Facial Challenges, Saving Constructions, and Statutory Severability

Article - Volume 99 - Issue 2

The doctrines that license “facial challenges” to the constitutionality of statutes are widely misunderstood. So are the two leading devices for limiting facial challenges’ potentially wrecking-ball effects: narrowing or saving constructions and severability doctrine.

This Article advances entwined theses about facial challenges, narrowing constructions, and statutory severability. Although the Supreme Court long maintained otherwise, facial challenges are commonplace. Besides being mandated by such familiar constitutional tests as strict judicial scrutiny, they reflect the nature of many constitutional rights. Given the pervasiveness of facial challenges, narrowing constructions and severance of otherwise-invalid statutes both play vital roles in preserving a myriad of statutes from total invalidation. But they also invite questions about whether courts impermissibly “rewrite” statutes in order to save them—or, conversely, about whether courts wrongly refuse to salvage statutes that they dislike for policy reasons.

Successful responses to these challenges require distinctions that the Supreme Court too often fails to observe, possibly due to confusion. Saving constructions are statutory interpretations, properly disciplined by canons of construction and theories of interpretation. In contrast, statutory severance occurs following a determination that a statute, as properly interpreted, is constitutionally invalid. Severing an invalid statute thus requires judicial agency, not interpretation, but agency that is restricted by separation-of-powers principles that this Article delineates.

This Article’s prescriptions concerning facial challenges, saving constructions, and statutory severance have neither a liberal nor a conservative valence. The Article’s analysis does, however, show the necessity for courts to act as sometime-helpmates to Congress in making statutes workable by severing them. The courts’ necessary role in severing statutes illuminates the inadequacies of some, though not all, textualist theories of statutory interpretation.


I. Introduction

After decades of denial, the Supreme Court has begun to acknowledge that facial challenges to statutes constitute the norm, not the exception, among constitutional cases on its docket.[1] As recently as 1987, United States v. Salerno[2] insisted that to succeed with a facial challenge, “the challenger must establish that no set of circumstances exists under which [a challenged statute] would be valid”[3]—a stringent though not always impossible standard to meet. Justice Scalia subsequently maintained that adherence to the Salerno rule precluded courts from facially invalidating anti-abortion statutes that could have any valid applications.[4] Roughly two decades after Salerno, in Sabri v. United States,[5] Justice Souter’s Court opinion still maintained that “facial challenges are best when infrequent.”[6] Quoting an earlier decision, Justice Souter asserted that “[f]acial adjudication carries too much promise of ‘premature interpretatio[n] of statutes’ on the basis of factually barebones records.”[7] Justice Souter acknowledged a “relatively few settings” in which the Court had “recognized the validity of facial attacks,”[8] including cases involving freedom of speech, the right to travel, abortion, and legislation enacted under Section Five of the Fourteenth Amendment.[9] But “[o]utside of these limited settings,” he wrote, “we do not extend an invitation to bring overbreadth claims.”[10]

Jeremiads against facial challenges have largely vanished in recent years,[11] perhaps as the Justices have come to recognize that not all facial challenges involve “overbreadth,”[12] a basis for challenge to legislation that the Supreme Court had developed under and sought to limit to the First Amendment.[13] Instead, the Justices have begun to confront, or at least accept, the reality that the “tests” of constitutional validity that dominate much of contemporary constitutional law frequently call for statutes to be judged on their faces.[14] A few examples should suffice to make the point. The strict scrutiny test asks whether statutes are narrowly tailored to compelling governmental interests.[15] The intermediate scrutiny formula also applies to statutes, not statutory applications.[16] So, even, does the rational basis test inquire into whether statutes are rationally related to legitimate state interests.[17] The Court has similarly promulgated tests for whether statutes are valid under the Commerce Clause[18] and Section Five of the Fourteenth Amendment.[19] In the domain of the First Amendment, the Court sometimes inquires into overbreadth,[20] but just as frequently it hinges outcomes on whether statutes (not applications of statutes) are narrowly tailored to compelling government interests or satisfy other measures of constitutional validity.[21] One could go on and on.[22] The conclusion is inescapable: doctrinal tests that gauge the constitutional validity of statutes invite, authorize, and frequently require facial challenges. The mystery is how the myth that facial challenges are rare, disfavored, and nearly impossible to mount successfully persisted for as long as it did.[23]

Conventional wisdom long sought an answer in a “presumption of severability”[24] that holds, roughly, that even if a statute has invalid parts or applications, courts can separate those invalid parts or applications from the rest and enforce the valid ones. Given the presumption of severability, it was said, courts normally should deal only with the personal rights of the parties before them. But purveyors of the conventional wisdom never offered convincing explanations of how the presumption of severability interacts with constitutional tests that measure statutes’ facial validity.[25] Consider Craig v. Boren,[26] in which the Supreme Court applied the intermediate scrutiny test—under which statutes are invalid unless “substantially related” to “important” governmental interests[27]—to invalidate an Oklahoma law that prohibited young men but not young women from purchasing low-alcohol beer. In defense of the statute, the state argued that young men would be more likely than young women to drive while drunk if given access to alcohol.[28] In rejecting that defense, the Court held out no prospect that the statute might be valid as applied to those young men who fit the stereotype on which the state had based its challenged legislation.[29] Instead, the Court invalidated the statute as to all.[30]

The statute-based focus of the Court’s analysis in Craig, which has parallels in cases under a plethora of constitutional tests, is by no means a modern innovation. Facial challenges have long hidden in plain sight. The Supreme Court invalidated over 200 statutes during the Lochner era.[31] In doing so, it only rarely suggested that the presumption of severability—which I agree exists and requires explication—made facial invalidations inappropriate.

Perhaps not surprisingly, the fading of the old myth that facial challenges are rare and the emergence of a new regime have brought problems of their own. As the Supreme Court apprehended back when it insisted that facial challenges should be infrequent, facial challenges have the potential to operate as “wrecking ball[s].”[32] In understandable revulsion from invalidating too many statutes on their faces—even when familiar tests of constitutional validity threaten them with condemnation—the Court has turned repeatedly to two limiting devices. One is “narrowing” or “saving” constructions. The other is statutory “separability” or “severability,” even though severability is not nearly as ubiquitous as some once imagined. To the contrary, as we shall see, the Justices appear to worry increasingly that judicial severance of statutes offends the separation of powers by producing laws that Congress did not enact.[33]

In cases involving federal statutes, the Supreme Court applies federal standards in furnishing narrowing constructions and in separating invalid statutes.[34] In challenges to state provisions, state law controls,[35] but the Court frequently and perhaps typically assumes that state courts would follow the same approach as federal courts.[36]

As between the judicial functions in offering narrowing constructions and separating valid from invalid statutory parts or applications, the basic distinction should be clear. Courts consider narrowing constructions before ruling definitively that a statute is constitutionally invalid. When anticipating that a statute might fail an applicable test of constitutional validity if construed one way, judges inquire, pursuant to the canon of constitutional avoidance,[37] whether they can save the statute or avoid the looming constitutional question by interpreting it another way.

A textbook example of a narrowing construction comes from Edward J. DeBartolo Corp. v. Florida Gulf Coast Building & Construction Trades Council.[38] The case arose when the Trades Council distributed handbills urging shoppers to boycott DeBartolo’s mall until it pledged to utilize only construction contractors who provided “fair wages and fringe benefits.”[39] DeBartolo complained that the handbilling violated a provision of the National Labor Relations Act that forbade unions to threaten or coerce anyone engaged in commerce.[40] On appeal, the Supreme Court ruled that if the statute were construed to forbid peaceful handbilling that advocated lawful action, a serious question would arise about its constitutional validity under the First Amendment.[41] Under those circumstances, the Court held unanimously that it should construe the statute’s prohibition of threats and coercion as not encompassing the defendant’s conduct.[42] The Court stated the applicable rule of statutory interpretation as follows: where “an otherwise acceptable construction of a statute would raise serious constitutional problems, the Court will construe the statute to avoid such problems unless such construction is plainly contrary to the intent of Congress.”[43] Subsequent cases have clarified that for a narrowing or saving construction to be permissible, it must be a linguistically plausible one that does not involve a judicial rewriting of a statute.[44]

By contrast with the judicial role in providing narrowing constructions, courts sever or separate statutes after determining that a provision, as written, violates the Constitution. Even when no linguistically plausible saving interpretation is available, courts will determine whether a statute’s invalid provisions or applications can be severed such that what remains thereafter satisfies constitutional norms.

An example comes from Ayotte v. Planned Parenthood of Northern New England,[45] which I shall discuss extensively below. The Court first ruled that a statute that restricted minors’ access to abortions was invalid as properly interpreted and only then considered whether—and if so, how—the statute’s invalid applications could be severed so that the statute’s valid prohibitions could remain in force.[46] “Generally speaking, when confronting a constitutional flaw in a statute, we try to limit the solution to the problem,” the Court explained.[47]

The gap between saving constructions and severing of invalid applications might diminish if the Supreme Court adopted narrowing or saving interpretations only after definitively determining that a statutory provision would be invalid if interpreted one way rather than another. At one point in the past, it took this approach with the result that saving constructions—like severance of invalid statutory parts—came only after a judicial decision of constitutional invalidity.[48] But even if the Court were to revert to the old protocol, a significant difference between saving constructions and statutory severance would remain: the Court has often prescribed the severance of statutes—as in Ayotte—to reach results that could not be described as involving statutory “interpretation.” When a court severs a statute, it determines the appropriate judicial response after acknowledging that a statute cannot be saved by interpretation, understood as an effort to determine what a statute, as enacted by Congress, means.[49]

Despite clear differences, the concepts of narrowing constructions and statutory separability have both provoked controversy and misunderstanding in recent cases, sometimes alone and sometimes in their connections to one another. Among the confusions are these:

After having found that no narrowing construction of a statutory provision is available, the Court sometimes proceeds immediately to invalidate the provision on its face, without further separate consideration of whether it might be severed. For example, in Iancu v. Brunetti,[50] the Court, in the course of finding that a statutory provision impermissibly discriminated on the basis of viewpoint, rejected a proposed saving construction that would have cured the problem by narrowing the provision’s reach.[51] So far, so good. But then the Court failed to consider whether valid parts or applications could be separated from, and survive the invalidation of, constitutionally impermissible ones.[52] The Court offered no explanation for this apparent oversight.

In cases involving the permissibility of judicial severance of a statute, dissenting Justices sometimes protest that the Court has no authority to “rewrit[e]” a statute by separating and enforcing valid parts after striking down invalid ones.[53] The insistence that courts should not rewrite statutes defines an important limit on the judicial role in cases involving statutory interpretation or construction. But in cases involving statutory separation—which occurs only after a court has determined that a statute as properly interpreted violates the Constitution—the objection to rewriting is almost invariably misleading. In one sense, no separation of valid from invalid parts of a statute rewrites it: the court merely gives effect to the valid parts or applications that Congress enacted. In another sense, every judicial severing of a statute produces a statute different from the one that Congress enacted—namely, one that excludes invalid parts or applications. As we shall see, the courts’ actual role in severing statutes is sometimes more complex than either of these stylized views acknowledges. Nevertheless, characterizations of the judicial role in severing statutes as involving an impermissible “rewriting” prove too much insofar as they imply that courts should never sever statutes with invalid applications that Congress sought to prescribe.

In determining whether the separation of a statute is permissible, the Supreme Court concluded in Murphy v. NCAA[54] that it must ask whether, if Congress had known that one or more provisions would be invalid, it would nevertheless have enacted the provisions that are constitutionally unobjectionable.[55] After answering in the negative, the Court invalidated the challenged law—which included multiple prohibitions spread over multiple provisions—in its entirety. In other contexts, however, some of the same Justices who joined the Court majority in Murphy rejected speculations about the intentions and attitudes of members of Congress in voting for legislation and about what they would have done if apprised that an issue would subsequently arise.[56] Under textualist principles of statutory interpretation, the law is defined by what Congress enacted, not by what it wanted or would have wanted to achieve.[57] If valid, that premise should apply to cases involving the consequences of a statute’s partial invalidity as much as to any others.

In this Article, I argue that these and other confusions are correctable if the Justices would adhere to the logic of doctrines that they have shaped. Absent such adherence, the Justices are likely to vote disproportionately to facially invalidate statutes whose underlying policies they dislike and to salvage statutes of whose policies they approve. Robert Stern found this pattern in a classic study of separability published in 1937.[58] Analogues persist today, as prominently displayed in disputes such as those involving the Patient Protection and Affordable Care Act (ACA). In National Federation of Independent Business v. Sebelius,[59] the Court divided five to four both about the permissibility of a saving construction and about whether the ACA could be severed following the invalidation of one or more sections.[60] Four liberals contributed to the majorities on both of these issues.[61] Four conservatives rejected what Chief Justice Roberts framed as a saving construction of the ACA’s mandate that individuals must purchase health insurance[62] and, having done so, would have invalidated the ACA in its entirety. Only the Chief Justice broke what otherwise appeared to be ideological ranks with regard to either issue.

The Court is likely to confront the issue of the ACA’s severability once again, following a congressional amendment of its terms and a decision by the Fifth Circuit that a key provision of the amended version violates the Constitution,[63] in California v. Texas.[64] A division of the Justices along liberal/conservative lines—with the liberal Justices voting to sever and the conservatives holding the statute nonseverable—would surprise few.

But issues involving the availability of facial challenges, narrowing constructions, and statutory separability do not have a consistent liberal versus conservative valence. In challenges to statutes involving abortion restrictions, liberal Justices have favored facial invalidation and rejected demands for statutory severance, while judicial conservatives have mostly aligned on the side of severability.[65] By contrast, conservative Justices have upheld facial challenges over liberal Justices’ protests that statutes were valid as applied to the dispute before them in Citizens United v. Federal Election Commission,[66] which invalidated central provisions of the Bipartisan Campaign Reform Act,[67] and in Shelby County v. Holder,[68] which struck down an important section of the federal Voting Rights Act.[69] Liberals have sometimes also embraced severability arguments that conservatives rejected in challenges to legislation as beyond congressional power to enact under Section Five of the Fourteenth Amendment.[70]

The confusions that plague the Supreme Court’s treatment of facial challenges, narrowing constructions, and statutory severability are not isolated, random mistakes. Rather, they occur within an ongoing dialectic that must itself be understood if soundly consistent analysis is to become the norm.

That dialectic begins with facial challenges. Although facial challenges are increasingly accepted, the grounds for their inevitability are perhaps still not fully understood. As Part II elaborates, facial challenges are entailments not just of judicial tests but also of the nature of constitutional rights.[71] Nevertheless, unavoidable though facial challenges are, their potentially devastating effects leave courts with no practical alternative but to develop limiting devices. Such devices have always existed, but they have necessarily evolved as understandings of constitutional rights have shifted and increasingly invited facial challenges. Today, the two principal limiting devices are saving constructions and statutory severability, both of which pose issues about the permissibility of efforts by the courts to rescue statutes from invalidation. About each, one can ask: Is this a permissible function for courts under Article III and the constitutional separation of powers? In debates about statutory interpretation, it is a familiar refrain that courts should interpret statutes as Congress wrote them, not revise, improve, or repair them.

Among this Article’s central premises is that clearing up the confusions that surround saving constructions and statutory separability requires toggling back and forth between two levels of constitutional analysis. On one, we need an account of the judicial role under Article III that both recognizes the unavoidability of facial challenges and accepts the practical imperative—endorsed by historical practice—to prevent facial challenges from devastating too many statutes.[72] At another level, it is important to grasp the distinctive structures and implicit logic of the doctrines that the Supreme Court has evolved to limit the systemic harms that facial challenges otherwise would inflict. The Court has sown confusion by conflating narrowing or saving constructions with the separation of statutes’ invalid parts or applications. By disentangling the doctrines that govern saving constructions and statutory severability, I argue, we can achieve understandings of both that jointly meet the wrecking-ball threat posed by facial challenges and that individually reflect sound, sustainable understandings of the judicial role.

In the current state of debate and confusion, this Article’s most practically important takeaways may involve statutory severability, about which the Supreme Court has divided recurrently, and often confusedly, in recent Terms: When courts sever statutes, they are not interpreting them; the rules that limit statutory interpretation in the provision of narrowing constructions do not apply. But courts that sever statutes do not or should not rewrite them either. The judicial role is one of sometimes-creative preservation. We need understandings of Article III that simultaneously accept and limit this judicial function.

The Article’s argument develops as follows. Part II provides a further introduction to the concepts of facial challenges, saving constructions, and statutory separability. It also highlights issues that arise from these concepts’ interactions with one another. Part III probes and resolves some of the confusions that have surfaced in debates about facial challenges. Among Part III’s central conclusions is that the availability of a facial challenge should depend on the nature of the constitutional right at issue. Part IV closely examines the conceptual logic of narrowing constructions of statutes. Part V exposes, and offers proposals for dissolving, confusion concerning statutory separability. Some of those confusions result from conflating narrowing constructions with separability, as the Supreme Court sometimes does. Taken together, Parts IV and V reveal the considerations proper to narrowing constructions, on the one hand, and statutory separability, on the other, and demonstrate that neither, when properly conceived and executed, involves objectionable judicial rewriting of statutes. Nevertheless, Part V emphasizes, the severing of statutes sometimes requires judicial creativity and initiative. Part VI, which functions as a conclusion, highlights the ongoing dialectic between the imperatives to embrace facial challenges and to limit their potentially destructive impact. It argues that any sound theory of the judicial role must acknowledge the courts’ responsibility to separate valid from invalid statutory parts and applications. As an implication of that conclusion, Part VI also rejects theories of statutory interpretation whose foundational premises would preclude such a role, as some versions of textualism purport to do.

II. A Conceptual Primer

This Part offers a bare-bones introduction to the three legal concepts with which this Article is centrally concerned: facial challenges, saving constructions, and statutory separability or severance.

A. Facial Challenges

For approximately fifty years, the Supreme Court has self-consciously differentiated between as-applied and facial challenges.[73] The terms are impossible to make wholly precise, but a rough cut will suffice for current purposes. In an as-applied challenge, a party maintains that the Constitution forbids a statute’s application to his or her case. In contrast, a facial challenge asserts that a statute—or, more commonly, a provision of a multipart statute—exhibits a defect that renders it invalid as applied to all cases, even if a more narrowly (or occasionally a more broadly[74]) framed provision could have prohibited the challenger’s conduct.

In one important sense, all challenges to statutes begin as as-applied challenges.[75] To have standing and a claim to relief, a party must argue that the Constitution forbids the application of a statutory provision to his or her case. In another sense, however, many and perhaps most constitutional challenges to the application of statutes are necessarily facial challenges. In order to prevail, a party challenging the application of a statutory provision must give reasons why the statute cannot be applied to her.[76] Very often the reasons, if valid, will establish that a provision is unconstitutional on its face.

1. Constitutional Tests and Their Implications

The widespread availability of facial challenges flows from judicially prescribed tests of constitutional validity that dominate modern constitutional law. As explained in the Introduction, many of these tests—including strict judicial scrutiny and rational basis review—apply to statutory provisions as written. The Supreme Court’s sometime-protestations that facial challenges are disfavored to the contrary notwithstanding, a survey of the Court’s cases during its 2009, 2004, 1999, 1994, 1989, and 1984 Terms revealed that the Justices ruled on more facial than as-applied challenges in each.[77]

Even so, not all tests of constitutional validity are framed to determine whether statutes are permissible on their faces. For example, parties can always maintain that statutory provisions are unconstitutionally vague as applied to them.[78] That formulation implies that a provision that is vague as applied to one person’s conduct might apply clearly to another’s. The Supreme Court has also established that statutes that are not unconstitutional on their faces might remain subject to challenge as applied to particular individuals. For example, in Crawford v. Marion County,[79] the Court ruled that a statute requiring would-be voters to present government-issued photo identification survived a facial challenge,[80] but the plurality opinion left open the possibility that the photo-identification requirement might be unconstitutional as applied to voters who could demonstrate that it would impose special burdens on them.[81] In Crawford, the applicable test—under which “‘evenhanded restrictions that protect the integrity and reliability of the electoral process’. . . . must be justified by relevant and legitimate state interests ‘sufficiently weighty to justify the limitation’”[82]—was one that could be used to gauge the validity either of statutes on their faces or of statutes as applied to particular classes of individuals.

2. Successful Facial Challenges: An Important Ambiguity

It is also important to note an ambiguity in references to the facial invalidity of statutes. Such references typically involve linguistically separable provisions of statutes, rather than multipart statutes as packaged by Congress and signed by the President. For example, Coleman v. Court of Appeals of Maryland[83] involved a facial challenge to a single section of the multipart Family and Medical Leave Act that purported to abrogate the states’ sovereign immunity from suit for violating a provision that mandates up to twelve weeks of unpaid leave for an employee to deal with personal healthcare issues.[84] Although the Court held the challenged provision unconstitutional on its face, no one suggested that the entire Family and Medical Leave Act, which applies to both public and private employers and creates rights to unpaid leave for a variety of reasons, was therefore invalid.[85] To the contrary, the Court had previously upheld another provision against a sovereign-immunity-based facial challenge in Nevada Department of Human Resources v. Hibbs.[86] More generally, virtually no one thinks that a single invalid item in a piece of omnibus legislation should provoke invalidation of all components.[87]

Nevertheless, the Supreme Court occasionally concludes—as in Murphy v. NCAA, for example—that an entire statute should be invalidated if it determines that Congress would not have wanted some provisions to continue to operate if it had known that courts would invalidate others.[88] In these cases, a successful facial challenge to a statutory provision supports a further successful facial challenge to a package of provisions or even an entire statute.[89]

It would undoubtedly enhance the clarity of analysis if discussions of facial challenges distinguished rigorously between facial challenges to statutory provisions and facial challenges to entire statutes, the latter of which assert a further, broader claim that is predicated on the invalidity of one or more provisions. Absent rigorous distinction, most facial challenges—in standard legal usage of the term—involve facial challenges to discrete provisions of multipart statutes.

