Abstract: Results of investigation on the colchicine-induced autotetraploids in rye, barley and rice as well as the wheat-rye amphidiploid made during 1951–1954 are summarised as follows: 1. Tetraploid rye. (1) In comparison with diploid rye, the autotetraploid has stouter vegetative growth, larger stomata, longer spikes, larger kernels, fewer tillers, shorter stems and lower fertilities. (2) In regard to physiological characteristics, the autotetraploid is more resistant to drought than the diploid as tested by the method of permanent wilting. On determining the monosaccharide and total sugar by Hassid's titraiton method, he autotetraploid shows more accumulation of monosaccharide and less total sugar as compared with the diploid. This, probably, means, that the autotetraploid is less efficient in monosaccharide transformation. (3) The autotetraploid is low in fertility (about 50% seed setting). But there seems no apparent connection between partial sterility and the tetravalents formation in the first meiotic devision. (4) Plants grown from self-pollinated seeds of autotetraploid rye decrease significantly in fertility and test weight of kernels. However, mixed sowing of the self lines improves the fertility and kernel weight. (5) The diploid rye has 7 pairs of chromosomes, and the autotetraploid has 14 pairs. Identification made during diakinesis of the first meiotic division shows that the tetraploid has an average number of 2.2 tetravalents per cell. (6) The partial sterility and poor seed plumpness of the autotetraploid rye are two major problems yet to be solved. 2. Wheat-rye amphidiploid. (1) Chinese spring, a variety of Triticum vulgare, morphologically related closely to the local variety Chengtu-beardless, was used as the female parent in the intergeneric cross with Pingwu rye (Secale cereale). This crossing was as successful as that of varieties of common wheat, and therefore hybrid seeds were easily obtained. The crossability of Chinese spring with rye was found to be an inheritable character since another variety of the common wheat, Ardito, was incompatible in crossing with the rye. When F1 of Chinese spring (♀) × Ardito (♂) was used as the female parent in crossing with rye, the percentage of seed-setting reached as high as 61.68% against 17.07% in the ease of crossing of F1 of Ardito (♀) × Chinese spring ( ♂ ) with rye. (2) The fertility of wheat-rye F1 hybrid is extremely low. The average seed setting is 0.044 kernel per spike. For the vegetatively reproduced F1 plants by dividing the tillers before shooting, it is 0.39 kernel per spike. However, the fertility of wheat-rye amphidiploid as produced by the colchicine treatment of young F1 plants, is comparatively much higher. The average seed-setting reaches 20 kernels per spike. For individual spikes, the fertility of amphidiploid varies from zero to more than 60 kernels per spike. Such variation of fertility exists among the spikes of a single plant. (3) Application of phosphorus and potassium during the heading time shows some improvement in fertility and seed plumpness of the wheat-rye amphidiploid. (4) The most interesting character of the amphidiploid is the larger spike. The average number of spikelets per spike is 20 and the largest spike has 32 spikelets. (5)Apparantly, no segregation could be observed among the progenies of amphidiptoid. The vegetative growth and the form of spikes showed great Uniformity. When the amphidiploid was crossed with a common wheat variety, Quality, the F2 showed great variation in seedling growth habit, form of spikes, fertility, and seed plumpness. In F2, there were a lot of individual plants with normal fertility and good seeds. Morphologically, these plants belong to the common wheat types. 3. Tetraploid barleys. (1) We have four varieties of autotetraploid barley induced by colchicines technique, namely: Chengtu Shyh-dah-duei, naked barley 37-2, hybrid lines 37B- 45A and 37B-54C. The diploid barley has 7 pairs of chromosomes and the autotetraploid 14 pairs. (2) In comparison with the diploid, the autotetraploid barley has stouter vegetative growth, shorter stems, larger spikes and kernels, and less tillers. (3) The F1 plants of the intravarietal crosses among the individual plants of autotetraploid Shyh-dah-duei, and of the intervarietal crosses between autotetraploid Shyh-dah-duei and autotetraploid naked barley 37-2, showed no improvement in fertility and kernel weight. However, the fertility of many F2 plants rose to more than 90%, and the test weight of 1000 kernels reached 50 g or more. There are more plants with normal fertility and good kernels in F2 of intervarietal crosses than those in intravarietal ones. Hence, the intervarietal crosses of autotetraploids are more favourable in fertility and seed plumpness. (4) The kernels of F1 plants of the