水稻肥料试验误差平均数经验分布及其在推荐施肥中的应用

THE EMPIRICAL ERROR MEAN DISTRIBUTION OF THE FERTILIZER EXPERIMENT AND ITS APPLICATION IN RICE FERTILIZER RECOMMENDATION

从水稻肥料试验误差（离均差d）数据库中随机抽样得到的误差平均数呈以0为原点，方差为的对称高狭峰分布，且随样本容量n的增加，逐渐接近于正态分市，当n＞8时，已与正态分布无显著差异。该分布在水稻推荐施肥中可用于实际产量平均数x与目标产量u_0的差异显著性检验，推断实际产量是否达到或超过目标产量，进而验证目标产量预测结果的准确性，并对目标产量加以修正。

The yield deviation means(d^-) presented symmetry distributions(d=0)with a high and narrow peak,but gradually tended toward normal distribution with increasing of simpling units(a).if n>8,the d distributions were not significantly different to the normal distribution.The d distributions could be applied to hypothesis test for yield mean(x~_)in fertilization recommendation of rice,by which could know if the real yield(x~_)achieve or s urpass the target yield(μ_0) so as to prove the precision of target yield (μ_0) and finally make the target yield (μ_0) preciser.