Concept
The Treaty Power
Treaty and foreign affairs powers
The Treaty Power
Article II, Section 2 lets the President make treaties on two conditions: the Senate must give its advice and consent, and two thirds of the Senators present must concur.1 The House of Representatives has no role. A treaty binds the whole nation to another sovereign, and the supermajority requirement ensures that the commitment rests on agreement far broader than a bare majority of one chamber.
The clause settles who may make a treaty and leaves most of what follows to be worked out. Whether a treaty operates as law inside the United States depends on its own terms: in Medellin v. Texas, a non-self-executing treaty (one that needs implementing legislation) carried no domestic force without it, and the President could not supply that force alone. A treaty that does operate domestically still ranks below the Constitution, as Reid v. Covert held. Two further questions remain: how a treaty ranks against federal statutes and state law, and how a treaty ends, since the text prescribes how treaties are made and says nothing about unmaking them.
The subsections below follow a treaty through its life in that order. They begin with formation and domestic effect, turn to what the President may do about a treaty that lacks domestic effect, and set the treaty against statutes and state law. They then test how far the treaty power reaches into matters otherwise left to the states, mark the ceiling the Constitution sets, and end with termination. Each stage assumes the one before it, because a rule about a treaty's rank or limits presupposes a treaty that was validly made.
Making Treaties
The Treaty Clause authorizes the President, “by and with the Advice and Consent of the Senate,” to make treaties, “provided two thirds of the Senators present concur.”2 A treaty, for constitutional purposes, is a compact between the United States and a foreign nation that satisfies those advice-and-consent requirements.3 The clause names two actors and gives the House no part, and that allocation shapes everything in this subsection: it determines who controls each stage of formation, why the vote threshold is so demanding, and why a treaty's subject matter raises a question that an ordinary statute does not.
The text names the actors but not the sequence, which tradition supplies. The executive branch negotiates, the Senate reviews the finished text and may consent or withhold consent, and only after consent may the President ratify. Signing commits the United States to nothing yet: the parties' signatures establish that the text is authentic and definitive, but a nation is bound only when it approves the treaty through ratification, which for the United States follows Senate consent.4 Ratification then makes the treaty binding internationally and, if it is self-executing, a law of the United States with the force of a federal statute.5
Each part of this division has a practical logic. A single negotiator can keep positions confidential, answer a counterpart's move at once, and speak for the nation without first assembling a legislative majority. The Senate acts last, on a finished text, so its power to refuse operates on a concrete commitment. And the two-thirds figure answers to the nature of the commitment: a treaty binds the whole nation to another sovereign, so the Constitution demands agreement much broader than a bare majority.
The vote is measured against the Senators present, a phrase easily misread as two thirds of the full Senate.
The same 60 votes that fail with 100 Senators present succeed with 90 present, so the requirement tracks attendance rather than a fixed share of the Senate's membership.6
A treaty between two governments is bilateral, and one among several is multilateral. A party to a multilateral treaty may attach reservations, which release it from certain provisions, together with understandings or declarations, which state its own interpretation of treaty terms; the three are often abbreviated RUDs.7 The device lets a nation join a treaty while setting one provision aside, where otherwise each government would have to accept every provision as drafted or stay out. The treaty's own text usually states what brings it into force, and for multilateral conventions that is often approval by a specified number of nations.8
Lawyers cite a treaty by its name, the date of its signing, the parties if there are only two, and references to its main sources of publication. For a bilateral treaty the Bluebook generally calls for an official U.S. source if one exists, and for a multilateral treaty a parallel citation to an international source, usually the United Nations Treaty Series, may be added.9 The official source for treaties after volume 35 of United States Treaties (1983–84) is the State Department's Treaties and Other International Acts Series (TIAS), which was printed through 2006 and is now published only online.10
The House's absence gives a President with strong Senate support and a hostile House a reason to prefer a treaty to a statute.11 A President who doubts the Senate will consent may consider an executive agreement instead, but that route carries its own risk: the danger that the President is usurping the Senate's role in approving treaties.12 Executive agreements receive their own section below. For now, the point is that the treaty route excludes the House, and the exclusion raises the question of what subjects a treaty may reach.
The argument that this treaty is invalid has surface appeal, since the President and Senate would be doing what Congress as a whole ordinarily does. It likely fails. The Court has never held a treaty invalid simply because it governs an area Congress could regulate under its enumerated powers, and in Edwards v. Carter the D.C. Circuit upheld a treaty transferring property to Panama even though Article IV gives Congress the power to dispose of property belonging to the United States.13 Overlap of this kind is common: a trade treaty can readily cover ground within Congress's power “[t]o regulate Commerce with foreign Nations.”14 If overlap alone invalidated treaties, a great many would fall.
