Thucydides’ History is one of my favorite books on international relations.  Kant’s Perpetual Peace is the other.  Kant was long neglected by ‘mainstream’ theorists put off by the words ‘perpetual peace’ that sound so fuzzily idealistic to the contemporary mind.  But the neglect has been reversed by a recognition that he was describing phenomena that have recently emerged in some parts of the world, notably in most of Europe.  The peace may not be perpetual, but it is actual in that most states with republican constitutions, substantial commercial exchange, and ties of international law and institutions—present-day equivalents of Kant’s three articles of perpetual peace—do not fight each other or expect to do so.  The characteristics of law (within republics, and commercial and international law between them) in his ‘pacific federation’ of sovereign states create norms and constraints for citizens and leaders.  Equally, the institutions establish mutual interests so that citizens and leaders need not be ‘angels’ guided solely by Kantian norms.  Rather, constraints of interest can make even self-seeking devils well-behaved ‘so long as they possess understanding’.  Kant is the Enlightenment’s empiricist and systematizer of hope for human freedom in an ordered world. 
Thucydides is a very different writer.  His discourse is that of the narrative historian rather than the philosophical analyst.  That makes him more engaging.  Also, those in the recently predominant realist school of international relations theory claim Thucydides as their patron.  Thucydides’ time and place fit the technical definition of anarchy, itself a good Greek word meaning “without a ruler.”  That does not mean chaos or the absence of some order, but rather that no legal or military power is able to enforce laws to provide basic security to the member units (city states in ancient Greece), and still fits much of the international system today.  States ultimately must rely on their own power, and recognition of its capabilities and limits, to survive in a fiercely competitive environment.  Acts undertaken for defensive motives may be seen by others as posing a real offensive threat whatever the intention.   Thucydides is the Greek historian recounting the tragedy of human action in a world of powerful forces beyond control.
Yet Thucydides does not fit a rigid realist paradigm.  He is far too perceptive. Like Kant, he is deeply concerned with norms of behavior, and how they include appropriate elements of self-interest.  But he knows that the world of Greek city-states is no Kantian pacific federation, so states will be far less restrained in their behavior, and perhaps less perceptive about their long-term interests.  Most importantly, Thucydides rejects a standard assumption of many realists that states can be considered as if they were rational unitary actors, carefully calculating (as best they can under conditions of uncertainty), the costs and benefits of alternative actions.  Changes in relative power provide the source of states’ insecurity, but then the perception of relative power, and choice about how to address it, form the heart of his story.  For Thucydides, domestic politics matters: democracies and demagogues, passions and perceptions.
To claim Thucydides as more than a realist is certainly not to claim him as a proto-Kantian, or early democratic peace (DP) theorist.  He does not see democracies as more peaceful than autocracies, and regards different elements of politics as making democracies either more or less effective in preserving their security.  Even as “simply” an historical narrative of fact, it is hard to find much evidence for a democratic peace in his volume.  While there is some evidence (Grasping the DP, ch. 3) that pairs of democratically-ruled states were somewhat less likely to fight one another than pairs composed of only one or no democracies, I make no more than that very weak claim.  Rather, I devote this presentation to some speculation, driven by evidence and theory for the democratic peace in the modern world, as to why we should not be able to garner much support for the DP in the experience of ancient Greece.
If the Greeks had democracy, why did their democratically-ruled states not behave as contemporary DP might be thought to predict?  A tentative answer lies in seeing how various parts of DP theory apply to both historical eras.  I first make some general comments on what DP theory most centrally claims, and then show how it applies to two questions that must be kept separate:  Are democracies peaceful in general?  Or are they peaceful only toward one another?  In discussing both these questions I consider how well the reasoning and evidence for each apply to both the modern system of states and the city-state system of Thucydides’ time, with some reference to their implications for understanding imperial expansion. All this depends on clarifying what we mean by democracy in both eras, and how democracy motivates decisions for war and peace.  I contend that of all the differences between ancient and modern democracy, the most relevant is in the degree to which institutions can enforce accountability on those who make and execute decisions.
Democratic Peace as a Probabilistic Theory
Despite great progress in refining the theoretical arguments and empirical support for democratic peace theory, the effort remains a work in progress.  Continuing the progress is important just because the theory is likely to be challenged in this historical moment, and that demands fully understanding its implications for both ancient and contemporary peoples.  To do so requires asking precisely what would constitute evidence for or against DP theory. 