B. Narrowing Constructions

The same tests of constitutional validity that can result in statutory provisions being held facially invalid can also impel judicial searches for narrowing constructions. In typical modern cases, consideration of a narrowing or saving construction occurs pursuant to the canon of constitutional avoidance, which the Supreme Court applies to refrain from resolving serious constitutional questions.[90] Insofar as the terms “narrowing construction” and “saving construction” are treated as synonymous, the notion of a “saving” construction therefore encompasses not merely saving a statute from unconstitutionality, but also saving the Court from needing to decide a constitutional question.

Three further points about saving constructions deserve emphasis. First, saving constructions are interpretations of a statute as written. Accordingly, the Supreme Court routinely insists when entertaining proposed saving constructions that it cannot rewrite a statute under the guise of interpreting it.[91] Rather, the Court can only adopt an interpretation that possesses independent plausibility. In this respect, statutes that have received narrowing or saving constructions differ from severed statutes, the meanings of which necessarily diverge from the meanings of the statutes that had to be severed in order to be rescued from total invalidity.

Second, because saving constructions involve statutory interpretation, the acceptability of a proposed saving construction sometimes depends on contestable issues of interpretive methodology. These include such matters as the permissibility of reliance on legislative history and speculations about how the legislature, as a psychological matter, intended or would have intended a statute to be applied. Textualists ordinarily maintain that courts should pay no heed to such considerations.[92] Purposivists and intentionalists sometimes disagree.[93]

Third, saving constructions presuppose that statutes are subject to successful facial challenges. The entire point of saving constructions is to rescue statutes from facial invalidation.

C. Statutory Separability or Severability

Questions about statutes’ separability come into play only after a court determines that a challenged provision, as properly interpreted, either fails or would likely fail a test of facial validity. For example, imagine that a court concludes that a statute or provision that Congress enacted pursuant to Section Five of the Fourteenth Amendment is not congruent and proportional to an identified pattern of constitutional violations, as legislation enacted under Section Five must be.[94] Having reached that conclusion, a court may then proceed to consider whether parts or applications can be severed such that remaining parts or applications satisfy the test and are therefore enforceable.

Sometimes, moreover, it may suffice for a court to determine that a statute would be severable, and could validly be applied to the challenger, even if it were held unconstitutional as applied to another party or to other conduct in a future case. Albeit without adverting specifically to separability, the Supreme Court effectively followed this course in United States v. Georgia.[95] The Court granted certiorari to determine “[w]hether Title II of the Americans with Disabilities Act [ADA] . . . is a proper exercise of Congress’s power under Section 5 of the Fourteenth Amendment, as applied to the administration of prison systems.”[96] Following the grant of certiorari, Georgia argued that the challenged provision swept far too broadly to satisfy the judicially prescribed test under which legislation enacted pursuant to Section Five must be “congruent and proportional” to a pattern of judicially cognizable constitutional violations.[97] According to the state, Title II was therefore invalid on its face.[98]

In an opinion by Justice Scalia, the Supreme Court unanimously found it unnecessary to resolve the question of the statute’s facial validity.[99] Instead, the Court emphasized that the plaintiff had alleged violations of the ADA that “independently” amounted to constitutional violations.[100] “[I]nsofar as Title II creates a private cause of action for damages against the States for conduct that actually violates the Fourteenth Amendment,” the Court reasoned, “Title II validly abrogates state sovereign immunity.”[101] In holding that Title II was valid and enforceable as applied to actual constitutional violations, the Court implicitly found that even if the challenged provision was facially invalid under the “congruence and proportionality” test, as Georgia maintained, it could be severed in a way that left Georgia subject to sanctions.[102]

The concept of statutory severability presents many complexities. It can apply to denominated provisions of a statute, to linguistic subunits within a provision, or to a single provision’s various applications.[103] With respect to differentiated provisions, the process is intuitive. For example, imagine that a statute provides:

It shall be unlawful either to:

(a) Threaten the president with physical violence; or

(b) Speak disrespectfully of the president.

Part (b) is constitutionally invalid, but Part (a) would be constitutionally unobjectionable standing alone. In this case, nearly all would agree that Part (b), which is invalid, could be separated from Part (a), which would thereafter be valid and enforceable despite having been enacted as part of the same legislative package as the invalid Part (b).

The same result would hold if the statute, without formal separation into Parts (a) and (b), read: “It shall be unlawful either to threaten the president with physical violence or to speak disrespectfully of the president.” The linguistic subunit that purports to prohibit speaking disrespectfully of the president could be severed.[104]

But separability does not always require terms that can be severed from a statute in their entirety. More typically, severability questions involve whether invalid statutory applications can be severed from valid ones. A clear example comes from Ayotte v. Planned Parenthood of Northern New England, which concerned a New Hampshire statute that prohibited doctors from performing abortions on a minor unless the minor’s parent or guardian had received at least forty-eight hours’ notice.[105] In an opinion by Justice O’Connor, the Supreme Court held the statute constitutionally invalid insofar as it applied to cases in which waiting forty-eight hours would pose a risk to the minor’s health.[106] Nevertheless, the Court contemplated that the statute could be severed through a judicial injunction barring its enforcement only in cases of health emergency (even though the statute included no specific bits of severable language).[107]

The Court in Ayotte remanded the case to the lower court to ascertain whether severing the statute would accord with the intent of the New Hampshire legislature.[108] But it left no doubt that federal separability doctrine called for invalidating the statute, by separating some of it and saving the rest, only insofar as specific applications would be invalid.[109] A similar premise underlies the decision in United States v. Georgia: the Court anticipated that it could sever valid from invalid applications of a provision (Title II of the ADA) that was challenged on its face, even in the absence of any linguistic subunit that it could discretely excise.[110]

One of the puzzles about severability, once it is introduced, is why statutes cannot always be separated with the result—as the Court put it in United States v. Salerno—that facial challenges could never succeed unless the challenger could “establish that no set of circumstances exists under which [a challenged statute] would be valid.”[111] I shall address this puzzle below. The short version of the solution is that for a statute to be severed successfully, the severing court must be able to articulate lines of severance—as it did explicitly in Ayotte[112] and impliedly in United States v. Georgia[113]—that would allow a statute’s surviving elements to pass constitutional muster.

For now, three points merit highlighting. First, when a court severs a statute, it does not purport to identify what the statute, as properly interpreted, means or meant. Rather, the court identifies parts or applications that can survive following a determination that others are invalid. When severing a statute, the Supreme Court sometimes says that the lines of severance must accord with the intent of the legislature.[114] But the relevant legislative intent in cases such as Ayotte is not the same intent as bears on determinations of a statute’s meaning before it is adjudged valid or invalid on its face. In Ayotte, there was no serious question that the challenged statute, as properly interpreted, included no exception for cases involving threats to a pregnant minor’s health. Rather, the legislative intent that bears on statutes’ severability involves whether the legislature had an expressed, an implied, or a standing intention that if a statute should be found unconstitutional as written, courts either should or should not sever invalid parts or applications from valid ones.[115]

Second, the question whether a court can, should, or does “rewrite” a statute when it severs it is more complicated than the partly analogous questions that arise when courts furnish saving constructions. Statutes that include severability clauses should normally pose relatively easy cases[116]—though the Supreme Court, as commentators have noted, has sometimes resisted giving effect to such clauses,[117] and a few cases may pose genuine constitutional puzzles.[118] If Congress enacts a severability clause, then a court, in enforcing it, follows Congress’s instructions. Applicable law directs that a court should sever the statute and enforce any valid remainder.

In the absence of a severability clause, it would be possible to adopt either of two views. According to one, courts always and necessarily rewrite statutes when they find the statute that Congress wrote to be invalid and, having done so, determine to enforce the altered law that survives once invalid parts or applications are severed.[119] According to an alternative view, statutory severing never involves the rewriting of a statute if done properly. When Congress enacts a statute, it enacts all of the statute’s parts and prescribes all of its applications. If analysis begins with this premise, a court that enforces a statute’s valid parts or applications after determining that others are invalid does not write into law any regulations or prescriptions that Congress had not adopted. Rather, it enforces the statute that Congress wrote, subject to constitutionally mandated carve-outs.[120]

In order to be intellectually honest, however, even this second characterization must acknowledge that courts sometimes play an active and potentially creative role when severing statutes. When severing statutory applications, in particular, a court—even though prescribing no applications that Congress had not previously enacted—must nevertheless articulate a line of severance that effectively formulates a rule of law, not previously articulated by Congress, that itself can survive applicable tests of constitutional validity. Determining exactly where and how to sever a statute can require judicial judgment and even creativity, aimed at crafting a surviving rule of law that would itself pass constitutional muster.[121]

Third, when a court severs a statute, there may be no obvious answer to the question of whether a facial challenge succeeded or failed. In one sense, severing becomes necessary only when a facial challenge succeeds: a court determines that if left unsevered, a statute would be invalid and, therefore, unenforceable on its face. From the perspective of the challenger, however, judicial severance of a statute can sometimes signal defeat. The challenger will experience defeat if a court, despite agreeing that a statute was facially invalid prior to severance, severs the statute in such a way that it remains enforceable against the challenger.

Barr v. American Ass’n of Political Consultants[122] provides a vivid illustration. The Telephone Consumer Protection Act of 1991 generally prohibits robocalls to cell phones but, following a 2015 amendment, excepted robocalls made to collect debts owed to or guaranteed by the United States.[123] In a challenge by plaintiffs who wished to make political robocalls, six Justices—though without a majority opinion—concluded that the statute discriminated impermissibly on the basis of content and therefore violated the challengers’ rights under the First Amendment.[124] With the Court having identified this “equal-treatment constitutional violation,”[125] seven Justices agreed that the appropriate judicial response was to sever the provision that exempted robocalls involving debts owed to or guaranteed by the government.[126] According to Justice Kavanaugh’s plurality opinion, “[i]nvalidating and severing the government debt-exception fully addresses that First Amendment injury” of unequal treatment.[127] From the perspective of the plaintiffs, who remained forbidden to place political robocalls under the severed statute, a successful facial challenge left them no better off than they would have been if the Court had ruled against them.

III. Some Confusions About Facial Challenges

This Part considers sources of confusion involving the nature of facial challenges and the grounds for their permissibility. It first responds to the Supreme Court’s longtime insistence that facial challenges should be rare and possibly limited to First Amendment cases.[128] That insistence reflected a misunderstanding of the nature of constitutional rights. Although the Court appears to have retreated from its resistance to facial challenges in recent years, it is important to understand the basis of prior misunderstanding in order first to grasp and then to resolve the hard issues of judicial role that the conjunction of facial challenges, requests for saving constructions, and questions of statutory severability presents. However daunting the challenges involving saving constructions and statutory severability may be, there is no escaping them when challengers assert a right not to be harmed by a constitutionally invalid statute.

Nevertheless, it does not follow that courts can never refuse to adjudicate facial challenges brought by challengers with standing to sue. First, courts can opt not to adjudicate facial challenges in any case in which the challenger can prevail on an as-applied challenge. Second, courts can forestall adjudication of facial challenges when they foresee that even if the challenger successfully established that a statute was unconstitutional on its face, the statute could be severed in such a way that the challenger would obtain no judicial relief.

In a final section, this Part also addresses the question—which has occasioned some confusion in its own right—whether the failure of a facial challenge to a statutory provision precludes subsequent as-applied challenges. The answer depends on the nature of the constitutional right at issue.

A. Facial Challenges and the Nature of Constitutional Rights

If one asks why judges, Justices, and commentators ever regarded facial challenges as suspect, the explanation lies partly in a mistaken conception of constitutional rights. In the traditional picture, constitutional rights are privileges of individuals to engage in “protected” speech or conduct that the Constitution immunizes from regulation.[129] Pursuant to this understanding, a party pressing an as-applied challenge seeks to enforce her privilege. By contrast, the Supreme Court often referred indiscriminately to facial challenges as a species of “overbreadth” challenge, modeled on the First Amendment overbreadth doctrine, under which a party whose own speech or conduct falls within a constitutionally unprotected category could seek facial invalidation of a statute on the ground that it would be unconstitutional as applied to someone else.[130] The Court regarded facial challenges as suspect partly because adjudicating them seemed to take the judiciary beyond its imagined, traditional role as solely one of vindicating the rights of individuals before the court.[131]

A second role-based anxiety was that facial challenges require a court to determine, in one case, how a statute would apply to other, as yet hypothetical, cases.[132] To determine whether a provision is valid on its face, a court must ascertain, at least to a rough approximation, how far it extends.

The Court voiced the first of these anxieties, on facts that made it seem plausible, in United States v. Salerno. Salerno involved a facial challenge to a statute denying bail to a category of people accused of serious federal crimes.[133] In emphasizing that facial challenges could not succeed if a statute had any valid applications, the Court sought to limit Salerno to arguing that he had a personal right or privilege not to be denied bail.[134] Even if the statute would be unconstitutional as applied to someone else, Chief Justice Rehnquist reasoned, that deficiency did not impair any right of Salerno’s, nor provide a sufficient reason for the Court to invalidate the statute as applied to people whose rights it did not violate.[135]

Salerno may have been rightly decided on its facts.[136] Nevertheless, it rested on a conception of the nature of constitutional rights that often fails to fit the facts of modern disputes that reach the Supreme Court. Many and perhaps most constitutional rights are not privileges to engage in particular actions, but protections against regulation or punishment under particular kinds of statutes. To borrow a phrase that Professor Matthew Adler coined, many constitutional rights are “rights against rules.”[137] Adler’s most arresting example came from the flag-burning case of Texas v. Johnson.[138] Johnson, who had burned a flag as a form of political protest, was prosecuted under a Texas statute that made it a crime publicly to “desecrat[e]” a “venerated object.”[139] The Supreme Court invalidated the statute on its face as not narrowly tailored to a compelling governmental interest.[140] As Adler emphasized, however, the Supreme Court did not determine that burning a flag, even as a form of political protest, necessarily involves constitutionally privileged conduct.[141] Imagine that Texas had a generally applicable statute forbidding the lighting of fires on public property, justified by interests in preserving public safety. If Johnson had been prosecuted for violating that statute, the Constitution would not have barred his conviction and punishment.[142] His constitutional right was not to engage in the privileged conduct of burning a flag for expressive purposes. Rather, the right that the Supreme Court vindicated in Johnson was a right not to be punished for burning a flag under the kind of statute that Texas sought to enforce—one that singled out the desecration of a venerated object for distinctive prohibition.[143] In light of the nature of the right at stake, Johnson’s challenge was necessarily a facial challenge.

Equal protection rights typically furnish more obvious examples of rights that are not privileges to engage in particular conduct but rights not to be regulated on impermissible bases or for inadequate reasons.[144] I could multiply examples, but the point should be clear, as should its implications.

Once it is recognized that many constitutional rights are rights not to be regulated or sanctioned under constitutionally invalid rules or statutory provisions, it follows inescapably that in many circumstances, parties must be able to challenge the constitutional validity of statutes as written, not merely as applied.[145] Moreover, in challenging a statutory provision as facially invalid, a party does not characteristically argue that a court should invalidate it because it would be unconstitutional as applied to someone else whose conduct would be privileged, even if the challenger’s is not. Rather, as in Texas v. Johnson, the party bringing a facial challenge more commonly argues, often rightly, that she has a right not to be sanctioned under a statutory provision precisely because it fails an applicable test of facial constitutional validity.

To be clear, not all constitutional rights are rights against rules.[146] Constitutional rights are diverse. Prior to the 1990 decision in Employment Division, Department of Human Resources of Oregon v. Smith,[147] the Supreme Court had held that statutes that substantially burdened the performance of religious obligations triggered strict judicial scrutiny, even if not targeted at religiously motivated conduct.[148] If the Court were to reverse Smith, its protection for free-exercise rights would take the form of a qualified privilege to perform religious duties, limited only insofar as restrictions are necessary to protect compelling governmental interests.

Of more immediate pertinence, the Constitution creates many rights against executive conduct, including conduct that no rule dictates. For example, unreasonable searches and seizures violate the Fourth Amendment, regardless of whether they occur pursuant to a statutory rule.[149] Similarly, individual government officials who discriminate on the basis of race or religion violate the Constitution even when no law or policy purports to compel them to do so.[150] Even if the rights involved are not categorical privileges, neither are they rights against rules.

As-applied challenges to statutes could, I suppose, be characterized as “rights against rules as applied.” But this usage would be misleading if it failed to recognize that the judicially constructed tests of constitutional validity through which constitutional rights are given practical effect do not always draw the entirety of a statutory rule into question. Consider once again the due process right not to be punished for violating a statute that is so vague in its application to particular cited conduct that a person of ordinary intelligence could not have known whether the conduct was prohibited. That right holds even if the statute is not vague in so many applications as to run afoul of the different standards that courts use to determine whether statutes are unconstitutionally vague on their faces.

To sum up, much opposition to facial challenges rests on a misapprehension about the nature of constitutional rights. Many rights, including a number of those most likely to be litigated in the Supreme Court, are rights not to be regulated or punished under constitutionally defective statutes or statutory provisions. And insofar as rights are rights against rules, it follows that statutory provisions—which are a form of rule—must be subject to tests that gauge the permissibility of rules or rule-like prohibitions on their faces. Once confusion about the nature of constitutional rights is dispelled, facial challenges emerge as inescapable in many if not most constitutional cases involving the application of statutes.

B. Possible Priorities for As-Applied Challenges and Prohibitions of Facial Challenges

Because of the close relationship between the nature of some constitutional rights as rights against rules and facial challenges as a means for drawing the validity of statutory rules into question, it might appear that courts could never refuse to adjudicate facial challenges brought by challengers who have standing to mount them. But this conclusion would be mistaken. Courts need not entertain a facial challenge if the challenger would prevail anyway based on an as-applied challenge. In addition, courts can decline to adjudicate facial challenges if it is evident that the challenger’s claim to judicial relief would fail, even if a statute is invalid on its face, as a result of severability doctrine.

1. Barring Facial Challenges in Cases also Presenting Meritorious As-Applied Challenges

Sometimes a single party may have dual grounds for maintaining that a statute cannot validly be applied to her. She may have arguments that a statute fails both a test of facial validity and another test that would mark it as invalid only as applied to her individually or to a narrower range of cases. As noted previously, vagueness doctrine furnishes a case in point. In recent years, the Supreme Court has made clear that statutes can sometimes be adjudged facially invalid due to excessive vagueness. In City of Chicago v. Morales,[151] the Court held an anti-loitering ordinance invalid on its face.[152] It more recently invalidated a statute for vagueness in Johnson v. United States,[153] with Justice Scalia—a formerly staunch proponent of the Salerno dictum that facial attacks must fail if a statute has any valid application[154]—writing for the Court that “our holdings squarely contradict the theory that a vague provision is constitutional merely because there is some conduct that clearly falls within the provision’s grasp.”[155] Subsequent decisions reaffirm and build on Johnson.[156]

Nonetheless, the long-recognized right of individuals not to be sanctioned under a statute that is vague as applied to them remains.[157] Accordingly, if the Supreme Court wished, it could insist that courts rule first on whether statutes are unconstitutionally vague as applied to particular conduct before considering, only if necessary, arguments that would call for a statute to be invalidated facially.

2. Severability as a Basis for Permitting Challengers to Raise Only “As-Applied” Challenges

As noted above, all challenges to statutes begin as as-applied challenges when a challenger argues that the Constitution forbids a statute’s enforcement against her; but sometimes the ground for objection will be that a statute is invalid on its face. Although substantive constitutional doctrine initially converts an as-applied challenge into a facial challenge in cases of this kind, the conversion need not always prove enduring. By deploying severability doctrine, courts can sometimes achieve the practical effect of converting a facial challenge back into an as-applied challenge. A reversal of this kind can occur if the court rules that even if a statute were invalid on its face, it could and should be severed in a way that would result in the challenger’s immunity from sanctions. Via such a ruling, a court can forestall a determination of whether a statute is invalid in all applications.

