1990 年 28 巻 2 号 p. 206-221
This paper focuses on the re-examination of the Raffles's population statistics of Java in early nineteenth century. Population statistics of these early periods have been either cited uncritically or simply abandoned as unreliable. The defects of these old statistics should be discussed explicitly before they are treated in either way. The figures compiled by Raffles in his History of Java were based on the reports by the local colonial officers and/or native rulers in Java. Underestimation in the statistics is due to a. oversight of certain regions, b. underestimation of household numbers, c. underestimation of household members. A considerable part of the underestimation derives from that of household members. In this paper, a detailed examination was done of the following aspects: a. errors in computation, b. distribution of household size by sub-district, c. distribution of sex-ratio by subdistrict, d. variation of age-structure by subdistrict, e. problems of Chinese population. Special attention was given to the consistency of the definition of categories and the distribution patterns of the reported figures. Various aspects of underestimation may reflect the opposing interests of colonial officers vs. native rulers, and rulers in general vs. the ruled. The diverse distribution of the statistical index figures seems to be a result of a balancing of these elements in diverse situations.