worth of a plant by the average condition of its offspring, instead of
by its own visible characters. If this determination of the "centgener
power," as Hays calls it, should prove to be the true principle of
selection, then indeed the analogy between natural and artificial
selection would lose a large part of its importance. We will reserve
this question for the last lecture, as it pertains more to the future,
than to our present stock of knowledge.

Something should be said here concerning hybrids and hybridism. This
problem has of late reached such large proportions that it cannot be
dealt with adequately in a short survey of the phenomena of heredity in
general. It requires a separate treatment. For this reason I shall limit
myself to a single phase of the problem, which seems to be indispensable
for a true and at the same time easy distinction between elementary
species and retrograde varieties. According to accepted terminology,
some crosses are to be considered as unsymmetrical, while others are
symmetrical. The first are one-sided, [21] some peculiarity being found
in one of the parents and lacking in the other. The second are balanced,
as all the characters are present in both parents, but are found in a
different condition. Active in one of them, they are concealed or
inactive in the other. Hence pairs of contrasting units result, while in
unbalanced crosses no pairing of the particular character under
consideration is possible. This leads to the principal difference
between species and varieties, and to an experimental method of deciding
between them in difficult and doubtful cases.

Having thus indicated the general outlines of the subjects I shall deal
with, something now may be said as to methods of investigation.

There are two points in which scientific investigation differs from