Some commentators have understood the Supreme Court as mandating reliance on severability doctrine to convert a facial challenge into an as-applied challenge in the First Amendment overbreadth case of Brockett v. Spokane Arcades.[158] “[W]here the parties challenging the statute are those who desire to engage in” speech with respect to which a statute reaches too broadly, the Court pronounced, the statute should “be declared invalid to the extent that it reaches too far, but otherwise left intact.”[159]

Whatever the merits of the Brockett approach (which would have substantially weakened the potency of First Amendment overbreadth doctrine), subsequent cases have generally not followed it in the absence of a relatively obvious saving construction or comparably obvious line of statutory severance. For example, United States v. Stevens[160] found a statute that criminalized the commercially motivated creation, sale, or possession of depictions of animal cruelty to be invalid and unenforceable in all cases without considering whether the statute might have had valid applications that could be separated from invalid ones, potentially including applications to the challenger in the case before the Court.[161] To the contrary, the Court concluded that facial invalidation became proper upon a determination that the law before it was substantially overbroad.[162]

By contrast, in United States v. National Treasury Employees Union,[163] the Supreme Court avoided determining whether a statute was unconstitutional in all possible applications by finding that the challengers would prevail even if the statute could be severed successfully.[164] At issue was a statute forbidding government employees from accepting “any compensation for making speeches or writing articles.”[165] Although the Court found that the statute as written violated the challengers’ rights under a First Amendment test of facial validity,[166] it stopped short of holding that the provision was therefore void in all possible applications.[167] Instead, taking account of severability doctrine, the Court ruled that even if the statute could be severed so as to have some valid applications, the parties in the case before it would remain outside the surviving remnant of constitutionally valid law.[168]

3. Refusals to Entertain Facial Challenges When Separability Doctrine Would Bar Relief to the Challenger

Even when constitutional rights are rights against rules, courts can refuse to entertain facial challenges to statutes when separability doctrine makes it evident that a challenger could not ultimately benefit from a ruling of facial invalidity. In this kind of case, too, a court’s deployment of separability doctrine can have the effect of converting a facial challenge into an as-applied challenge—though with the result that the challenger achieves no relief. As noted above, United States v. Georgia stands as a case in point.[169]

Apart from circumstances in which a challenger could not obtain relief anyway due to the operation of separability doctrine, in cases in which a challenger’s asserted constitutional right is a right not to be sanctioned under an invalid law, courts, after rejecting as-applied challenges, should have no license to refuse to consider facial challenges. Although there are older cases that appear to conclude otherwise, the reasoning of those cases is either mistaken or confused. If the rulings are sound, they are best understood as having relied impliedly on the same separability principles that best explain the Supreme Court’s decision in United States v. Georgia.

The leading exemplar of a case in which the Supreme Court refused to countenance a facial challenge by a party whose as-applied challenge failed is Yazoo & Mississippi Valley Railroad v. Jackson Vinegar Co.[170] The Yazoo case involved a Mississippi statute that required common carriers doing business in the state to “settle all claims” of less than $200 for “freight which has been lost or damaged between two given points on the same line or system” within a period of 60 days or “be liable to the consignee for twenty-five dollars damages . . . in addition to actual damages.”[171] When the
railroad failed to settle a claim within sixty days, a Mississippi state court awarded the plaintiff Jackson Vinegar Company $4.76 in actual damages plus the statutorily prescribed $25 penalty.[172] On appeal to the Supreme Court, the railroad appeared to have conceded its liability for $4.76 in
actual damages.[173] Nevertheless, it argued that the statutory penalty for failing to settle all claims for less than $200 was unconstitutional on its face under the Due Process and Equal Protection Clauses because it would penalize the railroad for contesting doubtful or excessive claims, not just meritorious ones.[174]


The Supreme Court refused to adjudicate the railroad’s facial challenge:

[T]his court must deal with the case in hand and not with imaginary ones. It suffices, therefore, to hold that, as applied to cases like the present, the statute is valid. How the state court may apply it to other cases, whether its general words may be treated as more or less restrained, and how far parts of it may be sustained if others fail are matters upon which we need not speculate now.[175]

To see why the Court’s analysis has struck many commentators as unsatisfactory,[176] consider a hypothesized set of facts. Imagine that in a subsequent case, a Mississippi court were to hold that (a) the statute in question in Yazoo requires railroads to settle frivolous and exorbitant claims as well as valid ones; (b) the statute as so interpreted violates the Due Process and Equal Protection Clauses in its application to frivolous and exorbitant claims; and (c) the statute is inseparable and thus invalid in whole as a matter of state law. If we now reconsider the Supreme Court’s ruling in the actual Yazoo case, the Court’s analysis seems to contemplate that despite the possibility that events might unfold in the way just imagined, the railroad was limited to challenging the statute as applied to it and could not mount a facial challenge that seemingly should have succeeded. That result seems wrong. Even before Professor Adler had advanced his right-against-rules thesis, Henry Monaghan had argued—persuasively, in my view—that under the premises of Marbury v. Madison,[177] a constitutionally invalid law is no law at all, and, accordingly, that everyone has a personal constitutional right not to be subjected to punishment except pursuant to a constitutionally valid rule of law.[178]

The Supreme Court implicitly affirms this thesis in cases in which it allows parties to challenge the enforcement against them of statutes that violate structural constitutional provisions, such as the Commerce Clause[179] or Article II.[180] In those cases, it is typically taken for granted that the Constitution forbids the government from taking coercive action that harms identifiable individuals without constitutionally valid authorization. If the “valid rule principle” needs a specific foundation in the Constitution’s text, the Due Process Clause should suffice.[181]

If the Constitution will not allow the imposition of sanctions except to constitutionally valid laws, then the Supreme Court’s opinion in Yazoo appears troubling. From the Court’s perspective, the case was complicated by considerations of constitutional federalism. The state courts had not passed on the question that the Railroad pressed in the Supreme Court. The Justices may have hesitated to interfere in state procedure by directing the state courts to rule on a facial challenge that presented issues of statutory interpretation and possibly statutory severance that the state courts wished to postpone. As the Court insisted in Yazoo, precedent also supported its ruling.[182] But the precedents were not persuasively reasoned. In one earlier decision, the Supreme Court had canvased several possible justifications for not inquiring into the severability of state statutes that defendants claimed were facially invalid and, without endorsing any of them, summarily concluded that “[w]hatever the reason” for the earlier cases being resolved as they were, “the decisions are clear.”[183]

One possible, attempted excuse for the Supreme Court’s ruling in Yazoo lies in the presumption that state statutes, unlike federal statutes, are separable.[184] In light of the presumption of severability, the Court might have reasoned that even if the challenged statute were held in a subsequent case to apply to insubstantial as well as valid claims, and if it was ruled unconstitutional in those applications, it could be separated at that point in a way allowing its continued application to valid claims such as the one in Yazoo.[185]

If interpreted as depending on a presumption of separability, Yazoo would permit a federal court to refuse to entertain facial challenges in some cases, but—crucially—only if it first ascertains that the would-be facial challenger would remain subject to sanction under a properly separated statute even if a facial challenge succeeded. But whether a state statute is severable is a state law question.[186] Accordingly, a federal court ruling on the separability of a Mississippi statute as a matter of state law would not have bound the state courts in a subsequent case anyway. Under those circumstances, the Supreme Court should have remanded the Yazoo case to the Mississippi courts with instructions to address the Railroad’s facial challenge unless it either determined that the challenged statute did not apply to insubstantial claims at all or that, if it did, those applications could be severed.

Taken seriously, the nature of many constitutional rights as rights against rules establishes obligations that a court must meet before refusing to adjudicate a facial challenge on the merits. At the time when a court relies on separability doctrine either to decline to adjudicate a facial challenge or to deem an otherwise invalid statute nevertheless severable and therefore enforceable against a challenger, the court should need to explain its conclusion that a statute could be severed in a way that would permit the surviving remnant to pass constitutional muster. The unpalatable alternative would be to accept that states and potentially the federal government could impose sanctions, including criminal punishments, based on constitutionally invalid rules of law—even, one presumes, after the statutes imposing such punishments had been held both invalid and nonseverable in subsequent cases in which courts confronted the question of statutory severability.

C. The Implications of Unsuccessful Facial Challenges

Confusion has also attended the question whether failed facial challenges preclude subsequent, successful as-applied challenges. The path to clarity begins with recognition that everything depends on the nature of the asserted constitutional rights in particular cases and on the available judicial tests for enforcing them.

Sometimes the only doctrinally available ground for challenging the enforcement of a statutory provision may be that it is invalid on its face. So far the leading, if not the only, example comes from Commerce Clause doctrine. In Gonzales v. Raich,[187] the Supreme Court rebuffed an argument that a federal statute forbidding possession of marijuana was unconstitutional as applied to homegrown medical marijuana in a state that permitted the use of marijuana for medical purposes.[188] In ruling as it did, the Court implied that the only way to argue that a prohibition against the possession of marijuana lay beyond congressional power to enact under the Commerce Clause was to mount a facial challenge, which the Court held could not succeed.[189] The Court’s precedents foreclosed arguments that Congress must exempt even purely local activities that, taken in the aggregate, could reasonably be expected to affect interstate markets.[190] In light of Raich, the rejection of a facial challenge under the Commerce Clause precludes subsequent as-applied challenges.[191]

By contrast, if the Supreme Court holds that a statute survives a facial challenge under constitutional provisions other than the Commerce Clause, its ruling does not necessarily preclude subsequent arguments that the statute is unconstitutional as applied to particular conduct. Under the Due Process Clause, for example, a determination that a statute is not unconstitutionally vague on its face does not bar arguments that the statute is impermissibly vague as applied to particular conduct. This result reflects the nature of rights under the Due Process Clause.[192]

Confusion is most likely when it is not clear whether the law recognizes or would recognize rights that would support as-applied challenges to statutes that courts have upheld on their faces. In Washington v. Glucksberg[193] and Vacco v. Quill,[194] the Supreme Court rejected facial challenges to statutes that barred physician-assisted suicide as well as as-applied challenges brought by particular classes of patients. In both cases, Justice Stevens wrote in concurring opinions that the possibility of as-applied challenges remained.[195] Chief Justice Rehnquist’s majority opinions responded with seeming perplexity. Although acknowledging that “[o]ur opinion does not absolutely foreclose such” claims, he noted that the Court’s “holding that the Due Process Clause of the Fourteenth Amendment does not provide heightened protection to the asserted liberty interest in ending one’s life with a physician’s assistance” made it hard to imagine how a successful as-applied claim might be framed.[196]

The Chief Justice’s reasoning highlights three points about the nature of constitutional rights and constitutional tests and their bearing on facial challenge doctrine. First, as emphasized throughout this Part, many and perhaps most constitutional rights are rights not to be sanctioned under particular kinds of rules or statutes. Second, because not all constitutional rights possess this character, the rejection of a facial challenge does not preclude the possibility of a subsequent, successful as-applied challenge in some cases. Third, it is in some ways a mark of the extent to which rights against rules and facial challenges predominate that the Supreme Court, in Washington v. Glucksberg and Vacco v. Quill, had a hard time imagining what as-applied challenges might look like after facial challenges had failed, even though the Court did not wholly rule out the possibility that a future as-applied challenge might succeed.[197]

IV. Narrowing Constructions

The Supreme Court’s provision of saving or narrowing constructions, typically occurring pursuant to the canon of constitutional avoidance,[198] have occasioned recurrent confusions. The most common involve mis-understandings of state court prerogatives and the conflation of statutory construction with statutory severance.

A. Narrowing Constructions, Rewriting Statutes, and State Court Prerogatives

The Supreme Court often says that it will not and must not “rewrite” a statute in order either to save it from invalidity or to avoid a difficult constitutional question.[199] Narrowing constructions of federal statutes can occur only when statutes are genuinely ambiguous, with more than one plausible interpretation.

But the federal rule that courts must not rewrite statutes in order to save them applies only to federal courts and federal statutes. Federal courts must almost always accept that state statutes mean whatever state courts say that they mean.[200] Sometimes state courts appear to conflate the provision of narrowing or saving constructions with decisions that statutes are severable after they have succumbed to facial challenges. In the case of state court action under state law, there is no federal law prohibition against judicial rewriting except insofar as the state court’s construction might deny fair notice to an adversely affected party or otherwise abridge a federal constitutional right.[201] Nor does it matter for federal purposes whether a state court describes its pronouncements about a statute’s meaning as involving a saving construction or a severance of invalid parts or applications. If state courts characterize decision-making that occurs after the determination of a challenged statute’s facial invalidity as involving the provision of a saving construction, the Supreme Court will accept the state court’s “construction” as binding on it.[202]

B. Supreme Court Conflation of Saving Constructions with Statutory Severing

In cases involving federal statutes, the Supreme Court has sometimes conflated the provision of a narrowing construction with the severing of a statute’s invalid parts or applications. An example comes from Skilling v. United States.[203] A federal statute, 18 U.S.C. § 1346, makes it a crime to fraudulently “deprive another of the intangible right of honest services.”[204] Skilling argued that the statute was unconstitutionally vague on its face and that, if it were properly interpreted, his alleged conduct lay beyond its reach.[205] In an opinion by Justice Ginsburg, the Court found that Skilling’s vagueness challenge “ha[d] force” in light of “disarray” among prior judicial opinions that Congress, in enacting the most recent version of the statute, had meant to codify.[206] But before invalidating the statute, the Court reasoned, it should consider whether the statute was susceptible of a narrowing construction, which it then provided.[207] “To preserve the statute,” the Court held that “§ 1346 criminalizes only” the bribery and kickbacks that constituted the “core of the pre-McNally caselaw.”[208]

Concurring Justices protested that Justice Ginsburg had rewritten the statute, not construed it.[209] To an extent, the Justices’ disagreement on that score may have reflected differences about appropriate statutory interpretation. Theories rooted in Congress’s purposes or intentions sometimes authorize interpretations without specific support in the language of a statute; from a textualist perspective, the majority’s approach bore the hallmarks of a judicial rewriting.[210]

Nevertheless, the charge of conflation finds support in other evidence. In defense of what she called a limiting construction,[211] Justice Ginsburg invoked a test more familiarly at home in cases in which the issue involves the permissibility of severing a statute than in cases involving a statute’s interpretation. As if conceding that the Court’s interpretation did not fit the statute as written, Justice Ginsburg wrote that if “[a]pprised that a broader reading of § 1346 could render the statute impermissibly vague, Congress, we believe, would have drawn the honest-services line, as we do now, at bribery and kickback schemes.”[212]

If the majority sought to draw the “honest-services line” not where Congress placed it but where Congress “would have drawn” it if apprised that the statute was invalid as written, then the majority can fairly be characterized as “rewriting” the statute under the guise of interpreting it. The inquiry into what Congress “would have” done under counterfactual circumstances has no obvious relevance to a conclusion about the line that Congress actually drew. By contrast, the Court’s precedents on statutory separation make it highly relevant whether separation would accord with congressional intent.[213] But the severability question arises only if no saving construction is available and a statute is therefore facially invalid. As if apprehending its own conflation of the logic of saving constructions with that of severance, the Skilling Court supported its reliance on speculation about Congress’s counterfactual wish to avoid invalidation on vagueness grounds with a “cf.” cite to its decision to sever an otherwise invalid statute in United States v. Booker.[214]

Via what the Court called a saving construction, it also achieved in one fell swoop the result that could have emerged from a series of decisions that found § 1346 unconstitutional as applied to cases not involving bribery and kickback schemes—because it was too vague to give fair notice in those cases—but that invalidated the statute only as so applied. At the end of a sequence of as-applied invalidations and severings of invalid applications, a prohibition against bribery and kickbacks would have remained as the sole enforceable aspect of the statute.

The Supreme Court’s use of what it called a narrowing construction to achieve a result that it could have realized through statutory severing raises the question: Do narrowing constructions ever achieve effects in saving statutes from invalidity that could not also be achieved by a judicial declaration that a statute is facially invalid, but that it can nevertheless be severed and some parts or applications sustained? And if not, what harm could ensue from conflating the two concepts? If a proper understanding of the judicial role permits a federal court to achieve a result by severing a statute after finding it facially invalid, how could considerations of judicial role preclude a court from reaching the same result via statutory interpretation?

Apart from considerations of conceptual coherence and doctrinal integrity, there are two practical reasons to maintain a distinction between providing federal statutes with narrowing or saving constructions and severing them to rescue valid parts after a determination of facial invalidity. First, the objection to rewriting statutes in the guise of interpreting them has significant practical implications in cases in which courts give narrowing constructions to avoid mere constitutional doubts about a statute’s validity. In that kind of case, severing would not be possible if a statute, though giving rise to constitutional questions, should ultimately be adjudged con-stitutionally valid. As a result, promiscuous reliance on the vocabulary of “narrowing” constructions can result in courts producing outcomes that they could not achieve by severing invalid applications.[215]

Second, the conflation of narrowing constructions with severing can lead to confusion and mistake in cases in which the Supreme Court, after rejecting a proposed narrowing construction as insupportable, then fails to ask the classic severability question of whether a statute that could not be sustained as written could nevertheless be severed such that some parts or applications could still be enforced. The Supreme Court appears to have succumbed to this fallacy in Iancu v. Brunetti. Dissenting in Iancu, Justice Sotomayor argued that a saving construction rendered the statute valid and enforceable.[216] If the majority had borne in mind the conceptual difference between saving constructions and statutory severing, it would have needed to consider whether—even if adopting that proposed construction would have constituted an impermissible rewriting of the statute—the statute could nevertheless be severed along similar lines.

V. Severing Statutes

The severing of statutes presents a myriad of conceptual and practical challenges that remain little understood not only by judges and Justices, but also by lawyers who frequently fail to make severability arguments that might shift the outcomes of their cases. This Part begins by addressing conceptual issues involving the nature of severing and its relationship to interpretation and rewriting. I then examine the distinctive issues presented by severance of valid from invalid applications of a statutory provision, on the one hand, and by severance of invalid provisions from multipart statutes, on the other hand. Each of these forms of severance presents multiple complexities. Properly executed, neither involves the “rewriting” of a statute, but both can require partly policy-based judgments that need to be cabined, even though they cannot be eliminated.

Throughout this Part, my aim is to discern an immanent logic in the Supreme Court’s pattern of decisions. My efforts admittedly confront two large obstacles. First, as I am hardly the first to notice, the Court’s holdings betray a number of inconsistencies.[217] Among other things, the Justices may be less likely to sever legislation that they disapprove of than legislation pursuing goals that they applaud.[218] Second, the Justices frequently fail to take note of severability issues.[219] They often either invalidate statutes on their faces or uphold them as applied to particular facts without referring to severability. Under these circumstances, my effort to identify controlling severability norms unavoidably depends on generalizations that admit counterexamples, rational reconstruction of analysis that the Supreme Court may have failed to conduct explicitly, or both.

A. Severing, Interpreting, and Rewriting

When the Supreme Court severs either invalid statutory applications or invalid statutory provisions, complaints that the Court has “rewritten” the statute are typically misplaced. As explained in Part I, the validity of such charges depends on a prior issue of characterization. On one view, when Congress enacts a statute, it adopts the totality of the statute’s provisions and applications as a unit. Accordingly, any severance constitutes rewriting. But historical and practical reasons require rejection of this conceptualization. With regard to history, the position that all severance entails rewriting impliedly repudiates the traditional understanding that courts should refuse to enforce statutes only insofar as those statutes are “repugnant to the Constitution.[220] Within the traditional understanding, courts judged constitutional repugnancy on an application-by-application basis. At the practical level, the view that separation constitutes rewriting would imply either that courts could never separate statutes’ invalid parts or applications and enforce the rest—in which case facial challenges would devastate the statute book—or that courts may rewrite statutes without obvious limits. The latter view would fit badly with traditional and still-sound understandings of the judicial role under the separation of powers.

The vastly preferable conceptualization of what courts do when severing statutes begins with the premise that when Congress enacts a statute, it votes to give legal effect to all of the statute’s provisions and applications unless it indicates otherwise. If this presumption of congressionally intended severability applies, as the doctrine indicates that it should, then the severance of invalid parts or applications results not in judicial lawmaking, but instead in the enforcement of the valid—and validly enacted—applications and provisions that remain after invalid ones have been stricken.[221]

As I noted above and shall describe more fully below, statutory severance requires judicial agency in cases in which a doctrinal test reveals a statute as having invalid applications. In severing a statute, a court must identify a line or lines that produce a constitutionally valid and enforceable rule of law that the legislature did not specifically formulate—as the Supreme Court did in Ayotte v. Planned Parenthood of Northern New England and United States v. Georgia, for example. Nonetheless, rejection of the view that all separation constitutes judicial rewriting frames the challenges for courts in manageable terms. By contrast, portrayal of any severance as a form of constitutionally improper judicial rewriting invites wrecking-ball effects in a doctrinal world rife with potentially meritorious facial challenges.[222]

In recent contributions to the literature on statutory severability, Eric Fish has adopted a more iconoclastic stance.[223] Fish maintains that judicial severance of statutes involves rewriting, but unlike most others who embrace that characterization, he defends statutory revision as a proper judicial function. In developing his position, Fish challenges the traditional view—which I generally accept—that statutory severance functions analogously to surgical excision: courts identify a line that separates valid from invalid applications or provisions, deny effect to the invalid, and enforce the remainder.[224] By Fish’s lights, that conceptualization is unduly formalistic. According to him, what counts as severance or excision depends on accidents of statutory drafting that should not determine judicial responses to constitutional violations.[225]

Fish’s argument relies heavily on cases involving federal statutes that violate equal protection principles by conferring benefits on one class but not another. Califano v. Westcott[226] presented a constitutional challenge to a federal statute that provided benefits to families whose dependent children were deprived of parental support due to the unemployment of the father but denied benefits in cases involving unemployment of the mother.[227] Finding that the discrepancy violated the Constitution,[228] the Supreme Court ordered extension of the more generous benefits to the families with unemployed mothers.[229] According to Fish, it is odd, if not untenable, to maintain that the extension of benefits that Congress did not authorize constitutes a severance of the statute in issue.[230]

The cases that draw Fish’s attention present genuine puzzles. Nevertheless, I would not leap from those cases to the broad conclusion that courts, generally, have a power to rewrite statutes. The distinctive aspect of Califano v. Westcott and similar cases is that after the challengers established an ongoing violation of their equal protection rights, a court could provide constitutionally adequate redress by ordering either an extension of benefits to the challengers or a withholding of benefits from others. Under these circumstances, the question is not whether the statute should be severed, but whether applicable remedial principles permit or require a court to extend more favorable treatment to a group that Congress attempted to treat less favorably.[231]

This analysis enables avoidance of two recurring confusions in discussions of statutory severability. First, we should not conflate the judicial role in severing statutes with the judicial role in remedying constitutional violations. It is easy to confuse the remedial question in cases such as Califano v. Westcott with the severability questions in other cases because of familiar references to statutory severing as a “remedy” for statutes’ constitutional defects. The Supreme Court has repeatedly used this terminology,[232] as have commentators.[233] But referring to severance as a remedy invites confusion.[234]

In ordinary legal parlance, a remedy is an award of relief to a party who has prevailed on the merits. According to Black’s Law Dictionary, for example, a remedy is “[t]he means of enforcing a right or preventing or redressing a wrong.”[235] Severance is not always a remedy in this sense, nor is it necessarily a prelude to the provision of a remedy. Recall United States v. Georgia, in which the possibility of statutory severance precluded Georgia—which had challenged a provision of the ADA on its face—from receiving any judicial relief whatsoever.

