Abstract
The sizses of F2 and F3 populations are investigated based on the two criteria, i. e., the probability that promising genotypes are selected in the F2 and F3 Selection, and the economic advantage which is expected to be gained by obtaining the promising genotypes. In formulating these criteria, a selection system is assumed where within and among cross selection is made in F2 regarding qualitative and some highly heritable continuous charaters, selection for these and other less heritable characters being operated in F3. The optimurn. sizes of the populations are differently estimated according to the quantity of breeding cost and the number of cross combinations available for the breeding programme at issue, but it may be safely adopted as a standard in a routine selection work that several hundred and a few to several thousand plants are grown for the F2 and. F3 selection respectively. The genetic potentiality of individual crosses may not be fully tested in this scale of selection, but the economic advantage in the long run will be enlarged. Only one out of hundreds of crosses can lead to a new variety in highly improved crops such as rice and wheat. With such a poor identification of desirable crosses, spending a large cost for a limited number of crosses may cause a serious waste of breeding cost unless the crosses are pre-confirmed to be more promising than others.