Concept
Limits on Congress's Power
Congressional power to define and limit
Limits on Congress's Power
The preceding sections asked how much control Article III gives Congress over the federal courts' dockets, and they left that question partly open: the Exceptions Clause may be plenary, or it may be bounded by a requirement that some federal court stay available, and the Supreme Court has not chosen between those readings. A second question can be answered with more confidence, and it is the subject here. Whatever the reach of the exceptions power and of the power to constitute lower courts, each is an enumerated power, and every power the Constitution confers has limits.1 Those limits take three forms. The power must be exercised within its defined scope, which is why the authority to make exceptions to appellate jurisdiction cannot be used to make exceptions to original jurisdiction. It must respect the structure of the Constitution, especially the separation of powers, which is why Congress cannot use that authority to vacate a decision the Court has already rendered. And it must respect external guarantees such as those in the Bill of Rights, which is why a statute excepting from the Court's jurisdiction every case brought by members of one religious denomination would violate the First Amendment religion clauses.2
The reason these limits deserve a section of their own is that the theoretical debate remains unresolved. The Court has not recognized any Article III restriction on jurisdiction stripping in express terms, McCardle admits of a narrow reading, and the modern Court appears divided about what the decision means, so the scope of the power is still fluid.3 An answer that rests on only one reading of the Exceptions Clause is therefore exposed to the other, while an answer that rests on limits holding under either reading is not.4 The limits also reach beyond the Supreme Court. Congress's authority over the lower federal courts comes from its power to constitute inferior tribunals, and that power is subject to the same scope, structural, and external boundaries as the exceptions power, so the limits collected here apply to Supreme Court and lower-court jurisdiction alike.5
The five subsections below take these questions in the same order, and each gives its question the rule, the reason, and a matched pair of examples. The first of them, Leaving Some Avenue Open, builds directly on the narrow reading of McCardle developed earlier in this chapter.
Leaving Some Avenue Open
The first question to put to a statute that narrows federal jurisdiction concerns forums: does the statute close one route to the courts, or every route? The answer is drawn from McCardle. Congress has broad power to make exceptions to the Supreme Court's appellate jurisdiction, including the power to repeal a particular statutory avenue of review, so long as it does not eliminate all avenues of Supreme Court review or all federal judicial review of constitutional claims.6 The permission half of that rule is the part the reader has already seen at work, since the 1868 repeal ended McCardle's appeal by withdrawing the single statutory route he had used and left the Court's habeas jurisdiction under the Judiciary Act of 1789 in place.7 The limiting half is the subject here, and it asks whether a constitutional claim can still reach a federal court somewhere after the statute takes effect.
The two halves rest on different reasons. The permission rests on the text of the Exceptions Clause, which confers the power in express words, together with the fact that a statute withdrawing one route leaves the claim reachable, so the federal courts keep their essential role in the constitutional plan of exercising judicial review.8 The limit rests on what judicial review is for. The Constitution restrains Congress along with every other actor, and if Congress could close every forum to a claim that it had violated a constitutional provision, it would decide for itself whether its own limits are ever enforced. Closing the courts to claims of constitutional violation therefore effectively denies the substantive right, and the judicial-guardian theory of Marbury v. Madison treats judicial review as the tool that keeps the majoritarian branches from overrunning constitutional limits.9
In practice the limit has rarely needed enforcing, because Congress usually leaves a forum and the Court reads restrictions so as to keep one open. The Court has not directly faced a constitutional challenge to a broad restriction on its appellate jurisdiction. Its strategy has been to read statutory restrictions narrowly and to reaffirm that alternative sources of appellate jurisdiction remain.10 Yerger, decided within about a year of McCardle, is the model. The repeal reached only appeals under the 1867 act, so the Court held that it could review the lower court's habeas decision through the original writ of habeas corpus, which the Court issues directly and which the Judiciary Act of 1789 continued to supply, and the writ served as the functional equivalent of the appeal Congress had repealed.11
The same approach reappeared more than a century later in Felker v. Turpin. Congress had restricted the Court's authority to review a federal appellate court's denial of leave to file a second or successive habeas petition, and the petitioner argued that the restriction unconstitutionally limited the Court's appellate jurisdiction. The Court avoided that question by emphasizing that appellate jurisdiction remained available through the original writ, and it upheld the withdrawal because it retained jurisdiction over original habeas petitions.12 The Court said that its narrow reading was informed by a desire to avoid the serious constitutional question that would arise if Congress repealed all appellate authority, and a concurring opinion observed that such stripping could interfere with the Court's constitutional supremacy.13
The two decisions share a surviving avenue to the Court, and they differ in the use the avenue served. In Yerger the original writ let the Court take up the prisoner's case after the repeal, and in Felker it let the Court sustain the restriction itself, because the statute shut one door and left another. Neither decision tested a statute that shut every door, and the Court has not directly faced one, so the outer edge of the rule rests on avoidance reasoning and constitutional principle.14
Congress's ordinary practice also leaves a forum. Measures that reroute a category of cases, such as the one in Lockerty v. Phillips giving the Emergency Court of Appeals exclusive jurisdiction over suits attacking wartime price-control regulations, change where the claim is heard and keep review available, since appeals there ran directly to the Supreme Court.15 When Congress strips only federal courts, state courts ordinarily remain open to federal claims, because federal jurisdiction is exclusive only where Congress has made it so, expressly or by fair implication.16 Whether a state forum alone satisfies the limit is contested, since the plenary reading accepts it and the mandatory readings taught earlier ask for an Article III court.