The question is more than complex enough without expanding into the whole Kantian Peace including the effects of commerce and international law/organizations.  While unpersuaded critics certainly remain, there is now scholarly near-consensus for the basic empirical claim that over the two centuries have rarely fought one another.  Depending on how one defines the key terms, full-scale war between established democratic states is somewhere between extremely rare and completely absent.  Militarized disputes in a range of severity from purely diplomatic threat to small-scale violence falling short of a thousand war dead are more common, but still much less so than between non-democratic dyads.
The major component of DP theory is the dyadic proposition that the more democratic any two states are, the less frequent and less severe will be any militarized disputes between them. Democracy and autocracy are best conceptualized not as a dichotomy, but at various degrees on a scale.  This point applies both within a particular historical context and between such contexts.  For example, Athens, with an enfranchised citizenry amounting to about 10 percent of the population, has been labeled a “radical” democracy by the standards of its day.  Sparta would be somewhere on the autocratic side as a militarized oligarchy, but short of the end characterized by tyrannies of the time.  A common typology for the 19th through 21st centuries employs a 21-point composite scale reflecting competitiveness of political participation, openness and competitiveness of executive recruitment, and the level of institutionalized constraints on the executive.  It ranges, for example, from Sweden to North Korea.  Even there, the standards for a full democratic franchise rise from the 19th century to the middle of the 20th.  Comparing Greek and modern systems of democracy is of course very difficult, as will be apparent shortly.
It is essential to recognize the probabilistic aspect of democratic peace theory.  Virtually all political phenomena are too complex, and thus too interesting, to support a law-like claim.  The other two Kantian articles reinforce democracy as a force for peace, and such elements of realist theory as relative power and alliances, as well as personal perceptions and passion, can weaken it.  The vast majority of analysts would not expect to find an absolute law that democracies will never fight each other.  Even a fairly long list of examples of fights between democracies, ancient or modern, cannot refute a probabilistic theory until the actual probabilities are known.  It is relatively easy to compute those probabilities for the modern era, and much harder for the Greek world in which we have hardly any information about the characteristics and behavior of many states, or even of many states’ existence.  Certainly Thucydides does not have a full record on all the wars and states in the system, and would not give it to us if he did.  He cares about a particular set of relationships, and especially wars, central to his narrative.
Are Democracies Peaceful in General? 
A more sweeping claim that democracies are more peaceful monadically—that is, in general, with non-democratic states as well as with fellow democracies—is more contested.  The empirical evidence for this, however, is at best weak, much less powerful than the dyadic claim and much less robust (that is, remaining solid in response to changes in the empirical domain, what variables are included in a more complex analysis, and how the variables are defined and measured). 
The most important refinement of the simple relationship is to look at which side starts or escalates the fight, moving a largely peaceful diplomatic dispute up to the level of a militarized one, and a low-level militarized dispute up to a full-scale war.  On this the evidence is stronger: even when democracies are involved in diplomatic disputes with dictatorships they are less likely than the dictators to initiate the use of violence, and less likely to escalate any violence to a high level.  In other words, it is the dictator’s action that tends to produce the fight.
While this keeps aspects of a generalized (monadic) version of the DP alive empirically, it doesn’t help with the problem of imperial powers who claim a right to preventive military action.  Some further theoretical points are necessary to dispel an expectation of finding it in all cases.  A big qualification is that great powers are always nasty.  Whether democratic or autocratic, because of their widespread interests and intervention capabilities they are more likely than small ones to get into conflict.  That was true of the United States and the Soviet Union, and of Athens and Sparta.  So the difference in simple conflict involvement between democratic and autocratic great powers may be minimal.
Another important qualification is that democracies are generally more successful in war than are autocracies.  This too is a well-established fact for the 19th and 20th centuries, during which democracies won nearly 80 percent of their wars (all but possibly one or two of them against autocracies).  Under those conditions, by simple math autocracies must lose more than half all their wars (half of those fought against other autocracies, and more than half of those against democracies).  Democracies are more likely to win when they initiate the conflict because they are somewhat more likely to choose their uses of force prudently.  They won 93 percent of the wars they initiated, but only 63 percent of those they did not initiate.  (Examples such as Athens against Syracuse and recent events do not disprove the generalization).  They also may be able to fight more effectively because they can deploy a better motivated and educated fighting force, with troops and commanders capable of taking initiative in battle.
