¹ See the discussion in Kaplan, Morton A., System and Process in International Politics, New York, 1957, pp. xi ff.Google Scholar (referred to hereinafter as System and Process); and in idem, “Toward a Theory of International Politics,” Journal of Conflict Resolution, 11 (December 1958), pp. 335–47. The skepticism expressed in the present paper was stated clearly and repeatedly in System and Process, but apparently was not understood by some commentators.

² For a recent detailed discussion of this problem, see Schwab, Joseph J., “What Do Scientists Do?” Behavioral Science, V (January 1960), pp. 1–27.Google Scholar

³ Do not contend, of course, that these types of theory exhaust the kinds of analysis that might be made. Obviously, the historian who attempts to explain a concrete sequence of events or the genesis of a specific event proceeds in ways that differ in part from those analyzed. The student of voting behavior interested in prediction of voting trends usually adopts a different form of analysis, although this need not invariably be the case. Attempts at statistical correlation of historic events are also of a different order. A case in point is Teggart's Rome and China, although one might contend that he used his statistical information to build a theory of a strictly biographical-causal type. Similarly, attempts to relate international activities to form of government or to trade rivalry may be restricted to explanations that are based on correlations, or may extend to causal sequences and to theories of one of the two specified types. I would Contend, however, that for the kinds of problems this paper attempts to deal with at the theoretical level—that is, problems involving systematic social theory—the two broad types of theory specified roughly exhaust the field.

⁴ Regularities may be thought of in many ways. For instance, the fact that 2 is the first integer greater than I is a regularity of our ordinary number system. That winter follows fall is a regularity of weather in temperate climates. That the candidate with a majority of the votes takes office is a regularity of our political system. That a system of at least five essential actors goes along with the essential rules of the “balance of power” system is a regularity of my “balance of power” model. Attention to regularities directly implies attention to differences. For instance, in System and Process six comparative types of international systems are specified and differences postulated with respect to three different sets of variables: the essential rules, the endogenous variables of the system, and the parameters or exogenous variables. The search for regularities in System and Process was itself responsible for the construction of a comparative international typology for perhaps the first time in the discipline.

I do not know how to separate the search for regularities from the search for differences in political science. Nor can I understand the assertion that the search for reguxslarities must operate “only at the level of wholes” (Hoffmann, Stanley J., ed., Contemporary Theory in International Relations, Englewood Cliffs, N.J., 1960, p. 42).Google Scholar Regularities in my theory always involve die values of variables of systems, not “wholes,” just as differences do. Of course, there are irregular events and processes, and also random ones. Reality cannot be forced into preconstructed molds. But the existence of regularities (and thus of regular differences) is necessary for the development of theory. The predisposition to search for regularities is essential to finding them and is desirable, provided it does not involve a dogmatic rejection of evidence to the contrary.

There is, of course, the important problem of level or precision of analysis. For instance, two thermo statically constant and equivalent temperatures may on analysis be discovered to have different patterns of variance around the mean temperature. Of two superficially identical ovens, one may have a concealed timer that halts its operation automatically after some given period of time. Of two presidential systems of government, one system may at a particular time have a strong president and the other a weak president; one may operate according to an item-veto principle and the other not. Some differences at the more precise level of analysis may affect the system in a way relevant to our inquiry; others may affect it but only after a certain period of time; others may affect it in no relevant way. In System and Process such problems are considered as related to the “levels of abstraction” problem and are linked to coupled systems and to engineering problems.

⁵ Kaplan, Morton A., Burns, Arthur L., and Quandt, Richard E., “Theoretical Analysis of the ‘Balance of Power,’” Behavioral Science, v (July 1960), pp. 241–52.Google Scholar See also the article by T. C. Schelling in the present symposium.

⁶ For example, in the prisoners’ dilemma, if the prisoners acted irrationally and remained silent instead of talking because they misunderstood the strategic situation, they would in fact obtain a jointly more desirable result than in the case of a rational decision and might be led to believe mat they had acted rationally. See Kaplan, Morton A., Some Problems in the Strategic Analysis of International Politics, Research Monograph No. 2, Center of International Studies, Princeton University, January 12, 1959Google Scholar; and System and Process, ch. 10.

⁷ I do not disagree so much with die qualifications concerning game theory as it now stands that Burns makes elsewhere in this symposium as with his tendency to throw the baby out with the bath. For efforts to modify game theory to increase its relevance to international politics, see Schelling, Thomas C., The Strategy of Conflict, Cambridge, Mass., 1960Google Scholar; Kaplan, , System and Process, ch. 11Google Scholar; and idem, Some Problems in the Strategic Analysis of International Politics.

⁸ See Kaplan, Morton A. and Katzenbach, Nicholas de B., The Political Foundations of International Law, New York, 1961.Google Scholar