Abstract
This paper examines the relationship of war to power concentration and alliance configuration among the major powers. It does so by developing a theoretical argument from the literature on bipolarity and multipolarity. The paper suggests that bipolarity has two components — power distribution and alliance clustering. The two, it is argued, have 'opposite' effects on warfare in the major power system: power bipolarity minimizes the magnitude of those wars that do break out, while alliance bipolarity increases the likelihood that a war will occur. Correlational evidence, drawn from the Correlates of War data set, is supportive of both hypotheses for the twentieth century. The power polarity hypothesis is also confirmed for the nineteenth century, but the alliance polarity hypothesis seems disconfirmed for that century, perhaps because of the effects of hostility on the balance of power alliance structures of that era. The con ceptual framework and findings of the present paper allow one the reexamine the classic debate on the relative probabilities of war in bipolar and multipolar systems. The insights of Deutsch and Singer on the one hand and Waltz on the other appear to be more consistent with each other than heretofore recognized. Key elements of both the Waltz and the Deutsch and Singer arguments are supported by the findings. On the other hand, more recent, empirical studies by Bueno de Mesquita and others are critiqued on the basis of the present paper's theoretical perspective and empirical findings.