本文应用数理统计的理论、方法对延寿跨断层短水准1988 年至1994 年计46 期观测成果进行了数理统计处理t 分布检验，对其检验结果结合测线的布设、场地情况，干扰情况以及施测条件等因素加以综合分析，认为观测成果中系统误差是客观存在的。据此采取相应的改进措施，利用改进措施后的观测成果进一步做t 分布检验系统误差得以减小。 In this paper, mathematical statistics theory, method of extension of short-span cross-fault short-term 1988 to 1994, 46 observations were t-test mathematical distribution of statistical processing, the test results combined with the layout of the line, the site conditions, interference and Shi Measuring conditions and other factors to conduct a comprehensive analysis, that the observations of the systematic error is an objective reality. Based on this, corresponding improvement measures are taken and the error of t-distribution is further reduced by using the observation results after the improvement measures.