Does the United Nations Charter permit one State (say, the United States or Turkey) to use armed force on the territory of another State (say, Syria), without the territorial State’s consent, targeting a non-State actor (say, ISIL or Kurdish militia)? Here is a bad argument that it does. Article 51 of the UN Charter says:

Nothing in the present Charter shall impair 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.

The text does not say “armed attack by a State.” Therefore, the argument goes, the text must mean “armed attack by a State or a non-State actor.” Since nothing in the Charter impairs self-defense under Article 51, the prohibition of inter-State force in Article 2(4) must give way. One State may use force on the territory of another State, without its consent, in self-defense against a non-State actor. The territorial State loses its legal protection from inter-State force even if it is not responsible for the armed attack carried out by the non-State actor. Or so the argument goes.

A number of States, scholars, and judges accept some version of this argument. But it is unsound.

The UN Charter permits the use of force in self-defense against an armed attack by another State. That limitation is not expressly stated, but it is presupposed by the text of Article 51, it is implied by the context of Chapter VII, and it is confirmed by the preparatory work. If States wish to change the law, that is their right. If they choose to exercise that right, they should do so responsibly. If they change the law irresponsibly, Syria will not be the last country vivisected before our eyes.

A treaty provision may mean more (or less) than it says. To determine what a treaty provision means, we must consider both what it says and what went without saying, what was implicit or presupposed. We must look to the whole text of the provision, in the context provided by the entire treaty. We may also look to the treaty’s preparatory work to confirm the meaning of the text in its context, to determine that meaning when text and context leave it ambiguous or obscure, or to establish the parties’ intent to give a term a special meaning.

For example, Article 2(4) of the Charter prohibits “the threat or use of force.” The text does not say “armed force,” but that is what the text means. Context and purpose make this clear. Articles 41, 44, and 46 use the terms “force” and “armed force” interchangeably. The purposes of the Charter include “to ensure . . .that armed force shall not be used, save in the common interest.” The preparatory work confirms the meaning of the text in its context, as Brazil’s proposal to prohibit economic coercion was rejected. This is how we know that Article 2(4) prohibits the threat or use of armed force, albeit not in so many words.

Similarly, the text, context, and preparatory work of Article 51 show that it permits only the use of armed force in self-defense against an armed attack by a State.

First, text: Article 51 says that self-defense is permitted “until the Security Council has taken measures necessary to maintain international peace and security,” that is, peace and security between States. The Charter consistently uses “international” to mean “inter-State” rather than, say, “transnational” or “cross-border.” Only an armed attack by one State against another would shatter inter-State peace, trigger inter-State war, and require collective measures to maintain inter-State peace and security.

Second, context: Article 51 is the final article of Chapter VII. The chapter heading, “Action with Respect to Threats to the Peace, Breaches of the Peace, and Acts of Aggression,” already suggests that self-defense against an armed attack is itself an action with respect to a threat to the peace, a breach of the peace, or (most plausibly) an act of aggression. The operative provisions confirm this suggestion. Chapter VII opens with Article 39, which provides that

The Security Council shall determine the existence of any threat to the peace, breach of the peace, or act of aggression and shall make recommendations, or decide what measures shall be taken … to maintain or restore international peace and security.

The Security Council maintains international peace by preventing States from threatening international peace, breaching international peace, or committing acts of aggression. Article 1(1) confirms this close relationship between means and ends. Importantly for our purposes, only States can directly threaten international peace, breach international peace, or (most importantly) commit acts of aggression.

What if the Security Council fails to prevent a State from threatening international peace, breaching international peace, or committing an act of aggression? The answer lies in the final provision of Chapter VII: Article 51. States may not resort to force in response to every threat to the peace or breach of the peace, or to every act of aggression. States may resort to force only in response to one form of aggression: an armed attack. Article 51 does not abruptly change the subject of Chapter VII from acts of aggression by States to armed attacks by non-State actors. Instead, Article 51 identifies the kind of act of aggression that States may respond to unilaterally, until the Security Council responds collectively.

Finally, and most interestingly, the preparatory work: The first draft, presented by the United States in May 1945, said:

Should the Security Council not succeed in preventing aggression, and should aggression occur by any state against any member state, such member state possesses the inherent right to take necessary measures for self-defense. The right to take such measures for self-defense against armed attack shall also apply to understandings or arrangements like those embodied in the Act of Chapultepec, under which all members of a group of states agree to consider an attack against any one of them as an attack against all of them.

This alone shows that “armed attack” was understood as a subcategory of “aggression … by any state.” Otherwise, the U.S. draft would permit individual self-defense only against aggression by a State, but permit collective self-defense both against aggression by a State and against armed attacks by non-State actors. Unsurprisingly, there is no evidence to support that interpretation.