Where no forum remains, the limit has two independent sources. The first is the Suspension Clause: Article I, section 9 forbids suspending the writ of habeas corpus except under very limited conditions, so Congress would offend that provision if it closed all federal courts to certain habeas claims.17 The second is due process, and two decisions mark its edge. In Yakus v. United States the Court found no due process violation when district courts were barred from hearing, in criminal prosecutions, defenses attacking price-control regulations, because the defendants could have protested to the Price Administrator and, if refused relief, sued in the Emergency Court of Appeals.18 In Battaglia v. General Motors Corp., Congress retroactively abolished a burden the Court had read into the Fair Labor Standards Act and purported to remove from all federal and state courts jurisdiction to enforce the original statute, yet the Court reached the merits, reasoning that Congress's power to limit federal jurisdiction fails when it clashes with due process, and it found no violation in the repealer itself.19 A bar on one forum survived in the first case because another forum was available, and a bar on every forum gave way in the second to the guarantee at stake.
One fact separates the two versions, which is whether any forum remains, and that fact decides whether the statute falls inside the permission McCardle supplies or outside it.
The two halves of the limit are not equally firm, and an answer should say so. The firmer half protects federal judicial review of constitutional claims as such: the repeal in McCardle left habeas jurisdiction standing, and a statute that leaves no jurisdictional basis for any federal court to hear a claim is one that the decision cannot be said, with certainty, to validate.20 The half protecting Supreme Court review specifically is less settled. One account holds that removing every appellate avenue to the Supreme Court would only make the lower federal courts the final decision makers, which is less unusual than it sounds because limited certiorari review already makes the courts of appeals the practical last word in most cases.21 Against that account, the essential-role theory would forbid depriving litigants altogether of access to the Supreme Court in cases involving the supremacy of federal law, though it has little direct authority, and the avoidance reasoning in Felker shows that the Court takes the concern seriously.22 Because even adherents of the plenary view accept that Congress may not violate other constitutional provisions in controlling jurisdiction, the rule holds whichever reading of the Exceptions Clause an answer adopts.23
Answers go wrong when they stop at the observation that Congress may strip jurisdiction and never ask whether any forum remains. That question decides whether McCardle supports the statute at all, so it belongs at the start of the analysis. A statute that clears it can still fail, as the next limit shows, because leaving a forum open does not allow Congress to dictate the outcome of a case that forum will hear.
The Klein Limit
The second question in the checklist asks whether a statute withdraws jurisdiction as a device for dictating how pending cases come out. It matters even after a forum has been found, because a statute can leave a federal court fully open and still tell that court what to decide. The Constitution gives the legislature no such power, since the branch that makes law and the branch that decides cases must stay distinct.25 Earlier in this chapter, United States v. Klein appeared only as a foil to McCardle; here it is the subject. The rule it supplies is the Klein limit: Congress may not decide the merits of a case under the guise of limiting jurisdiction.26
The facts show how the statute worked. Congress had allowed owners of property seized during the Civil War to recover it on proof that they had not aided the rebellion, and the Supreme Court had held that a presidential pardon was enough to supply that proof.27 Klein won in the Court of Claims on the strength of a pardon, and the government appealed. While the appeal was pending, Congress added a proviso with three parts: pardons were not to be considered on appeal, a pardon accepted without disclaiming the acts it recited was conclusive evidence that the recipient had committed them, and whenever a Court of Claims judgment rested on a pardon without other proof of loyalty, the Supreme Court would have no further jurisdiction and had to dismiss the case.28 The proviso thus converted the same pardon the Court had treated as proof of loyalty into proof of disloyalty, and it ended jurisdiction at the moment the pardon appeared.
