The Computation and Adjustment of Probabilities derived from Observation

Abstract

All our empirical knowledge depends not only upon the exactness, trustworthiness, and frequency of the observations on which it is founded, but quite as much upon the possibility of drawing general conclusions from the observations. The great variety of form in which the results of observation appear, even independently of the diversity of the nature and the method of observation, will easily explain why the range of the conclusions to be drawn from the observations, differs widely according to the field of the science to which they refer. While the natural philosopher must be satisfied with the mere knowledge of a fact in a field which has not previously been investigated, he goes in other realms of science far beyond this first step of observation; he obtains an explanation about the causes of the phenomenons and the connection of the forces and the laws which govern the events that come under his observation. Such explanations cannot in general be drawn directly; they require a number of intermediate steps in which the observed result has to be critically examined and analyzed into its single elements. The proceeding to be adopted for this purpose depends upon the nature and form of the observation, and it would be a mistake to suppose that we can follow the same method for all kinds of observation.