There is more. What was the Act of Chapultepec, as referenced in the U.S. draft? It was an Inter-American defense pact, adopted in March 1945, just two months before the drafting of Article 51. The Act of Chapultepec provided:

That every attack of a State against … an American State, shall … be considered as an act of aggression against the other States which sign this Act.

In response, the Act contemplated the “use of armed force to prevent or repel aggression.” By defining collective self-defense in terms of “understandings or arrangements like those embodied in the Act of Chapultepec,” the U.S. draft incorporated by reference the Act’s limited scope to attacks of a State.

So, in the U.S. draft, aggression by any state triggers individual self-defense, while armed attack of a State triggers collective self-defense. Why was the language changed?

Two reasons. First, the U.K. representative (Eden) complained that the draft was “clearly of Latin American origin. It would result in regionalism of the worst kind.” He said that “if such a provision as this were included in the Charter he would not be able to sign it. It would make it a Latin American document.”

Second, and more persuasively, the U.K. representative complained that “no one had been able to define aggression in thirty years,” suggesting it would be unwise to make ‘aggression’ the trigger for individual self-defense. In response, the U.S. representative (John Foster Dulles) stated that “the United States proposal attempted to define aggression in terms of ‘armed attack’ and in this way it was hoped to avoid the problem of trying to define aggression as such.”

So, the terms “aggression … by any state” were dropped, and armed attack became the trigger for both individual and collective self-defense. But the understanding that an armed attack was an act of aggression, carried out by a State, was never questioned. As the United States assured its Latin American colleagues:

If the Security Council fails to stop aggression and there is an attack on any American republic or on the United States, adequate counter-measures can be taken.

The deletion of the terms “by any state” was never understood to permit the use of force against non-state actors in the territory of another State. The final text was supposed to clarify that only the clearest and gravest form of aggression—armed attack—would trigger the right of self-defense, individual or collective, under the Charter. The limiting words “by any State” went without saying.

Subsequent Practice: The Rio Treaty

Article 51 permits self-defense against an armed attack by a State. That is what the text means in its context, as the preparatory work confirms. Is that the end of the story? No. Subsequent practice in the application of a treaty matters as well, in one of two ways. Subsequent practice which establishes the agreement of the parties regarding its interpretation must be taken into account together with the context. Subsequent practice which does not establish the agreement of the parties may be used as a supplementary means of interpretation together with the preparatory work.

One example of subsequent practice deserves special attention. The Inter-American Treaty of Reciprocal Assistance (Rio Treaty) opened for signature in 1947, just two years after the UN Charter. In its preamble, the parties reaffirm “their adherence to the principles of inter-American solidarity and cooperation, and especially to those set forth in the preamble and declarations of the Act of Chapultepec.” Article 3 provides:

1. The High Contracting Parties agree that an armed attack by any State against an American State shall be considered as an attack against all the American States and, consequently, each one of the said Contracting Parties undertakes to assist in meeting the attack in the exercise of the inherent right of individual or collective self-defense recognized by Article 51 of the Charter of the United Nations.

In other words, the Rio Treaty requires what its parties understood the UN Charter to permit: collective self-defense against an armed attack by any State. As Joseph Kunz observed at the time, the Rio Treaty, “making full use of Article 51 of the UN Charter, distinguishes ‘armed attack’ from other forms of aggression.”

While the Rio Treaty contemplates “an aggression which is not an armed attack,” it does not even hint at the possibility of an armed attack which is not an aggression. Similarly, the Rio Treaty lists “[u]nprovoked armed attack by a State against … another State” as one of two acts that “shall be” considered aggression in all cases. But nothing suggests that armed attack by a non-State actor—though it cannot be considered aggression—triggers the right of collective self-defense under the UN Charter but not under the Rio Treaty.

We should not be surprised that the Rio Treaty, adhering to the principles set forth in the Act of Chapultepec, confirms that Article 51 only permits self-defense against an armed attack by a State. As we have seen, Article 51 was, truth be told, “clearly of Latin American origin.”

Subsequent Practice Since 9/11

Subsequent practice in the application of the Charter did not stop in 1945 or 1947. I will not attempt to summarize all the relevant practice here, let alone evaluate it. Instead, I will touch on a few points relevant to the analysis above, and then briefly sketch the current landscape.