The Court began by granting everything Congress could rightly claim. The legislature has complete control over the organization of the Court of Claims and may confer or withhold appeals from it, and a statute that simply denied appeal in a particular class of cases would have been an exercise of the power to make exceptions, which the Court would have been bound to enforce.29 The proviso, however, withheld appellate jurisdiction only as “a means to an end,” and the end was to deny pardons the effect the Court had already adjudged them to have.30 The structure of the proviso gave that purpose away. The Court had jurisdiction of the cause “to a given point,” and once it found that a pardon existed, its jurisdiction was to cease and it was required to dismiss.31 A jurisdictional rule that switches off at the moment the court learns which way settled law points operates as an instruction about the outcome, and the Court asked as much: “What is this but to prescribe a rule for the decision of a cause in a particular way?”32 It also asked whether it could comply without “allowing one party to the controversy to decide it in its own favor.”33 The proviso applied “because and only because” the Court's decision under settled law “must be adverse to the government and favorable to the suitor,” and the Court concluded that Congress had “inadvertently passed the limit which separates the legislative from the judicial power.”34
The decision rests on two separate separation-of-powers defects, and each has its own logic. The first is the usurpation of a judicial function. Legislation makes rules that govern cases generally, and courts decide which side wins under those rules. If an instruction to rule for one side could be relabeled a withdrawal of jurisdiction, the label would defeat every other limit on legislative power, and a government litigant would be writing the rule for its own case, with an obvious temptation to do so whenever settled law runs against it. The second defect is the encroachment on the President. The Constitution entrusts the pardon power to the Executive alone, and the Court reasoned that the legislature cannot change the effect of a pardon any more than the executive can change a law, yet the proviso required courts to disregard pardons granted on conditions that had been fulfilled and to treat them as evidence of guilt.35 The Court therefore invalidated a statute that tried both to dictate the result in pending cases and to limit the effect of a pardon.36 Because most jurisdictional statutes have nothing to do with a pardon, the rule-of-decision defect is the one an exam problem will usually test.
McCardle and Klein both involved statutes that ended pending appeals by withdrawing jurisdiction, so the difference between them is the lesson. The McCardle repeal closed the 1867 appeal route to the government and to the private party alike, left the Court's other habeas jurisdiction in place, and told the Court nothing about how any habeas case should be decided.37 The Klein proviso ended jurisdiction only in the one situation where the Court's own precedent compelled a ruling against the government, and it paired that trigger with a command to treat the pardon as proof of disloyalty.38 Neutrality is therefore the working test: a limit that applies evenly to both sides helps assure that Congress is not trying to alter the results of specific cases, and a limit tied to the outcome lacks that assurance.39 Congress's motives were not the problem in Klein, since the Court attributed the proviso to inadvertence, and what failed was the command the proviso gave the courts, visible on its face.40
The rationale of the opinion has puzzled scholars, and it can be read in several ways: as a bar on reversing a decided result, as protection for the pardon power, or as a bar on prescribing a rule of decision in a pending case. The Court itself stressed the last of these. At a minimum the case holds that Congress cannot simply denominate legislation as jurisdictional and avoid constitutional scrutiny.41 One account adds that, unlike the McCardle repeal, the proviso closed every avenue of relief, so that it denied the claimants' substantive rights.42 Students lose this issue when they treat a jurisdictional label as the end of the analysis.
