IMPROVEMENT OF WHEAT BY SELECTION 85 
fat parents, and some from parents only accidentally 
fat. This is the general fate of mass selection. Some 
improvement is. made, but it is an improvement of 
the average not of the whole, and progress towards 
improvement is slow and tedious. Now, suppose the 
sheep breeder recognized the difficulties of his task, and 
instead of mixing up all the selected breeding sheep and 
looking at the average of the resultant lambs, that he 
mated the sheep separately and kept their offspring 
separately under observation while giving them the same 
feed. He would still get fat lambs and lean ones, but 
he would be able to pick the parents that had produced 
the fattest lambs, and know that these parents were 
congenitally fat because they had transmitted the 
character. Thus he would be able to pick out an early fat- 
tening strain in a single generation, and would have only 
to multiply that strain. The whole of the resulting flock 
would then be early fatteners. This is the homologue 
of ‘‘single ear selection.’’ The problem above supposed 
confronts the animal breeder only when he is buying 
breeding stock and does not know how the animals have 
been fed, never wher he is selecting from his own stock 
for breeding. But the problem confronts the plant 
breeder in every first selection he makes. He can never 
know if the apparent excellence of a head of wheat is an 
inherent character or due only to an extra favourable 
position in the soil, relative to manure, moisture, space, 
ete. If all the heads are threshed together the resultant 
offspring is a mixture of good and bad as the parents 
were, and although repeated re-selection is bound to 
produce improvement, the process is tedious and slow. If, 
however, each head is threshed separately, and its seed 
sown separately under similar conditions under strict 
control, then the head whose progeny yields well will be 
a congenitally good head, and that whose progeny is 
indifferent will be an indifferent head, and look good only 
