The present authors have long engaged in crossing experiments between diploid and autotetraploid plants, in order to obtain triploid plants in rice. As already reported, the results were not successful although enormous flowers were pollinated. From the above crossing many F1 seeds were obtained, but a few of them were perfect seeds and the others were imperfect seeds. The perfect seeds were germinable but almost all of them were selfed seeds of female plants. On the other hand, the imperfect seeds which developed very retardedly, although confirmed to be true hybrid seeds, were less germinable and could not grow to seedlings. In this connection, artificial culture method was conducted and results were as follows. For culture, modified White's medium was used, and seeds were sown in aseptic condition. The imperfect seeds that obtained from 4X×2X germinated at the rate of 54.4% on the artificial medium. This percentage was strikingly high compared with the 16.4％ which were shown in germination test using water. And 11.2% of the seeds sown grew to adult plants. All of them were ascertained to be triploid plants. On the other hand, the imperfect seeds obtained from the crosses 2X×4X germinated in very low percentage of 0.2%, and effects were not expected. The perfect seeds that obtained from the crosses of 2X×4X and 4X×2X were also sown on artificial media. Out of 55 seeds sown, 53 germinated and 44 grew to adult plants. And 6 triploid plants (10.9％) were obtained. These results were relatively effective comparing with the data resulted from sowing on soil nursery.