The Case for War: What Counts as an Act of War?

What has counted as an act of war - and who has ever decided

There is no definition.

That is the first thing to understand, and almost nobody knows it. There is no list, anywhere, in any treaty, of the things that constitute an act of war. No threshold of casualties. No category of weapon. No specification of what a state may do to another state before the doing of it becomes the doing of war.

The United Nations Charter, which governs the use of force between nations and to which almost every country on earth is a party, permits a state to defend itself "if an armed attack occurs." It does not say what an armed attack is. The North Atlantic Treaty, which binds thirty-two countries to treat an attack on one as an attack on all, uses the same phrase and offers the same silence.

The single most consequential term in international law is undefined, and it has been undefined for eighty years, and that is not an oversight.

This piece is about how we got here, what the phrase used to mean, and who benefits from its remaining vague.

•••

When war needed no excuse at all

Begin with the strange fact that for most of history, the question this article asks was meaningless.

War was a sovereign right. A state could go to war for any reason it found sufficient, and no external authority existed to say otherwise. The Latin term casus belli, literally, the case for war, described not a legal test but a justification offered to other rulers, to one's own nobles, and to God.

The classical list of acceptable grounds is worth reading because of how ordinary it is. Dynastic claims. Denial of trade. Mistreatment of envoys. Unpaid debts. Sovereignty over disputed territory. A slight to national honour. Any of these could licence an invasion, and all of them regularly did.

Rome formalised the etiquette. A body of priests, the fetiales, conducted a ritual procedure, a formal demand for redress, a waiting period, and finally a declaration, designed to establish that Rome's war was a bellum iustum, a just war. The procedure did not restrain Roman expansion in any measurable way. It made the expansion legible: correct in form, properly announced, sanctioned by ceremony.

That is worth holding onto, because it is the pattern the whole history follows. The rules have almost never prevented the war. They have determined what the war would be called.

•••

Three attempts to abolish it

The twentieth century tried three times to make war illegal, and the sequence of failures is instructive.

The Hague Convention (III) of 1907 required states to formally declare war before commencing hostilities, a modest procedural rule intended to prevent surprise attacks. It is still technically in force. It has been almost universally ignored since 1939, and formal declarations of war have become vanishingly rare; the United States, to take the most active belligerent of the modern era, last formally declared war in 1942.

The Covenant of the League of Nations required members to submit disputes to arbitration and observe a cooling-off period before fighting. It failed comprehensively, in part because the United States never joined and the major powers ignored it when inconvenient.

The Kellogg-Briand Pact of 27 August 1928, formally the General Treaty for Renunciation of War as an Instrument of National Policy, went furthest. Its signatories, beginning with Germany, France and the United States and eventually including most of the world, declared that they condemned recourse to war for the settlement of international controversies and renounced it as an instrument of national policy.

They renounced war. In writing. Eleven years before the largest war in human history.

The Pact had no enforcement mechanism whatsoever, and it is still in force today, which tells you something about the relationship between a treaty's ambition and its effect.

But it was not worthless. Kellogg-Briand established the norm that conquest cannot create legal title, that territory taken by force does not become yours. That principle survived into the UN Charter and is the reason annexations are still, eighty years later, not recognised. The Pact failed at stopping war and succeeded at making its winnings unbankable, which is a smaller achievement than intended and a larger one than nothing.

•••

1945, and the narrowing

The Charter of the United Nations did something genuinely radical, and it is worth stating precisely because most people have no idea it happened.

Article 2(4) prohibits all member states from the threat or use of force against the territorial integrity or political independence of any state. Not aggressive war, force. That is the default rule of the international system, and it reversed several thousand years of practice in a single sentence.

Against that prohibition, the Charter permits exactly two exceptions.

Article 51 preserves "the inherent right of individual or collective self-defence if an armed attack occurs against a Member of the United Nations, until the Security Council has taken measures necessary to maintain international peace and security."

Chapter VII allows the Security Council to authorise force.

That is the whole list. Everything else, honour, dynastic claim, unpaid debt, mistreatment of nationals, denial of trade, the entire classical catalogue, was demoted overnight from legal justification to political rhetoric. A state may still say those things. They no longer authorise anything.

This is a real achievement and I want to give it its weight before I take it apart. The default position of international law is now that you may not attack another country, and the burden of justification sits with whoever uses force. For most of human history the burden sat nowhere.

•••

The hole in the middle

Now read Article 51 again, slowly, and notice what it does not say.

If an armed attack occurs. Not if a threat exists. Not if hostile action is taken. Not if you are provoked, undermined, sabotaged, infiltrated, blockaded, or robbed. An armed attack.

And nowhere in the Charter, in any annex, in any subsequent binding instrument, is that term defined.

