finally becomes hereditary, _i.e._, develops in the
descendants in the absence of the stimuli, becomes in our sense
embryonic" (p. 180, 1881). Again, "form-characteristics which were
originally acquired in post-embryonic life through functional adaptation
may be developed in the embryo without the functional stimulus, and may
in later development become more or less completely differentiated, and
retain this differentiation without functional activity or with a
minimum of it. But in the continued absence of functional activity they
become atrophied ... and in the end disappear" (p. 201, 1881).

This conception of the nature of hereditary transmission is an important
one, and constitutes the first big step towards a real understanding of
the historical element in organic form and activity. It supplies a
practical criterion for the distinguishing of "heritage" characters from
acquired characters, of palingenetic from cenogenetic--a criterion which
descriptive morphology was unable to find.[484] The introduction of a
functional moment into the concept of heredity was a methodological
advance of the first importance, for it linked up in an understandable
way the problems of embryology, and indirectly of all morphology, with
the problem of hereditary transmission, and gave form and substance to
the conception of the organism as an historical being.

It is this element in Roux's theories that puts them so far in advance
of those of Weismann. Weismann did not really tackle the big problem of
the relation of form to function, and he left no place in his mechanical
system of preformation for functional or second-period development; he
conceived all development to be in Roux's sense embryonic, and due to
the automatic unpacking of a complex germinal organisation. Roux himself
was to a certain extent a preformationist, for the development of his
first-period characters is conditioned by the inherited organisation of
the germ-plasm, and is purely automatic. It was indeed his experiments
on the frog's egg (1888) that supplied some of the strongest evidence in
favour of the mosaic theory of development. The number of _Anlagen_
which he postulates in the germ is however small, and the germ-plasm in
his conception of it has a relatively simple structure (p. 103, 1905).

The transmission of acquired characters forms, of course, an integral
part of Roux's conception of heredity and development, for without this
transmission second-stage