Second, characterization of severance as a remedy may promote the mistaken view that courts, upon finding a statute facially invalid, have the authority to do anything necessary to “remedy” its defects. They do not. Under Marbury v. Madison, courts must refuse to enforce statutes insofar as they violate the Constitution. A determination that identified statutory applications or provisions violate the Constitution and are therefore invalid thus hews closely to the traditional judicial function. Califano v. Westcott exhibits another traditional and necessary judicial power to remedy the ongoing violation of an identifiable party’s constitutional rights. There is no need for a dramatically expanded account of judicial power.

Adherence to a conception of the judicial role that equates severance of invalid statutory applications and provisions with quasi-surgical excision and allows judicial remedies only as necessary to vindicate the rights of identified individuals involves a kind of separation-of-powers formalism that Fish rejects. But the denial of a further judicial power to revise statutes in response to the identification of constitutional defects serves a purpose by maintaining an identifiable line of separation between the legislative and judicial functions. The need to mark a limit on judicial power becomes greater, not smaller, once it is recognized that courts, in severing statutes, take on a role of assisting Congress in producing workable, judicially enforceable, constitutionally valid law. That assistance-providing role becomes especially visible when severing a statute requires a court to articulate what the surviving remnants of a severed statute can constitutionally validly prohibit—as, for example, in Ayotte and United States v. Georgia.

B. Severing Provisions vs. Severing Applications

The previous subpart established the conceptual possibility that courts can sever otherwise invalid statutes without rewriting them. Against that background, we confront the questions of when and how courts should do so. In order to forestall the wrecking-ball effects that facial challenges could have otherwise, courts need to exercise agency, judgment, and sometimes creativity. A mechanical jurisprudence would prove ruinous. At the same time, the judicial role needs bounding. In searching for limiting principles, it is important to distinguish between cases involving the severance of invalid statutory applications, on the one hand, and invalid statutory parts or provisions, on the other.

1. Severing of Statutory Applications

In a legal-doctrinal world dominated by tests that judge statutory provisions on their faces, identification of legal principles to govern the severance of invalid from valid applications is a matter of urgent priority. The task, moreover, is multifaceted. Successfully theorizing the process of severing invalid from valid statutory applications poses conceptual mysteries as well as practical challenges.

Consider Ayotte, in which the statute forbade minors to have abortions without parental notification. Absent an exception for emergency cases, the statute was unconstitutional on its face. Yet if one imagines severing as analogous to excision, it may appear that there is nothing to cut when the same statutory language yields invalid as well as valid applications.

To give the process of severing invalid applications an intuitive foundation, I have proposed conceptualizing statutory provisions as comprising a multitude of subrules.[236] For example, we could imagine the law in Ayotte as encompassing the subrules (a) nonemergency abortions on minors are unlawful absent parental notification and (b) emergency abortions on minors that are necessary to the mother’s health are unlawful absent parental notification. On this model, subrule (b), which would be invalid, could be separated from the valid subrule (a), which a court could therefore enforce.

As thus conceptualized, the severance of valid from invalid statutory applications approximates—but also supplements—the judicial practice that Professor Kevin Walsh describes as having prevailed through early American history.[237] On Walsh’s account, early judges did not affirmatively separate invalid from valid applications. Rather, when confronting constitutional challenges to statutes’ applications, they assumed that the Constitution rendered statutes unenforceable to the extent, but only to the extent, that they were repugnant to the Constitution.[238] In the traditional model of “displacement” of invalid constitutional applications, a court, after holding a statute invalid as applied, takes no “next step” of ascertaining whether a statute continues to have valid applications.[239] Instead, the court would leave it to challengers in subsequent cases to establish that the statute violated the Constitution as applied to them.

In his concurring opinion in Murphy v. NCAA, Justice Clarence Thomas suggested that it might be preferable to return to the old practice, sketched by Professor Walsh, in which courts, after finding that a challenged statute violates the Constitution in its application to a particular case, would pursue no further inquiry into lines of permissible severance. Justice Thomas cloaked his proposal to return to early historical practice—which Justice Gorsuch also supported in Seila Law LLC v. Consumer Financial Protection Bureau[240]—in a veneer of judicial modesty: it would uphold the traditional judicial role of vindicating the rights of parties before the courts and avoid creative, policy-based judgments about whether and how to sever statutes.[241] In practice, however, the proposal’s adoption would have staggering implications when applicable tests of constitutional validity require the appraisal of statutes on their faces. Now-familiar examples emerge from the anti-abortion statute that triggered the “undue burden” test in Ayotte[242] and the provision of the ADA that, as written, likely would have failed the applicable “congruence and proportionality” test in United States v. Georgia.[243] In those cases, a finding by the Supreme Court that the challenged provisions were repugnant to the Constitution, when coupled with stare decisis, would have established that the statutes could not be enforced against other parties in other cases, either. As a result, in cases in which Justice Thomas’s approach functions differently in practice from now-traditional severability analysis, it would threaten the de facto facial invalidation of any statute that failed a facial test of constitutional validity. No one should welcome that result.[244]

If it is practically imperative for courts sometimes to separate statutes’ valid from their invalid applications, one might ask why courts cannot or should not always separate a statute’s valid applications from its invalid ones. The answer inheres in the nature of modern constitutional litigation, which frequently involves doctrinal tests that gauge the validity of statutory provisions on their faces. After failing a test of facial validity, a statutory provision has, and can have, no valid applications until it has been pared down so that a valid rule remains. Accordingly, in order to sever invalid applications, a court must articulate lines of severance that define a new, reformulated rule that itself survives constitutional scrutiny.[245] In other words, modern tests of constitutional validity, reflecting modern understandings of the nature of constitutional rights, render the severability doctrine functionally indispensable to avert the total invalidation of all statutory provisions that fail a facial test of constitutional validity.

Charting a path forward requires painstaking, nuanced analysis. Different kinds of cases present different challenges, some surmountable and others not. Sometimes how to sever a statute to produce a constitutionally valid rule of law will be relatively obvious based on extant constitutional tests. In Ayotte, for example, severance of emergency cases from a categorical restriction of minors’ abortions defined a new, narrower prohibition that satisfied the applicable “undue burden” test under settled Supreme Court precedent.[246]

In other cases, by contrast, there may be no imaginable line of severance that would produce a constitutionally valid rule of law (or a court may be unable to identify one). Consider Texas v. Johnson and the statutory prohibition against public desecration of venerated objects. If we imagine that Johnson’s act of flag-burning had posed a fire hazard, and if we further suppose that Texas could validly prohibit the public burning of any object under circumstances that created a serious threat of conflagration, we confront the question whether the application of the statute to Johnson would constitute a valid application. The answer should be no. A statute framed as a prohibition against the desecration of venerated objects will not pass constitutional muster under the strict-judicial-scrutiny test even as limited to desecrations that pose fire hazards. Texas has a valid and possibly compelling interest in eliminating fire hazards, but a prohibition limited to desecrations of venerated objects that pose such hazards would fail the narrow tailoring requirement.[247] It would leave too many fire hazards unregulated.

The harder question is whether there are situations in which a court—though potentially able to identify a line of statutory severance—should nevertheless decline to separate a statute and should instead declare it unenforceable in all cases. In a number of cases, the Supreme Court has maintained that judicial lines of severance must be traceable to Congress’s intentions or purposes in enacting a statute and should not represent judicial innovations.[248] But restrictions on severability that are framed in these terms are often confused, conflating what a court does when interpreting a statute with what a court does in severing one. Interpretation may sometimes yield the conclusion that Congress wanted a statute to be either severable or nonseverable (possibly including a matter of what textualists sometimes call “objectified” intent).[249] By contrast, there are likely to be few if any cases in which a court should infer—absent specific instructions in a severability clause—that Congress intended for a statute to be severable only if it could be severed in a particular way.

That said, considerations of judicial role should matter crucially to determinations of whether, and if so how, otherwise invalid statutes should be severed. Although I have maintained that Congress enacts all of a statute’s applications, I have also emphasized that Congress does not formulate all of the rules of law—which I have proposed conceptualizing as subrules—that would be necessary for some applications to survive constitutional objection if other applications were adjudged invalid. For a partially invalid statute to survive, the court, rather than Congress, must perform that crucial function. In doing so, a court thus functions as a helpmate to Congress in producing an enduring, workable body of statutory law that avoids constitutional objection—a role that, if carried too far, would invite protestation under the separation of powers. Moreover, in contrast with Ayotte, there are cases in which identifying lines for severing valid from invalid applications could require complex decision-making of a quasi-discretionary character that courts reasonably might feel incompetent to perform.

Justice Thomas has recently highlighted concerns of this kind. In Seila Law, he suggested that modern severability doctrine requires quasi-legislative judgments that Article III forbids courts to make.[250] But that blunderbuss argument overlooks myriad relevant distinctions among three categories of cases, respectively involving (a) federal statutes that have neither separability clauses nor inseparability clauses, (b) federal statutes with separability clauses, and (c) state statutes.

a. Federal Statutes That Have Neither Severability nor Nonseverability Clauses.—In cases involving neither severability nor nonseverability clauses, the Supreme Court’s emerging, sensible practice appears to be to sever valid from invalid applications in all cases—such as Ayotte—in which applicable tests of validity mark clear, relatively obvious lines along which successful severance could occur. As the Court explained in Ayotte and recited more recently in Seila Law: “Generally speaking, when confronting a constitutional flaw in a statute, we try to limit the solution to the problem, severing any problematic portions while leaving the remainder intact.”[251]

Prior to Ayotte, post-Salerno debates about whether statutes restricting abortion are subject to facial or only to as-applied challenge may have clouded thinking about the separability issues that remain after a constitutional test marks an unsevered provision as unconstitutional on its face.[252] If Ayotte was rightly decided, then at least some Roe-vintage statutes that too broadly prohibited abortions might, in principle, have been severed to satisfy the strictures that Roe laid down.

Sometimes, in contrast with Ayotte, potential lines along which to sever valid from invalid applications may not be immediately obvious to judges or Justices who determine that a statutory provision as written by Congress fails a test of constitutional validity. In such cases, the Supreme Court generally, and I believe rightly, feels no obligation to sever statutes in imaginative ways.[253]

In United States v. Lopez,[254] for example, the Court invalidated a federal statute barring possession of a gun in a school zone on the ground that it lay beyond Congress’s power to enact under the Commerce Clause. Although that law would almost surely have been constitutionally permissible as applied to guns that had traveled in interstate commerce,[255] the Court took no note of that possibility.[256] It is hard to know whether the Court might have responded differently if the parties had framed a severability issue. In Ayotte,[257] Justice O’Connor’s opinion stated that the Court would have severed a statute and held it only partially invalid in Stenberg v. Carhart,[258] rather than pronouncing it wholly invalid, if the parties had called that option to the Court’s attention.

That pronouncement frames both an invitation and a challenge that too many constitutional lawyers—including those arguing in the Supreme Court—have ignored. In light of modern severability doctrine, good lawyers defending statutory provisions against facial challenges should advance “fallback” arguments identifying appropriate lines of severance if severance should prove necessary. In the absence of such arguments, courts may have good, role-based reasons not to devise lines of severance that have not been tested by adversarial argument. But if the parties identify a clear line of severance or if the court itself is confident that it can mark one, legal insightfulness should not disqualify an otherwise available result.[259]

b. Federal Statutes with Separability Clauses.—When Congress includes a severability clause in a statute, courts should strive to implement it.[260] In these cases, any suggestion that the Constitution categorically forbids courts to exercise judgment in determining how best to sever a statute with invalid applications claims far too much. In addition to long-ensconced precedents calling for judicial severance of otherwise unconstitutional statutes, settled doctrine also permits Congress to vest the federal courts with responsibility to engage in federal common lawmaking.[261] If Congress can validly mandate federal common lawmaking, it also should be able to call upon the judiciary to sever statutes in order to save them from facial invalidation, even if doing so requires elements of policy judgment.[262]

Against this background, the Supreme Court frequently pays lip service to the notion that severability clauses bind the courts to do their bidding.[263] In practice, however, the Court has often failed to give effect to severability clauses.[264] There are two possible explanations. First, the Court may believe that severability clauses, as properly interpreted, sometimes include implicit exceptions for particular kinds of cases.[265] Second, some congressional instructions to sever invalid statutory applications might violate Article III as applied to particular facts.

In my view, federal courts, including the Supreme Court, should indulge a strong presumption that severability clauses are constitutionally valid and binding and should make every reasonable effort to sever statutes’ valid from their invalid applications when directed by Congress to do so. Nonetheless, precedent, including some modern decisions, establishes the need for case-by-case inquiries to determine whether severability clauses, as applied, might call for courts to perform functions for which they lack constitutional competence.

Shelby County v. Holder[266] furnishes a possible example. Shelby County facially invalidated the “coverage provision” of the Voting Rights Act, which identified jurisdictions, mostly in the South, that could not change their voting procedures without seeking preclearance from the Department of Justice. Congress initially enacted the coverage formula in 1965.[267] In Shelby County, the Court reasoned that a 2006 reenactment had failed to take account of changed circumstances and was no longer closely enough tailored to an identified pattern of constitutional violations to pass muster under Section Two of the Fifteenth Amendment.[268] I disagree with that conclusion. Nevertheless, if we accept the Court’s determination of insufficiently close tailoring as a predicate for the severability analysis that ensued, I agree that judicial severance of the statute to render it valid in its application to some jurisdictions would have tested limitations on the permissible judicial role under Article III, even though the Voting Rights Act included a severability clause.[269]

Especially if we assume that Congress might have crafted multiple constitutionally permissible rules that would have subjected some voting jurisdictions to preclearance requirements, for the Court to formulate one rather than another would have required sensitive policy judgments based on information to which the Court would not have had ready or possibly reliable access.[270] Under those circumstances, the Court’s conclusion that severing the statute exceeded its competence strikes me as at least a reasonable one.

The Court also struggled with whether and if so how to implement a severability clause in Reno v. American Civil Liberties Union.[271] But its analysis in that case spotlights confusions that future Courts should seek to avoid. Reno presented a First Amendment challenge to two anti-indecency provisions of the 1996 Communications Decency Act (CDA). One provision criminalized the “knowing” transmission of “obscene or indecent” messages to any recipient under eighteen years of age.[272] The other prohibited the “knowin[g]” sending or displaying to a person under eighteen of any message “that, in context, depicts or describes, in terms patently offensive as measured by contemporary community standards, sexual or excretory activities or organs.”[273] The CDA also included a broadly written severability clause: “If any provision of this chapter or the application thereof to any person or circumstance is held invalid, the remainder of the chapter and the application of such provision to other persons or circumstances shall not be affected thereby.”[274]

In an opinion by Justice Stevens, the Supreme Court ruled that both challenged provisions violated the First Amendment.[275] After having done so, the Court found that although the first provision was severable, the second was not. Severance of the first provision was straightforward: the Court excised the words “or indecent” from the ban on “obscene or indecent” com-munications.[276] By contrast, the Court refused to separate valid from invalid applications of the provision that barred knowing displays or transmissions to persons under eighteen of “patently offensive” depictions of “sexual or excretory activities or organs.”[277] “In considering a facial challenge, this Court may impose a limiting construction on a statute only if it is ‘readily susceptible’” to such a construction, the Court asserted.[278]

That reasoning was fallacious. To repeat a now-familiar point, what is true of permissible saving constructions is not necessarily true of severances of statutes’ invalid provisions and applications. Indeed, the Supreme Court implicitly recognized as much in Reno when it severed the words “or indecent” from the other challenged provision of the CDA. It is not plausible to construe “obscene or indecent” as meaning “obscene” only. The real and difficult question in Reno involved the scope of Congress’s power to require courts to engage in judgment-based severance of statutes’ valid from their invalid applications in the absence of relatively clear, doctrinally marked lines.

Reno never faced up adequately to that question. Dissenting in part, Justice O’Connor, joined by Chief Justice Rehnquist, did.[279] Justice O’Connor would have severed the prohibition against knowing sending or display of material containing certain patently offensive depictions and allowed its application “to the transmission of Internet communications where the party initiating the communication knows that all of the recipients are minors.”[280]

It is impossible to appraise the merits of Justice O’Connor’s proposed line of severance without deep immersion in substantive First Amendment law that I shall forgo here. But two points seem clear. First, the majority’s conflation of principles governing statutory interpretation with principles governing statutory severance provided no adequate response to Justice O’Connor. Second, as I have meant to acknowledge, the Court—while giving as much effect as possible to severability clauses—must insist that some demands for judicial severance would go too far. As Justice O’Connor herself recognized in her opinion for the Court in Ayotte, “‘[i]t would certainly be dangerous if the legislature could set a net large enough to catch all possible offenders, and leave it to the courts to step inside’ to announce to whom the statute may be applied.”[281]

Overall, in response to the daunting question of when the demands of a severability clause such as that in Reno might go further than Article III permits, case-by-case appraisal seems the prudent course. Congress cannot reasonably demand that the courts trim down every vague and overbroad federal law so that its remnants will satisfy all pertinent constitutional tests. At the same time, just as Congress can authorize federal common law-making that would not otherwise be permissible under Article III,[282] Congress should be able to charge courts with some responsibilities to sever valid from invalid statutory applications that they would not sever otherwise.

c. Separability of State Statutes.—Because issues involving the separability of state statutes present state law questions on which state courts ordinarily have the authoritative last word,[283] it is unsurprising that federal courts have often exhibited uncertainty and confusion when addressing facial challenges to state statutes.[284] To make analytical progress, it will help to distinguish suits for injunctions against state statutes that are filed in the lower federal courts from cases in the Supreme Court on appeal from state court decisions.

In the former category of cases, the federal courts—up to and including the Supreme Court—should attend to state law severability rules more alertly than they sometimes have in the past. After finding a state statute invalid as written, a federal court entertaining a suit for an injunction or declaratory judgment should presumptively follow state severability law.[285] In doing so, however, federal courts need to test state severability law—like all other state law—against federal constitutional norms.

In cases on appeal to the Supreme Court from state court rulings, the premise that a facially invalid statute is not law at all, which section III(B)(3) discussed at length, mandates that state courts must entertain facial challenges to state statutes whenever defendants in enforcement actions plausibly claim that statutes are not severable and therefore invalid in all applications. There may not be many cases within that category, but it is past time for the Supreme Court to clear up the confusion that the old chestnut of Yazoo & Mississippi Valley Railroad v. Jackson Vinegar Co. has long engendered.

Separability clauses drafted by state legislatures also pose different questions from those that federal separability clauses present. In Whole Woman’s Health v. Hellerstedt,[286] the Supreme Court, by a divided vote of 5–3, determined that the district court had correctly upheld a facial challenge to two provisions of Texas law on the ground that they constituted undue burdens on abortion rights: (1) a requirement that doctors performing abortions have admitting privileges at a hospital within 30 miles of the abortion facility,[287] and (2) a mandate that all abortion facilities meet “the minimum” statutory standards applicable to “ambulatory surgical centers” under state law.[288] The Court ruled the provisions facially invalid despite a severability clause providing that if “any application of any provision . . . is found by a court to be invalid, the remaining applications of that provision to all other persons and circumstances shall be severed and may not be affected.”[289]

Writing for the Court, Justice Breyer rejected arguments that the Court must implement the severability clause:

A severability clause is not grounds for a court to “devise a judicial remedy that . . . entail[s] quintessentially legislative work.”

. . . .

. . . Texas’ attempt to broadly draft a requirement to sever “applications” does not require us to proceed in piecemeal fashion when we have found the statutory provisions at issue facially unconstitutional.[290]

Although the severability of a state statute normally depends on state law,[291] Justice Breyer seems correct that by enacting a severability clause, a state legislature cannot conscript a federal court to blue-pencil a facially invalid statute without clearer guidelines than the Texas legislature had provided.[292] State law cannot compel federal courts to perform functions that Article III forbids them to perform or that Congress has not validly imposed on them.[293] In my judgment, the extensive and potentially creative surgery that a court would have had to perform in order to conform the Texas statute in Whole Woman’s Health to constitutional norms plainly fell within the latter category and possibly within the former as well.

