论文部分内容阅读
采用随机区组设计进行品种区域试验,在多重比较时,对照品种与参试品种的差异显著性可用最小显著差数浊(L.S.D)测验。由L.S.D可推导出CV≤D·(n/2/t_α)~(1/2)。由此可推算出一定概率平准保证下的精确度要求。按重复数3、4、5、6、推算了一定平均差数百分率(D=(B-A)/x×100%)下,5%概率平准容许的最大变异系数,列于表1。对江苏省近几年来小麦、大麦区试的历史资料进行分析,其平均CV约为6%,这样水平的CV仅适于测验D>20%的试验。而这样大小的差异,在大麦和小麦近年来的区域试验中,分别仅占17%和30%以下。结果表明,各试验点现在的CV水平过高,必须予以降低。对改进区域试验精确度的途径进行了讨论。增加重复次数也许是最有效的途径之一。
The randomized block design was used to carry out the regional trials of varieties. In the case of multiple comparisons, the significant difference between the control varieties and the tested varieties can be tested by the L.S.D. CV ≤ D · (n / 2 / t_α) ~ (1/2) can be derived from L.S.D. This can be calculated under a certain probability of accuracy under the assurance requirements. According to the numbers of repeats 3, 4, 5, and 6, the maximum coefficient of variation allowed by 5% probability leveling under the certain percentage of average difference (D = (B-A) / x × 100%) is estimated. The historical CV of wheat and barley in Jiangsu Province in recent years is analyzed. The average CV is about 6%. Such a level of CV is only suitable for the test with D> 20%. And such differences in size, in the recent regional trials of barley and wheat, accounted for only 17% and 30% respectively. The results show that the current CV levels in all test sites are too high and must be reduced. The ways to improve the accuracy of regional experiments are discussed. Increasing the number of repetitions may be one of the most effective way.