The overlap has edges. A good argument can be made that some functions the Constitution assigns to Congress cannot be performed by treaty, at least without congressional authorization.15 Article I, Section 9 permits money to leave the Treasury only “in Consequence of Appropriations made by Law,” so a treaty that tried to appropriate funds without an act of Congress would likely violate that textual limit. A treaty that purported to make certain conduct a crime would probably violate the Fifth Amendment's Due Process Clause, which traces to the old rule that a person could be deprived of life, liberty, or property only through the common law or an act of the legislature. It has also been argued that a treaty cannot substitute for a declaration of war, because the Constitution gave Congress that power to keep the President from taking the nation into war alone.16
The naturalization power differs from these examples: nothing in its purpose makes the House's participation indispensable, which is why the quota treaty is the easier case. Students tend to err in one of two directions. One reasons from the House's absence that any treaty on a legislative subject is void; the other reasons from the same overlap that a treaty can do anything Congress can, including spend money. Both shortcuts skip the real question, which is whether the Constitution commits a particular function to Congress for a reason the treaty process cannot satisfy.
Self-Executing Treaties
A ratified treaty binds the United States toward other nations, but whether it also binds a court deciding a case, or a state government, is a separate question. Medellin v. Texas supplies the framework. A self-executing treaty operates of its own force as domestic law; a non-self-executing treaty can be enforced only through legislation that carries it into effect.19 The Court drew the first definition from Foster v. Neilson, which treated a treaty as equivalent to an act of the legislature when it “operates of itself without the aid of any legislative provision.”20 A treaty can therefore bind the United States in its dealings with the world while carrying no weight in an American courtroom until Congress acts.
The label matters because of the Supremacy Clause, which lists treaties among the supreme law of the land that judges in every state must follow.21 A self-executing treaty carries the force of a federal statute, so it overrides conflicting state law and also displaces an earlier inconsistent federal statute.22 A non-self-executing treaty does neither. An obligation can be fully binding in international law without automatically becoming federal law enforceable in United States courts, and the question in Medellin was whether the International Court of Justice's Avena judgment had that automatic domestic effect.23
The line has a structural logic. The President negotiates and two thirds of the Senators present consent, but the House never votes. If every ratified treaty operated as domestic law automatically, those two actors could make binding rules for private parties and states without the chamber whose agreement an ordinary statute requires. Requiring legislation for treaties that do not say otherwise sends the creation of domestic law back through the ordinary lawmaking process.
The treaty's text therefore carries the weight. Whether a treaty is self-executing is a question of interpretation: the Court starts with the treaty's language and looks to related materials, including the drafting history and the parties' understanding after ratification, for evidence of what the negotiating President and ratifying Senate intended.24 The Executive Branch's long-held view receives great weight.25 The practical result is a presumption that a treaty silent about domestic enforcement is non-self-executing, which clear treaty language or implementing legislation can overcome.26 The treaties in Medellin were silent about judicial enforcement of ICJ judgments, and that silence decided which side of the line they fell on.27
Two ordinary treaties show the line before any litigation. A treaty giving the citizens of each party the freedom to engage in business in the other's territory on a nondiscriminatory basis creates the right itself and needs nothing further from Congress, so it is enforceable as domestic law at once and preempts a conflicting state licensing law.28 A treaty that merely sets international standards for nondiscriminatory treatment of foreign businesses and calls on each party to implement them through domestic legislation sits on the other side, because its words hand the next step to the legislatures; it is non-self-executing and has no domestic force until Congress acts.29 Both treaties address the same subject, so the fact that moves the answer is whether the treaty creates the right itself or directs the parties to create it.