First, subsequent practice certainly exists that interprets “threats to the peace” to include both direct threats to the peace, posed by States, and indirect threats to the peace, which may take many forms. In the words of former UN Secretary General Kofi Annan,

The threats to peace and security in the twenty-first century include not just international war and conflict but civil violence, organized crime, terrorism and weapons of mass destruction. They also include poverty, deadly infectious disease and environmental degradation since these can have equally catastrophic consequences.

Of course, Annan’s “more comprehensive concept of collective security” was to be implemented collectively, through the United Nations, not through the unilateral use of armed force by individual States. Still, as Christian Tams observes, if States converge on a new interpretation of “threats to the peace,” nothing prevents them from adopting a new interpretation of “armed attack” as well. But one evolution in the Charter’s legal meaning would not directly or automatically cause the other. As we have seen, armed attacks are a subset of acts of aggression, not of threats to the peace, and the inter-State understanding of aggression remains firmly in place. If States wish to adopt a new interpretation of “aggression” or “armed attack,” they must do so directly.

Second, in the aftermath of Al Qaeda’s attacks on New York and Washington, both the U.N. Security Council and the Organization of American States passed resolutions which, in their preambles, reaffirm or recall the right of individual and collective self-defense under the Charter and the Rio Treaty, respectively. These resolutions never mention Al Qaeda, the Taliban, or Afghanistan. At the time, it was unclear who carried out the attacks, whether they were attributable to a State, or where in the world force might be used in response. These resolutions do not establish the agreement of the parties to the Charter or the Rio Treaty that “armed attack” includes armed attacks not attributable to any State. As we shall see, more recent practice shows that no such agreement exists today.

Finally, subsequent State practice in the application of Article 51 lacks the clarity, specificity, consistency, and breadth to carry much interpretive weight one way or the other.

The United States, United Kingdom, Turkey, Canada, and Australia apparently interpret “armed attack” to include armed attack by any non-State actor.

Germany and Belgium interpret “armed attack” to include armed attack by a non-State actor that occupies part of a State’s territory, over which the State lacks effective control. (Note that whether the right of self-defense, once triggered, continues to justify the use of armed force after such a non-State actor loses its “consolidated territorial basis” is a separate question.)

France “does not recognize the extension of the right to self-defense to acts perpetrated by non-state actors whose actions are not attributable, directly or indirectly, to a State.” However, in exceptional cases, “an armed attack perpetrated by an actor having the characteristics of a ‘quasi-State,’” such as Daesh (ISIL), may qualify as an armed attack under Article 51.

Finally, Brazil interprets “armed attack” to include only armed attack undertaken by or attributable to a State. Similarly, Mexico interprets “armed attack” to exclude armed attacks by non-State actors in a third State that is allegedly unwilling or unable to suppress them.

This practice is clear and specific, but neither consistent nor broad.

Notably, the Community of Latin American and Caribbean States (CELAC), representing thirty-three countries, has expressed concern

at the increasing number of letters sent to the Security Council pursuant to Article 51 of the Charter of the United Nations—most often after the fact—with a view to using force to combat terrorism. The use of force in a manner inconsistent with the Charter was illegal, unjustifiable and unacceptable.

Referring to “military action, in the context of counterterrorism,” CELAC has conveyed

underlying concerns stemming from attempts to reinterpret the law on self-defense and to de facto expand an exception to the general prohibition to the use of force contained in Article 2.4 of the Charter, in an irregular manner.

These statements are neither clear nor specific, but they establish consistent and broad lack of agreement that “armed attack” includes armed attack by any non-State actor. These statements reject something as an irregular reinterpretation and expansion of the law on self-defense. Rejecting the qualified interpretations of France, Germany, or Belgium would logically entail rejecting the unqualified interpretation of the United States and others. So, directly or indirectly, these statements must reject the unqualified interpretation. If CELAC countries reject the qualified interpretations of France, Germany, or Belgium as well, then they should say so more clearly.

[Editor’s note: For more on CELAC’s concerns, see Pablo Arrocha Olabuenaga, “An Insider’s View of the Life-Cycle of Self-Defense Reports by U.N. Member States.”]

The vast majority of parties to the United Nations Charter have said and done nothing to establish their agreement with any of the interpretations described above. The subsequent practice that exists establishes ongoing disagreement, as we have seen. All of this reduces the weight of subsequent practice as a supplementary means of interpreting the Charter.

Text, context, and purpose remain the primary determinants of the Charter’s legal content. Perhaps that will change. Perhaps the parties will converge on a single interpretation, settling what is unsettled or, in my view, resettling what was once settled. For now, there is still time to avoid the extremely serious implications of the reckless interpretations of the United States, Turkey, and others before Article 51 takes on new meaning.

Images: United Nations photos at founding of the UN – San Francisco Conference, April – June 1945