長江三角洲是清代經濟活動的精華所在地，蘇州府則是當時位於其中的全國最大米糧市場，其糧價資料收集在十八世紀時也較為完整可靠，我們選取這個最具有代表性的地方來作初步的糧價統計分析，分別由趨勢性、季節性、週期性和不規則性等四個因素來模式化（modeling）當時糧價的變動，其中有線性迴歸分析、無重複二因子變異數分析法（ANOVA for two-way classification without replications）、週期頻譜法（periodogram）和Box and Jenkins所發展的ARMA（Auto Regressive and Moving Average）模型等常用的統計分析方法， 引用來解決相關的模式化和模型參數估計等問題。除此以外，研究歷史文獻，資料不完整往往是難以避免的問題，我們運用了較為穩健而簡單的方法來填補資料遺漏值部份，做為初步的資料分析和估計模型參數; 進一步可以用模型預測值來填補原資料遺漏值，並且利用疊代法繼續進行分析和估計，以評估模型的適用性。我們所分析的結果，除了和Wang（1990）中較為直觀性的觀察現象大體上一致之外，從週期性分析方面，我們發現了顯著約二十六年左右之長週期存在，很值得做進一步的研究。

　　　　　In an agrarian society like the Qing Dynasty of China, grains were the most important commodities in domestic trade in which food consumption made up more than half of the average household budget. Grain prices were therefore the leading indicator in the market. A clear knowledge of their trends and behavior patterns will provide a key to understanding the state of economy and society during that time. The Yangzi delta was the center of economic activity in late imperial China. We conduct a time series analysis of the prices of rice for Suzhou Prefecture which is the principal grain market in the delta. The decomposition model is chosen for studying the seasonal variations, the trend movements, the cyclical fluctuations and the irregularity in the grain prices. Specifically, the statistical methods, including regression analysis, Analysis OfVAriance(ANOVA) for two-way classification without replications, periodogram and AutoRegrssive Moving Average(ARMA), are all used. The pattern or systematic changes may reflect conditions of inflation, deflation, or crises of major proportion. Preliminarily, nonparametric regression is used for imputing the missing part in the original data for subsequent analysis. Once the modeling is done, an iterative procedure can be applied for smoothing the missing data and the results of consequent analysis may be used for evaluating the model fitness. The conduction results are generally consistent with the conclusions of Wang (1990) and indicate that a distinct 26-year cycle shown in this data set is invaluable for further study.