Some subjects nearly always require legislation. A treaty that requires an appropriation of money or the criminalization of specified conduct is generally treated as non-self-executing,30 for reasons rooted in the Constitution: money may leave the Treasury only under an appropriation made by law, and a treaty that purported to make conduct a crime would probably offend the Due Process Clause.31
The facts of Medellin show the inquiry at its hardest. The International Court of Justice held in Avena that the United States had violated Article 36(1)(b) of the Vienna Convention on Consular Relations by failing to inform 51 named Mexican nationals, including José Medellin, of their Vienna Convention rights, and it directed review and reconsideration of their state-court convictions and sentences regardless of the state rules that ordinarily bar such challenges.32 Medellin argued that the treaties requiring compliance with Avena were already supreme law binding Texas courts. The Texas Court of Criminal Appeals dismissed his second habeas application as an abuse of the writ, concluding that neither Avena nor the President's Memorandum was binding federal law that displaced state limits on successive habeas filings.33
The Court read the relevant treaties one at a time, and each reading pointed the same way. The Optional Protocol to the Vienna Convention provides for compulsory ICJ jurisdiction but does not obligate the parties to treat ICJ decisions as domestically binding.34 The duty to comply with ICJ judgments comes from Article 94(1) of the U.N. Charter, in which each Member “undertakes to comply”; the Court read those words as a promise to take further steps through the political branches, not as an instruction that courts give a judgment immediate legal effect.35 Article 94(2) confirmed the reading, because it makes referral to the Security Council the remedy for noncompliance, a diplomatic mechanism rather than a judicial one.36
That remedy matters because of the bargain it reflects. The political branches ratified the Charter knowing that the United States can veto Security Council resolutions, so the scheme leaves them the option of declining to comply with an ICJ judgment. Automatic enforcement in American courts would remove that option and move sensitive foreign-policy choices from the elected branches to judges.37 The Court added that an individual such as Medellin is not a party before the ICJ, whose decisions bind only the nations that are parties in the particular case, and that the Executive Branch's long-held view that these treaties create no directly enforceable domestic law deserved great weight.38
On the Avena facts the steps end at the fourth. The Optional Protocol grants jurisdiction and Article 94(1) promises compliance, so both address governments and not courts; the Security Council remedy confirms that enforcement was left to the political branches; and Congress had enacted no implementing legislation. The judgment therefore had no domestic force, and Texas's procedural rules stood. Students err in two directions here. One reasons from the Supremacy Clause that any ratified treaty binds state courts, which skips the earlier question whether the treaty operates without legislation. The other reasons from non-self-execution that the United States owes nothing, when the international obligation remains and only domestic enforceability is missing.39
Justice Stevens concurred in the judgment without joining the majority's method. He thought the self-execution question for the Vienna Convention was close and saw no support in the Supremacy Clause or the Court's treaty cases for a presumption against self-execution, but he found Article 94(1) most naturally read as a promise of further steps through the political branches.40 Justice Breyer, joined by two other Justices, dissented and would have held the Avena judgment enforceable in state court without further legislation.41 He rejected any presumption against self-execution and described the Court's precedents as applying a practical, context-specific inquiry that weighs text, drafting history, subject matter, whether the provision confers rights an individual can enforce, and whether judicial enforcement would intrude on the political branches.42 He also objected that demanding an explicit statement of self-execution is misplaced, because many treaties the Court has found self-executing contain no such statement and multilateral drafters are unlikely to agree on uniform language about domestic effect.43
The majority answered that a multifactor approach would replace the expectations of the President and Senate, as expressed in the negotiated text, with ad hoc judicial judgments, and would make a treaty's meaning depend on later court rulings.44 The dissent is a competing methodology and not the law, but it explains why close cases generate argument. The majority treats the text as the primary evidence of what the political branches intended; the dissent treats it as one factor among several. On an exam the text-first method controls, and the dissent supplies the strongest argument against a presumption of non-self-execution.
Presidents and Non-Self-Executing Treaties
Having held that Avena was not directly enforceable, Medellin turned to whether the President could supply the missing domestic force alone. The President's Memorandum purported to require States to provide the review and reconsideration Avena called for, in disregard of state procedural rules, and Medellin relied on it in his second habeas application.46 The Court held that neither Avena nor the memorandum was directly enforceable federal law that displaced Texas's limits on successive habeas petitions.47
The Court tested the memorandum under Justice Jackson's Youngstown framework, which sorts presidential action by its relation to Congress. Action with express or implied congressional authorization is at its maximum; action in the absence of a congressional grant or denial falls in a twilight zone; and action contrary to the expressed or implied will of Congress is at the President's lowest ebb.48 The Government offered two main arguments, that the treaties authorized the President, with congressional acquiescence, to implement Avena and that he held an independent claims-settlement power; Medellin added that the Take Care Clause supported the memorandum.49
