在实验研究工作中,通常的情况是,被测量的真值和测量仪器的标准误差都不知道,且是独立的等精度的多次测量。本文指出:就测试结果的报导而言,我们不能忽视t分布的重要性;在测量次数有十几次的情况下,对68.3％的置信概率,可以用正态分布报导测试结果;但在测量次数只有几次的情况 In the experimental work, the usual situation is that the measured true value and the standard error of the measuring instrument are unknown, and it is independent and equal accuracy multiple measurements. The paper points out that we can not ignore the importance of t distribution in terms of the test results. In the case of more than ten times of measurement, the test results can be reported in normal distribution for the confidence probability of 68.3%. However, Only a few times the situation