of distinct tissue, {283} namely, the
epidermic, sub-epidermic, spongy, intermediate, and the hard protective
layer formed of curiously thickened woody cells, and, lastly, the central
mass abounding with starch-granules on which the larvae feed.

Galls are produced by insects of various orders, but the greater number by
species of Cynips. It is impossible to read M. Lucaze-Duthier's discussion
and doubt that the poisonous secretion of the insect causes the growth of
the gall, and every one knows how virulent is the poison secreted by wasps
and bees, which belong to the same order with Cynips. Galls grow with
extraordinary rapidity, and it is said that they attain their full size in
a few days;[704] it is certain that they are almost completely developed
before the larvae are hatched. Considering that many gall-insects are
extremely small, the drop of secreted poison must be excessively minute; it
probably acts on one or two cells alone, which, being abnormally
stimulated, rapidly increase by a process of self-division. Galls, as Mr.
Walsh[705] remarks, afford good, constant, and definite characters, each
kind keeping as true to form as does any independent organic being. This
fact becomes still more remarkable when we hear that, for instance, seven
out of the ten different kinds of galls produced on _Salix humilis_ are
formed by gall-gnats (_Cecidomyidae_) which, "though essentially distinct
species, yet resemble one another so closely that in almost all cases it is
difficult, and in some cases impossible, to distinguish the full-grown
insects one from the other."[706] For in accordance with a wide-spread
analogy we may safely infer that the poison secreted by insects so closely
allied would not differ much in nature; yet this slight difference is
sufficient to induce widely different results. In some few cases the same
species of gall-gnat produces on distinct species of willows galls which
cannot be distinguished; the _Cynips fecundatrix_, also, has been known to
produce on the Turkish oak, to which it is not properly attached, exactly
the same kind of gall as on the European oak.[707] These latter facts
apparently prove that the nature of the poison is a much more powerful
{284} agent in determining the form of the gall than the specific character
of the tree which is acted on.

As the poisonous secretion of insects belonging to various orders has the
special power of affecting the growth of various plants;--as a slig