The first argument fails because of how the Constitution divides the treaty power. The President may make a treaty with the Senate's consent, so a President who wants a treaty to operate as domestic law of its own force should secure that effect in the making, through language plainly providing for domestic enforceability and the required two-thirds vote.50 The timing has a reason. The Senate votes on a text, and the text is the best evidence of what the Senate approved, so domestic effect written into the treaty has been approved by the body the Constitution assigns to approve it. After ratification without such language, a non-self-executing treaty can be enforced only through legislation carrying it into effect, and that task belongs to Congress.51
Under Youngstown the Government's position fares no better. A non-self-executing treaty is ratified on the understanding that it will not have domestic effect of its own force, which rules out reading the Senate's consent as authority for the President to create that effect alone. Because the treaties here are non-self-executing and no implementing legislation exists, the President's attempt to make them self-executing is inconsistent with the ratifying Senate's implicit understanding and falls in the third category, where his authority is at its lowest ebb.52 The Government's appeal to congressional acquiescence added nothing: earlier controversies did not involve a claim of unilateral authority to turn an international obligation into domestic law displacing state law, and statutes on representing the United States before international bodies address international responsibilities, not any power to create domestic law.53
The directive's target also matters. The memorandum would have compelled state courts to reopen final criminal judgments and set aside neutrally applicable state law.54 Separation of powers doctrine is not limited to disputes between Congress and the President; the Court has also relied on it to prevent congressional and executive encroachment on the judicial branch.55
The second argument rests on the claims-settlement cases, which involve a narrow set of circumstances: executive agreements settling civil claims between American citizens and foreign governments or foreign nationals.56 In Dames & Moore, the American hostages in Iran were released under an executive agreement stating the parties' purpose to terminate litigation between each government and the other's nationals and to settle all such claims through binding arbitration, and the agreement called for a claims tribunal to carry that out.57 The Court found the IEEPA and the Hostage Act “highly relevant in the looser sense of indicating congressional acceptance of a broad scope for executive action in circumstances such as those presented in this case.”58 Practice of that kind can carry weight: a systematic, unbroken executive practice, long pursued with Congress's knowledge and never questioned, raises a presumption that Congress consented.59 But past practice does not by itself create power, and the Government identified no precedent in which a President directed state courts to set aside neutrally applicable state criminal laws and reopen final judgments.60 A first-of-its-kind order cannot borrow authority from a settled practice of a different kind.
The third argument, Medellin's own, invoked the President's duty under Article II, Section 3 to “take Care that the Laws be faithfully executed.”61 The clause empowers the President to execute the laws, not to make them, and Avena was not domestic law. Without implementing legislation there was no law for the President to execute, so the clause could not carry the ICJ's judgment across into domestic law.62
An exam question on these facts asks what the President could have done to make Avena bind Texas, and the answer depends on when, and by whom, the choice about domestic effect is made.
The two versions differ in where domestic effect comes from. In the first it would come from a decree issued after the Senate acted; in the second it comes from the treaty the Senate approved.
The President is not left without options. The Executive may pursue other means of complying with treaty obligations consistent with the Constitution, but it may not rely on a non-self-executing treaty to establish binding rules of decision that pre-empt contrary state law.63 Congress, in which Article I vests the legislative powers,64 can give such a treaty domestic force by statute, and it has enacted implementing statutes for certain international judgments in other contexts.65 The international obligation also survives the ruling: Avena creates an obligation of the United States even though it is not binding federal law of its own force.66 The usual exam error is to reason from the President's leading role in foreign affairs directly to power over state courts, which skips the Youngstown question of what the ratifying Senate and Congress authorized.
Treaties Versus Statutes
The Supremacy Clause of Article VI names treaties and acts of Congress together as “the supreme Law of the Land,” and that shared wording fixes how each ranks against the other kinds of law it meets.69 Against state law the ranking runs one way: a valid treaty preempts conflicting state law, because the Supremacy Clause makes the treaty superior to it.70 Against a federal statute the ranking is parity, since a treaty is of equal authority with a statute approved by both houses and signed by the President.71 Students often carry the treaty's victory over state law across to federal statutes, yet the two comparisons call for different answers. This subsection assumes a treaty that operates domestically, whether because it is self-executing or because Congress has implemented it, since only such a treaty carries the force of a federal statute and so overrides all state law as well as prior federal statutory law.72
The parity follows from how each kind of law is made. The Constitution lets the President make treaties with the advice and consent of two thirds of the Senators present, and it makes statutes through passage by both Houses and presentment to the President.73 The Supremacy Clause names both products as supreme law without placing either above the other. A conflict between two sources of equal rank cannot be settled by asking which is higher, so it needs a tie-breaker that favors neither source and neither branch's work.
The tie-breaker the Court chose is the calendar. When a treaty and a federal statute conflict, the most recently adopted prevails, as Whitney v. Robertson held.74 The Chinese Exclusion Case expressed the idea in a phrase borrowed from conflicts between statutes: “the last expression of the sovereign will must control.”75 The Court has also described an Act of Congress as on full parity with a treaty, so that a later inconsistent statute renders the treaty null to the extent of the conflict.76 A date works as a tie-breaker because it asks only which lawmaking act came last, which requires no judgment about whether the President and Senate or the two Houses deserve more weight.
