The aim of this article is to give some explanations about the factors that support the accuracy of the statistical data in the late Meiji period Japan. I find some facts on the training courses for statistical enumerators held by the prefectural governments. Using journal articles written by contemporary statisticians as well as some prefectural documents of the Fukushima area, I find the following facts :
1) Students : During 1900-1911 there are more than 16,000 students all over Japan, most of whom are enrolled in the courses ex officio and, thereafter, become instructors at similar courses held by local governments of the lower level, namely canton(gun), town(machi)or village(mura) governments.
2) Instructors : Some 20-30 top class statisticians who worked at the Statistical Bureau(Naikaku Toukei Kyoku), the Imperial Universities or some ministries voluntarily worked as instructors.
3) Textbook : Though the courses are not authorized by the central government but held by local governments independently of each other, many of those courses used the same textbook and had quite similar curriculum. Masao Yokoyamas "An Introductory Guide to Statistics(Toukei Tsuron)" is so to say the defacto standard textbook in Japan(moreover, even in China too), which we can read today. It implies that we can know about the theoretical backbone and its application to statistical surveys in that period.