The limit is narrow, and the opinion itself marks its edge. In Pennsylvania v. Wheeling Bridge Co., the Court had decreed that a bridge was a nuisance to be abated, Congress then legalized the structure and made it a post-road, and the Court denied a motion for process to enforce the decree because the bridge had ceased to be a nuisance through Congress's exercise of its constitutional powers.43 The Klein Court left that decision undisturbed because no arbitrary rule of decision had been prescribed there: the court was left to apply its ordinary rules to the new circumstances the act created. In Klein no new circumstances had been created, and the court was directed to give evidence an effect precisely contrary to the one it would otherwise have.44
The modern decision on the permitted side of the line is Bank Markazi v. Peterson. Terrorism victims held unpaid judgments against Iran, and a 2012 statute made specified assets held in New York for Iran's central bank available to satisfy them, identifying the pending enforcement proceeding by its docket number. The Court held, by a vote of six to two, that Congress may enact new legislation that changes the outcome of a pending civil case, and that the prohibition in Klein does not apply because a statute does not trespass on judicial power “when it directs courts to apply a new legal standard to undisputed facts.”45 Both decisions rest on a familiar practice: courts apply the law as it stands when they decide, and so long as a decision has not become final on appeal, Congress's authority to change the governing rule remains intact.46 The test therefore asks what Congress changed. A statute that supplies new law, which the court then applies by its ordinary methods, is legislation, whereas a statute that leaves the law where the courts put it and dictates how that law applies to the facts of a pending or decided case directs the judgment itself.47
The second step does the work that separates Klein from McCardle, since it asks the neutrality question in a form that can be answered from the text of the statute. The first step marks the boundary drawn by Wheeling Bridge and Bank Markazi, where a change in the law left the courts with real work to do. The fourth step looks ahead, because the protection against a rule of decision covers cases still pending, and the next subsection takes up the separate rule that shields a judgment already final.
Final Judgments Are Beyond Reach
The Klein limit protects cases that are still pending from a legislative instruction about how they must come out. A second separation-of-powers rule protects cases the courts have already finished, and it turns on one variable, which is whether the judgment is final. Congress may change the law that governs pending and future cases, but it may not require a federal court to reopen a judgment that has become final.49 A judgment is final for this purpose when the court has decided or dismissed the case and no appeal remains available, so the matter stays closed unless a higher court in the Article III hierarchy reviews it.50
The line falls at finality because of what the judicial power consists of. The federal judiciary has the "power not merely to rule on cases, but to decide them," and a final judgment is the point at which the deciding is done.51 A case still on appeal has not reached that point, and courts apply the law in force when they decide, so a statute that changes the governing rule before the decision is final hands the court a new rule to apply and takes no decision away from it.52 A command to reopen a final judgment works differently, because it means the legislature has reviewed a court's decision and found it wanting, and the Constitution leaves that kind of review to a superior court within the Article III hierarchy.53 The incentive problem points the same way. If Congress could revive lost cases, every litigant who lost under a rule that Congress later disliked would have reason to petition the legislature instead of accepting the judgment, and no outcome in any federal court would be safe from reconsideration.
The decision that fixes the rule is Plaut v. Spendthrift Farm, Inc. The Supreme Court had held that certain securities fraud actions were subject to a one-year statute of limitations, and lower federal courts followed that holding by dismissing pending cases and entering final judgments against the plaintiffs, none of which were appealed.54 Congress answered with legislation that lengthened the limitations period and also gave the dismissed plaintiffs a new window in which to refile their cases.55 The Court held that the reopening violated both judicial finality and the separation of powers, and it observed that Congress, by commanding the federal courts to reopen final judgments, had violated a fundamental principle.56 The legislation therefore had two parts that came out differently. Congress was free to alter the limitations period for all pending and future cases, but it could not require courts to revive the cases that were already closed.57
The pair of outcomes isolates the fact that matters. A plaintiff whose securities suit was still pending when the longer period took effect gets the benefit of it, because the court applies the law as it stands when the case is decided. A plaintiff whose suit was dismissed and whose time to appeal had run gets nothing, because the court has already decided that case. The first result matches the freedom the Court recognized in Bank Markazi v. Peterson, where the enforcement proceeding was still pending and Congress could therefore change the law that governed it, as the previous subsection explained.58
Injunctions are the recognized exception in practice, and the reason is built into what an injunction is. A court that orders ongoing conduct keeps supervising the order, and it may modify the decree when circumstances change, including a change in the underlying law.59 In Pennsylvania v. Wheeling & Belmont Bridge, the Court had ruled that a bridge should be torn down as an impediment to navigation, Congress then passed legislation confirming the bridge's legality, and the Court treated the new statute as requiring the injunction to end.60 The Court also upheld parts of the Prison Litigation Reform Act in Miller v. French, because the statute left the courts in control of how far changed circumstances required existing decrees to end.61 A money judgment is the contrast case. It is a completed decision, so a later-enacted federal law cannot affect the plaintiff's right to those damages.62
This rule is the fourth step of the Klein checklist from the previous subsection, and it completes the separation-of-powers picture. Congress may not dictate the outcome of a pending case under the label of jurisdiction, and it may not undo the outcome of a case that is already final. Answers go wrong when they stretch Congress's freedom to change the law for pending cases into a power to revive cases already dismissed with no appeal left, so the first question to put to any retroactive statute is whether each affected judgment is final or still open.