An exam stem that sets a ratified treaty against a federal statute is asking for the order of adoption, and the result can run in either direction: a treaty can trump an earlier statute, and a statute can trump an earlier treaty.77 The Canadian quota example from the previous subsection shows the first direction. A federal law capped the number of Canadian immigrants who may apply for citizenship each year, and a later treaty with Canada raised the cap; because the treaty is the later expression of the sovereign will, it controls, and the earlier statute gives way to the extent of the conflict. A statute enacted while the 1954 Mutual Defense Treaty with the Republic of China (Taiwan) remained in force, providing that under no circumstances may American forces be used to defend Taiwan, would run the other way. It would not formally terminate the treaty, but to the extent the two conflict, the statute, as the more recent enactment, prevails.78 Both pairs put a treaty and a statute on the same subject, and the only fact that moves the answer is which came last. The usual exam error is to announce that treaties outrank statutes because the Supremacy Clause lists them, or that statutes always win because Congress writes them; each shortcut ranks the two sources and skips the date.
A statute that prevails over a treaty does so inside the American legal system. The statute barring the defense of Taiwan would trump the treaty for domestic purposes, yet that de facto repudiation may have repercussions for the United States under international law.79 The two systems answer different questions: a United States court asks which federal law governs the case in front of it, and a foreign government asks whether the United States kept its promise. International law may not recognize the constitutional sequencing of later-in-time, but in American courts the Constitution is ascendant over international law.80 The split echoes the one seen earlier with non-self-executing treaties, where a binding international obligation lacked domestic force. Here the gap opens the other way, and a domestic override leaves the international obligation standing.
The later-in-time rule applies only to a genuine conflict, because courts first try to interpret the treaty and the statute so that both can coexist.81 A forced collision can leave the United States in breach of an obligation to another nation, and the decision to break a promise of that kind belongs to the political branches. A judge who can read the two texts together avoids making that decision for them. Only when the conflict is real and unavoidable does the later in time prevail.
Because different opponents produce different rules, a problem of this kind runs best as a short sequence that fixes the treaty's domestic force before comparing it to anything else.
Both treaties and statutes rank below the Constitution, so a treaty that violates a constitutional restraint fails whatever its date.82 The next subsections take up that ceiling.
Missouri v. Holland and Federalism
Article II, Section 2 delegates the treaty power in express terms, and that delegation is what makes treaties a special problem in a federal system.85 Congress may legislate only on subjects the Constitution lists, and the Tenth Amendment reserves the rest to the states, so a treaty on a subject outside Congress's list raises the question whether the President and Senate can reach what Congress could not. Missouri v. Holland answered yes. Missouri sued to stop a federal game warden from enforcing the Migratory Bird Treaty Act of 1918, arguing that the statute invaded rights the Tenth Amendment reserved to the states.86
The treaty, between the United States and Great Britain, recited that many species of birds cross parts of the United States and Canada in their yearly migrations, that the birds are valuable as food and for destroying insects that damage vegetation, and that they were in danger of extermination for lack of adequate protection. The 1918 Act, passed to give the convention effect, prohibited killing, capturing, or selling the covered birds except as permitted by regulations of the Secretary of Agriculture.87 Missouri's argument had a respectable pedigree. Two lower courts had held invalid an earlier act of Congress that tried to regulate the killing of migratory birds on its own, with no treaty behind it, on the theory that the states owned the birds in their sovereign capacity and Congress could not displace that control.88 Missouri contended that the same objection applied with equal force to the statute implementing the treaty.
Justice Holmes began with the text. The Tenth Amendment reserves only powers the Constitution has not delegated to the United States, and Article II, Section 2 delegates the treaty power expressly, while Article VI declares treaties made under the authority of the United States, along with laws made in pursuance of the Constitution, the supreme law of the land.89 A bare reference to the Tenth Amendment therefore cannot defeat a treaty, because the Amendment reserves nothing the Constitution has already handed to the national government. The second step follows from the first: if the treaty is valid, the statute is valid under Article I, Section 8 as a necessary and proper means of executing the powers of the Government.90 Congress's authority to carry out a treaty thus does not depend on whether Congress could have legislated on the subject by itself.