No casualty threshold. No specification of weapon. No rule about whether it must be territorial, whether it must be attributable to a state, whether damage must be physical, whether a single incident suffices or a pattern is required.

The same hole sits in the North Atlantic Treaty. Article 5, the collective defence clause, the entire reason the alliance exists, provides that an armed attack against one member shall be considered an attack against all, and offers no definition of the term either. It has been invoked once in seventy-seven years.

The technical name for the NATO provision is casus foederis, the occasion for the alliance, as distinct from casus belli, the case for war. One is a cause you assert for yourself; the other is an obligation triggered in your friends. Both, in the modern instruments, hang on the same undefined phrase.

So the most important word in the architecture governing when nations may kill each other has never been given a meaning.

•••

The one attempt at a standard, and where it came from

There is exactly one widely accepted test in this area, and its origin is almost comically small.

On the night of 29 December 1837, during a rebellion in Upper Canada, British and Canadian loyalist forces crossed into United States territory at Fort Schlosser, New York, seized an American steamboat called the Caroline that had been ferrying men and supplies to insurgents, set it on fire, and sent it over Niagara Falls. An American named Amos Durfee was killed.

The diplomatic correspondence that followed, between US Secretary of State Daniel Webster and the British representatives Henry Fox and later Lord Ashburton, produced the formulation that still governs. For a preemptive use of force to be lawful, Webster wrote, the necessity of self-defence must be "instant, overwhelming, leaving no choice of means, and no moment for deliberation", and the response must not be unreasonable or excessive.

Two requirements: necessity and proportionality. A threat that is genuine but distant does not qualify. A response that continues after the danger has passed does not qualify. The test distinguishes lawful anticipatory defence from reprisal, deterrence and preventive war.

That standard, arising from a burned boat and one dead man on the Niagara River, was cited by the International Military Tribunal at Nuremberg in 1946 when it rejected Germany's claim that the 1940 invasion of Norway had been a defensive measure.

It remains the benchmark. It is also, increasingly, ignored, because the word imminent has been doing an enormous amount of quiet work.

The progression runs like this. Imminent originally meant what Webster meant: no moment for deliberation. It was then stretched to cover threats that were certain but not immediate. It was stretched again, notably after 2001, to cover capabilities rather than intentions, the argument that a state possessing a weapon constitutes an imminent threat regardless of whether it has shown any intention to use it. At that point the word has stopped meaning imminent and started meaning eventually, and preventive war has been smuggled inside a doctrine designed to exclude it.

•••

The oldest trick, which is not a conspiracy theory

There is a recurring feature of this history that is uncomfortable and thoroughly documented, and it deserves stating plainly rather than being left to insinuation.

When a state wants a war, it has frequently manufactured or misrepresented the incident that justifies it.

The sinking of the USS Maine in Havana harbour in 1898 was taken as the American case for war with Spain. The cause of the explosion has never been established, and no evidence of Spanish involvement ever emerged.

On 31 August 1939, SS operatives in Polish uniforms staged an attack on a German radio station at Gleiwitz. Hitler cited it the following morning as the provocation for the invasion of Poland.

In August 1964, a reported second attack on American destroyers in the Gulf of Tonkin produced a congressional resolution authorising the escalation of the Vietnam War. That second attack is now widely assessed not to have occurred.

I am not claiming that every war begins in fabrication; most do not. The claim is narrower and better supported: the pretext is a standard instrument, it works, and the vagueness of the threshold is what makes it work. If an act of war were precisely defined, a manufactured incident would have to be manufactured to specification and would be correspondingly easier to check. Because it is not defined, almost anything can be presented as sufficient, and the presentation is the whole of the process.

•••

Why nobody will define it

Which brings us to the question the piece exists to answer. The hole has been there for eighty years. Lawyers have noticed. Why has it never been filled?

The honest answer has two parts, and the first is a genuinely good argument.

A published threshold is a published permission. If the world agreed that an armed attack means, say, an incursion causing more than a certain number of deaths, then everything beneath that line becomes free. An adversary could calibrate precisely, enough sabotage to hurt, never enough to trigger. The ambiguity is a deterrent, because an opponent who cannot calculate the edge must assume it is closer than it is. Serious people hold this view and they are not wrong.

And a defined threshold creates obligations. This is the less noble half. A phrase that means nothing specific can be interpreted, in the moment, by the people who have to decide, which means they retain the freedom to act when they wish and to refrain when they wish, and to describe either choice as compliance. A definition would remove that discretion. It would mean that when the specified thing happened, the alliance would be obliged, and the obligation would arrive whether or not it was convenient.

Nobody in a position to write the definition wants to be bound by it. That is the whole of the reason, and it has held for eighty years across every configuration of world power.

•••

The grey zone, which is now most of it

The consequence is a category of hostile action that has grown to swallow the ordinary conduct of international relations.