* * *

Overall, the severance of valid from invalid statutory applications presents challenges that are irreducibly complex. No algorithmic approach could do justice to the difficulty of the subject matter. Nevertheless, it would dispel a multitude of confusions if the Supreme Court would recognize that five central principles should control. First, if courts can identify lines along which to sever an otherwise facially invalid statute that would rescue it from invalidity, it is presumptively within their constitutional power to do so. Second, there may be Article III-based limits on courts’ permissible roles in severing statutes along creative, judgment-suffused lines, but those limits are best identified on a case-by-case basis. Third, the existence of Article III judicial power to sever otherwise-invalid statutes does not imply a judicial obligation to articulate lines of severance in every case in which such lines could possibly be drawn, especially in the absence of relevant argument from the parties. Fourth, Congress, by enacting severability clauses, can at least marginally expand the judicial obligation to sever otherwise-invalid statutes, but some Article III limits are irreducible. Fifth, state law provisions mandating the severance of valid from invalid statutory applications cannot commandeer federal courts into rewriting state statutes as a condition of their being able to vindicate federal constitutional rights.[294]

2. Severance of Statutory Provisions

The Supreme Court recurrently affirms that invalid statutory provisions should be severed and that others should continue in force unless (1) the remaining parts of a statute could not function coherently as law or (2) the evidence establishes that Congress “would not have enacted those provisions which are within its power, independently of that which is not.”[295] The first criterion should engender little controversy. To take a hypothetical example, imagine that Congress forbade universities to employ professors who refused to sign loyalty oaths and, in order to monitor compliance, required all professors to register their employment with the government.[296] If the bar to employment of professors who refused to sign loyalty oaths were held invalid, the requirement that professors register with the government would no longer advance any rational purpose and should be deemed nonseverable from the oath requirement.[297]

By contrast, the Supreme Court’s second articulated ground for deeming statutes nonseparable, involving whether Congress would have enacted some parts in the absence of others, is a recipe for confusion.[298] The question of what Congress would have done under counterfactual circumstances could conceivably be interpreted in various ways. But if cashed out as calling for a psychological and political investigation into how individual members of the House and Senate would have voted under counterfactual circumstances and for a tallying of their imagined votes, it mandates misguided inquiries. The practical difficulty is self-evident. There are too many members of Congress, subject to too many psychological dispositions and political pressures, as well as too many contingencies, for the hypothetical vote-tallying to be done rigorously.[299]

The deeper problem involves the legal irrelevance of a counterfactual vote-count. In analysis of whether a statute should be deemed severable or nonseverable, the question whether the presumption of severability has been overcome is one of statutory interpretation.[300] And the leading theories of statutory interpretation converge in denying that interpretive outcomes should depend on appraisals of what Congress might have done if provided with information that it did not have. Modern textualists insist that statutory meaning depends on how a reasonable person would understand a statute’s language in its linguistic context.[301] Purposivists interpret statutes based on the premise that they reflect attempts by reasonable people to pursue reasonable goals in a reasonable way.[302] For either theory to make separability turn on a counterfactual inquiry into whether Congress might have enacted a statute that it did not enact under unforeseen circumstances would be bizarrely anomalous.[303] If Congress had known that harsh penalties for drug offenses would have devastating effects on urban families, it might not have enacted some of the laws that it did enact. But even if it could be shown that existing drug laws have unwanted effects, and that Congress would not have enacted those laws if it knew what their effects would be, the law that Congress enacted would not cease to be the law, nor should applicable principles of interpretation change.[304]

Nevertheless, in National Federation of Independent Business v. Sebelius,[305] a dissenting opinion jointly authored by four Justices resurrected the test, cited above,[306] that makes counterfactual congressional intent the measure of statutory separability. Even though part of the Patient Protection and Affordable Care Act was amended and, thus, was codified in a chapter of the U.S. Code that included a severability clause,[307] severability was improper, according to the joint dissenting opinion, unless “Congress would have enacted” the otherwise valid provisions of a partially invalidated law “standing alone.”[308] With respect to a number of “minor provisions,” the joint opinion quoted a statement by the Senate Majority Leader that “[o]ften, a minor provision will be the price paid for support of a major provision. So, if the major provision were unconstitutional, Congress would not have passed the minor one.”[309] According to the four dissenting Justices, “[a]n automatic or too cursory severance of statutory provisions risks ‘rewrit[ing] a statute’” and “impos[ing] on the Nation, by the Court’s decree, its own new statutory regime.”[310]

This analysis is muddled, especially as it applied to the part of the ACA that was codified in a chapter that included a separability clause. As I have argued repeatedly, when Congress votes for a statute, it enacts all of the statute’s valid parts and applications absent persuasive evidence to the contrary.

Confusions notwithstanding, the position of the dissenters in NFIB v. Sebelius commanded a majority in Murphy v. NCAA.[311] In an opinion by Justice Alito, the Court ruled that a provision of the Professional and Amateur Sports Protection Act violated the “anticommandeering doctrine,” under which Congress may not single out state governmental officials for federal regulation, by forbidding state legislatures to authorize gambling on sporting events.[312] Having done so, the majority refused to sever and enforce other provisions of the statute that prohibited states along with other regulated parties from sponsoring sports gambling and barred private gambling enterprises from operating pursuant to state law. Justice Alito reasoned that if Congress had known that states could authorize gambling by private entities, it would neither have “wanted” to bar state lotteries, which “were thought more benign” than private gambling, nor prohibit only those private gambling operations that the states had authorized.[313]

That analysis holds up no better under critical scrutiny than that of the joint dissenting opinion in the NFIB case—as, fortunately, most of the Justices who joined the severability holding in Murphy v. NCAA have subsequently appeared to recognize. In Barr v. American Ass’n of Political Consultants, Justice Kavanaugh’s plurality opinion, in which Chief Justice Roberts and Justice Alito joined, recited that “some of the Court’s cases declare that courts should sever” a statute’s unconstitutional provisions “unless the statute created in its absence is one that Congress would not have enacted.”[314] But the opinion then immediately acknowledged that “this formulation often leads to an analytical dead end . . . because courts are not well equipped to imaginatively reconstruct a prior Congress’s hypothetical intent.”[315] Instead, Justice Kavanaugh continued, “the Court’s cases have . . . developed a strong presumption of severability,” which he endorsed as a preferable, more workable alternative.[316] In Murphy v. NCAA itself, Justice Thomas joined the majority opinion, but he also authored a concurrence in which he attacked severability doctrine generally and called for its reconsideration in a future case. “Because we have a Government of laws, not of men, we are governed by legislated text, not legislators’ intentions—and especially not legislators’ hypothetical intentions,” he wrote.[317] Finally, Justice Gorsuch joined Justice Thomas’s protest in Seila Law LLC v. Consumer Financial Protection Bureau[318] that “[a] text-based” approach to statutory interpretation, to which he subscribes, “does not allow a free-ranging inquiry into what Congress, faced with the limitations imposed by the Constitution, would have preferred had it known of a constitutional issue.”[319]

If the Supreme Court is prepared to reject the formula that ties severability to counterfactual congressional intent, this will be a major step in the right direction. Abandonment of that “analytical dead end” should restore the role of severability doctrine as a device for saving statutes from devastation, not for wreaking havoc on laws all of whose provisions Congress has enacted.

In abandoning inquiries into counterfactual congressional intent, however, the Justices should not throw out the baby—which in this case is the concept of statutory severability itself—with the bathwater. As with respect to the severance of invalid statutory applications, Justice Thomas has hinted that courts should renounce all inquiries into the severability of statutory provisions.[320] Instead, he has suggested—again echoing Professor Walsh—that when a party prevails in a constitutional challenge to a statute’s enforcement, the court should bar the statute’s enforcement against the party before it and eschew further analysis of statutory separability.[321] In Justice Thomas’s view, the court should leave the burden on other parties in other cases to demonstrate that the statute offends the Constitution as applied to them.

The practical defects of this approach—measured in terms of potential devastating effects—show most clearly in cases in which, as a logical matter, the constitutional difficulty with a statute arises from the interaction of multiple provisions.[322] Seila Law exemplified the danger. After a majority of the Justices held that the Dodd–Frank Wall Street Reform and Consumer Protection Act violated Article II by limiting the President’s power to remove the Director of the Consumer Financial Protection Bureau (CFPB), the Court concluded that the provision that restricted presidential removal authority should be severed and the remainder allowed to remain in effect.[323] By contrast, Justice Thomas maintained—logically and persuasively, I believe—that the unconstitutionality in the operation of the CFPB could not be traced to any single provision.[324] Rather, the infirmity lay in the conjunction of the restriction on the President’s authority to remove the CFPB’s Director with other provisions that purported to empower the CFPB to conduct investigations and demand the production of documents. According to Justice Thomas, which provision or provisions to sever therefore represented a policy choice that, he worried, might lie beyond the constitutional competence of courts to make.[325] He would therefore have held simply that the Dodd–Frank Act was unenforceable against the challenger—with the practical effect, emerging through the operation of stare decisis, that it could not be enforced against any other target of a CFPB investigation or enforcement action, either.[326]

As is true of Justice Thomas’s arguments against judicial severance of valid from invalid statutory applications, his proposal to forbid judicial judgments about how best to sever some statutory provisions from others sweeps too far. As I emphasized above, precedents involving federal common lawmaking as well as statutory separability make it untenable to maintain that Article III categorically bars courts from exercising judgment in determining how best to sever a statute. Moreover, in Seila Law—as I would guess would be true in most cases involving statutes whose defects inhere in the combined operation of multiple provisions—any policy choice that the Court needed to make to sever the statute was relatively small and interstitial. That choice was also consistent with the judgments most sensibly ascribed to Congress in enacting the Dodd–Frank Act.[327] As seven of the Justices recognized, many and perhaps most of the Act’s purposes could be realized with a Director subject to an unrestricted power of presidential removal.[328] By contrast, a ruling that effectively abolished the CFPB would have “trigger[ed] a major regulatory disruption and [done] appreciable damage to Congress’s work in the consumer-finance arena.”[329]

With a total eschewal of separability analysis being as practically problematic as inquiries into counterfactual congressional intent are unworkable, the Supreme Court should resolve questions involving the separation of valid from invalid provisions of federal statutes based on four clear, workable, and tested principles. First, Congress enacts all elements of statutory packages into law. This conceptualization provides the best explanation of why a presumption of separability ordinarily applies. Having been enacted by Congress, valid provisions can stand even after invalid provisions fall.

Second, the presumption of separability should prevail unless a statute is properly interpreted to displace it. Absent a statute that specifically negates the presumption of separability, to object to separation in some cases on the ground that it rewrites a statute, but to embrace separability in other cases, is logically inconsistent.

Third, although it may sometimes be possible to determine as a matter of statutory interpretation that a statute is not severable—either because it contains a nonseverability clause or because the remaining parts of a severed statute could not function coherently following the severance of invalid parts—whether Congress would have enacted some provisions in the absence of others is irrelevant. “Ordinary” principles of statutory interpretation, whether textualist or purposivist, should apply against the backdrop of a presumption that invalid provisions should be severed unless the remaining parts of a statute could not operate coherently.[330]

Fourth, when a constitutional violation results from the interaction of multiple statutory provisions, courts may need to exercise judgment to determine which to sever, but Article III creates no categorical bar to their doing so. In cases of this kind, “the default fallback option should be that which least upsets the operation of the statutory scheme as enacted.”[331]

Cases involving the separability of state statutes again present special problems because state, rather than federal, severability norms presumptively control. As with respect to issues involving the severance of invalid statutory applications, however, states—by enacting severability clauses—cannot effectively conscript federal courts into engaging in severability analysis that overreaches their authority under Article III or imposes obligations not contemplated by Congress when it conferred federal jurisdiction over constitutional challenges to state laws. Once again, Whole Woman’s Health v. Hellerstedt illustrates relevant limits.[332]

VI. Conclusion: Facial Challenges, Statutory Severing, and the Judicial Role

In the preceding Parts of this Article, I have discussed confusions that surround facial challenges, saving constructions, and statutory severing mostly in analytical terms. By way of conclusion, I now want to offer a further perspective by identifying the most prominent confusions’ etiology in untenable views about Article III limits on judicial power.

Not long ago, as the Introduction emphasized, the Supreme Court worried about whether the adjudication of facial challenges accorded with the historic and proper role of the Article III judiciary. But the Court’s onetime resistance to facial challenges rested partly on a misconception about the nature of constitutional rights as privileges to engage in particular, protected conduct and not, as is more frequently the case, as rights against rules. The Court also worried that broad authorization of facial challenges might result in premature judicial invalidation of a vast swath of statutes.

More recent Supreme Court practice reflects the reality that many constitutional rights are rights against rules, often defined by doctrinal tests that contemplate challenges to statutes on their faces. As the Court recurrently confronts facial challenges to statutory rules, severability questions have grown pressing.

As the Court deals with questions of statutory separability, at least three considerations have exerted influence. First, as Kevin Walsh’s historical scholarship demonstrates, early understandings of the role of the Article III courts rarely called for judicial invalidation of statutes on their faces.[333] Rather, Walsh concludes, courts held statutes invalid only insofar as they were repugnant to the Constitution.[334] Later, as judicial doctrine began to reflect an implicit if not an explicit recognition that many rights are rights against rules, courts developed separability doctrines that, according to Walsh, they had not had occasion to apply before.[335] Nonetheless, modern separability doctrine exhibits continuity with earlier judicial practice in one important respect: it seeks to preserve traditional understandings that judicial decisions upholding constitutional objections to the application of statutes should not routinely result in statutes being rendered wholly unenforceable.

Second, as Part V noted, powerful practical considerations support a continued judicial role in separating valid from invalid statutory provisions and applications. No Justices of the Supreme Court and few commentators have advocated a total abandonment of statutory separability, though Justice Thomas has announced his openness to a broad reconsideration of separability doctrine,[336] as has Justice Gorsuch.[337]

Tension emerges when a third consideration enters the picture. In modern debates about statutory interpretation, textualists have advanced strong claims that the proper judicial role is to interpret statutes as Congress wrote them, not to effect improvements or repairs.[338] According to textualists, courts should not conceive themselves as partners of Congress in making legislation workable and effective. Rather, courts should restrict themselves to interpreting and applying the text that Congress enacted. Insofar as the severing of statutes involves statutory repair, then the severing of statutes fits uneasily with textualist assumptions about the role of courts vis-à-vis that of Congress.

As should be clear from my analysis in earlier Parts of this Article, it is not obvious whether to classify statutory severance as a form of statutory repair. On the one hand, I have insisted, Congress enacts every part and prescribes every entailed application of a statute. On the other hand, once it is recognized that many rights are rights against rules, the severing of statutes in order to permit their continued partial enforcement becomes an exercise not merely in excision, but also in the articulation of statutory rules or subrules, not previously formulated by Congress, that satisfy constitutional requirements. The Supreme Court implicitly so acknowledges when, in cases such as Shelby County v. Holder and Whole Woman’s Health v. Hellerstedt, it determines that some forms of severing and rule revision more appropriately call for legislative judgment than judicial decision-making. (It did so in those cases despite the inclusion of severability clauses in the challenged statutes.) The question is often one of degree: how much judicial blue-penciling of which kinds should courts undertake in the absence of a severability clause, or how much blue-penciling can a severability clause permissibly require them to provide? In the view of Supreme Court majorities—with which I am inclined to agree—the severing in Ayotte and United States v. Georgia lay comfortably within the bounds of the permissible, while the line-drawing and rule-articulation exercises that the Court eschewed in Shelby County and Whole Woman’s Health did not.

Because the severing of statutes is a different function from the provision of narrowing constructions, one can embrace a textualist approach to statutory interpretation, decrying any judicial role in shaping legislation in the performance of that function, yet welcome a judicial role in seeking to render otherwise invalid legislation workable at least in part through statutory severance. In cases involving severability clauses, a good textualist would of course implement the statutory directive to the full extent that the Constitution permits.[339] Even in the absence of a severability clause, Justice Scalia, who was a leading champion of textualism, appeared to adopt a creative, statute-saving stance toward judicial severance in his opinion for a unanimous Court in United States v. Georgia. In it, he identified a line of severance that the Court could not plausibly have picked out via an exercise in direct or initial statutory interpretation, aimed at identifying the meaning of the statute that Congress had enacted. As section III(B)(2) emphasized, when litigants advance facial challenges, a preference for as-applied rather than facial invalidation often requires severability analysis.

To some, the sharp distinction that I have posited between the judicial role in interpreting statutes and the judicial role in severing them may seem pointlessly formalistic in many cases. If a court can act as a presumptive junior partner of Congress in the process of statutory severance—in seeking to rescue statutes from total invalidation by articulating narrowed rules that will survive applicable tests of constitutional validity—then why can a court not play a similar role when interpreting a statute in the first place?[340] Or, to work from the other side, if Article III and the separation of powers forbid courts to act as helpmates to Congress in construing statutes, why should the same strictures not apply when courts are asked to separate statutes’ valid from their invalid parts and applications?

In this Article, I have argued that there are important practical and conceptual reasons—bearing on the proper judicial role under Article III—to maintain a sharp analytical distinction between judicial provision of narrowing constructions and judicial severance of otherwise invalid statutes. But I have not meant to deny that, depending on the conception of the judicial role that underlies one’s theory of statutory interpretation, one may experience more or less tension in accepting that judges function as helpmates to Congress in severing otherwise invalid statutes.

Justice Clarence Thomas highlighted that tension in a concurring opinion in Murphy v. NCAA in which he argued that “the severability doctrine does not follow basic principles of statutory interpretation” and suggested that it should therefore be abandoned.[341] Although I agree wholeheartedly that “the severability doctrine does not follow basic principles of statutory interpretation,” my argument in this Article has pointed toward a different conclusion. Holding issues involving facial challenges, saving constructions, and statutory severability simultaneously in view, I have argued that there is no sensible replacement for an active judicial role in severing statutes that are vulnerable to facial attack and that could not be saved by statutory interpretation. The alternative would be to allow facial challenges to ravage the statute book. If an active and occasionally creative judicial role in severing statutes is incompatible with the vision of the separation of powers that underlies robust versions of textualism, the incompatibility furnishes good reason to reject those versions of textualism, not the other way around.