Holmes did not claim the treaty power is unlimited. He said that qualifications exist but must be ascertained in a different way, and he left open whether “the authority of the United States” means more than the formal acts required to make a convention.91 He grounded the result in function: there may be matters of the sharpest exigency for the national well-being that an act of Congress could not deal with but that a treaty followed by such an act could, and it is not lightly to be assumed that, in matters requiring national action, “a power which must belong to and somewhere reside in every civilized government” is not to be found.92 He called Missouri's claim of title to the birds “a slender reed,” since wild birds are possessed by no one, and he framed the Tenth Amendment question as one about the present nation: “We must consider what this country has become in deciding what that Amendment has reserved.”93
The two steps explain how a statute that failed on its own can succeed once a treaty stands behind it. The statute is no longer measured against Congress's enumerated powers; it is measured as a necessary and proper means of executing the treaty, so the live question becomes whether the subject is a proper one for a treaty at all.94 Environmental protection shows the contrast. Congress's Property Clause power over “the Territory or other Property belonging to the United States” is important there because the federal government owns land nearly the size of India,95 but a treaty can supply reach that no other federal power does. A later case makes the point concrete: the Court held in United States v. Morrison that Congress lacked power to enact the Violence Against Women Act remedy under the commerce power.96
The answer on that question is deferential. The President and Senate could rationally conclude that violence against women is of sufficient international concern to warrant a treaty, and given the generally deferential review afforded treaties, the Court would likely accept that judgment, which would make the reenacted statute a necessary and proper means of implementing the treaty.97 The reasoning is easy to supply: if freedom from gender-motivated violence is a basic part of human rights, any nation seeking leadership among nations has an interest in eradicating such violence worldwide.98
Holland does not say that a treaty may violate the Constitution. Holmes observed that the treaty before him contravened no prohibitory words in the Constitution, and that the only question was whether the Tenth Amendment forbade it through some “invisible radiation from the general terms” of the Amendment.99 The line therefore runs between a prohibition written into the text, which binds treaties just as it binds statutes, and a reservation of powers to the states, which a valid treaty overcomes. The Reid plurality read Holland the same way, noting that the treaty there was inconsistent with no specific constitutional provision and that, to the extent the United States validly makes treaties, the people and the states have delegated their power to the national government, so the Tenth Amendment is no barrier.100 In practice the subject-matter check is light: the Court has never held a treaty invalid because it addressed a subject beyond the treaty power.101 The usual exam error is to read Holland as meaning that a treaty can do anything, which skips the prohibitory provisions that still bind.
The breadth of the holding has drawn pushback, and it starts from structure. The President and Senate make a treaty without the House, and if a made treaty can expand the subjects on which Congress may legislate, treaty-making becomes a way to enlarge federal power without the amendment process Article V prescribes.102 The worry has more force today than in 1920, because early treaties were typically bilateral and addressed relations between the two states and each other's citizens, while modern multilateral instruments regulate how states treat their own citizens and even the conduct of private persons.103 The operative limit the Court has chosen works at the level of statutory interpretation, and Bond v. United States shows it. Congress had enacted a criminal statute to implement the Chemical Weapons Convention, and prosecutors applied it to a neighborly feud born of romantic jealousy, in which chemicals spread on a doorknob and mailbox caused a minor thumb burn.104 The Court read the statute not to reach that far, reasoning that doing so would turn a law concerned with acts of war, assassination, and terrorism into a massive federal regime covering the simplest assaults. It added: “Absent a clear statement of that purpose, we will not presume Congress to have authorized such a stark intrusion into traditional state authority.”105
Holland and Bond differ in what the enacted text said about domestic reach. The treaty and statute in Holland addressed the killing of designated migratory birds and nothing else, so no clear-statement problem arose. The statute in Bond gave no sign that Congress meant to turn a local assault into a federal crime, so the ordinary assumption that traditional police matters stay with the states controlled. The demand for clarity protects the states where their protection actually sits: their defense against federal overreaching runs through Senate ratification and the federal lawmaking process, and vague language that swept in traditional state domains would weaken it.106 The canon governs how a court reads a treaty-implementing statute and says nothing about the outer limit of the treaty power, which Holland left where it was.
The Bond majority avoided the larger question Carol Bond pressed, namely that if the statute reached her conduct, Holland would have to be limited or overruled because it would remove all limits on federal authority once a treaty is ratified.107 Justice Scalia, joined by Justice Thomas and in part by Justice Alito, concurred only in the judgment. He would have reached the constitutional question and held that the treaty power did not extend to the statute, accusing the majority of “result-driven antitextualism” and concluding that the law's unavoidable meaning covered Bond's conduct.108 His argument starts from text. The Necessary and Proper Clause lets Congress carry into execution the “foregoing Powers, and all other Powers vested” in the Government, and one of those is the President's power to make treaties, but in his words, “a power to help the President make treaties is not a power to implement treaties already made.”109 On that view, once the treaty is made, Congress must rely on its own Article I, Section 8 powers to legislate compliance. Whether a majority will ever adopt this position remains to be seen. If it did, the domestic reach of non-self-executing treaties, which as the earlier subsections showed need implementing legislation to operate at home, could be seriously curtailed.110 For now the working law remains Holland, with the clear-statement canon as the practical check.
Treaties and the Constitution
Article VI lists three kinds of supreme law: the Constitution itself, the laws of the United States made in pursuance of it, and all treaties made under the authority of the United States.113 The earlier subsections placed a treaty above state law and level with a federal statute, which leaves one comparison open: whether a treaty may also override the Constitution. Reid v. Covert answered that it may not, because treaties and executive agreements are not co-equal with the Constitution and cannot authorize action that violates specific constitutional protections. Holland showed that a valid treaty removes the Tenth Amendment as an obstacle. Reid marks where that tolerance ends, at a prohibition written into the constitutional text.
