Polarity and Free Trade er in a multipolar system than in a bipolar that these agreements are more likely to one:the relative scarcity of exit options in occur within than between military alli- the latter would seem to create ample op- ances.More importantly,the evolution of portunities for exploitation.As McKeown alliances into free trade coalitions is more points out in a more general context,for probable in a bipolar than in a multipolar example,Britain could impose "highly system.Crucial to the advantage of a favorable commercial treaties with [sic] two-power system are the lower threat of the South American states";it was less exit and the stronger incentives for intra- successful with Spain and Portugal,who alliance altruism that it confronts. could seek French support if British de- This analysis sheds some new light on mands became excessive(McKeown 1982, several debates about the causes of free 214).The rivalry-to-rule constraint, and controlled trade during the nineteenth though,does not arise exclusively from and twentieth centuries.It effectively allies'opportunity costs of exit.It inheres assigns a large role in the opening of post- as well in the competition between or war Western markets to the transition among alliances as a whole that an anar- from a multipolar to a bipolar interna- chic international structure typically pro- tional security system that occurred duces. simultaneously.It also implicitly attrib- Although it is theoretically possible for utes the breakdown of trade in the inter- the great powers of any system to collude war period partly to the existence of a in the exploitation of their allies,it is more multipolar security dilemma.It provides a likely that the noncollusive outcome will good fit to arguments that Britain was emerge(Snyder 1984,462).In its attempt neither a political hegemon nor a strong to balance the power of the opposing advocate of lowering trade barriers else- bloc,each great power is more likely to where during the nineteenth century pursue the individually rational but col- (McKeown 1983;Russett 1985). lectively suboptimal strategy of investing Of course,the argument can also be in- in the welfare of its allies.The pattern of dicted on several counts.Any attempt to investment is likely to vary across sys- test it empirically will confront many tems,however. problems familiar to students of interna- Any great power member of a coalition tional relations:for example,the small in a multipolar world has strong incen- number of international systems during tives to discriminate among its allies in the period of modern international trade terms of its investment in them,as its precludes statistical analysis,and it is dif- allies presumably confront divergent op- ficult to use comparative case studies portunity costs of exit.In contrast,the in- because of large variations in important centive to discriminate in a bipolar world variables between or among the cases.In is much weaker,as allies are much more addition,the problem of the single case- uniformly and securely locked into coali- that of bipolarity-arises.One way to ad- tions.Thus,in the trade arena,for exam- dress the latter problem might be to break ple,investments in allies in a multipolar apart multipolar systems,since these sys- system are likely to be expressed in the tems vary in terms of alliance flexibility: form of discriminatory trade preferences; ones that have experienced an"integrative in a bipolar system,free trade is the more spiral"should more closely resemble bi- likely outcome.10 polar systems than ones that have not. Problems also exist at the analytic level. Conclusion Vulnerable to challenge are my implicit or explicit assumptions that incentives to ex- The security externalities of agreements ploit,the utility of allies,and the impor- to open borders to trade,I argue,imply tance of security externalities can be held 1253Polaritv and Free Trade er in a multipolar system than in a bipolar one: the relative scarcity of exit options in the latter would seem to create ample op￾portunities for exploitation. As McKeown points out in a more general context, for example, Britain could impose ''highly favorable commercial treaties with [sic] the South American states"; it was less successful with Spain and Portugal, who could seek French support if British de￾mands became excessive (McKeown 1982, 214). The rivalry-to-rule constraint, though, does not arise exclusively from allies' opportunity costs of exit. It inheres as well in the competition between or among alliances as a whole that an anar￾chic international structure typically pro￾duces. Although it is theoretically possible for the great powers of any system to collude in the exploitation of their allies, it is more likely that the noncollusive outcome will emerge (Snyder 1984, 462). In its attempt to balance the power of the opposing bloc, each great power is more likely to pursue the individually rational but col￾lectively suboptimal strategy of investing in the welfare of its allies. The pattern of investment is likely to vary across sys￾tems, however. Any great power member of a coalition in a multipolar world has strong incen￾tives to discriminate among its allies in terms of its investment in them, as its allies presumably confront divergent op￾portunity costs of exit. In contrast, the in￾centive to discriminate in a bipolar world is much weaker, as allies are much more uniformly and securely locked into coali￾tions. Thus, in the trade arena, for exam￾ple, investments in allies in a multipolar system are likely to be expressed in the form of discriminatory trade preferences; in a bipolar system, free trade is the more likely outcome.10 Conclusion The security externalities of agreements to open borders to trade, I argue, imply that these agreements are more likely to occur within than between military alli￾ances. More importantly, the evolution of alliances into free trade coalitions is more probable in a bipolar than in a multipolar system. Crucial to the advantage of a two-power system are the lower threat of exit and the stronger incentives for intra￾alliance altruism that it confronts. This analysis sheds some new light on several debates about the causes of free and controlled trade during the nineteenth and twentieth centuries. It effectively assigns a large role in the opening of post￾war Western markets to the transition from a multipolar to a bipolar interna￾tional security system that occurred simultaneously. It also implicitly attrib￾utes the breakdown of trade in the inter￾war period partly to the existence of a multipolar security dilemma. It provides a good fit to arguments that Britain was neither a political hegemon nor a strong advocate of lowering trade barriers else￾where during the nineteenth century (McKeown 1983; Russett 1985). Of course, the argument can also be in￾dicted on several counts. Any attempt to test it empirically will confront many problems familiar to students of interna￾tional relations: for example, the small number of international systems during the period of modern international trade precludes statistical analysis, and it is dif￾ficult to use compaiative case studies because of large variations in important variables between or among the cases. In addition, the problem of the single case￾that of bipolarity-arises. One way to ad￾dress the latter problem might be to break apart multipolar systems, since these sys￾tems vary in terms of alliance flexibility: ones that have experienced an "integrative spiral" should more closely resemble bi￾polar systems than ones that have not. Problems also exist at the analytic level. Vulnerable to challenge are my implicit or explicit assumptions that incentives to ex￾ploit, the utility of allies, and the impor￾tance of security externalities can be held