Thucydides admired Pericles, and recounts his funeral oration.  Pericles’ message (2.37-40) is one of praise for the culture of Athenian democracy as the source of Athenian greatness.  Its institutions are for ‘the many and not for the few’; ‘everyone has equal access to the law’.  The culture of Sparta relies on discipline, ‘painfully training’ its young men, whereas education into the culture of Athens trusts ‘less to our equipment and guile than to our personal courage in action’.  Athenians are ‘especially daring in our analysis and performance of whatever we undertake’.
In oligarchic Sparta the elite hoplites were renowned for their discipline and motivation, but not for their initiative.  Sparta’s greater weakness, however, was in the helots, a subjugated class in near-slavery, who could not be relied on to fight except under compulsion.  Indeed, Spartan generals hesitated to extend their forces too far from home lest the helots take that as an opportunity to rebel.  In the Athenian democracy free citizens were enfranchised and had a stake in the political and economic system.  Free soldiers formed a much larger share of the population than in Sparta, and were better educated.  Not only was Athens a great power, it was the richest of all Greek states per capita.  The Athenian navy was manned not by galley slaves subject to the lash, but by free men, well-paid thanks to proceeds from the empire, who rowed willingly and could be depended on to fight vigorously in the hand-to-hand combat characterizing much of naval warfare.  They could take remarkable initiatives, as after the landing on Pylos reported by Thucydides (4.4).  The Athenian position was exposed to Spartan attack.  The officers did nothing, so the rower-soldiers, unwilling to stay idle, took it upon themselves to build a strong defensive wall, at great effort without proper tools.
Democracies also seem more able and willing to mobilize resources from their economies in wartime, because the general population supports the war effort and will get to keep some of the winnings.  Citizens and foreigners will be more ready to lend to a democratic state in wartime because their chances of being paid back are better than with an autocracy.  Consequently, democracies often can spend relatively more on military forces than can most of their autocratic counterparts.  True, Stalin’s command economy and totalitarian political system devoted a very high proportion of the Soviet economy to military purposes.  But once World War II began Britain outmobilized Germany because Hitler was reluctant until very late to impose sufficient sacrifices on the German people. 
Kant makes some arguments about the culture of republican government, but his central one is about political institutions.  He contends that in states with republican constitutions the voting masses, both bearing the costs of wars and making the decisions about whether to fight, would be more reluctant to go to war than would the leaders of an oligarchy.  Leaders of oligarchies, by contrast, could keep most of any benefits of war to themselves but make the masses pay most of the costs.  Indeed, some of the benefits could be used to pay off key members of the oligarchy and the state security forces on which they depend for their rule.  This insight is behind a contemporary institutional argument as to why oligarchies and other autocracies go to war more often than do democracies.  I will return to it.
While there is much evidence for this formulation, it nonetheless misses a key point.  A democracy’s military success may then prime it for further efforts to exert its military power.  This leads to the matter of democratic imperialism, which cannot be ignored in ancient Athens, in the history of Western colonialism, or in the present day.  The same wide franchise that may restrain the impetus to war allows democracies to mobilize human and material resources more fully than oligarchies can.  Though other Greek democracies had some of the mobilizing advantages Athens did, none had remotely the level of resources available to it through its empire. Athens’ mobilizational advantage gave it incentives for expansionist or imperialist wars, raising both the probability of winning and the expected benefits to both elites and wider population.  I have seen this theoretical argument with illustrations from ancient Greece, Rome, Renaissance Italy, and more recent history.
For Greek democracies other than Athens the net effect of cost and benefit could be very different.  Most democracies are not imperialist.  But some are, at some times.  That in turn leads to further questions:  Are expansionist democracies expansionist because a particular kind of democracy feeds expansionism, or because expansionism promotes and supports a particular kind of democracy?  Or is it simply that most great powers, democratic or not, are expansionist in certain periods of their history?  Or does becoming a great power merely signify a successful policy of expansionism?  Untangling this complex interplay of influences should prove enlightening about questions of imperialism and democratic foreign policy in many eras.