Constitutional Rights Limits
The first three questions in the checklist concern which forum remains, whether the statute dictates an outcome, and whether it reaches a final judgment. The fourth question leaves the structure of Article III behind and asks whether the statute collides with a guarantee found elsewhere in the Constitution. This is the last of the three kinds of limit that apply to any enumerated power: the power must stay within its defined scope, it must respect the structure of the Constitution, and it may not transgress external guarantees such as those in the Bill of Rights.64 A jurisdictional statute can pass the scope question by being a true exception to appellate jurisdiction, and it can avoid the Klein and Plaut problems, and still fail here, because a guarantee in the Bill of Rights restrains Congress whatever the statute is called. The reason follows from what the label would otherwise accomplish. If describing a measure as jurisdictional were enough to put it beyond a constitutional guarantee, Congress could achieve through the docket what it is forbidden to achieve by direct regulation, and every right would depend on how skillfully a statute was drafted. For that reason even those who accept the plenary reading of the Exceptions Clause hold that Congress may not violate other provisions of the Constitution in controlling jurisdiction.65 The same reasoning supports the rule that Congress cannot deny jurisdiction on any basis that would itself violate an individual constitutional right, since closing the courts to claims of constitutional violation effectively denies the substantive right.66
The clearest violations are the direct ones, where the statute itself does what a guarantee forbids. A jurisdictional statute that excepted from the Supreme Court's jurisdiction every case brought by members of a particular religious denomination would violate the First Amendment religion clauses.67 The same is true of a statute that barred Supreme Court review by excluding litigants because of their race or political beliefs, since legislation that singles out classes of litigants on the basis of race or another suspect classification is presumptively invalid.68 These cases are easy because the statute classifies people by the very characteristic the guarantee protects, and moving the classification into a jurisdictional provision changes only the forum in which the discrimination operates. The one question the plenary view asks, whether the limitation is an exception to appellate jurisdiction, is answered yes, and the statute falls at the next step anyway. The Court is quite likely to strike down an exception that directly and unequivocally violates a provision of the Bill of Rights.69
Rules of general application sit at the other pole. Congress may provide that district courts lack diversity jurisdiction when $75,000 or less is at stake, and it may bar the Supreme Court from granting certiorari petitions in civil cases that are filed out of time.70 Each rule applies to every litigant alike, sorts no one by a protected characteristic, and burdens no particular right, so no external guarantee is implicated. The variable that separates these rules from the religious-denomination statute is whether the line the statute draws follows a trait or an interest that the Constitution protects.
The harder cases are the indirect ones, in which the statute commands nothing forbidden but makes it harder to enforce a right that already exists. The standard illustrations come from the school prayer and abortion controversies, which have produced repeated proposals to withdraw whole categories of constitutional cases from the federal courts.71 Consider a statute providing that the Supreme Court shall have no jurisdiction to review, by appeal, certiorari, or otherwise, any case arising out of a state law that proscribes or regulates abortions. The statute does not itself restrict anyone's access to an abortion, so it directly violates no constitutional prohibition. Yet by insulating challenges from Supreme Court review, it may make it significantly more difficult to challenge state laws that themselves impose undue burdens, and on that ground the statute might impose an undue burden of its own.72 The argument presupposes that the underlying right exists in the form the example assumes. Where the Court recognizes no such right, a jurisdictional statute has nothing to burden, which shows how much of the indirect analysis rests on the right itself. A jurisdiction-stripping measure in the school prayer setting presents the same uncertainty. A Justice with a broad conception of the exceptions power could treat it as a legitimate exception, premised on the violation of no one's rights, while a Justice with a narrower conception might find a structural defect, because the statute undermines the Court's ability to carry out its essential function, and an external defect, because it promotes a religious practice in violation of the Establishment Clause.73
The Court's response to indirect cases is uncertain because two unresolved questions feed the answer. The first is how broadly the Justice reads the exceptions power, which the earlier discussion of the plenary and mandatory readings left open. The second is how the Justice regards the underlying right, since a Justice more concerned to protect a right is more likely to find that a jurisdiction-stripping measure violates it.74 The Court has not recognized any Article III restriction on jurisdiction stripping in express terms, so neither question has an authoritative answer.75 The practical concern behind the uncertainty is that constitutional rights may be substantially diluted if only state-court judges can vindicate them, because those judges almost universally serve limited terms and face some form of popular election.76 A related and contested question is whether sending classes of issues, rather than classes of litigants, to state instead of federal courts is as vulnerable as sorting litigants by race.77 Evenhandedness in the sense taught earlier does not settle these cases, because the school prayer and abortion statutes close the Supreme Court to every litigant in the category and still raise the rights question.
