During 5 years from 1959 to 1963, many reciprocal pollinations between diploid and autotetraploid rice plants using Japanese and foreign varieties were artificially carried on by hot water emasculation method, and results were as follows (Tables 2 and 3). From these crossings, besides plumply, well developed seeds, many shriveled and ill-developed seeds which contained no endosperm and were very light, 1/10～1/20 of normal seed in weight, were obtained. For convenience' sake, they were called perfect and imperfect seeds respectively. The perfect seeds set in very low percentage, that is about 1%, but they germinated well and grew to maturity. From the cytological observations, it was found that almost all of them were selfed seeds of female plants. As to the imperfect seeds, seed setting ratio was comparatively high, but they germinated in low percentage and if germinated they could not grow to seedlings. For the question whether these imperfect seeds were true hybrid seeds or parthenocarpically developed ones, some knowledges were already given in the first report. When diploid and tetraploid plants were castrated by hot water emasculation method and not pollinated afterwards, imperfect seeds did not set on diploid plants, and on tetraploid plants only 1.1% were set. This ratio was very low compared with the crossed results as shown in Tab. 2 and 3, so that the greater part of imperfect seeds in crossed case were seemed to be true hybrid seeds. Degree of seed set was typically different between reciprocal crossings, and some differences were also found between varieties. In Japanese varieties, when diploid was used as female parent imperfect seeds were less easily obtained, but in reversed crossings they were more easily obtained and more heavy in weight and germinable. On the other hand, in foreign varieties imperfect seeds were more easily obtained in the crossing 2X×4X than in reversed crossings. This tendency was very striking in the varieties O. sativa f. spontanea and Ba Thut Nuho T. T. B., but in the crosses with O. sativa f. spontanea germination was better than that in 4X×2X cross as in Japanese varieties. In foreign varieties, Ba Thut Nuho T. T. B. and Songa T. T. B. belong to the so-called Indica type, and O. sativa f. spontanea belongs to wild species. So that they were genetically different from Japanese varieties. From above facts, it was considered that the tendency of seed setting was not constant within varieties, and that this phenomenon was affected by genetic differences.