The Dyadic Democratic Peace: Why So Little in Ancient Greece?
Evidence for a democratic peace between states in the era of the Peloponnesian War is weak at best.  Democratic states or factions often felt strong ties of affinity—a sense of common norms of behavior, maybe linked to constitutional principles—with other democratic states or factions within them.  Those ties seem to have been expressed in greater trust, and sometimes in alliances or at least a sense of solidarity against common adversaries.  But they were not strong enough to overcome more immediate forces such as competing alliances, past injury, or the risk that another democracy might nevertheless become an adversary if its relative power increased.  In other words, the realist principle that today’s friends or allies can readily become tomorrow’s enemies all too often triumphed.  Democracies often did not come to each other’s aid.  It was a self-help system, and long-term alliances were not reliable. 
Modern democracies are more likely to conclude formal alliances with each other than one would expect without knowing anything about their political regimes.  But they are not more likely to come to each other’s aid when one of them gets into a war.  Great powers fairly often come to the aid of weak states, but weak ones usually do their best to stay out of the way if they can.  Small or weak states individually can contribute little to the likelihood that a strong state will win in wartime, but a strong state can readily make all the difference for a weak one.  Thus small states have great incentive to free-ride on or ‘hide’ from the big state’s military efforts, even when they have stakes in the big state’s efforts.  In short, material and security interests typically trump ties of shared democratic norms when it comes to decisions about joining an ongoing war.  Kant expected his ‘federation’--more correctly, only a weak confederation--of commercial republics to serve as a collective security system both for solidarity within and protection from outsiders.  But that probably requires a higher degree of institutionalization than existed among the Greeks or among many modern allies. 
The need to add other influences that affect war and peace decisions before concluding that democratic peace does or does not exist in any international system has wider applications.  Distant states rarely fight each other, as they usually lack both the capability and the incentive to do so.  Only great powers are frequent exceptions, as discussed already.  So both distance and power should be included in any dyadic analysis.  This has not yet been done in systematic analyses of the Greek city-state system, and it should be before reaching conclusions either way.  Distance is easy, but a reliable measure of relative power is not.  It would be equally important to add the rest of Kant’s prescription, for republican governments, commercial ties, and international law/organizations as mutually supporting influences within his federation.  To look at the effect of democracy alone is to miss the degree to which the other elements may be conflated with it.  In the Greek system international organizations other than the Delian League (under Athenian hegemony) did not exist, and not enough data on trade patterns is available for a system-wide analysis.  Greek trade data may, however, be adequate to analyze its role in promoting peace or war in specific cases. 
A better understanding of why we see little evidence for the DP in Thucydides’ time, however, emerges from looking closely at the causal arguments offered for the contemporary democratic peace.  Theories often distinguish between cultural/normative explanations and structural/institutional ones.  Normative explanations generally say that leaders and citizens of democratic states expect political conflicts of interest to be settled largely without violence by means of negotiation, mediation, and arbitration.  Democratic peoples have a cultural aversion to violence, and prefer to settle international conflicts by similar means.  When they see other states whose regimes are founded on similar principles, they can expect conflicts of interest between their states similarly to be subject to peaceful settlement rather than the use or threat to use military force.  By contrast, they expect states governed by autocratic principles not to share a cultural preference for peace and so behave more assertively in their interactions with such states.
Democratic norms provided only weak restraints on interstate behavior in the Greek system. I also am coming to believe that while shared democratic culture and norms significantly influence peace between democracies in the modern era, their effect is less powerful than are democratic institutions.  Some institutional theories emphasize deliberative institutions and a separation of powers that may slow down decision making to avoid rash actions and provide time for diplomacy to address the conflicts. 
A more influential version of institutional theory emphasizes open political competition and regular free elections in keeping decision-makers accountable to a wide electorate.  I regard this as the most compelling explanation of the DP.  By this view what matters is the invention of institutions of representative government constrained by free elections.  That means a wide franchise and institutions which produce truly competitive elections that the incumbent can actually lose.  In a long or losing war, large parts of the electorate pay severe costs in money and blood.  If that happens, the democratic leader and his party can expect to be thrown out of office, especially if it was a “war of choice” rather resisting a clear attack.  So a wise democratic leader will anticipate electoral retaliation and avoid war in the first place.  In the United States, this is most likely when at least one branch of the legislature is controlled by the opposition party, giving it a more informed and credible platform if it chooses to oppose him.
