DESCENT WITH MODIFICATION 165 
at almost anything that may happen in individual cases, from 
which we infer that we shall never be able to predict from the 
parents what a particular offspring will be, but that we can tell 
very close as to what they will be in the long run and on the 
average ; and that the more uniform the ancestry, the more 
accurate will the prediction be, and the more uniformity will 
there be among individuals. We see, too, by the principle of 
progression, that under selection the correspondence between 
parent and offspring becomes rapidly closer. 
Summary. The individual offspring is seldom like the parent. It may 
be better (progression) or it may be worse (regression), but in general the 
offspring is like the parentage as a whole. More exceptional offspring arise 
from common parentage than from exceptional parentage because mediocrity 
is the common lot, yet the proportion of exceptional offspring is higher from 
the exceptional parent than from the mediocre. 
Reversion shows that as long as even a trace of old-time characters 
exists, the gametes are not absolutely pure, and an occasional appearance of 
such ancient characters is inevitable. Being ancient, they are correlated 
with others, and are likely to creep upon the breeder unawares, like the high 
shoulders and thin flanks of cattle, or the inferior hams of pigs. 
The offspring, therefore, is to be considered not so much the product of 
his personal parents as of his parentage as a whole. 
