抄録
1. From 120 rice varieties of various districts of Eastern Asia (same to those used in the preceding work), seven varieties were chosen as analysing varieties, and were crossed with each of the 120 to examine which of the analysing varieties would produce fertile hybrids with it. 2. The percentage of good pollen, by which the degree of sterility was measured, was different with the combination of parental varieties and ranged from nearly 100% to 0%. It was not affected so greatly by outer conditions. 3. No difference was found between reciprocal cfosses . 4. Among three varieties A. B and C there were often found such cases where A x B and B x C were fertile and A xC was semi-sterile. 5. By denoting a cross-combination higher or lower than 87. 5 % in good pollen percentage by a plus or minus sign, the reaction type of each variety was represented by a combination of seven signs. Arnong 128 (27) kinds of possible reaction types, 31 kinds were fcund actualiy. Having arranged them according to the number' and position of plus sings, they were dividable into two groups, and in each again into two minor groups (Fig.1). 6. Group I was nearly equivalent to "Continental"l variety group, which had been established inthe preceding work on the ground of a definite tendency in character combination. Groups II a and 11 b seemed to bc comparable to "Tropicai Insular" and "Temperate Insular" variety groups, respectively . 7. Formosan wild rice (Oryza sativa sponetanea var. formosana) was largest in the number of fertile analysing varieties. It was pointed out, after some considerations on the genetical mechanism of the intervarietai hybrid sterility, (refer to OKA 1953) that the above-mentioned classification of reactiol types of hybrid sterility would serve as, a model of the pattern of differentiation of rice varieties .