The case arose from two killings overseas. Two civilian wives were tried by court-martial for murdering their husbands, members of the armed forces: Mrs. Covert killed her husband, an Air Force sergeant, in England, and Mrs. Smith killed her husband in Japan.114 Each was tried under Article 2(11) of the Uniform Code of Military Justice, which reached civilians accompanying the armed forces overseas, and a court-martial gives the accused no jury trial and none of the other Bill of Rights protections of a civilian court.115 The Court first upheld the trials, then granted rehearing and held that neither woman could constitutionally be tried by military authorities.116
The Government defended the trials by pointing to international agreements. It argued that Article 2(11) was necessary and proper to carry out the nation's obligations under agreements with Great Britain and Japan, the reasoning that had sustained the bird statute in Holland.117 Justice Black's plurality opinion rejected the argument in direct terms: “The obvious and decisive answer to this, of course, is that no agreement with a foreign nation can confer power on the Congress, or on any other branch of Government, which is free from the restraints of the Constitution.”118
The plurality's reasons are layered, and each supports a different part of the rule. The first concerns the source of federal power. The United States is entirely a creature of the Constitution and can act only in accordance with the limitations the Constitution imposes.119 The treaty power is one of those constitutionally conferred powers, and a power that exists only because the Constitution grants it cannot sensibly include authority to disregard the Constitution's other terms. The Court added that these limits do not disappear when the government acts against a citizen overseas, since the protection of the Bill of Rights should not be stripped away because a citizen happens to be in another land.120
The second reason answers a textual objection. Article VI makes laws supreme only when made “in Pursuance” of the Constitution, while treaties need only be made “under the Authority of the United States,” and a party might read that difference as freeing treaties from constitutional prohibitions.121 The Court found nothing in the language or the ratification debates to suggest that treaties need not comply with the Constitution, and it explained the wording by history: treaties were not tied to the word “pursuance” so that agreements made under the Articles of Confederation, including the peace treaties ending the Revolutionary War, would remain in effect.122 The wording had a transitional job and offers no evidence that the drafters meant to exempt treaties from constitutional prohibitions.
The third and fourth reasons are structural, and the third does the heavier work. If Article VI allowed the nation to exercise power under an international agreement without observing constitutional prohibitions, the Constitution could be amended outside the process Article V prescribes.123 A treaty needs only the President and two thirds of the Senators present, a far narrower route than the amendment process, so letting a treaty override a prohibition would let those officials do what only an amendment can do. The fourth reason uses the hierarchy from the previous subsection. A later statute that conflicts with a treaty overrides it, and a statute must conform to the Constitution, so it would be anomalous to say that a treaty need not comply when a statute that must comply can displace it.124
Reid and Holland fit together because they involve different kinds of constitutional objection. The plurality observed that the Holland Court carefully noted that the treaty before it was inconsistent with no specific provision of the Constitution and that its only concern was the Tenth Amendment, which reserves undelegated powers to the states or the people.125 A reservation of undelegated power yields to a valid treaty, because the treaty power is itself a delegation. A specific prohibition denies the power to every branch, and it does not yield. The two women invoked exactly that kind of provision, since the court-martial did not meet the requirements of Article III, Section 2 or the Fifth and Sixth Amendments, which guarantee jury trial and grand jury indictment.126 The ban on export duties works the same way: a treaty may not impose a duty on exports without violating Article I, Section 9, and generally may not violate or alter other constitutional provisions, including those of the Bill of Rights.127
The two treaties were made by the same officials in the same way, and neither the date of the agreement nor the size of the vote affects the result. The deciding fact is the kind of constitutional objection raised against the action the treaty is offered to support.
The second step carries the doctrine, and students go wrong at it in two directions. One error is to reason from Article VI's listing of treaties as supreme law that a ratified treaty overrides anything, including the Constitution, which skips the prohibitions that bind every branch. The other is to read Reid as barring treaties from subjects the states would otherwise control, which skips the plurality's own statement that the Tenth Amendment is no barrier to a valid treaty.
