中文摘要：

文中对不同学者利用不同来源中国历史文献资料重建的温度变化代用序列进行了比对,分析了同一地区不同序列之间及不同地区间序列的异同,以及造成这些异同的主要原因。结果表明：（1）不同学者重建的同一地区温度（或冷暖）变化序列具有较高的相似性;不同学者所估计的30 a平均温度相对变幅完全一致（即在99.9%置信水平下,二者通过无显著性差别的检验）的时段平均占所有时段的73.4%,还有8.6%的时段虽在幅度大小上有差异但冷暖定性一致,二者共计占82%。（2）不同地区间的温度变化序列也具有较高的相关性,且重建区域相距越近其相关系数也越高,序列的相似程度也越大;而不同学者所选择的重建方法与代用指标不同并不影响不同地区之间序列的相似程度,说明不同地区序列的不一致应是由于地区的气候变化差异造成的。（3）虽然不同学者所估计的中国过去千年以上的温度变化趋势及波动幅度等存在一定差异,但各家所揭示的中国东部过去2000年主要冷暖阶段的出现时间则基本一致。这些结果进一步说明：以前学术界对各家序列差异的认识可能并没有或很少考虑地区间气候变化的差异,因而夸大了不同学者因重建方法与原始证据不同而造成的重建结果差异,这是不客观的。

英文摘要：

China has abundant continuous historical documents with a large number of records describing the weather and climate information. Much effort has been made in the last few decades to reconstruct the historical climate over China using a variety of historical documents, such as histories, local gazettes, archives of government, the private diaries, and so on. Consequently, a lot of differences exist in the reconstructed time series when researchers use either the data from same sources or records from different documents. To address the issue, we have conducted a detailed analysis of 21 time series reconstructed using different historical documents by different researchers recently, and differences among these time series and their causes are also discussed. It is found that the temperature series in the same region by different researchers have higher similarity; and time periods with a completely consistent relative change range of temperature for 30-year mean temperature magnitudes estimated by different series from different researchers account for 73.4% of the total （P〈0.01, Student＇s t test）, and time periods with a consistent cold/warm grade for 8.6 % of the total, leading to 82 % consistent periods in total. The temperature series in different regions are also closer associated, and the nearer the reconstructed regions, the closer the association of temperature series, and the higher the similarities. Differences in reconstructed approaches and in proxy indices used by different researchers do not have any impacts on the similarities of different series for different reconstructed regions, suggesting that differences of time series reconstructed in different regions are resulted from the regional differences of climate change. These comparative analyses also indicate that the 30-year temporal resolution is the most creditable period for using the historical documents in temperature change studies in China. Moreover, although there are still some differences in trends and amplitudes of the temperatu