多元分析的误差传递需要一种简单、准确、数值化的表达方法.向量空间中,线性多元混合信号的随机误差可表述成真值子空间中随机向量的表现;由体系多元变量对应的向量构成的真值子空间中,被关注向量和其他向量子空间的空间角θ是描述多元体系的重要参数.如果被关注向量和其他向量子空间关系确定,体系总体误差呈正态分布,那么,被关注向量上误差也是正态分布,其多元统计分析结果的标准差与体系误差标准差的比值为1/(2·sin(θ/2),结论在构造算例和邻、间、对苯二酚混合体系的紫外光度分析中得到验证. The error propagation in multivariate analysis needs a simple, accurate and numerical expression method. In vector space, the random errors of linear multivariate mixed signals can be expressed as the representation of random vectors in true value subspaces. The vector corresponding to multivariate variables , The spatial angle θ of the attention vector and other vector subspaces is an important parameter for describing the multivariate system.If the relationship between the attention vector and other vector subspaces is determined and the overall system error is normally distributed, The error of the attention vector is also a normal distribution, and the ratio of the standard deviation of the multivariate statistical analysis to the standard error of the systematic error is 1 / (2 · sin (θ / 2)). Conclusions In constructing the example and the ratio of o, Phenol mixed system of UV spectrophotometry was verified.