n open plantation;
and the seedlings varied in almost every single character, both in their
flowers and foliage, to a degree which {259} I have never seen exceeded;
yet they could not have been exposed to any great change in their
conditions.

With respect to animals, Azara has remarked with much surprise,[614] that,
whilst the feral horses on the Pampas are always of one of three colours,
and the cattle always of a uniform colour, yet these animals, when bred on
the unenclosed estancias, though kept in a state which can hardly be called
domesticated, and apparently exposed to almost identically the same
conditions as when they are feral, nevertheless display a great diversity
of colour. So again in India several species of fresh-water fish are only
so far treated artificially, that they are reared in great tanks; but this
small change is sufficient to induce much variability.[615]

Some facts on the effects of grafting, in regard to the variability of
trees, deserve attention. Cabanis asserts that when certain pears are
grafted on the quince, their seeds yield more varieties than do the seeds
of the same variety of pear when grafted on the wild pear.[616] But as the
pear and quince are distinct species, though so closely related that the
one can be readily grafted and succeeds admirably on the other, the fact of
variability being thus caused is not surprising; we are, however, here
enabled to see the cause, namely, the different nature of the stock with
its roots and the rest of the tree. Several North American varieties of the
plum and peach are well known to reproduce themselves truly by seed; but
Downing asserts,[617] "that when a graft is taken from one of these trees
and placed upon another stock, this grafted tree is found to lose its
singular property of producing the same variety by seed, and becomes like
all other worked trees;"--that is, its seedlings become highly variable.
Another case is worth giving: the Lalande variety of the walnut-tree leafs
between April 20th and May 15th, and its seedlings invariably inherit the
same habit; whilst several other varieties of the walnut leaf in June. Now,
if seedlings are raised from the May-leafing Lalande variety, grafted on
another May-leafing variety, though both stock and graft have the same
early habit of leafing, yet the seedlings leaf at various times, {260} even
as late as the 5th of June.[618] Such facts as these are well fitted to
show, on what obscure and s