THE LAW OF ANCESTRAL HEREDITY 169 
upon the offspring, and this chance grows rapidly less as we go 
backward, never, however, becoming zero ; so that it is possible 
that resemblances to any ancestor, no matter how far removed, 
may crop out in individual cases from time to time, giving 
strange but not unaccountable cases of reversion. These are ex- 
tremely noticeable, first, from their variety; and second, from the 
fact that complete ignorance generally surrounds all ancestry 
more than a generation or two back. What chance is there, for 
example, for knowing much about the separate characters of 
each of the thirty individuals involved in the first four genera- 
tions only ? The next generation backward would add thirty-two 
more, showing how rapidly the transmission becomes compli- 
cated, particularly when we remember that all the ancestry has 
contributed to the individual. 
The individual a composite. This makes it look as if the 
individual were pretty well distributed among his ancestry from 
his parents backward, and that is exactly the condition of matters. 
The individual is a kind of mosaic, taking a portion (on the 
average one half) of his resemblances from his parents, others 
from his grandparents, and still others from earlier ancestors, 
even to the remote past. 
At first thought this may seem impossible, but upon careful 
research we find that racial characters are but loosely held to- 
gether,! and it is only upon reflection that we realize the extent 
to which combinations and recombinations take place and how 
resemblances come and go in a long line of ancestry. 
In this way an individual may seem in some particular to re- 
semble, we will say, the paternal grandsire, whereas the actual 
resemblance is not only to him but to perhaps a score or more 
of similar ancestors still further back and long forgotten, but 
whose blood lines combined with and intensifying those of 
1 Shown by the fact that the “ correlation” or bond that compels characters 
to move together is very low, seldom as much as 50 per cent, so that almost 
literally it is a free-for-all contest when matters of hereditary resemblances 
are being determined. For a full discussion of Correlation, see “ Principles of 
Breeding,” chap. xiii. 
