loss and recovery of wings in various generations of the same species,
he has concluded that the gap between the exopterygote and the
endopterygote method of development may have been bridged by an
anapterygote condition; that the ancestors of those insects with
complete transformations were the wingless descendants of primitive
insects which grew their wings from visible external rudiments, and
that in later times re-acquiring wings, they developed these organs in a
new way, from inwardly directed rudiments or imaginal buds.

This theory of Sharp's is original, daring, and ingenious, but the loss
and re-acquisition of wings which it presupposes is difficult to imagine
in large groups during a prolonged evolutionary history, while the
sudden appearance of a totally new mode of wing-growth in the offspring
of wingless insects would be an extreme example of discontinuity in
development.

On the whole the most probable suggestion which can be made as to the
origin of 'complete' transformation in insects is that the instar in
which wings were first visible externally became later and later in the
course of the evolution of the more highly organised groups. In this way
a gradual transition from the exopterygote to the endopterygote type of
life-story is at least conceivable. It will be remembered that a may-fly
(p. 33) undergoes a moult after acquiring functional wings, emerging
into the air as a 'sub-imago.' In not a few endopterygote insects, the
pupa shows more or less activity, swimming through water intermittently
(gnats) or just before the imago has to emerge (caddis-flies); working
its way out of the ground (crane-flies) or coming half-way out of its
cocoon (many moths). The pupa of the higher insects almost certainly
corresponds with the may-fly's sub-imago, and the facts just recalled as
to remnants of pupal activity suggest that in the ancestors of