f samples of different sorts and stocks on
your own grounds. Having formed such a conception, write out the
clearest possible description of exactly what you want and the ideal
plant you are aiming at, stating as fully and minutely as possible every
desirable quality and also those to be avoided. I consider not only the
formation of an exact ideal, but the writing out of a most minute and
exact description of precisely what in every particular the ideal plant
should be and the rigid adherence to that exact ideal in selection, as
the most important elements of successful seed breeding. Without it one
is certain to vary from year to year in the type selected and in just so
far as he does this, even if it be toward what might be called
improvements or in regard to an unimportant quality, he undermines all
his work and makes it impossible to establish a strain which can be
relied upon to produce an exact type.

With this description in hand, search out one or more plants which seem
the nearest to the ideal. In doing this it should be kept in mind that
the character of the seed is determined by the plant rather than by the
individual fruit. Therefore, a plant whose fruit is most uniformly of
the desired type should be chosen over one having a small proportion of
its fruits of very perfect type, the others being different and
variable. Save seed from one or more fruits from each of the selected
plants, keeping that from each fruit, or at least each plant, separate.
Give it a number and make a record of how nearly, in each particular,
the plant and fruit of each number come to the desired ideal. I regard
the saving of each lot separately and recording its characters as very
important, even when all have been selected to and come equally close to
precisely the same ideal. Quite often the seed of one plant will produce
plants precisely like it, while that of another, equal or superior, will
produce plants of which no two are alike and none like that which
produced the seed, so that often the mixing of seed from different
plants of the same general type, and seemingly of equal quality,
prevents the establishment of a uniform type.

The next year from 10 to 100 plants raised from each lot are set in
blocks and labeled. As they develop the blocks are studied and compared
with the original description of the desired type and that of each plant
from which seed was saved, and the block selected in which all the
plants come the neares