我们把纯粹由化学分析中的实验误差组成的变量称为“误差变量”。分类问题中常常把这些“误差变量”去掉以浓缩信息,本文认为:“误差变量”往往对分类问题是有贡献的。我们对化学分析中的实验误差与分类的关系作了比较详细的分析;并用一个例子来说明我们的观点。 We refer to variables that are purely composed of experimental errors in chemical analysis as “error variables.” Classification problems often remove these “error variables” to condense information. This paper argues that “error variables” often contribute to the classification problem. We analyze the relationship between experimental error and classification in chemical analysis in more detail and use an example to illustrate our point of view.