American Political Science Review Vol.83 historical record suggests that such win- eth centuries (Downs,Rocke,and Siver- dows can open surprisingly quickly son1985,119-20). (Trachtenberg 1988). Thus,although states can in theory suc- An obvious alternative is to negotiate cessfully confront the problems inherent trade and security agreements simultan- in the security externalities of free trade eously.Suppose,for example,that the agreements,analysis and evidence suggest Soviet Union and the United States that such successes will be rare events in wanted to trade freely with each other but world politics.It is,as a result,not sur- were concerned that the balance of power prising that students of trade regimes between them might shift as a result.In either explicitly or implicitly link their theory,the two countries could resolve analyses to the postwar Western alliance their fears by simultaneously negotiating (cf.Haggard and Simmons 1987,134; an arms control accord.In effect,this Keohane 1984;Yarbrough and Yarbrough solution compresses the sequential proc- 1986).Nor is it surprising that some ob- ess just outlined into a single stage. servers argue that tariff cuts are more like- The costs of implementing this solution ly between allies than between states be- may be prohibitive,however.As the longing to different military coalitions opening shots in the Uruguay Round (cKeown1982,225). make clear,the negotiation of a trade Not all military alliances are equally agreement alone,even among allies,is likely to evolve into free trade coalitions, very complex;its success is uncertain at however.I shall argue that alliances pro- best.If the merger of negotiations itself duced by bipolar systems are more likely, does not affect the incentives to conclude and ones formed within multipolar sys- either accord,an explicit linkage of trade tems are less likely,to so evolve. and security issues between potential or actual adversary states raises the costs and lowers the probability of success by Great Powers,Alliances, several orders of magnitude.Thus, and Free Trade assuming that gains from trade do not International systems are distinguished vary widely across potential negotiating here in terms of individual states rather partners,it is cheaper for states to negoti- than in terms of alliances (Rapkin 1988). ate with their allies on trade alone than to Thus,the definition of multipolar and bi- link trade and security issues in a compre- hensive negotiation involving possible or polar systems used here follows that of Synder and Diesing:the structure of an actual adversaries. The historical record demonstrates that international system more powerful incentives than the gains is defined by the number of major actors in the from trade have been necessary to control system and the distribution of military power arms races.Analyses of the nineteenth and potential among them.In a multipolar sys- and twentieth centuries conclude that the tem there are several (more than two)'Great Powers'whose military power is roughly equal, majority of arms races that have ended and whose rivalry and cooperation dominate peacefully have done so because of politics in the system.,··A bipolar system is changes in the macropolitical order- one with only two Great Powers and a number specifically,the emergence of a threat by of smaller states.(Snyder and Diesing 1977, 419-20) a third state to both participants in the race (Downs,Rocke,and Siverson 1985). If the security externalities of any free The rise of German power,for example, trade accord render military alliances the ultimately ended the British-French naval natural basis of such agreements,the PD race of the midnineteenth-to-early-twenti- incentive structure confronting states that 1248American Political Science Review Vol .83 historical record suggests that such win￾dows can open surprisingly quickly (Trachtenberg 1988). An obvious alternative is to negotiate trade and security agreements simultan￾eously. Suppose, for exdmple, that the Soviet Union and the United States wanted to trade freely with each other but were concerned that the balance of power between them might shift as a result. In theory, the two countries could resolve their fears by simultaneously negotiating an arms control accord. In effect, this solution compresses the sequential proc￾ess just outlined into a single stage. The costs of implementing this solution may be prohibitive, however. As the opening shots in the Uruguay Round make clear, the negotiation of a trade agreement alone, even among allies, is very complex; its success is uncertain at best. If the merger of negotiations itself does not affect the incentives to conclude either accord, an explicit linkage of trade and security issues between potential or actual adversary states raises the costs and lowers the probability of success by several orders of magnitude. Thus, assuming that gains from trade do not vary widely across potential negotiating partners, it is cheaper for states to negoti￾ate with their allies on trade alone than to link trade and security issues in a compre￾hensive negotiation involving possible or actual adversaries. The historical record demonstrates that more powerful incentives than the gains from trade have been necessary to control arms races. Analyses of the nineteenth and twentieth centuries conclude that the majority of arms races that have ended peacefully have done so because of changes in the macropolitical order￾specifically, the emergence of a threat by a third state to both participants in the race (Downs, Rocke, and Siverson 1985). The rise of German power, for example, ultimately ended the British-French naval race of the midnineteenth-to-early-twenti￾eth centuries (Downs, Rocke, and Siver￾son 1985, 119-20). Thus, although states can in theory suc￾cessfully confront the problems inherent in the security externalities of free trade agreements, analysis and evidence suggest that such successes will be rare events in world politics. It is, as a result, not sur￾prising that students of trade regimes either explicitly or implicitly link their analyses to the postwar Western alliance (cf. Haggard and Simmons 1987, 134; Keohane 1984; Yarbrough and Yarbrough 1986). Nor is it surprising that some ob￾servers argue that tariff cuts are more like￾ly between allies than between states be￾longing to different military coalitions (McKeown 1982,225). Not all military alliances are equally likely to evolve into free trade coalitions, however. I shall argue that alliances pro￾duced by bipolar systems are more likely, and ones formed within multipolar sys￾tems are less likely, to so evolve. Great Powers, Alliances, and Free Trade International systems are distinguished here in terms of individual states rather than in terms of alliances (Rapkin 1988). Thus, the definition of multipolar and bi￾polar systems used here follows that of Synder and Diesing: the structure of an international system is defined by the number of major actors in the system and the distribution of military power and potential among them. In a multipolar sys￾tem there are several (more than two) 'Great Powers' whose military power is roughly equal, and whose rivalry and cooperation dominate politics in the system. . . . A bipolar system is one with only two Great Powers and a number of smaller states. (Snyder and Diesing 1977, 419-20) If the security externalities of any free trade accord render military alliances the natural basis of such agreements, the PD incentive structure confronting states that