Plant Breeding. 251 
with which we started (341). Our varieties of fruits, pota- 
toes, geraniums and many flowering plants, and of many of 
our finest ornamental trees and shrubs are fixed in this way. 
It is well known that varieties propagated in this way rarely 
“come true” from seed, i.e., their seed does not usually 
produce plants of the same variety us the parent. But it is 
not practicable to propagate all plants by division. With 
plants more conveniently propagated from seed, as the 
cereals, Indian corn and most garden vegetables, we may 
fix varieties to a certain extent, 
(2) By persistent selection toward an ideal type. For ex- 
ample, if we discover a single pea plant, in a row of peas, 
that produces earlier pods than any other plant, and we 
desire to fix this variation, we would save all the peas from 
this plant and sow them the next spring. Most of the plants 
from this seed will probably be later than the parent, but 
two or three of them may very likely equal it in earliness. 
We would save the seeds from the very earliest plant again, 
and continue this system of selection through several sea- 
sons. It would be well to note the incidental characters of 
the earliest plants, i. e., whether the pods were borne singly 
or in pairs, and if they were straight or crooked, and whether 
the plants were tall or dwarf. Having decided on the char- 
acters that seem to accompany the extreme earliness, we 
should save seeds from no plants that do not show all these 
characters. After following this kind of selection eight or 
ten years, we may be able to introduce a new variety of pea. 
It is impossible to so fix variations in plants grown from 
seed that they will continue to come true without a certain 
amount of selection, hence varieties propagated by seed 
continually tend to “run out,” i. e., to lose their distinctive 
characters. Seed growers find it necessary to use the utmost 
