Emory Law Journal

Severability as Conditionality
Eric S. Fish Ph.D. candidate in Law, Yale Law School. For helpful feedback on prior drafts I would like to thank Kiel Brennan-Marquez, William Eskridge, Heather Gerken, Emily Gerrick, Abbe Gluck, Daniel Hemel, Mike Knobler, Lisa Ouellette, Alexander Schwab, Gordon Silverstein, Kate Stith, Maggie Wittlin, my Ph.D. classmates, and the participants in the Yale Doctoral Colloquium. I would also like to thank the editors of the Emory Law Journal for their excellent editing.

The Supreme Court currently operates under the premise that if it finds one part of a law unconstitutional, it can strike down other parts as well. In National Federation of Independent Business v. Sebelius, four justices would have exercised this power to strike down the entire Affordable Care Act on the basis of one unconstitutional provision. But it is not clear where the Court finds this power to declare laws inseverable. And that lack of clarity has created a doctrinal muddle wherein the Court applies several inconsistent tests. In this Article, I seek to clarify the scope of the inseverability power by considering several different theories of its source. Three such theories are implicit in the current judicial doctrine and academic debate about severability: (1) that it is an equitable remedial power, akin to the power to issue a civil injunction; (2) that it is a variant of intentionalist statutory interpretation, wherein courts strike down further provisions of a partially unconstitutional law so as to preserve the legislators’ hypothetical intentions; and (3) that it is a judicial contract remedy applied to legislative deals. This Article explores these three theories, teasing out their respective logics and showing that they are implausibly broad and inconsistent with Article III of the Constitution.

This Article then develops and defends a fourth, narrower theory: that a court can declare a statute inseverable only where the legislature has made one part of the statute conditional on the continued validity of another. Such conditionality can most easily be found through explicit inseverability clauses. But it can also be found implicitly (analogous to the implied repeal and implied preemption doctrines) where severing a provision would make nonsense of a statute’s language, or where otherwise valid parts of a statute cannot have legal effect or do not serve any purpose without the unconstitutional provision. The main cost of this theory is that it only permits inseverability in limited circumstances, so it does not allow courts to rewrite statutes to avoid the perverse consequences of judicial review. The benefits are that the theory is consistent with Article III, and that it prevents judges from acting too much like legislators. Two further implications follow from the conditionality theory: that there must be a party with standing to challenge an inseverable provision before a federal court can strike it down, and that the proper unit of analysis for severability questions is the entire legislative code (rather than a single act or bill).

Introduction

In National Federation of Independent Business v. Sebelius (NFIB), the Supreme Court upheld a provision of the Affordable Care Act (ACA) requiring individuals to purchase health insurance. 1132 S. Ct. 2566 (2012). In their dissent, Justices Scalia, Kennedy, Thomas, and Alito declared not only that they would have held this individual mandate unconstitutional, but also that they would have struck down the rest of the ACA in its entirety. 2Id. at 2668 (Scalia, Kennedy, Thomas, & Alito, JJ., dissenting) (“In our view, both these central provisions of the Act—the Individual Mandate and Medicaid Expansion—are invalid. It follows, as some of the parties urge, that all other provisions of the Act must fall as well.”). Thus the dissenters would have invalidated provisions of the law establishing health insurance exchanges, raising the income cutoff for Medicaid, prohibiting insurance companies from turning away clients with preexisting conditions, providing benefits for sufferers of black lung disease, and granting the FDA authority to approve biosimilars, among many other things. 3See id. at 2675–76. The dissenters would have done so despite the fact that these provisions were perfectly constitutional, that they could have been enacted and enforced without the individual mandate, and that none of the parties before the Court had standing to challenge them. 4See id. at 2671.

