1. In 1934, one of the authors found a semi-sterile rice plant of 46.9% sterility. The next generation individuals of that plant showed, without exception, a strong tendency of making polycaryopses, the average percentage of sterility being reduced to 12.7%. The character producing polycaryopses bred true for the later generations.
2. The hybrid between the mutant above mentioned and the normal varieties produced normal caryopses. The segregation ratios of the two parental characters actually obtained in the second generation closely approximated in some cases to 3 to 1, but in others they deviated fairly remarkably from that ratio owing probably to the weakness of the mutant type.
3. When two caryopses are found in a spikelet, one is situated on the side of lemma, while the other is on the side of palea. In unhulled condition, the caryopsis on the lemma side germinates normally, but the one on the side of palea does not, or take very long time for germination.
4. The portion of the lemma against the embryo of the normal caryopsis has a special nature and construction to be split easily by the swelling pressure of the embryo. No such a structure is found at the base of the palea. Germination of the palea side caryopsis is possible only through that germination part of the lemma or through between the lemma and palea.