Even today, it is a fundamental problem in statistics to form our own theory of the scientific nature of statistics. Also in order to overcome the criticism which asserts that the content of statistical analysis in agricultural economics and its related fields is getting poorer and poorer, and in order to develop our statistical analysis to a higher level, we are still obliged to solve this fundamental problem. According to previous studies, the theories of the scientific nature of statistics can be classified into five large groups, excluding mixtures between them. The first is called the "methodology of universal science" theory, the second the "methodology of subjectively rational decision making" theory, the third the "methodology of social science" theory, fourth the "independent social science" theory, and the fifth the "historical description of statistical activities" theory. These five theories have existed separately without being fully related to each other, and the word "statistics" has various meanings corresponding to these theories and the mixtures between them. Therefore we have to form our own theory of the scientific nature of statistics for ourselves. To put it another way, forming our own theory of the scientific nature of statistics is forming our own theory of the relation between statistics and other field of science. Therefore to form our own theory of the scientific nature of statistics, we must find our own answer to such questions as what science is, by what criterion science is divided into various fields, and what scientific method is. From this standpoint, mainly following the studies by Braithwaite (1964), Taketani (1968a, b) and Tosaka (1966b), I form my own theory of the scientific nature of statistics, which asserts that statistics is the activity of studying some part of the method of science. When we discuss the scientific nature of statistics, the concept of probability plays a fundamental and important role. But there is no single interpretation of the concept of probability, and now, broadly speaking, the concept of logical probability, the concept of subjective probability, and the concept of objective probability have been distinguished, apart from mathematical and formal concept of probability measure. Therefore we also have to form our own theory of the concept of probability. In this respect, from my theory of the scientific nature of statistics, I assert that the concept of objective probability is best suited to statistics. On the basis of the above considerations, I come to the following conclusion. The first and the third of the five theories mentioned above are concrete and important cases in my theory of the scientific nature of statistics, the fifth is a supplementary yet indispensable part of my theory, but the second and the fourth are incompatible with my theory. Therefore the content of statistics based on the second or the fourth theory should be classified as part of other fields of science.