  1. .See, e.g., Murphy v. NCAA, 138 S. Ct. 1461, 1482–84 (2018) (refusing to sever and uphold secondary statutory provisions after invalidating a more central one); Johnson v. United States, 135 S. Ct. 2551, 2556–57, 2561 (2015) (upholding a facial challenge to a criminal statute on vagueness grounds and explaining that “our holdings squarely contradict the theory that a vague provision is constitutional merely because there is some conduct that clearly falls within the provision’s grasp”); Sessions v. Dimaya, 138 S. Ct. 1204, 1213 (2018) (extending Johnson’s holding that statutes can be facially invalid due to vagueness to apply to some statutes imposing civil liability); City of Los Angeles v. Patel, 135 S. Ct. 2443, 2449 (2015) (observing that although facial “challenges are ‘the most difficult . . . to mount successfully,’ the Court has never held that these claims cannot be brought under any otherwise enforceable provision of the Constitution” (quoting United States v. Salerno, 481 U.S. 739, 745 (1987))).
  2. .481 U.S. 739 (1987).
  3. .Id. at 745.
  4. .Ada v. Guam Soc. of Obstetricians and Gynecologists, 506 U.S. 1011, 1011–12 (1992) (Scalia, J., dissenting from the denial of certiorari).
  5. .541 U.S. 600 (2004).
  6. .Id. at 608.
  7. .Id. at 609 (quoting United States v. Raines, 362 U.S. 17, 22 (1960)).
  8. .Id.
  9. .Id. at 610.
  10. .Id.
  11. .But see United States v. Sineneng-Smith, 140 S. Ct. 1575, 1585 (2020) (Thomas, J., concurring) (calling for a reevaluation and rejection of First Amendment overbreadth doctrine that “exacerbates the many pitfalls” of facial challenges that constitute “a ‘disfavored’ method of adjudication”); June Med. Servs. L.L.C. v. Russo, 140 S. Ct. 2103, 2175 (2020) (Gorsuch, J., dissenting) (insisting that “the path for a litigant pursuing a facial challenge is deliberately difficult” and that “[t]ypically, a plaintiff seeking to render a law unenforceable in all of its applications must show that the law cannot be constitutionally applied against anyone in any situation” (citation omitted)).
  12. .See generally Marc E. Isserles, Overcoming Overbreadth: Facial Challenges and the Valid Rule Requirement, 48 Am. U. L. Rev. 359, 363 (1998) (differentiating “overbreadth facial challenge[s]” from “valid rule facial challenge[s]”). Obvious examples of non-overbreadth facial challenges include those alleging that statutes have discriminatory purposes, see, for example, Romer v. Evans, 517 U.S. 620, 632 (1996); prohibit speech on the basis of content, see, for example, Reed v. Town of Gilbert, 135 S. Ct. 2218, 2227 (2015); and attempt to commandeer state and local government officials to perform federally prescribed functions, see, for example, Murphy v. NCAA, 138 S. Ct. 1461, 1478 (2018).
  13. .See, e.g., United States v. Williams, 553 U.S. 285, 292 (2008) (“According to our First Amendment overbreadth doctrine, a statute is facially invalid if it prohibits a substantial amount of protected speech.”); United States v. Salerno, 481 U.S. 739, 745 (1987) (“[W]e have not recognized an ‘overbreadth’ doctrine outside the limited context of the First Amendment.” (citation omitted)).
  14. .See Richard H. Fallon, Jr., Fact and Fiction About Facial Challenges, 99 Calif. L. Rev. 915, 917–18 (2011) (demonstrating the connection between doctrinal tests and facial challenges and documenting, based on an empirical study of all constitutional cases decided by the Supreme Court during its 2009, 2004, 1999, 1994, 1989, and 1984 Terms, that “facial challenges to statutes are common, not anomalous”).
  15. .See, e.g., Roe v. Wade, 410 U.S. 113, 155 (1973) (noting that under strict scrutiny, “legislative enactments must be narrowly drawn to express only the legitimate state interests at stake”).
  16. .See Craig v. Boren, 429 U.S. 190, 197 (1976) (“[S]tatutory classifications that distinguish between males and females . . . must serve important governmental objectives and must be substantially related to achievement of those objectives.” (citation omitted)).
  17. .See, e.g., Trimble v. Gordon, 430 U.S. 762, 770–71 (1977) (ruling that a statutory provision discriminating against illegitimate children lacked a rational basis); Reed v. Reed, 404 U.S. 71, 76–77 (1971) (invalidating a statute that discriminated against women on grounds of irrationality).
  18. .United States v. Lopez, 514 U.S. 549, 558 (1995) (“[W]e have identified three broad categories of activity that Congress may regulate under its commerce power.” (citation omitted)).
  19. .Bd. of Trustees of the Univ. of Ala. v. Garrett, 531 U.S. 356, 365 (2001) (“Accordingly, § 5 legislation reaching beyond the scope of § l’s actual guarantees must exhibit ‘congruence and proportionality between the injury to be prevented or remedied and the means adopted to that end.’” (citation omitted)).
  20. .E.g., United States v. Williams, 553 U.S. 285, 297 (2008) (“We now turn to whether the statute, as we have construed it, criminalizes a substantial amount of protected expressive activity.”).
  21. .E.g., Barr v. Am. Ass’n of Political Consultants, 140 S. Ct. 2335, 2346 (2020) (plurality opinion) (affirming that “[c]ontent-based laws are subject to strict scrutiny”); R.A.V. v. City of St. Paul, 505 U.S. 377, 395 (1992) (noting that “‘the danger of censorship’ presented by a facially content-based statute requires that that weapon be employed only where it is ‘necessary to serve the asserted [compelling] interest’” (citations omitted)).
  22. .For a more extensive catalogue, see Fallon, supra note 14, at 935–40.
  23. .Id. at 926–31 (tracing the mistaken “conventional wisdom” concerning the rare and disfavored status of facial challenges to a relatively few high-profile and much-cited cases).
  24. .See, e.g., Barr, 140 S. Ct. at 2350 (“The Court’s cases have instead developed a strong presumption of severability.”); Henry Paul Monaghan, Overbreadth, 1981 Sup. Ct. Rev. 1, 6 (“[T]he normal judicial course is to approach the issue of constitutional validity with a presumption of separability.”); Michael C. Dorf, Facial Challenges to State and Federal Statutes, 46 Stan. L. Rev. 235, 250 (1994) (“Thus, the decision in Yazoo established a presumption that a statute’s constitutional and unconstitutional applications are severable.” (footnote omitted)).
  25. .See Fallon, supra note 14, at 948–49 (“Regardless of the test of constitutional
    validity that a court applies, the Supreme Court’s cases falsify any claim that, outside of the First Amendment overbreadth doctrine, the Justices consistently apply a presumption of severability . . . postpon[ing] . . . how, precisely, the severing . . . would occur.”).
  26. .429 U.S. 190 (1976).
  27. .Id. at 197.
  28. .Id. at 200–01.
  29. .Id. at 204, 210.
  30. .Id.
  31. .See Barry Friedman, The Will of the People: How Public Opinion Has Influenced the Supreme Court and Shaped the Meaning of the Constitution 181 (2009) (“Felix Frankfurter . . . compiled a list that indicated more than 220 state laws struck down between 1897 and 1938. A modern recount confirms this.” (footnote omitted)).
  32. .See Murphy v. NCAA, 138 S. Ct. 1461, 1489 (2018) (Ginsburg, J., dissenting) (characterizing the majority’s approach to statutory severability, which resulted in total invalidation of the Professional and Amateur Sports Protection Act, as a “wrecking ball”).
  33. .See infra notes 220–222 and accompanying text.
  34. .See, e.g., Richard H. Fallon, Jr., John F. Manning, Daniel J. Meltzer & David L. Shapiro, Hart & Wechsler’s The Federal Courts and the Federal System 172 (7th ed. 2015) [hereinafter Hart & Wechsler] (“The separability of a federal statute is, of course, a purely federal issue.”).
  35. .See, e.g., Dorchy v. Kansas, 272 U.S. 306, 308 (1926) (finding a state supreme court decision on the separability of parts of a state statute binding on the Supreme Court); Hart & Wechsler, supra note 34, at 171 (“[Q]uestions about the meaning and thus the separability of state statutes are primarily questions of state law.”).
  36. .See, e.g., Ayotte v. Planned Parenthood of N. New Eng., 546 U.S. 320, 330–31 (2006) (reciting the federal rule that severability depends on legislative intent and remanding for the lower court to determine the intent of the New Hampshire legislature without referencing New Hampshire severability rules).
  37. .See Hart & Wechsler, supra note 34, at 79–81 (discussing the canon). The canon has been traced, variously, to Mossman v. Higginson, 4 U.S. (4 Dall.) 12, 14 (1800), see United States v. Davis, 139 S. Ct. 2319, 2350 (2019) (Kavanaugh, J., dissenting) (“That uncontroversial principle of statutory interpretation dates back to the Founding era. See Mossman v. Higginson.”), and to Murray v. Charming Betsy, 2 U.S. (2 Cranch) 64, 118 (1804), see Skilling v. United States, 561 U.S. 358, 406 n.40 (2010) (“This cardinal principle has its roots in Chief Justice Marshall’s opinion for the Court in Murray v. The Charming Betsy . . . .” (citations omitted)).
  38. .485 U.S. 568 (1988).
  39. .Id. at 570–71.
  40. .Id. at 571–72.
  41. .Id. at 575–76.
  42. .Id. at 584.
  43. .Id. at 575 (citation omitted).
  44. .United States v. Stevens, 559 U.S. 460, 481 (2010) (“[T]his Court may impose a limiting construction on a statute only if it is ‘readily susceptible’ to such a construction.” (quoting Reno v. ACLU, 521 U.S. 844, 884 (1997))); Iancu v. Brunetti, 139 S. Ct. 2294, 2301 (2019) (noting that the constitutional avoidance canon “applies only when ambiguity exists. ‘[The Court] will not rewrite a law to conform it to constitutional requirements.’” (quoting Stevens, 559 U.S. at 481)).
  45. .546 U.S. 320 (2006).
  46. .Id. at 328.
  47. .Id.; see also Free Enter. Fund v. Pub. Co. Accounting Oversight Bd., 561 U.S. 477, 508 (2010) (quoting this formulation); Seila Law LLC v. Consumer Fin. Prot. Bureau, 140 S. Ct. 2183, 2209 (2020) (same).
  48. .Presser v. Illinois, 116 U.S 252, 269 (1886); Murray v. Charming Betsy, 6 U.S. (2 Cranch) 64, 118 (1804); see also Adrian Vermeule, Saving Constructions, 85 Geo. L.J. 1945, 1949 (1997) (distinguishing classical and modern avoidance).
  49. .The Supreme Court took clear note of the distinction in Chaplinsky v. New Hampshire, 315 U.S. 568 (1942), in which the petitioner argued that the state statute under which he had been convicted violated the First Amendment. Id. at 571. In rejecting that claim, the Court noted that the New Hampshire Supreme Court had authoritatively found that one provision of the statute, which had no application to Chaplinsky’s case, was separable from the provision under which he had been convicted, id. at 572, and that the latter, which literally forbade “offensive” speech, had been given an adequate saving construction by the state court. Id. at 573.
  50. .139 S. Ct. 2294 (2019).
  51. .Id. at 2301–02.
  52. .Id. at 2302.
  53. .See, e.g., Nat’l Fed’n of Indep. Bus. v. Sebelius, 567 U.S. 519, 691 (2012) (Scalia, Kennedy, Thomas, and Alito, JJ., dissenting in part) (noting that the authority to rewrite a statute lies with the legislature, rather than the judiciary).
  54. .138 S. Ct. 1461 (2018).
  55. .Id. at 1482.
  56. .See, e.g., Henson v. Santander Consumer USA Inc., 137 S. Ct. 1718, 1725 (2017) (“[I]t is never our job to rewrite a constitutionally valid statutory text under the banner of speculation about what Congress might have done had it faced a question that, on everyone’s account, it never faced.” (citation omitted)); Magwood v. Patterson, 561 U.S. 320, 334 (2010) (“We cannot replace the actual text with speculation as to Congress’ intent.”); W. Va. Univ. Hosps., Inc. v. Casey, 499 U.S. 83, 100–01 (1991) (explaining that it is not the Court’s role to reconstruct how Congress would have decided an issue if it had considered it).
  57. .See infra notes 312–320 and accompanying text.
  58. .Robert L. Stern, Separability and Separability Clauses in the Supreme Court, 51 Harv. L. Rev. 76, 101–02 (1937).
  59. .567 U.S. 519 (2012).
  60. .Id. at 574–75, 586, 588, 645, 691–92.
  61. .See id. at 530–34, 589 (describing the two issues the Court would be addressing in its opinion: the individual mandate and the Medicaid expansion).
  62. .See id. at 646–48, 661–63 (characterizing the Court’s ruling as a “blatant violation of the constitutional structure”). Although the four Justices normally characterized as liberals would have upheld the mandate under the Commerce Clause, the other five Justices disagreed. Chief Justice Roberts then offered a saving construction, which the liberals joined him in accepting, that interpreted the mandate as a tax on those who failed to purchase health insurance.
  63. .Texas v. United States, 945 U.S. 355 (5th Cir. 2019).
  64. .140 S. Ct. 1262 (2020).
  65. .Examples include Gonzales v. Carhart, 550 U.S. 124, 133, 169 (2007), in which a conservative majority ruled, over a vehement dissent by Justice Ginsburg, that a prohibition against so-called partial birth abortion could not be tested on its face, and Whole Woman’s Health v. Hellerstedt, 136 S. Ct. 2292, 2330–31 (2016), in which the conservative Justice Samuel Alito, writing in dissent, angrily protested the majority’s refusal to sever and thereby partially preserve an otherwise facially invalid statute.
  66. .558 U.S. 310 (2010).
  67. .Id. at 365.
  68. .570 U.S. 529 (2013).
  69. .Id. at 557.
  70. .See, for example, Florida Prepaid Postsecondary Education Expense Board v. College Savings Bank, 527 U.S. 627, 664 (1999), in which Justice Stevens, dissenting, thought that a statute that the Court invalidated on its face should have been deemed separable. For discussion, see infra note 250.
  71. .See Richard H. Fallon, Jr., The Nature of Constitutional Rights: The Invention and Logic of Strict Judicial Scrutiny 96–111 (2019). My analysis of the nature of rights builds on and adapts, but does not fully embrace, the argument of Matthew D. Adler, Rights Against Rules: The Moral Structure of American Constitutional Law, 97 Mich. L. Rev. 1 (1998) that all constitutional rights are rights against rules.
  72. .See generally Kevin C. Walsh, Partial Unconstitutionality, 85 N.Y.U. L. Rev. 738 (2010) (arguing that early American courts routinely enforced partially invalid statutory provisions except insofar as they were “repugnant” to the Constitution without framing questions of separability in the modern sense).
  73. .See Scott A. Keller & Misha Tseytlin, Applying Constitutional Decision Rules Versus Invalidating Statutes in Toto, 98 Va. L. Rev. 301, 308–09 (2012) (chronicling the Supreme Court’s initial adoption and use of the terms “facial challenge” and “as-applied challenge”).
  74. .See, e.g., R.A.V. v. City of St. Paul, 505 U.S. 377, 392 (1992) (finding the statute facially invalid because of its content-based selectivity in banning some but not all “fighting words”).
  75. .See Richard H. Fallon, Jr., As-Applied and Facial Challenges and Third-Party Standing, 113 Harv. L. Rev. 1321 (2000) (“To succeed with a constitutional claim . . . the challenger must always show that the statute has no valid subrule that can, under applicable law, be enforced
    against her.”).
  76. .See id. at 1336–37 (noting that arguments in support of constitutional claims often involve doctrinal tests).
  77. .See Fallon, supra note 14, at 941, 966 (summarizing survey results).
  78. .See, e.g., United States v. Williams, 553 U.S. 285, 304 (2008) (noting that “[a] conviction fails to comport with due process if the statute under which it is obtained fails to provide a person of ordinary intelligence fair notice of what is prohibited”); Palmer v. City of Euclid, 402 U.S. 544, 545 (1971) (invalidating statute “as applied to Palmer”).
  79. .553 U.S. 181 (2008).
  80. .Id. at 203.
  81. .Id. at 199 n.19.
  82. .Id. at 189–91 (citation omitted).
  83. .566 U.S. 30 (2012).
  84. .Id. at 34.
  85. .Id. at 34, 36, 43–44.
  86. .See 538 U.S. 721, 740 (2003) (upholding the family-care provision of the Family and Medical Leave Act as within Congress’s power to enact under Section Five of the Fourteenth Amendment).
  87. .See Michael C. Dorf, Fallback Law, 107 Colum. L. Rev. 303, 370 (2007) (arguing that “courts never face a choice of whether to sever invalid provisions or applications from valid ones, but instead must always decide how much to sever,” and that “[e]ven in contexts in which we say the Constitution forbids severability—such as the First Amendment overbreadth doctrine—what we really mean is that the Constitution forbids total severability” (emphasis omitted) (footnote omitted)).
  88. .See, e.g., Murphy v. NCAA, 138 S. Ct. 1461, 1484 (2018) (“[W]e hold that no provision of PASPA is severable from the provision directly at issue in these cases.”); R.R. Ret. Bd. v. Alton R.R., 295 U.S. 330, 362 (1935) (“[S]ome of the features we hold unenforcible [sic] . . . so affect the dominant aim of the whole statute as to carry it down with them.”).
  89. .The Supreme Court’s grant of certiorari in Texas v. California and California v. Texas, see supra note 64 and accompanying text, may confront it with a severability question of this kind. The question arises under the ACA. After the Court upheld the Act’s individual mandate to purchase health insurance as an exercise of Congress’s taxing power, see supra notes 59–62 and accompanying text, Congress, in 2017, repealed the tax on nonpurchase of insurance but left the remainder of the ACA intact. The first question is whether the individual mandate, which the Court held beyond Congress’s power to enact under the Commerce Clause in National Federation of Independent Business v. Sebelius, 567 U.S. 519 (2012), see supra note 62, remains constitutionally valid if no longer supported by a tax penalty. If not, the second question is whether the invalidity of the individual mandate would require invalidation of the entire ACA on grounds of nonseverability.
  90. .See Edward J. DeBartolo Corp. v. Fla. Gulf Coast Bldg. & Constr. Trades Council, 485 U.S. 568, 575 (1988) (“[W]here an otherwise acceptable construction of a statute would raise serious constitutional problems, the Court will construe the statute to avoid such problems unless such construction is plainly contrary to the intent of Congress.” (citing NLRB v. Catholic Bishop of Chi., 440 U.S. 490, 499–501, 504 (1979))).
  91. .See, e.g., Jennings v. Rodriguez, 138 S. Ct. 830, 836 (2018) (“Under the constitutional-avoidance canon, when statutory language is susceptible of multiple interpretations, a court may shun an interpretation that raises serious constitutional doubts and instead may adopt an alternative that avoids those problems. But a court relying on that canon still must interpret the statute, not rewrite it.”); Clark v. Martinez, 543 U.S. 371, 381 (2005) (Avoidance “is a tool for choosing between competing plausible interpretations of a statutory text”); Int’l Ass’n of Machinists v. Street, 367 U.S. 740, 749–50 (1961) (“[T]his Court will first ascertain whether a construction of the statute is fairly possible by which the question may be avoided.” (emphasis added) (quoting Crowell v. Benson, 285 U.S. 22, 62 (1932))).
  92. .See, e.g., John F. Manning, What Divides Textualists from Purposivists?, 106 Colum. L. Rev. 70, 84 (2006) (noting that “textualists generally forgo reliance on legislative history as an authoritative source of [a statute’s] purpose”).
  93. .See, e.g., Stephen Breyer, On the Uses of Legislative History in Interpreting Statutes, 65 S. Cal. L. Rev. 845, 848 (1992) (arguing that “[l]egislative history helps a court understand the context and purpose of a statute”).
  94. .See City of Boerne v. Flores, 521 U.S. 507, 520 (1997) (noting that there “must be congruence and proportionality between the injury to be prevented or remedied and the means adopted to that end”).
  95. .546 U.S. 151 (2006).
  96. .Brief for the United States as Petitioner at I, United States v. Georgia, 546 U.S. 151 (2006).
  97. .Bd. of Trustees of the Univ. of Ala. v. Garrett, 531 U.S. 356, 374 (2001) (“[T]he remedy imposed by Congress must be congruent and proportional to the targeted violation.”).
  98. .Brief for Respondents at 38–40, United States v. Georgia, 546 U.S. 151 (2006).
  99. .Georgia, 546 U.S. at 159.
  100. .Id. at 157.
  101. .Id. at 159.
  102. .Id. at 162 (Stevens, J., concurring).
  103. .The doctrine’s application to these diverse categories is longstanding, as recognized in the classic discussion of statutory severability. Stern, supra note 58, at 78–79 (“Questions of separability fall into two general classes. One relates to situations in which some applications of the same language in a statute are valid and other applications invalid; the other to statutes containing particular language—whether words, phrases, sentences or sections—which is invalid, and other language entirely constitutional.”); see also Gillian E. Metzger, Facial Challenges and Federalism, 105 Colum. L. Rev. 873, 885 (2005) (recognizing the difference between the severance of provisions and the severance of applications).
  104. .Cf. Reno v. ACLU, 521 U.S. 844, 883 (1997) (severing the term “or indecent” from a statutory prohibition against “obscene or indecent” communications).
  105. .Ayotte v. Planned Parenthood of N. New Eng., 546 U.S. 320, 323–24 (2006).
  106. .Id. at 328.
  107. .Id. at 332.
  108. .Id. at 331.
  109. .Id. at 328–29 (“[W]hen confronting a constitutional flaw in a statute, we try to limit the solution to the problem. We prefer, for example, to enjoin only the unconstitutional applications of a statute while leaving other applications in force, or to sever its problematic portions while leaving the remainder intact.” (internal citations omitted)).
  110. .See 546 U.S. 151, 158–59 (2006) (holding that Title II’s abrogation of a state’s sovereign immunity is valid in instances when a state’s conduct “actually violates the Fourteenth Amendment”).
  111. .481 U.S. 739, 745 (1987).
  112. .See supra note 105 and accompanying text.
  113. .See supra note 110 and accompanying text.
  114. .See Minnesota v. Mille Lacs Band of Chippewa Indians, 526 U.S. 172, 191 (1999) (“The inquiry into whether a statute is severable is essentially an inquiry into legislative intent.” (citing Regan v. Time, Inc., 468 U.S. 641, 653 (1984) (plurality opinion))); Reno v. ACLU, 521 U.S. 844, 883–84 (1997) (declining to sever a challenged statutory provision and observing that the Court has “declined to ‘dra[w] one or more lines between categories of speech covered by an overly broad statute, when Congress has sent inconsistent signals as to where the new line or lines should be drawn.’” (quoting United States v. Treasury Emps., 513 U.S. 454, 479 n.26 (1995))).
  115. .See, e.g., Ayotte v. Planned Parenthood of N. New Eng., 546 U.S. 320, 331 (2006) (remanding the case to the lower court and instructing it to determine whether severing the statute would accord with the intent of the state legislature).
  116. .See Barr v. Am. Ass’n of Political Consultants, 140 S. Ct. 2335, 2349 (2020) (plurality opinion) (“When Congress includes an express severability or nonseverability clause in the relevant statute, the judicial inquiry is straightforward. At least absent extraordinary circumstances, the Court should adhere to the text of the severability or nonseverability clause.”).
  117. .See, e.g., John Copeland Nagle, Severability, 72 N.C. L. Rev. 203, 222–24 (1993) (providing examples of the Supreme Court holding statutes nonseverable despite the presence of a severability clause).
  118. .The most characteristic problem arises when the Supreme Court believes that judicial severance of a statute would require a quasi-legislative rewriting with insufficient legislative guidance about how to implement the severability clause. For discussion, see infra notes 220–222 and accompanying text.
  119. .See Eric Fish, Judicial Amendment of Statutes, 84 Geo. Wash. L. Rev. 563, 566 (2016) (“[W]hen a judge finds a statute unconstitutional, the judge then issues a remedial order that changes the statute’s meaning so as to make it constitutionally valid.”); see also Laurence H. Tribe, The Legislative Veto Decision: A Law by Any Other Name, 21 Harv. J. on Leg. 1, 22 (1984) (articulating but not ultimately endorsing the view that statutory severance, even when effected pursuant to a severability clause, “disregard[s] the absence of any actual enactment of the severed law in accord with Article I’s strictures” with the result that “[t]he constitutional safeguards of bicamerality and presentment are . . . abandoned, and a new law is created by judicial fiat”).
  120. .See Tribe, supra note 119, at 25.Rather than conceiving of the court as enforcing the law “minus” its invalidated provision—a “law” the legislature never enacted—perhaps one should simply understand the court as resolving the controversy before it in terms of the entire body of law applicable to that controversy, the entire Act of Congress (not the Act “minus” any offending portion) plus the Constitution.

    Id. (footnote omitted).