This then becomes the basis for a theory of strategic interaction, as leaders contemplate the incentives and constraints on each other’s choices.  A democratic leader will especially avoid war with another democracy, since she knows that other democracies, like her own, are able to mobilize their resources and populations for long and effective resistance.  Autocratic leaders, however, have much less fear of being driven from office if their wars go badly, and whether or not they are successful in wars they may be able to make the general populace pay most of the costs and able to retain enough gains to hold the loyalty of their narrow base of supporters in the oligarchy and the domestic security forces.  Much systematic evidence supports this theory even though one may find exceptions where democratic leaders are yet able to hang onto office.
As I said earlier, comparisons between ancient and modern forms of government are hazardous.  We already know that it is a simplification to think of contemporary democracy and autocracy on a unidimensional scale, or to regard all democracies as having equally accountable executives and all autocracies as equally unaccountable.  A free press is highly correlated with democracy, and contributes importantly to democratic accountability. But the degree of press freedom varies within contemporary democracies and autocracies, and so may its effect on foreign policy behavior.  Other institutions and competitive practices vary notably across democracies and autocracies.  Such differences can elucidate rather than undermine how the democratic peace may operate.  And while there is little evidence of a corresponding autocratic peace, we know that certain types of autocracies get along better with each other than do other types.  Institutional distinctions among autocracies and democracies are complex and hardest to make in the mid-range of the scale.  Correspondingly, the effect of regime type on behavior is weakest there, and much stronger for the more uniform types at the highest and lowest ends of the scale.  That gives all the more reason for caution in generalizing about the equally great variation within simple dichotomy of democracy and autocracy in the Greek world, or about similarities between ancient and modern eras.   
By these interpretations, the failure to find much democratic peace in ancient Greece should not be surprising.  One reason could be that the Athenians had many slaves, and only men voted in what was by our standards a narrow electorate.  More important, however, may have been the lack of institutional constraints in the form of separation of powers and representatives subject to periodic and regular electoral removal. The assembly of citizens met at least 40 days per year.  There was no higher authority than the assembly, and its decisions were final. Only it could reverse its own decisions (as it did hastily reverse its initial hasty decision to slaughter the Mytilenians).  Generals and treasurers were chosen by election, but the nine chief administrative officers (archons) were chosen by lot without any special capacity or training.  So too was a Council of 500, whose members each served for a year.  Its executive committee changed composition ten times a year.  So there was no professional bureaucracy or representative institution resist rash action or to be held responsible.  The free citizens of the assembly, with their stake in retaining and expanding the system, were not especially pacifist.  That they, rather than their elected representatives, took the decision to go to war made it harder to blame leaders (even demagogues) as responsible. 
Kant expects the citizens of a republic to hesitate greatly before taking on a dangerous enterprise like war, for that ‘would mean calling down on themselves all the miseries of war, such as doing the fighting themselves, supplying the costs of the war from their own resources,’ and carrying the burden of debt.  But he distinguishes a republic, where ‘the executive power (the government) is separated from the legislative power,’ from the possible despotism of ‘badly organized constitutions of ancient and modern times (e.g., of democracies without a system of representation)’.  The Athenian combination of citizens who had the potential to gain from war, with the lack of representative institutions, was a citizenry with neither pacific preferences nor restraint. 
Very possibly, representative institutions and constrained executives are best produced by the wealth and culture of modernity in modern republics, though modernity is neither a sufficient nor necessary condition for such institutions.  If they are especially common under the conditions of modernity, they constitute not just another aspect of modernity but a greater perfection of democracy itself, one far better suited to a largish contemporary state which might have served even the Athenians better.  Modernity would therefore be a major ‘cause’ of democratic institutions, yet it could remain true that those institutions constitute the direct cause of peace among democracies.  How well the combination of broad suffrage and republican institutions can sufficiently restrain the resort to war by powerful democracies against nondemocracies is less clear.  The better we can understand how the differences between Greek democracies and modern affected their foreign policies, the better we will be able to understand the purposes and consequences of those policies, including imperialism.