cross between autotetraploids Shy-dah-duei and 37-2 were all with husks. In F2 only one plant was observed with naked seeds. The fertility of this single plant was 90.5%, and the test weight 48.87g. This test weight is 100% over the diploid 37-2 and 18.0% more than the highest test weight of 37-2 autotetraploid plants. (5) The fertility of the first and second generations of autotetraploid Shyh-dah-duei was about 50%, and the test weight of the kernels was about 50% heavier than that of diploid. Of the third generation, the fertility of many individual plants reached higher than 90%, and the test weight of the kernels increased to 100% more than that of diploid. (6) The highest test weight of the kernels is the autotetraploid 37B-45A, which has 67.18g per 1,000 kernels. After selection, the test weight could reach as high as 81.6g. But there was no one F1 plant of the crosses between autotetraploids Shyh-dah-duei and 37B-45A with a test weight of kernels over 60g. Four out of 9 selected F1 plants have a test weight of over 50g, and the remaining five have over 40g. (7) It could be then concluded that the problems of fertility and seed plumpness in tetraploid barleys have been solved adequately. 4. Tetraploid flees. (1) We have three varieties of autotetraploid rice, namely: Indica type 422, Japonica type Ningfong, and Shoei-yuan. (2) The autotetraploids have shorter stem, less spikelets per spike, larger kernels, and a much lower fertility as eompared with the diploids. The fertility of autotetraploid Japonica types is even lower than that of the autotetraploid Indica type, the former being below 20%, and the latter around 50%. It is, however, very surprising that after colchicine treatment, a spike of Ningfong had both diploid and autotetraploid spikelets and the fertility of the tetraploid spikelets reached as high as 93.94%, practically the same as the diploids. (3) The diploids of the three rice varieties are all awnless, while all of their autotetraploids are with short awn. (4) It is generally known that the F1 plants of the Indica-Japonica crosses are highly sterile. Four out of five combinations of such crosses made by L. F. Chao have fertilities below 6%, and the highest one only reaches 24.73%. On the contrary, the F1 plants of the autotetraploid Indica-Japonica crosses are quite fertile. The lowest fertility of our 7 F1 plants reaches 17.95%, and the highest, 61.72%. Individual spike even reaches a fertility as high as 74.58% which is practically normal as compared with the diploid ones heading at the same time. (5) Four out of 7 F1 plants have kernels with a test weight of over 40g, which is 20% heavier than that of the autotetraploid 422 parent, and 60%, than that of diploid 422. Five out of 7 F1 plants have long awns, and the remaining two, short awns. (6) The environmental conditions seem to have profound influence upon the fertility and seed plumpness of the rice autotetraploids. In 1954 some of autotetraploid 422 seeds were sown in greenhouse on January 5. The seedlings headed in the last days of April. Surprisingly enough, these plants had a fertility of as high as 76.68%, and a kernel test weight of 50g. 5. The problems of fertility and kernel plumpness of the polyploids. It is generally observed that the artificially produced polyploids are usually low in fertility and poor in kernel plumpness. From the analysis of the known facts, it is revealed that these general defects of the artificially produced polyploids are most probably due to the internal hindrance of transportation of nutrients or metabolites from the vegetative parts to the reproductive ones. Therefore, any practical technique which has any beneficial effect, on the transportation of metabolites, such as application of phosphorus and potassium at the heading period, and the spray of minor element boron, would be favourable to the fertility and kernel plumpness of the polyploids. It is possible that by training in this way for some generations, the polyploids might be getting better successively and finally have a normal fertility and plump kernels. On the artificial induction of autopolyploids or amphidiploids by polyploidogenic agents such as colchicine, two daughter nuclei from a mitotic dividing mother cell are going to reunite. It is evident that the reunited nuclei are nearly identical in property. And consequently, the artificially induced polyploid cell is apparently highly homozygous. The highly homozygous individuals as the inbred lines are frequently showing retrogressive defects as semisterility and poor kernels. Crosses between the varieties of autopolyploids or amphidiploids will restore the heterozygous state of the plants. This might be the reason why crosses between the varieties of polyploids are usually very effective for improving the fertility and kernel plumpness.