, and the tail is much less elevated and
expanded than in our improved birds; so that the Java bird forms a link
between a first-rate fantail and the rock-pigeon.

Occasionally a breed may be retained for some particular quality in a
nearly unaltered condition in the same country, together with highly
modified offshoots or sub-breeds, which are valued for some distinct
property. We see this exemplified in England, where the common tumbler,
which is valued only for its flight, does not differ much from its
parent-form, the Eastern tumbler; whereas the short-faced tumbler has been
prodigiously modified, from being valued, not for its flight, but for other
qualities. But the common-flying tumbler of Europe has already begun to
branch out into slightly different sub-breeds, such as the common English
tumbler, the Dutch roller, the Glasgow house-tumbler, and the long-faced
beard tumbler, &c.; and in the course of centuries, unless fashions greatly
change, these sub-breeds will diverge through the slow and insensible
process of unconscious selection, and become modified, in a greater and
greater degree. After a time the perfectly graduated links, which now
connect all these sub-breeds together, will be lost, for there would be no
object and much difficulty in retaining such a host of intermediate
sub-varieties.

The principle of divergence, together with the extinction of the many
previously existing intermediate forms, is so important for understanding
the origin of domestic races, as well as of species in a state of nature,
that I will enlarge a little more on this subject. Our third main group
includes carriers, barbs, and runts, which are plainly related to each
other, yet wonderfully distinct in several important characters. According
to the view given in the last chapter, these three races have probably
descended from an unknown race having an intermediate character, and this
from the rock-pigeon. Their characteristic differences are believed to be
due to different breeders having at an early period admired different
points of struct