Due process supplies the clearest example of a court taking these limits seriously, in the litigation that followed the Portal-to-Portal Act. In the mid-1940s the Supreme Court held that time spent in work that had never been paid for, such as underground travel in iron ore mines, was compensable under the Fair Labor Standards Act, and the ruling created immense retroactive liability for many businesses.78 Congress responded by amending the statute to make that work noncompensable and by withdrawing the jurisdiction of the courts over cases pending or later brought under the old statute.79 Employees argued that the jurisdictional limit took their property without due process or just compensation. The Supreme Court never ruled on the question, and the lower courts that heard it all upheld the limit, but several dismissed the constitutional argument in summary fashion because the law was drafted as a jurisdictional limitation.80 The Second Circuit in Battaglia v. General Motors Corp. examined the argument on its merits and found no taking, because the employees held no vested property right to overtime pay until their claims were reduced to judgment, and a court applies the law in force at the time of appeal.81 By considering the claim and then rejecting it, the court explicitly recognized that congressional power over the jurisdiction of the courts is limited by the due process clause.82 Its reasoning, that the power to limit federal jurisdiction fails when it clashes with due process, applies equally to other constitutional provisions.83
The decision turns on the same line of finality that decided Plaut in the previous subsection. A claim that has been reduced to a money judgment and collected is vested property, and Congress may not use a jurisdictional limitation to avoid its duty to pay just compensation or its duty not to take property without due process.84 A claim still pending is an expectancy that a change in the law may extinguish.
On an exam the direct and indirect cases are best kept apart. A statute that sorts litigants by religion, race, or political belief, or that itself violates a particular guarantee, is likely to fall, and the answer can say so with confidence. A statute that withdraws review of a category of cases and thereby makes an existing right harder to enforce raises a real question, so the answer should set out both lines of argument, as with the plenary and mandatory readings, and should avoid presenting the school prayer and abortion statutes as settled either way. Because external guarantees restrain every congressional power, the same analysis applies to a statute narrowing the lower federal courts' jurisdiction as to one narrowing the Supreme Court's.85 A final practical question remains for statutes that never mention constitutional claims at all: how should a court read language that is unclear about whether Congress meant to cut off review of them? That question is the subject of the final subsection, Clear Statement and Avoidance.
Clear Statement and Avoidance
The first four questions in the checklist all assume a statute that plainly narrows federal jurisdiction. The fifth question concerns a statute whose words leave it unclear whether Congress meant to reach constitutional claims at all, and the Court's answer to it explains why the debate between the plenary and mandatory readings so seldom has to be resolved. Whichever reading of the Exceptions Clause is correct, the Court will not lightly assume that Congress intended to limit the authority of the Supreme Court or the lower federal courts to hear constitutional claims, so the working principle is the clear-statement rule: where Congress intends to preclude judicial review of constitutional claims, its intent to do so must be clear.87 The practical consequence is that a court facing an ambiguous restriction usually decides the case by construing the statute, and the choice between the traditional and mandatory interpretations is never reached.88
The rule has the shape it does for reasons that can be read off its structure. A narrowing construction decides only what Congress said, so it leaves unanswered the constitutional question of how much Congress could take away, a question on which no text, history, or precedent supplies a definitive answer.89 Requiring clarity also places the burden on the branch that wants review cut off. A bar on constitutional claims then results from a deliberate legislative choice, made in words that legislators and the public can read, and not from a general phrase drafted with other disputes in mind. Finally, a narrowing construction lets the Court describe the jurisdiction it keeps as jurisdiction Congress left in place, so the Court never has to claim authority to override a congressional limit.