A cyberattack that disables a hospital network. A drone that crosses a border and is shot down. Sabotage of an undersea cable. Assassination on foreign soil. A blockade. Election interference. The financing and arming of a proxy militia. The seizure of shipping.

None of these has an agreed status. Each has been described by someone as an act of war and by someone else as something short of one, and in every case the description has followed the political need rather than preceding it.

The NATO position on cyber is instructive: the alliance has said a cyberattack could trigger Article 5, that this would be decided case by case, and that it will not specify what would qualify. Which is not a policy about cyberattacks. It is a policy about not having a policy, stated openly, and defended on exactly the deterrence grounds set out above.

•••

What this actually means

Strip away eight centuries of jurisprudence and the position is this.

There is no natural boundary between an incident and an attack, sitting out there in the world, waiting to be discovered by careful people. There is a decision, taken in a room, by governments, about whether the thing that has just happened is the thing the treaties mean, and that decision is made after the event, in light of what those governments would like to do about it.

The classical world understood this and did not pretend otherwise: a casus belli was a case you made, not a fact you found. The modern world has built an elaborate legal architecture on top of the same reality and uses it to describe political decisions in the vocabulary of law.

That is not a reason to abandon the architecture. The prohibition in Article 2(4) is real, the principle that conquest confers no title is real, and both constrain behaviour in ways that are visible when you compare the world before 1945 with the world after.

But the citizen reading that a hostile act has or has not crossed a line should understand precisely what they are being told: not that a threshold was measured, but that a judgement was made, by people with interests, at a moment of their choosing, about an event they have already decided how to feel about.

The line does not exist until someone says it does. It never has.

•••

•••

Sources

The UN Charter. Article 2(4) prohibits the threat or use of force against the territorial integrity or political independence of any state; Article 51 preserves "the inherent right of individual or collective self-defence if an armed attack occurs against a Member of the United Nations, until the Security Council has taken measures necessary to maintain international peace and security"; Chapter VII provides for Security Council authorisation of force. Neither the Charter nor any subsequent binding instrument defines "armed attack." That only two lawful justifications for the use of force survive the Charter, Article 51 self-defence and Chapter VII authorisation, and that all other historical grounds have been demoted to political rhetoric without legal standing, is the standard account in the international law literature.

Earlier attempts to restrict war. The Hague Convention (III) of 18 October 1907, Relative to the Opening of Hostilities, obliged state parties to declare war formally before taking up arms. The Covenant of the League of Nations required arbitration and a cooling-off period. The Kellogg-Briand Pact (General Treaty for the Renunciation of War as an Instrument of National Policy) was signed on 27 August 1928 by Germany, France and the United States and subsequently by most other nations, with signatories condemning recourse to war for the solution of international controversies and renouncing it as an instrument of national policy; it lacked any enforcement mechanism, remains in force, and is generally credited with establishing the norm that territorial acquisition by force is unlawful.

The Caroline test. The destruction of the steamboat Caroline at Fort Schlosser, New York, on the night of 29–30 December 1837 during the Upper Canada Rebellion, and the death of the American Amos Durfee, are documented in the standard accounts. The formulation that the necessity of self-defence must be "instant, overwhelming, leaving no choice of means, and no moment for deliberation," and that the response must not be "unreasonable or excessive," is from the correspondence of US Secretary of State Daniel Webster with Henry Fox and Lord Ashburton in 1841–42. The International Military Tribunal at Nuremberg cited Webster's formulation in 1946 in rejecting Germany's claim that its 1940 invasion of Norway was a defensive measure.

Historical pretexts. The sinking of the USS Maine in Havana harbour in February 1898 was treated by the United States as the casus belli for the Spanish-American War; the cause of the explosion has never been established. The Gleiwitz incident of 31 August 1939 and the Gulf of Tonkin incident of August 1964 are documented in the standard historical literature; the second reported Tonkin attack is now widely assessed not to have taken place.

NATO. Article 5 of the North Atlantic Treaty provides that an armed attack against one or more members in Europe or North America shall be considered an attack against them all, and explicitly ties the right exercised to Article 51 of the UN Charter. It does not define "armed attack" and has been invoked once, following the attacks of 11 September 2001. NATO's position that a cyberattack could trigger Article 5, to be assessed case by case without a specified threshold, is stated in its own published policy.

Formal declarations of war have become rare since 1939; the United States last formally declared war in 1942. The absence of a declaration does not affect the applicability of the law of armed conflict.

The piece's central analytical claim, that the central term of the modern legal architecture governing the use of force has been left undefined deliberately, because a published threshold would both create obligations for those who would have to honour them and specify for adversaries what may be done for free, is the piece's own. The facts on which it rests are in the public record.

•••

The Kade Frequency is an independent investigative publication. No sponsors. No filters. No propaganda.