  121. .See, e.g., United States v. Grace, 461 U.S. 171, 182–83 (1983) (invalidating a ban against expressive displays on Supreme Court grounds as applied to sidewalks but not otherwise).
  122. .140 S. Ct. 2335 (2020).
  123. .Id. at 2344–45.
  124. .Id. at 2343–44.
  125. .Id. at 2354.
  126. .Id. at 2343.
  127. .Id. at 2355.
  128. .See, e.g., United States v. Salerno, 481 U.S. 739, 745 (1987).
  129. .See Fallon, supra note 71, at 10.
  130. .E.g., Sabri v. United States, 541 U.S. 600, 609 (2004) (“[W]e have recognized the validity of facial attacks alleging overbreadth (though not necessarily using that term) in relatively few settings . . . .”).
  131. .See, e.g., Younger v. Harris, 401 U.S. 37, 52 (1971) (asserting that “[t]he power and duty of the judiciary to declare laws unconstitutional is in the final analysis derived from its responsibility for resolving concrete disputes” about the application of statutes to particular individuals and that “this vital responsibility, broad as it is, does not amount to an unlimited power to survey the statute books and pass judgment on laws before the courts are called upon to enforce them”).
  132. .See Sabri, 541 U.S. at 609 (“Facial challenges of this sort . . . invite judgments on fact-poor records [and] call for relaxing familiar requirements of standing, to allow a determination that the law would be unconstitutionally applied to different parties and different circumstances from those at hand.” (citation omitted)).
  133. .481 U.S. 739, 742–43.
  134. .Id. at 745, 751 (“A facial challenge to a legislative Act is, of course, the most difficult challenge to mount successfully, since the challenger must establish that no set of circumstances exists under which the Act would be valid.”).
  135. .Id. (holding that Salerno’s due process rights were not violated and “[t]he fact that the Bail Reform Act might operate unconstitutionally under some conceivable set of circumstances is insufficient to render it wholly invalid, since we have not recognized an ‘overbreadth’ doctrine outside the limited context of the First Amendment”).
  136. .See Isserles, supra note 12, at 397–99 (explicating the steps in Salerno’s analysis establishing that the challenged statute satisfied possibly applicable tests of facial validity under the Due Process Clause other than mere overbreadth).
  137. .Adler, supra note 71, at 3. Adler overstated the point by claiming that this is true of all constitutional rights. See, e.g., id. at 7 (claiming his view holds “for the entire array of substantive constitutional rights that figure in modern constitutional law”).
  138. .491 U.S. 397 (1989).
  139. .Id. at 400.
  140. .Id. at 420.
  141. .Adler, supra note 71, at 4–5.
  142. .See Clark v. Cmty. for Creative Nonviolence, 468 U.S. 288, 297–99 (1984) (holding that the First Amendment does not entitle those engaging in expressive conduct to protection from generally applicable, conduct-restricting rules designed to serve governmental “interest[s] unrelated to the suppression of expression”).
  143. .Adler, supra note 71, at 5–6.
  144. .See, e.g., Craig v. Boren, 429 U.S. 190, 208–10 (1976) (invalidating a statute that barred young men but not young women of the same age from purchasing low-alcohol beer).
  145. .Cf. Adler, supra note 71, at 157 (arguing that “the concept of unconstitutionality does not attach to the treatment of particular litigants; it attaches . . . to the enactment of statutes and other rules”).
  146. .See Fallon, supra note 71, at 111–22 (describing judicial tests that the Supreme Court has used to protect constitutional rights against government action that does not involve the writing or enforcement of rules).
  147. .494 U.S. 872 (1990) (holding that the Free Exercise Clause was not offended by application of Oregon drug laws to ceremonial ingestion of peyote and that the state could therefore deny unemployment compensation based on unlawful drug use).
  148. .See, e.g., Sherbert v. Verner, 374 U.S. 398, 409 (1963) (holding that a state could not withhold unemployment compensation from a claimant who had refused employment because it would have required her to work on her faith’s Sabbath Day).
  149. .See Fallon, supra note 71, at 114–15 (describing the categorical test adopted by the Supreme Court to enforce the Fourth Amendment guarantee against unreasonable searches and seizures).
  150. .See id. at 112–13 (discussing strict tests that the Supreme Court has applied when considering challenges to “discretionary decisions by government officials” that are based on race or religion).
  151. .527 U.S. 41 (1999).
  152. .Id. at 64.
  153. .135 S. Ct. 2551 (2015).
  154. .See, e.g., Ada v. Guam Soc’y of Obstetricians & Gynecologists, 506 U.S. 1011, 1011 (1992) (Scalia, J., dissenting from the denial of certiorari) (concluding that the Ninth Circuit’s determination that a statute was unconstitutional on its face “seems . . . wrong, since there are apparently some applications of the statute that are perfectly constitutional”).
  155. .Johnson, 135 S. Ct. at 2561. Justice Scalia had joined the majority opinion in Salerno and had previously protested that statutes regulating abortion could not be challenged on their faces if they had any valid applications. See supra note 4 and accompanying text.
  156. .See generally Sessions v. Dimaya, 138 S. Ct. 1204, 1210 (2018) (“[a]dhering to [the] analysis in Johnson” to hold that a statute’s definition of “crime of violence” was impermissibly vague); United States v. Davis, 139 S. Ct. 2319, 2326–27, 2236 (2019) (following Johnson and Dimaya to hold the residual clause of a statutory definition of “crime of violence” unconstitutionally vague).
  157. .See supra note 78 and accompanying text.
  158. .472 U.S. 491 (1985) (holding that the Court of Appeals erred in facially invalidating a Washington statute employed an impermissibly broad definition of prohibited obscenity); see Richard H. Fallon, Jr., Daniel J. Meltzer & David L. Shapiro, Hart & Wechsler’s The Federal Courts and the Federal System 191 (5th ed. 2003) (suggesting that Brockett “announce[s] the apparently ironic conclusion that a litigant whose speech is constitutionally unprotected enjoys advantages over a litigant whose speech is constitutionally privileged in seeking to bar a statute from being enforced at all”); Isserles, supra note 12, at 454 (characterizing Brockett as having “held that a litigant whose own conduct is protected under the First Amendment, and who thus can succeed on an as-applied challenge, is not entitled to assert an overbreadth challenge” (footnote omitted)).
  159. .Brockett, 472 U.S. at 504.
  160. .559 U.S. 460 (2010).
  161. .The challengers in Stevens were indicted for disseminating videos of dogfighting. Id. at 466. Despite the government’s having proposed a construction of the statute under which it would cover only “crush” videos, depictions of animal fighting, and other depictions of extreme animal cruelty, the Court noted specifically that it “need[ed] not and d[id] not decide whether a statute limited to crush videos or other depictions of extreme animal cruelty would be constitutional.” Id. at 482.
  162. .Id. at 473.
  163. .513 U.S. 454 (1995).
  164. .Id. at 477–78.
  165. .Id. at 457.
  166. .Id. at 457, 470.
  167. .Id. at 477.
  168. .See id. at 477–78 (acknowledging that “the occasional case requires [the Court] to entertain a facial challenge in order to vindicate a party’s right not to be bound by an unconstitutional statute” but holding that “[i]n this case, granting full relief to respondents . . . does not require passing on the [statute’s] applicability . . . to [other] Executive Branch employees” (citations omitted)).
  169. .See supra notes 95–102 and accompanying text.
  170. .226 U.S. 217 (1912).
  171. .Id. at 218.
  172. .Id. at 219.
  173. .Id.
  174. .Id.
  175. .Id. at 219–20.
  176. .See, e.g., Hart & Wechsler, supra note 34, at 170 (asserting that the tenability of the Yazoo approach depends on assumed premises of statutory severability that the Court did not articulate).
  177. .5 U.S. (1 Cranch) 137 (1803).
  178. .Monaghan, supra note 24, at 8 (“[I]n addition to a claim of privilege, a litigant has always been permitted to make another, equally ‘conventional’ challenge: He can insist that his conduct be judged in accordance with a rule that is constitutionally valid.” (footnote omitted)); see United States v. Davis, 139 S. Ct. 2319, 2323 (2019) (“In our constitutional order, a vague law is no law at all.”). Most commentators agree with Monaghan. See, e.g., Metzger, supra note 103, at 887 n.60 (“Scholars generally agree that the valid rule requirement is a basic constitutional principle.”); Isserles, supra note 12, at 367 (recognizing “a litigant’s right to be judged in accordance with a constitutionally valid rule of law”); Dorf, supra note 24, at 243 n.29 (“Monaghan correctly concludes that a litigant has a right not to be judged by an unconstitutional rule of law . . . .”). But see Adler, supra note 71, at 1391–1406 (reasserting that the valid rule requirement is not reflected in Supreme Court doctrine and that the application of invalid rules does not violate personal constitutional rights).
  179. .See, e.g., United States v. Lopez, 514 U.S. 549, 551 (1995) (holding that the Gun-Free School Zones Act of 1990 exceeded Congress’s powers under the Commerce Clause).
  180. .See, e.g., Seila Law LLC v. Consumer Fin. Prot. Bureau, 140 S. Ct. 2183, 2196 (2020) (“Our precedents have long permitted private parties aggrieved by an official’s exercise of executive power to challenge the official’s authority to wield that power while insulated from removal by the President.” (citation omitted)).
  181. .Cf. Nathan S. Chapman & Michael W. McConnell, Due Process as Separation of Powers, 121 Yale L.J. 1672, 1807 (2012) (arguing that due process historically required a valid legal authorization for executive action).
  182. .Yazoo & Miss. Valley R.R. v. Jackson Vinegar Co., 226 U.S. 217, 219–20 (1912).
  183. .New York ex rel. Hatch v. Reardon, 204 U.S. 152, 160–61 (1907).
  184. .See, e.g., Monaghan, supra note 24, at 6 (“[I]n reviewing state court decisions, the Supreme Court will presume that the state statute is separable unless the contrary otherwise clearly appears.”); Dorf, supra note 24, at 250 (“[T]he decision in Yazoo established a presumption that a statute’s constitutional and unconstitutional applications are severable.” (footnote omitted)).
  185. .In St. Louis Iron Mountain & Southern Railway v. Wynne, 224 U.S. 354 (1912), decided less than a year prior to Yazoo, the Court held a similar Arkansas statute unconstitutional as applied to require settlement of an excessive claim (for animals killed by railroads). Id. at 359–60.
  186. .See, e.g., Dorchy v. Kansas, 272 U.S. 306, 308 (1926) (acknowledging that the question of severability is one to be “passed upon by the state court”); Hart & Wechsler, supra note 34, at 171 (“[Q]uestions about the meaning and thus the separability of state statutes are primarily questions of state law.”).
  187. .545 U.S. 1 (2005).
  188. .Id. at 22.
  189. .See David L. Franklin, Facial Challenges, Legislative Purpose, and the Commerce Clause, 92 Iowa L. Rev. 41, 43 (2006) (“The reasoning and result of Raich strongly suggest that ‘as-applied’ challenges under the Commerce Clause will not receive a friendly reception at the Court, and that ‘facial’ challenges are now the norm.”).
  190. .Raich, 545 U.S. at 18–19.
  191. .It would, however, be just as accurate to say that a judicial ruling of facial validity does not strictly preclude subsequent as-applied challenges since as-applied challenges could never have succeeded in the first place.
  192. .See supra notes 151–158 and accompanying text.
  193. .521 U.S. 702, 735 (1997).
  194. .521 U.S. 793, 800 (1997).
  195. .Glucksberg, 521 U.S. at 739 (Stevens, J., concurring); Vacco, 521 U.S. at 809.
  196. .Id. at 735 n.24.
  197. .See generally Fallon, supra note 14, at 940–41 (documenting that facial challenges were both more frequent and more frequently successful than as-applied challenges during selected Supreme Court Terms).
  198. .See supra note 90 and accompanying text.
  199. .E.g., Azar v. Allina Health Servs., 139 S. Ct. 1804, 1815 (2019); Nat’l Ass’n of Mfrs. v. Dep’t of Def., 138 S. Ct. 617, 629 (2018); Puerto Rico v. Franklin Cal. Tax–Free Tr., 136 S. Ct. 1938, 1949 (2016).
  200. .See, e.g., Osborne v. Ohio, 495 U.S. 103, 120 (1990) (“We, moreover, have long respected the State Supreme Courts’ ability to narrow state statutes so as to limit the statute’s scope to unprotected conduct.”); Hart & Wechsler, supra note 34, at 171 (describing “the meaning . . . of state statutes” as “primarily” a “question[] of state law”).
  201. .See, e.g., Indiana ex rel. Anderson v. Brand, 303 U.S. 95, 100, 107 (1938) (rejecting a state court interpretation of state law in order to enforce rights under the Contracts Clause).
  202. .See R.A.V. v. City of St. Paul, 505 U.S. 377, 381 (1992) (characterizing the state court as having offered a saving “construction” when it had severed a statute after finding that it would be overbroad otherwise).
  203. .561 U.S. 358 (2010).
  204. .Id. at 402 (quoting 18 U.S.C. § 1346 (2006)).
  205. .Id. at 402, 414.
  206. .Id. at 405.
  207. .Id. at 405–07.
  208. .Id. at 408–09.
  209. .Id. at 423–24 (Scalia, J., concurring in part and concurring in the judgment).
  210. .See, e.g., Manning, supra note 92, at 91–93 (noting that whereas textualists determine statutory meaning based on “evidence about the way a reasonable person conversant with relevant social and linguistic practices would have used the words,” purposivists emphasize words’ policy contexts).
  211. .Skilling, 561 U.S. at 405.
  212. .Id. at 408 n.42.
  213. .In such cases, the Supreme Court frequently frames the applicable test as depending on a hypothetical, counterfactual congressional intent. See, e.g., Exec. Benefits Ins. Agency v. Arkison, 134 S. Ct. 2165, 2173 (2014) (“We ordinarily give effect to the valid portion of a partially unconstitutional statute so long as it remains fully operative as a law, and so long as it is not evident from the statutory text and context that Congress would have preferred no statute at all.” (citations omitted) (internal quotations omitted)). For critical discussion of this formula even as applied to resolve severability issues, see infra notes 248–249 and 316–321 and accompanying text.
  214. .See Skilling, 561 U.S. at 408 n.42 (“We seek to determine what ‘Congress would have intended’ in light of the Court’s constitutional holding.” (quoting United States v. Booker, 543 U.S. 220, 246 (2005))).
  215. .See, e.g., Richard A. Posner, Statutory Interpretation—in the Classroom and in the Courtroom, 50 U. Chi. L .Rev. 800, 816 (1983) (arguing that the effect of the avoidance canon is “to enlarge the already vast reach of constitutional prohibition beyond even the most extravagant modern interpretation of the Constitution—to create a judge-made constitutional ‘penumbra’ that has much the same prohibitory effect as the . . . Constitution itself”); Frederick Schauer, Ashwander Revisited, 1995 Sup. Ct. Rev. 71, 74, 98 (concluding that the avoidance canon permits judges to use disingenuous interpretations of statutes to “substitute their judgment for that of Congress” without assuming responsibility for rendering a constitutional holding).
  216. .Iancu v. Brunetti, 139 S. Ct. 2294, 2312–13, 2318 (2019) (Sotomayor, J., concurring in part and dissenting in part).
  217. .See, e.g., David H. Gans, Severability as Judicial Lawmaking, 76 Geo. Wash. L. Rev. 639, 652 (2008) (arguing that the Supreme Court employs three distinguishable approaches to severability that have yielded varying outcomes); Stern, supra note 58, at 78 (pointing out inconsistencies in the Supreme Court decisions involving severability of statutes).
  218. .See supra note 58 and accompanying text; see also Stern, supra note 58, at 101–02 (“[T]he Court avails itself of one [severability] formula or another in order to justify results which seem to it to be desirable for other reasons.”).
  219. .See, e.g., Fallon, supra note 14, at 953–54 (identifying the Supreme Court’s “inadvertence, inconsistency, or confusion” with regard to severability issues); Metzger, supra note 103, at 793 (noting that the Supreme Court “has not frequently acknowledged the important role played by severability”).
  220. .See Stern, supra note 58, at 79 (“In the few opinions before 1870 in which any mention was made of the fact that only some of the provisions of a statute were being held invalid, the Court assumed as obvious that ‘full effect will be given to such as are not repugnant to the Constitution.’” (quoting Bank of Hamilton v. Dudley, 27 U.S. 492, 526 (1829))); Walsh, supra note 72, at 768–69 (arguing that early courts declined to enforce statutory provisions insofar as they were “repugnant” to the Constitution, but did not frame the further “severability” question of whether a partially unconstitutional statute could survive as a problem for analysis in its own right).
  221. .See Barr v. Am. Ass’n of Political Consultants, 140 S. Ct. 2335, 2351 n.8 (2020) (plurality opinion) (explaining that when the Supreme Court invalidates a law as unconstitutional, it “does not formally repeal the law” but rather “recognizes that the Constitution is a ‘superior paramount law,’ and that a ‘legislative act contrary to the constitution is not law’ at all” (quoting Marbury v. Madison, 5 U.S. (1 Cranch) 137, 177 (1803))). 
  222. .See Dorf, supra note 87, at 370 (asserting the inevitability of statutory severance).
  223. .See Eric S. Fish, Choosing Constitutional Remedies, 63 UCLA L. Rev. 322, 322 (2016) [hereinafter Fish, Choosing] (arguing that courts should exercise discretion in adding or striking down statutory language, with the goal of ensuring that the resulting statute approximates the enacting legislature’s preferences as closely as possible); Eric S. Fish, Constitutional Avoidance as Interpretation and as Remedy, 114 Mich. L. Rev. 1275, 1279 (2016) (maintaining that the constitutional avoidance canon is partly remedial in character and “will allow judges to actually change a statute’s meaning by creatively reinterpreting it to render it constitutionally valid”). See generally Fish, supra note 119 (advancing a model of judicial review under which courts have a permissible role in amending statutes to repair constitutional defects).
  224. .See Fish, supra note 119, at 573 (advancing an approach under which “judges can change unconstitutional statutes in any way that will render them constitutional”).
  225. .See Fish, supra note 119, at 565–66.
  226. .443 U.S. 76 (1979).
  227. .Id. at 79–81.
  228. .Id. at 89.
  229. .Id. at 91.
  230. .Fish, supra note 119, at 593–94.
  231. .The Court so affirmed in Sessions v. Morales-Santana, 137 S. Ct. 1678 (2017). “The choice between” expanding and contracting the class of statutory beneficiaries “is governed by the legislature’s intent, as revealed by the statute at hand,” Justice Ginsburg wrote for the Court’s majority. Id. at 1698–99. On the facts of the case, which involved a provision of the immigration laws that discriminated impermissibly by making it easier for unwed citizen mothers than for either unwed citizen fathers or for married couples to pass on their citizenship to children born abroad, the Court opted to eliminate the exception favoring unwed citizen mothers. Id. at 1686. “Put to the choice, Congress, we believe, would have abrogated [the] exception.” Id. at 1700.
  232. .E.g., United States v. Booker, 543 U.S. 220, 245 (2005); Thornburgh v. Am. Coll. of Obstetricians & Gynecologists, 476 U.S. 747, 764 (1986); Heckler v. Mathews, 465 U.S. 728, 738 (1984).
  233. .E.g., Matthew D. Adler & Michael C. Dorf, Constitutional Existence Conditions and Judicial Review, 89 Va. L. Rev. 1105, 1171 (2003); Kent Barnett, To the Victor Goes the Toil—Remedies for Regulated Parties in Separation-of-Powers Litigation, 92 N.C. L. Rev. 481, 518–21 (2014); Fish, Choosing, supra note 223, at 327; Gans, supra note 217, at 643–45; Ryan Scoville, The New General Common Law of Severability, 91 Texas L. Rev. 543, 569–70 (2013); Evan H. Caminker, Note, A Norm-Based Remedial Model for Underinclusive Statutes, 95 Yale L.J. 1185, 1208 (1986); Robert L. Nightingale, Note, How to Trim a Christmas Tree: Beyond Severability and Inseverability for Omnibus Statutes, 125 Yale L.J. 1672, 1736 n.275 (2016).
  234. .Others who have rejected the characterization of severability as a remedy include Brian Charles Lea, Situational Severability, 103 Va. L. Rev. 735, 755–57 (2017); John Harrison, Severability, Remedies, and Constitutional Adjudication, 83 Geo. Wash. L. Rev. 56, 57–59, 88–89 (2014) and Erik R. Zimmerman, Supplemental Standing for Severability, 109 Nw. U. L. Rev. 285, 320–21 (2015).
  235. .Remedy, Black’s Law Dictionary (11th ed. 2019).
  236. .See Fallon, supra note 75, at 1331.
  237. .See generally Walsh, supra note 72 (“Prior to severability doctrine’s rise, courts routinely held an unconstitutional law void to the extent of repugnancy . . . . After severance emerged, partial unconstitutionality . . . became dependent on the satisfaction of a legislative-intent test.”).
  238. .Id. at 777.
  239. .Id.
  240. .140 S. Ct. 2183, 2211 (2020) (Thomas, J., dissenting).
  241. .Murphy v. NCAA, 138 S. Ct. 1461, 1486–87 (2018) (Thomas, J., concurring).
  242. .Ayotte v. Planned Parenthood of N. New Eng., 546 U.S. 320, 325 (2006) (internal quotation omitted).
  243. .United States v. Georgia, 546 U.S. 151, 162 (2006) (Stevens, J. concurring).
  244. .But cf. Thomas Campbell, Severability of Statutes, 62 Hastings L.J. 1495, 1514 n.92 (2011) (arguing that severability doctrine should be abolished and that statutes with any invalid application should be deemed unenforceable).
  245. .See, e.g., Reno v. ACLU, 521 U.S. 844, 884 (1997) (holding a statute nonseverable and therefore invalid in the absence of judicially identifiable lines defining a severed, enforceable statute).
  246. .See Ayotte, 546 U.S. at 327–28 (citing Planned Parenthood of Se. Pa. v. Casey, 505 U.S. 833, 879 (1992) (plurality opinion)) (suggesting that a statute requiring parental notification prior to abortions on minors would satisfy constitutional requirements if severed to exempt health-related emergencies).
  247. .See City of Ladue v. Gilleo, 512 U.S. 43, 52 (1994) (recognizing that exceptions to a speech restriction “may diminish the credibility of the government’s rationale for restricting speech in the first place”).
  248. .See Reno v. ACLU, 521 U.S. 844, 884–85 (1997) (“This Court ‘will not rewrite a . . . law to conform it to constitutional requirements.’” (quoting Virginia v. Am. Bookseller’s Ass’n, 484 U.S. 383, 397 (1988))); Ayotte, 546 U.S. at 329 (observing that “we restrain ourselves from rewrit[ing] state law to conform it to constitutional requirements” (quoting Virginia, 484 U.S. at 397)).
  249. .See, e.g., Antonin Scalia, Common-Law Courts in a Civil-Law System: The Role of United States Federal Courts in Interpreting the Constitution and Laws, in A Matter of Interpretation: Federal Courts and the Law 3, 17 (Amy Gutmann ed., 1997) (“We look for a sort of ‘objectified’ intent—the intent that a reasonable person would gather from the text of the law, placed alongside the remainder of the corpus juris.”); John F. Manning, Textualism and Legislative Intent, 91 Va. L. Rev. 419, 423 (2005) (“[T]extualists have sought to devise a constructive intent that satisfies the minimum conditions for meaningfully tracing statutory meaning to the legislative process.”).
  250. .See Seila Law LLC v. Consumer Fin. Prot. Bureau, 140 S. Ct. 2183, 2219–20 (2020) (Thomas, J., concurring in part and dissenting in part) (suggesting that statutory revision and excision are not judicial remedies, but rather methods of legislative reconstruction).
  251. .Id. at 2209.
  252. .Compare, for example, the dueling memoranda of Justices Stevens and Scalia about the availability of facial challenges to anti-abortion statutes in Janklow v. Planned Parenthood, 517 U.S. 1174, 1175 (Stevens, J.), 1176–80 (Scalia, J., dissenting from denial of certiorari). See also Ada v. Guam Soc’y of Obstetricians and Gynecologists, 506 U.S. 1011, 1011–12 (1992) (Scalia, J., dissenting from denial of certiorari) (arguing that a facial challenge must be rejected unless the statute is unconstitutional in all its applications).
  253. .Ayotte, 546 U.S. at 329.
  254. .514 U.S. 549 (1995).
  255. .See Metzger, supra note 103, at 909 (noting that “it seems especially likely that Lopez would have come out differently had the School Zones Act included a jurisdictional element”). In the aftermath of Lopez, Congress amended the Gun-Free School Zones Act to require proof that a gun “has moved in or . . . otherwise affects interstate or foreign commerce.” 18 U.S.C. § 922(q)(2)(A) (2018). 
  256. .Cf. Metzger, supra note 103, at 930 (observing that severing arguably “would stray over the line from judicial narrowing to judicial rewriting of a challenged statute”).
  257. .546 U.S. at 331.
  258. .530 U.S. 914, 929 (2000).
  259. .See United States v. L. Cohen Grocery Co., 255 U.S. 81, 88 (1921) (raising “on our own motion” an interpretation of a challenged statute that “would render a consideration of the constitutional questions unnecessary”). An example of relatively creative lines of proposed severance may come from Justice Stevens’s dissenting opinion in Florida Prepaid Postsecondary Education Expense Board v. College Savings Bank, 527 U.S. 627 (1999), which sustained a facial challenge to a federal statute that sought to abrogate state sovereign immunity in suits for patent infringement. Id. at 647–48. According to the Court’s majority, the statute was not congruent and proportional to an identified pattern of constitutional violations, as required for valid legislation under Section Five of the Fourteenth Amendment, in part because, even though patents are a species of property, merely negligent deprivations of property do not violate the Due Process Clause. Id. at 645. Dissenting, Justice Stevens implied that the Court could have drawn a line between negligent and willful patent infringements and upheld the statute as applied to the latter. Id. at 653–54 (“The question presented by this case, then, is whether the Patent Remedy Act . . . may be applied to willful infringement.”).
  260. .See Barr v. Am. Ass’n of Political Consultants, 140 S. Ct. 2335, 2349 n.6 (2020) (plurality opinion) (recognizing that regardless of whether “Congress enacts a law with a severability clause and later adds new provisions to the statute” or “subsequently enact[s] a severability clause that applies to the existing statute . . . , the text of the severability clause remains central to the severability inquiry”).
  261. .See, e.g., Textile Workers v. Lincoln Mills, 353 U.S. 448, 456 (1957) (holding that the law applicable to suits under a provision of the Labor Management Relations Act was federal common law, “which the courts must fashion from the policy of our national labor laws”).
  262. .See Lea, supra note 234, at 781 (identifying “the text, structure, legislative history, and purpose of the statutory scheme, as well as the practical effects of the possible fallback options” as relevant factors to consider).
  263. .See, e.g., Barr v. Am. Ass’n of Political Consultants, 140 S. Ct. 2335, 2349–51 (2020) (plurality opinion) (noting that the Court should generally defer to severability clauses); Seila Law LLC v. Consumer Fin. Prot. Bureau, 140 S. Ct. 2183, 2209 (2020) (noting that the Court’s task is simplified when Congress supplies a severability clause).
  264. .See, e.g., Nagle, supra note 117, at 222–25 (noting decisions that construed severability clauses as creating mere presumptions of severability); Stern, supra note 58, at 78–79, 117 (identifying instances in which statutes containing severability clauses “have been held inseparable and totally void”); see also Eric S. Fish, Severability as Conditionality, 64 Emory L.J. 1293, 1300–09 (2015) (canvassing the Court’s varied historical approaches to severability).
  265. .See, e.g., Hill v. Wallace, 259 U.S. 44, 70–71 (1922) (holding that the severability clause of the Futures Trading Act “did not intend the court to dissect an unconstitutional measure and reframe a valid one out of it by inserting limitations it does not contain”); cf. Barr, 140 S. Ct. at 2350–52 (plurality opinion) (noting but rejecting an argument to this effect).
  266. .570 U.S. 529 (2013).
  267. .Id. at 553, 557.
  268. .Id. at 557.
  269. .See 52 U.S.C. § 10313 (2012) (“If any provision of [this Act] or the application thereof to any person or circumstances is held invalid, the remainder of [the Act] and the application of the provision to other persons not similarly situated or to other circumstances shall not be affected thereby.”).
  270. .Shelby County, 570 U.S. at 554.
  271. .521 U.S. 844, 884 (1997). Although Reno is probably the leading modern case, the Supreme Court’s struggles with issues involving when to give full effect to federal severability clauses go back for more than a century. See generally Nagle, supra note 117, at 222 (noting that “[t]he first severability clauses appeared late in the nineteenth century, and they became much more common around 1910”); id. at 234–46 (discussing judicial interpretation of severability clauses).
  272. .Reno, 521 U.S. at 844.
  273. .Id. (quoting 47 U.S.C. § 233(d) (1994 ed., Supp. II)).
  274. .47 U.S.C. § 608 (1994).
  275. .See Reno, 521 U.S. at 885 (affirming the decision of the district court).
  276. .See id. at 883 (“Therefore, we will sever the term ‘or indecent’ from the statute, leaving the rest of §233(a) standing.”).
  277. .See id. at 884–85 (explaining that when Congress has not set clear lines, severability may be an improper invasion of the legislative domain).
  278. .521 U.S. at 884 (quoting Virginia v. Am. Bookseller’s Ass’n, 484 U.S. 383, 397 (1988)).
  279. .See id. at 895–97 (O’Connor, J., concurring in part and dissenting in part).
  280. .Id. at 895.
  281. .Ayotte v. Planned Parenthood of N. New Eng., 546 U.S. 320, 330 (2006) (quoting United States v. Reese, 92 U.S. 214, 221 (1876)).
  282. .See supra note 261 and accompanying text.
  283. .See, e.g., Dorchy v. Kansas, 272 U.S. 306, 308 (1926); Hart & Wechsler, supra note 34, at 171.
  284. .The uncertainty and confusion have sometimes extended to the question whether federal courts should “abstain” from adjudication of facial challenges to state statutes pending authoritative interpretation of the challenged statutes by states’ highest courts. See generally Hart & Wechsler, supra note 34, at 1110–11 (discussing when an issue of state law is sufficiently uncertain to warrant abstention under Pullman abstention doctrine).
  285. .See Ayotte, 546 U.S. at 331 (remanding to lower court to ascertain whether issuance of an injunction prohibiting particular applications of a state statute would accord with the state legislature’s intent).
  286. .136 S. Ct. 2292 (2016).
  287. .Id. at 2310–11.
  288. .Id. at 2314, 2318.
  289. .Id. at 2319 (citations omitted).
  290. .Id. at 2319 (quoting Ayotte, 546 U.S. at 329). 
  291. .Leavitt v. Jane L., 518 U.S. 137, 139 (1996) (per curiam) (“Severability is of course a matter of state law.”); see also Scoville, supra note 233, at 564–69 (concluding that after Erie Railroad Co. v. Tompkins, 304 U.S. 64 (1938), “the Court generally settled upon the rule that the sovereign whose statute is at issue dictates the severance test”).
  292. .Most of the Court’s substantive analysis in Whole Woman’s Health rests on the ground that the challenged provisions are invalid based on their effects, 136 S. Ct. at 2310–18, but the opinion also quoted language affirming that statutes can be invalid if they have “the purpose . . . of presenting a substantial obstacle to a woman seeking an abortion.” Id. at 2300 (quoting Planned Parenthood of Se. Pa. v. Casey, 505 U.S. 833, 878 (1992) (plurality opinion)). The decision regarding facial challenges and nonseverability might have fit better with seemingly settled doctrine if the Court had concluded that the challenged provisions had the forbidden purpose of unduly burdening abortion rights and that that forbidden purpose rendered them invalid in all possible applications.
  293. .See, e.g., Hollingsworth v. Perry, 570 U.S. 693, 715 (2013) (“[S]tanding in federal court is a question of federal law, not state law. . . . [T]he fact that a State thinks a private party should have standing . . . cannot override . . . [t]he Article III requirement that a party invoking the jurisdiction of a federal court seek relief for a personal, particularized injury . . . .”).
  294. .Federal courts might appropriately withdraw or modify any injunctions against the enforcement of state statutes of declarations of their invalidity if a state court subsequently severed an otherwise facially invalid state statute successfully. See David L. Shapiro, State Courts and Federal Declaratory Judgments, 74 Nw. U. L. Rev. 759, 768–69 (1979).
  295. .Buckley v. Valeo, 424 U.S. 1, 108 (1976) (per curiam) (quoting Champlin Ref. Co. v. Corp. Comm’n of Okla., 286 U.S. 210, 234 (1932)). See Exec. Benefits Ins. Agency v. Arkison, 573 U.S. 25, 37 (2014) (“We ordinarily give effect to the valid portion of a partially unconstitutional statute so long as it remains fully operative as a law, and so long as it is not evident from the statutory text and context that Congress would have preferred no statute at all.” (citations omitted)). The Court appears first to have stated the applicable test in essentially its current form in Champlin Refining Co., 286 U.S. at 234. See Nagle, supra note 117, at 214.
  296. .Cf. Speiser v. Randall, 357 U.S. 513, 514–15, 527–29 (1958) (invalidating a state statute that denied tax exemptions to otherwise-eligible veterans who refused to sign loyalty oaths).
  297. .There are a number of nonhypothetical examples. See, e.g., Minnesota v. Mille Lacs Band of Chippewa Indians, 526 U.S. 172, 191–94 (1999) (finding executive order nonseverable because the unconstitutional part was integral to its single, coherent policy); Williams v. Standard Oil Co. of La., 278 U.S. 235, 242–45 (1929) (same, as applied to statute); Hill v. Wallace, 259 U.S. 44, 70–72 (1922) (holding that “Section 4 with its penalty to secure compliance with the regulations of Boards of Trade [was] so interwoven with those regulations that they [could] not be separated,” although not striking down the entire statute).
  298. .See, e.g., Lea, supra note 234, at 774–75 (pointing out the potential difficulties of ascertaining the legislature’s intent); Walsh, supra note 72, at 753–54 (noting the absurd results that such an approach could yield).
  299. .See Barr v. Am. Ass’n of Political Consultants, 140 S. Ct. 2335, 2350 (2020) (plurality opinion) (noting that “experience shows that” the formula that calls for courts to inquire into whether Congress would have enacted some provisions in the absence of others “often leads to an analytical dead end”).
  300. .See Mille Lacs Band of Chippewa Indians, 526 U.S. at 191 (quoting Regan v. Time, Inc., 468 U.S. 641, 653 (1984) (plurality opinion)) (stating that an inquiry of whether a statute is severable is, in essence, an inquiry into the legislature’s intent).
  301. .See Antonin Scalia & Bryan A. Garner, Reading Law: The Interpretation of Legal Texts 16 (2012) (“In their full context, words mean what they conveyed to reasonable people at the time they were written.”); Frank H. Easterbrook, The Role of Original Intent in Statutory Construction, 11 Harv. J.L. & Pub. Pol’y 59, 65 (1988) (“The meaning of statutes is to be found not in the subjective, multiple mind of Congress but in the understanding of the objectively reasonable person.”); John F. Manning, The Absurdity Doctrine, 116 Harv. L. Rev. 2387, 2392–93 (2003) (“[Textualists] ask how a reasonable person, conversant with the relevant social and linguistic conventions, would read the text in context.”).
  302. .See Henry M. Hart, Jr. & Albert M. Sacks, The Legal Process: Basic Problems in the Making and Application of Law 1374, 1378 (William N. Eskridge, Jr. & Philip P. Frickey eds., 1994) (postulating the interpretive assumption that the legislature consists of reasonable people seeking to promote reasonable goals in reasonable ways); Philip P. Frickey, From the Big Sleep to the Big Heat: The Revival of Theory in Statutory Interpretation, 77 Minn. L. Rev. 241, 249 (1992) (observing that the paradigmatically purposivist Legal Process theory of Henry Hart and Albert Sacks calls for judges to “conjure up plausible organizing purposes for” statutes, rather than discover them, and then to interpret statutes in light of the ascribed purposes).
  303. .See Nagle, supra note 117, at 206 (characterizing an approach that depends on projections of legislative votes as “non-textualist” and the questions that it poses as “unanswerable”). The textualist Justices Scalia and Thomas once appeared to recognize as much. See Rita v. United States, 551 U.S. 338, 383 n.7 (2007) (Scalia, J., joined by Thomas, J., concurring in part and concurring in the judgment) (“The judicial role when conducting severability analysis is limited to determining whether the balance of a statute that contains an unconstitutional provision is capable of functioning independently.” (internal quotation omitted)).
  304. .Recent Supreme Court precedents have also affirmed that “[w]e cannot replace the actual text with speculation as to Congress’ intent.” Magwood v. Patterson, 561 U.S. 320, 334 (2010); see also Henson v. Santander Consumer USA Inc., 137 S. Ct. 1718, 1725 (2017) (“[I]t is never our job to rewrite a constitutionally valid statutory text under the banner of speculation about what Congress might have done had it faced a question that, on everyone’s account, it never faced.”); W. Va. Univ. Hosps., Inc. v. Casey, 499 U.S. 83, 100–01 (1991).
  305. .567 U.S. 519 (2012).
  306. .See supra note 53 and accompanying text.
  307. .See 42 U.S.C. § 1303 (2018) (“If any provision of this chapter, or the application thereof to any person or circumstance, is held invalid, the remainder of the chapter, and the application of such provision to other persons or circumstances shall not be affected thereby.”).
  308. .Sebelius, 567 U.S. at 693 (Scalia, Kennedy, Thomas, and Alito, JJ., dissenting).
  309. .Id. at 704.
  310. .Id. at 692.
  311. .138 S. Ct. 1461 (2018).
  312. .Id. at 1478.
  313. .Id. at 1482; see also Randall v. Sorrell, 548 U.S. 230, 262 (2006) (plurality opinion of Breyer, J., joined by Roberts, C.J., and Alito, J.) (declining to sever an invalid state law prohibition on campaign contributions and thereby permit enforcement of a statute’s remaining provisions, even though other provisions might have remained fully operative, since resulting consequences suggested that the legislature “would have intended” the Court not to sever).
  314. .Barr v. Am. Ass’n of Politcal Consultants, 140 S. Ct. 2335, 2350.
  315. .Id.
  316. .Id.
  317. .Murphy v. NCAA, 138 S. Ct. 1461, 1487 (2018) (Thomas, J., concurring) (citation omitted).
  318. .140 S. Ct. 2183 (2020).
  319. .Id. at 2222 (Thomas, J., concurring in part and dissenting in part) (citation omitted).
  320. .See Seila Law, 140 S. Ct. at 2219–20 (Thomas, J., concurring in part and dissenting in part); Murphy, 138 S. Ct. at 1487 (Thomas, J., concurring).
  321. .See Murphy, 138 S. Ct. at 1486 (Thomas, J., concurring) (“[W]hen early American courts determined that a statute was unconstitutional, they would simply decline to enforce it in the case before them. . . . ‘[T]here was no “next step” . . . .’” (quoting Walsh, supra note 72, at 755–66, 777) (citation omitted)).
  322. .For an insightful discussion of cases of this kind, see Lea, supra note 234, at 781–82.
  323. .Seila Law, 140 S. Ct. at 2207–10.
  324. .Id. at 2222–24 (Thomas, J., concurring in part and dissenting in part).
  325. .Id. at 2224.
  326. .Id.
  327. .See Lea, supra note 234, at 781 (identifying “the text, structure, legislative history, and purpose of the statutory scheme, as well as the practical effects of the possible fallback options” as relevant factors to consider).
  328. .See Seila Law, 140 S. Ct. at 2209–10 (framing the question before the Court as whether “Congress would have preferred a dependent CFPB to no agency at all” and concluding that “the answer seems clear”).
  329. .Id.
  330. .As Nagle, supra note 117, at 229, points out, different theories of statutory interpretation could imaginably call for “at least four different inquiries,” involving examinations of:(1) the meaning of the statute itself (a textualist approach); (2) the purpose of the statute (a purposive approach); (3) the legislature’s intent when it enacted the statute (an intentionalist approach); or (4) what the legislature would have done had it considered the issue (an imaginative reconstruction approach).