Demore v. Kim shows the rule at work on a general bar. A federal statute authorized the Attorney General to detain without bail certain lawful permanent residents during deportation hearings, those found deportable because of convictions for specified crimes, and it precluded judicial review of the Attorney General's decision to detain them.90 The question was whether a bar on review of that decision also closed habeas corpus review, and the Court held that it did not, because the clear text of the statute disclosed no intention to reach habeas.91 The Court relied on the principle, drawn from Webster v. Doe, that "where Congress intends to preclude judicial review of constitutional claims its intent to do so must be clear."92 The demand rises when the claim arrives through habeas, because precluding habeas review requires an especially clear statement of congressional intent.93
The same technique appears in INS v. St. Cyr and its companion Calcano-Martinez v. INS, where the Court construed a purported jurisdiction-stripping measure about deportation proceedings so as to leave open the possibility of federal-court habeas review.94 The pressure behind such readings is easy to see in the habeas setting. If state courts were also unable to hear habeas petitions on behalf of federal detainees, as the government might argue under Tarble's Case, a restriction nominally addressed to federal courts could operate to foreclose all judicial review, and that prospect pushes federal courts to find review under some alternative jurisdictional provision.95 In general the Court has struggled to avoid the conclusion that Congress has acted to foreclose all judicial review.96
Felker and Hamdan v. Rumsfeld apply the technique to the Supreme Court's own docket, and they differ in what the narrow reading rested on. In Felker v. Turpin Congress had restricted review of a court of appeals' denial of leave to file a second or successive habeas petition, and the Court avoided the constitutional challenge by emphasizing that appellate jurisdiction remained available through the original writ, a narrow reading it said was informed by a desire to avoid the serious constitutional question that would arise if Congress repealed all appellate authority.97 Hamdan turned instead on whether the statute reached pending cases at all. Salim Hamdan challenged the military commission system, the D.C. Circuit rejected his claims in 2005, and the Supreme Court granted certiorari before the Detainee Treatment Act was adopted that December.98 The Act contained a sweeping restriction on habeas authority, but its timing provisions pointed in different directions: one said the statute took effect on enactment, another said certain aspects applied to claims pending on that date, and some legislative history suggested that members of Congress saw the exception as preserving jurisdiction over some pending challenges.99 The government moved to dismiss in what amounted to a replay of McCardle, and the Court held that its appellate jurisdiction remained intact, explaining that ordinary principles of interpretation sufficed to answer the government's argument, at least as to pending claims.100 Because those principles were enough, the Court had no occasion to rest on the constitutional question, though it did not rule out using avoidance norms in a different setting, and it went on to decide the merits.101
The McCardle repeal stands at the opposite pole. The 1868 act repealed in terms the provision of the 1867 act affirming Supreme Court jurisdiction over habeas appeals, and the Court observed that it was "hardly possible to imagine a plainer instance of positive exception."102 Nothing in that language was ambiguous, so there was no narrowing reading to adopt, and the Court went straight to the question of Congress's power, which it found given by express words.103 Congress knows how to supply that clarity. In 2006, after the decision in Hamdan, the Military Commission Act restated and expanded the habeas restrictions and declared that it would "apply to all cases, without exception, pending on or after the date of the enactment of this Act."104 A statute that plain leaves no narrowing construction, and the constitutional question that avoidance had postponed must then be faced under the limits taught earlier in this chapter.
One fact separates the two versions, which is whether the text says that constitutional claims are covered. The first version leaves the constitutional question unasked, and the second presents it in full. That is why the rule gives a litigant a real but limited advantage: it can win a case on the statute's wording, and it cannot protect a claim that Congress has plainly excluded unless one of the limits taught above applies.
On an exam, this rule belongs at the start of the analysis of any statute that narrows review, because it often ends the analysis early. Ask whether the statute plainly says that it reaches constitutional claims. If its language is general or ambiguous, and especially if habeas is involved, read it narrowly, say that the constitutional question is therefore not reached, and note any alternative route that remains open. If the statute is express, as the repeal in McCardle was, move on to the competing readings of the Exceptions Clause and the limits that survive under either. Answers go wrong when they describe the rule as a ban on jurisdiction stripping, because it governs how the Court reads a statute and says nothing about whether Congress may enact one: Congress may withdraw jurisdiction, but when constitutional claims are at stake it must say so unmistakably.105