The case should be cited with its voting pattern in mind. Justice Black wrote for a plurality, so the broad statement that no agreement can confer power free of constitutional restraints did not command a majority. Justices Frankfurter and Harlan concurred in the result on narrower grounds limited to capital offenses: Justice Frankfurter concluded that in capital cases the constitutional guarantees must be observed, and Justice Harlan held that Article 2(11) could not constitutionally be applied to civilian dependents overseas in peacetime when the offense is capital.128 The proposition that a treaty must conform to the Constitution is nonetheless generally accepted, and it drew on the Court's older recognition in Geofroy v. Riggs that the treaty power is “in terms unlimited except by those restraints which are found in that instrument against the action of the government or of its departments.”129 No case has squarely held a treaty invalid on this basis, because the Court has never been confronted with the issue.130 Reid itself involved executive agreements and the statute invoked to carry them out, and the agreements could not support trials that violated the Fifth Amendment. The decision put to rest the concern that treaties might support domestic action overriding individual rights, but it left open whether federalism was adequately protected, which is why the clear-statement canon discussed in the previous subsection carries so much practical weight.131
Terminating Treaties
Article II, Section 2 tells you how the United States makes a treaty: the President negotiates, the Senate advises and consents, and two thirds of the Senators present must concur.134 The clause covers only the making of a treaty, and the Constitution is silent about how one ends.135 That silence supports two opposite arguments. One holds that ending a treaty should mirror making one, so the Senate, and perhaps the House, must take part. The other holds that because the Constitution requires no Senate or congressional approval to end a treaty, the authority rests with the President as part of the executive power.136 Each side treats the same missing text as proof for its own reading, which is why the text alone cannot settle the question.
Practice does not break the tie. The United States has ended treaties by a statute directing the President to send notice of termination, by a statute without separate presidential notice, by the President with the approval of both Houses, by the President with the consent of the Senate, and by the President alone.137 Because every method has precedent, a President who wants to withdraw unilaterally and a Congress that wants a voice can each point to history.
The dispute that tested these readings arose from the recognition of China. In 1978, when President Carter recognized the People's Republic of China as the sole government of China, he announced that he was terminating the 1954 Mutual Defense Treaty with the Republic of China (Taiwan). Members of Congress sued in Goldwater v. Carter, claiming that a President may end a treaty only with the consent of two thirds of the Senate or a majority of both Houses.138 The Supreme Court dismissed the challenge in 1979 and treated the case as not justiciable, meaning not suited for a court to decide.139 Six Justices voted to dismiss, but they did not share a reason. Four concluded that how a treaty may be ended is a political question, an issue the Constitution leaves to the elected branches. Another found the challenge not ripe, meaning the dispute had not developed far enough for a court to resolve it. The sixth did not disclose a rationale.140
Because no rationale commanded a majority, the case produced no rule, and it is better remembered for what it left open than for anything it decided. Whether a President may terminate a treaty alone remains unresolved, to be worked out by the political branches themselves.141 The four-Justice group's reasoning explains why courts hold back. They were reluctant to enter a dispute between coequal branches, each of which has constitutional powers to assert and protect its interests outside the courtroom, and those powers are not available to private litigants.142 Congress can legislate and control funding, while the President directs diplomacy and decides what notice to send, so each branch holds real leverage over the other. A court that declared a winner would end a contest the branches are equipped to carry on themselves. The same instinct explains why foreign-policy disputes so often reach court already labeled political questions.143
A separate opinion read the text differently, and it shows that the political-question label was contestable. It reasoned that a political question requires a textual commitment of the issue to another branch, and no constitutional provision explicitly gives the President power to terminate treaties. Article II, Section 2 gives the President power to make treaties only with Senate consent, and Article VI makes treaties part of the supreme law of the land, so the text does not unquestionably commit termination to the President alone.144 That opinion treated the question as one of ordinary constitutional interpretation and rejected the idea that every case touching foreign relations lies beyond judicial reach, the same point the Court later made in Zivotofsky v. Clinton.145 It added that if Congress, by appropriate formal action, had challenged the President's authority, the Court would have a duty to decide the issue.146 The two rationales leave different futures open: a political-question dismissal closes the courthouse to the subject, while a ripeness dismissal suggests that a suit brought after formal action by Congress itself, rather than by individual members, could be heard.
Termination should not be confused with a conflict between a statute and a treaty, and the Taiwan treaty supplies both situations. Suppose that while the 1954 treaty was in force, Congress approved and the President signed a bill providing that under no circumstances may American armed forces be used to defend Taiwan. The bill does not formally end the treaty, but to the extent the two conflict, the statute, as the more recent enactment, prevails in United States courts, although this de facto repudiation may have repercussions for the United States under international law.147 The two questions get different answers. Whether the President may end a treaty by notice asks which branch holds the power to act, and courts have declined to answer it. Which of two conflicting laws governs a case asks only which was adopted last, and courts answer that by the calendar. Congress can thus effectively end a treaty's operation in domestic courts by enacting a contradictory statute, even while the question of formal termination stays open.148 The usual exam error is to apply the later-in-time rule to a President's announcement of withdrawal, or to call a conflicting statute a termination, because each move mixes the two questions together.