  331. .Lea, supra note 234, at 781.
  332. .On the consequences that would ensue if a state court were to sever the challenged statute with the effect that it thereafter satisfied federal constitutional norms, see Shapiro, supra note 294, at 768–69.
  333. .Walsh, supra note 72, at 756–57.
  334. .See id. at 755–61 (contrasting judicial refusals to give effect to laws insofar as they were invalid with judicial decisions to “strike down” statutes).
  335. .See id. at 766–77 (tracing the development of modern severability analysis).
  336. .Murphy v. NCAA, 138 S. Ct. 1461, 1487 (2018) (Thomas, J., concurring) (“[O]ur modern severability precedents are in tension with longstanding limits on the judicial power. And, though no party in this case has asked us to reconsider these precedents, at some point, it behooves us to do so.”); United States v. Sineneng-Smith, 140 S. Ct. 1575, 1585 (2020) (Thomas, J., concurring) (advocating reconsideration of First Amendment overbreadth doctrine in light of “pitfalls” of facial challenges).
  337. .See Barr v. Am. Ass’n of Political Consultants, 140 S. Ct. 2335, 2367 (2020) (Gorsuch, J., concurring in the judgment in part and dissenting in part) (citing the Court’s decision to sever a provision of a federal statute with the result that it prohibited more speech, not less, as furnishing “all the more reason to reconsider our course”).
  338. .See, e.g., King v. Burwell, 576 U.S. 473, 514 (2015) (Scalia, J., dissenting) (maintaining that “[t]his Court . . . has no free-floating power to rescue Congress from its drafting errors” and that “[o]nly when it is patently obvious to a reasonable reader that a drafting mistake has occurred may a court correct the mistake” (internal quotation marks omitted)). See generally Manning, supra note 301, at 2846 (arguing that “negating perceived absurdities that arise from clear statutory texts in fact entails the exercise of judicial authority to displace the outcomes of the legislative process” and that “the Court should permit such displacement only when the legislature’s action violates the Constitution”).
  339. .See Barr, 140 S. Ct. at 2349 n.6 (recognizing that regardless of whether “Congress enacts a law with a severability clause and later adds new provisions to that statute” or “subsequently enact[s] a severability clause that applies to the existing statute,” “the text of the severability clause remains central to the severability inquiry”).
  340. .For discussion of the judiciary’s junior partnership role in matters of statutory interpretation, see generally Richard H. Fallon, Jr., On Viewing the Courts as Junior Partners of Congress in Statutory Interpretation Cases: An Essay Celebrating the Scholarship of Daniel J. Meltzer, 91 Notre Dame L. Rev. 1743 (2016).
  341. .Murphy, 138 S. Ct. at 1486; Seila Law LLC v. Consumer Fin. Prot. Bureau, 140 S. Ct. 2183, 2219–20 (2020) (Thomas, J., joined by Gorsuch, J., concurring in part and dissenting in part).