¹ History as a discipline has been excepted for two reasons: (1) It is frequently considered by its practitioners not as a scientific field but as one of the “humanities,” and more importantly (2) in so far as it has sought to be scientific it has either (a) kept itself largely to a descriptive rather than analytical level (see below) or (b) has indulged in non-empirical analysis or empirical analysis classifiable under one or more of the headings above. In this there is no intention to contemn history as a discipline. Its record for objectivity is at least as great and probably greater than the disciplines listed above, and its empirical theories have often been most fruitful and rewarding. In general, however, most of the historical efforts aimed primarily at generalized systems of abstractions have been of the non-empirical variety (e.g. Spengler and Toynbee).

² If the chain is shown unclear or nonexistent, the instrument is revised or discarded.

³ M. Cohen and E. Nagel, An Introduction to Logic and Scientific Method, Harcourt, Brace and Co., New York, 1934, p. 191. See also M. Cohen, Reason and Nature, Harcourt, Brace and Co., New York, 1931, pp. 83–115.

⁴ Parsons, T., The Structure of Social Action, McGraw-Hill Book Co., New York, 1937, p. 16.

⁵ See Poincaré, H., Dernières Pensées, Paris, 1913, pp. 223–226.

⁶ I am aware that there is considerable respectable philosophical opposition to the position taken here about the difference between values and factual statements. Examples of this are to be found in the works of John Dewey (e.g., Logic, The Theory of Inquiry), and the position has been most cogently put to me by Professor W. T. Stace of Princeton University. Professor Stace has argued that he has no objection to the clear distinction between ethics and (say) sociology—the distinction between what ought to be the case and what is the case, but that he objects to the banishment of ethics altogether from the field of what can be established by empirical observation. His position boils down, I believe to the point that the two statements, “x ought to be y” and “x is y,” are both equally factual empirical statements. I am not convinced of this point, but even granting its validity, three points can be made: 1) The two types of statements are different in kind and not degree, and the natural sciences have characteristically dealt only in statements of the second type. 2) It is possible for social science to do the same, and 3) if the social sciences are going to involve both types of statements on this level of consideration, they will involve an activity different in kind from that involved in the natural sciences.

⁷ There is no intention to single out this field especially. There is hardly any field in the literature which would not serve as well.

⁸ Not entirely, however. The so-called “medical” portion of biology is, perhaps, the best example of a natural science in which this problem persists in important respects.

⁹ The difficulty is not characteristic of the philosophically competent members of the pragmatic school. Scholars such as John Dewey close the circle by assuming useful any contribution to knowledge, and yet even in his work (e.g. Logic, the Theory of Inquiry, pp. 490–491) one may see a possible basis for misinterpretation.

¹⁰ By “arbitrary definition” it is meant that the specific denotation of any given term is not determined by previous usage or any authority other than the person using the term. This refers to the terminological aspect of definitions. The test of the usefulness and adequacy of a definition is something else again. If scientist A chooses to use the term “mota” for the empirical unit called “atom” by other scientists his work is not invalidated thereby. He may be cursed by others for his whims, but scientifically speaking it is neither better nor worse to call that unit “mota” rather than “atom.” By “precision” reference is made to two factors: (1) the degree of difficulty of confusing the denotations of a given term with those of other terms conceived of as different and (2) the closeness of relation between the denotation of the term and the referent to which it is applied. If “dog” is defined merely as a carnivorous mammal in a universe of discourse in which all mammals are a part, it is relatively imprecisely defined unless one intends as its referent all those mammals which are carnivorous. If the referent intended is the usual one for the term “dog,” the term is here not precisely enough defined to rule out cats, etc. and does not correspond solely to the referent intended.

¹¹ John Dewey and Arthur F. Bentley in an article, “Definition,” The Journal of Philosophy, Vol. XLIV, No. 11 (May 22, 1947), pp. 291–305, have a different view of the status of undefined predicates. I am not able to see how this article puts a stop to the possibility of an unending sequence of “and what do you mean by that” questions save by an undefined predicate.

¹² See Parsons, T., The Structure of Social Action, pp. 60–69.

¹³ There has been some social science application of mathematical analysis, largely in the field of economics, but even here the requirements of definition have been such as virtually to transport the effort from the realm of the empirical sciences to that of the formal sciences.

¹⁴ But even so this will not destroy the validity of the proposition relating Chinese familism and industrialization.

¹⁵ The mathematical statisticians are well aware of the problem of few cases and have devised a statistics applicable to small samples. See G. U. Yule and M. G. Kendall: An Introduction to the Theory of Statistics, Charles Griffin and Co., Ltd., London, 1946, pp. 434–62.

¹⁶ The reader must grant the author, at least for the sake of argument, precise empirical definitions of the terms used in these generalizations. To go into the elucidation of the generalizations themselves would carry the paper far beyond its bounds.