頁籤選單縮合
題 名 | 臺灣水稻田間試驗之適當重複次數=On the Minimum Replication Number of Rice Paddy Experiment in Taiwan |
---|---|
作 者 | 張魯智; | 書刊名 | 中華農學會報 |
卷 期 | 43 民52.09 |
頁 次 | 頁9-26 |
關鍵詞 | 水稻; 田間試驗; |
語 文 | 中文(Chinese) |
英文摘要 | In this paper, the generalized error mean squares of the grain yield were computed and the frequency distribution of the differences between every two experimental varieties was tabulated from the record of 196 rice paddy experiments which were carried out in the recent 5 years in this island. With these results, the author made further calculation and a table of minimum replication numbers was obtained, and discussed the demerits of the replication numbers of rice paddy experiments now used in Taiwan. Following his conclusions that the errors of experiments at every place in Taiwan are somewhat different, but if has no uniform tendency. However, it is evident that there is a highly positive correlation between the error mean squares and the number of plots containing in every block. When 36, 18, 12 and 6 plots are contained in one block, the error mean squares are 11.94, 10.72, 9.44 and 8.49 kg/ha repectively. In table 4, the minimum replication numbers under the different percentage differences detected from 1 to 30% between the grain yield of two different experimental varieties or treatments is resulted from the formula of test of significance in which the error mean squares are limited only in the probability of Type 1 Errors are 5% & 1%. Table 5 is the table of the least percentage differences with the sighificant conclusions when the replication numbers are limited from 1 to 20. In table 6 & 7, the error mean squares are limited in the probability of Type I Errors are 5% & 1% and Type II Error is 20%. In table 8, the frequency distribution of 11,309 percentage differences grain yield of every two experimental varieties which are the data of the 99 regional variety experiments of the promising Ponlai rice in recent 5 years at 16 places around the island. In the first class of that table there are 7768 frequencies with the percentage differences from 0.0% to 9.9% which are 68.6% to the total (11,309). There are 9756 frequencies from 0.0% to 14.9% which are 86.27% to total. There are only five replications have been generally setted in this kind of experiments in Taiwan. It is based on the least accuracy, that is, the probability of Type II Error is unlimited and the probability of Type I Error is 5%, the significant conclusion can only be gotten when the difference between every two varieites is greater than 13.02%. All these kinds of experiments in Taiwan, there are only 19.55% of total number of differences between every two varieties is greater than 13.02%. This percentage is too small to be obtained. According to this fact, the author advises that there would have to set eight replications instead of 5 replications. Thus the significant conclusion can be gotten under the percentage of the significant conclusion of 5% and less than 5%. |
本系統中英文摘要資訊取自各篇刊載內容。