Knowledge requires both observation and interpretation, neither being more scientific than the other. And each branch of science must be worked out in its own way, which is mainly to be found in the actual search for truth rather than by _a priori_ methodology. Sociology has as ample a field of verifiable fact as any subject, and it is not clear that the interpretations are more unsettled than they are elsewhere. The chief reason why it has developed late and still appears uninviting to many is the very abundance and apparent confusion of the material, which seems to take away the hope of simple, sure, and lasting results. One purpose in our study of principles is to restore this hope and give order to this abundance. And while there are certainly special difficulties, as in all sciences, our own is coming to afford, I think, as great intellectual attraction as can be found in other studies, along with a human and social character peculiar to itself. It will be strange if an increasing proportion of good minds do not give themselves to it. While I ascribe the utmost importance to precision in preparing the data for social science, I do not think its true aim is to bring society within the sphere of arithmetic. Exact prediction and mechanical control for the social world I believe to be a false ideal inconsiderately borrowed from the provinces of physical science. There is no real reason to think that this sort of prediction or control will ever be possible. Much has been made of the fact that human phenomena, when studied statistically on a large scale, often show a marked numerical uniformity from year to year; and some have even inferred that human spontaneity is an illusion, and that we are really controlled by mathematical laws as precise as those which guide the course of the planets. But I take it that such uniformities as are to be observed in births, marriages, suicides, and many other human phenomena do not indicate underlying principles analogous to the laws of gravitation or chemical reaction. They merely show that under a given social condition the number of persons who will choose to perform certain definite acts within the year may remain almost the same, or may be increased or diminished by certain definite changes, such as the advent of war or economic hardship. They no more prove that human conduct is subject to numerical law than does the fact that I eat three meals a day, or that I shall spend more money if my salary is raised, and less if it is diminished.