92 CONTROLLED NATURAL SELECTION 
was present in the ancestors. Is it not reason- 
able to suppose that only those characters are 
retained which are useful, or that of a given 
character only so much as is useful: or else 
why are so many completely lost ? 
The Tadpole (Rana temporaria) has first 
external gills, then internal gills, and finally 
lungs, all of which are put to use: so too it 
is possible that the gill clefts or their rudi- 
ments found in the human embryo then serve 
some useful purpose. Natural Selection must 
deal with ova, embryos, juveniles, as it does 
with adults, and produce embryonic or juvenile 
characters. Ova and embryos, like adults, 
have environments to which they must adapt 
themselves. 
We are accustomed to study chiefly adult 
characters and to forget that adult is but a 
stage, merging on the one side into youth, and 
on the other, old age; each of which have 
important characters. Some of the characters 
of old age may be looked upon as means by 
which the species rids itself of an individual 
which has shed all its germ cells. 
Selection must occur with reference to a 
character, at the time of, or after, the appear- 
ance of the character, and not before. Thus 
the characters that go to form an embryo or 