Where did the dissenters find this awesome power? It is taken for granted in American legal thought that federal judges have the authority to strike down an entire statute because part of it is unconstitutional. But it is not clear where such a power comes from. Why should the invalidation of one part of a statute let a court strike down other, perfectly constitutional provisions? In doing so is the court employing an intrinsic judicial power to mark its red pen all over partially invalid laws? Or is it merely engaged in statutory interpretation, trying to preserve the legislators’ intentions after deleting part of their product? Because we lack a settled account of why judges can make statutes inseverable, judges have had little guidance in determining the scope and proper exercise of this power. This has created a doctrinal muddle. The Supreme Court has developed several different tests to determine severability, without any apparent unifying logic. It sometimes looks to whether the statute still works the way the legislature intended, 5See Alaska Airlines, Inc. v. Brock, 480 U.S. 678, 685 (1987) (“The more relevant inquiry in evaluating severability is whether the statute will function in a manner consistent with the intent of Congress.”); see also NFIB, 132 S. Ct. at 2668 (Scalia, Kennedy, Thomas, & Alito, JJ., dissenting) (“First, if the Court holds a statutory provision unconstitutional, it then determines whether the now truncated statute will operate in the manner Congress intended.”); United States v. Booker, 543 U.S. 220, 247 (2005) (“Hence we must decide whether we would deviate less radically from Congress’ intended system (1) by superimposing the constitutional requirement announced today or (2) through elimination of some provisions of the statute.”). sometimes to whether the legislature would hypothetically have enacted the law without the invalid provision, 6See Champlin Ref. Co. v. Corp. Comm’n, 286 U.S. 210, 234 (1932) (“Unless it is evident that the legislature would not have enacted those provisions which are within its power, independently of that which is not, the invalid part may be dropped if what is left is fully operative as a law.”); see also Randall v. Sorrell, 548 U.S. 230, 262 (2006) (“Given these difficulties, we believe the Vermont Legislature would have intended us to set aside the statute’s contribution limits, leaving the legislature free to rewrite those provisions in light of the constitutional difficulties we have identified.”); Denver Area Ed. Telecomms. Consortium, Inc. v. FCC, 518 U.S. 727, 767 (1996) (plurality opinion) (“The question is one of legislative intent: Would Congress still ‘have passed’ § 10(a) ‘had it known’ that the ‘remaining provision[s were] invalid’? If so, we need not invalidate all three provisions.” (alteration in original) (quoting Brockett v. Spokane Arcades, Inc., 472 U.S. 491, 506 (1985))); accord NFIB, 132 S. Ct. at 2607; Ayotte v. Planned Parenthood of N. New Eng., 546 U.S. 320, 330 (2006); Minnesota v. Mille Lacs Band of Chippewa Indians, 526 U.S. 172, 191 (1999); Regan v. Time, Inc., 468 U.S. 641, 653 (1984) (plurality opinion); Buckley v. Valeo, 424 U.S. 1, 108 (1976) (per curiam); United States v. Jackson, 390 U.S. 570, 585 (1968). sometimes to the presence of a severability clause, 7See Alaska Airlines, 480 U.S. at 686; Brockett, 472 U.S. at 506; INS v. Chadha, 462 U.S. 919, 931–34 (1983) (“Here, however, we need not embark on that elusive inquiry since Congress itself has provided the answer to the question of severability in § 406 of the Immigration and Nationality Act . . . .”); Champlin Ref. Co., 286 U.S. at 235; see also NFIB, 132 S. Ct. at 2607 (“The chapter of the United States Code that contains § 1396c includes a severability clause confirming that we need go no further.”). and sometimes to whether the provisions are capable of functioning independently. 8United States v. Reese, 92 U.S. 214, 220–21 (1876); see also Free Enter. Fund v. Pub. Co. Accounting Oversight Bd., 561 U.S. 477, 509 (2010) (noting that Sarbanes-Oxley “remains fully operative as law” and that “[t]he remaining provisions are not incapable of functioning independently” (quoting, in turn, New York v. United States, 505 U.S. 144, 186 (1992) and Alaska Airlines, 480 U.S. at 684) (internal quotation marks omitted)); Hill v. Wallace, 259 U.S. 44, 70 (1922) (“Section 4 with its penalty to secure compliance with the regulations of Boards of Trade is so interwoven with those regulations that they cannot be separated. None of them can stand.”). Yet when one follows the Court’s analysis in particular cases, all it seems to be doing is deciding whether severability is desirable as a matter of policy. As Robert Stern’s classic 1937 article observed, “the Court can easily hold any statute separable or inseparable, as it chooses.” 9Robert L. Stern, Separability and Separability Clauses in the Supreme Court, 51 Harv. L. Rev. 76, 111 (1937).

Academics have also treated the inseverability power as a given, without accounting for its source. It is a legal Beetlejuice—say its name and it appears, but no one knows where it comes from. There is vigorous debate about the proper scope of severability. Some scholars call for the courts to sever unconstitutional provisions absent a clear legislative statement to the contrary. 10See Mark L. Movsesian, Severability in Statutes and Contracts, 30 Ga. L. Rev. 41, 79–80 (1995) (arguing for a default presumption of severability); John Copeland Nagle, Severability, 72 N.C. L. Rev. 203, 206 (1993) (same); Michael D. Shumsky, Severability, Inseverability, and the Rule of Law, 41 Harv. J. on Legis. 227, 272–78 (2004) (same); Laurence H. Tribe, The Legislative Veto Decision: A Law by Any Other Name?, 21 Harv. J. on Legis. 1, 21–27 (1984) (arguing that every unconstitutional provision should be held severable unless Congress has provided an explicit inseverability clause); Kevin C. Walsh, Partial Unconstitutionality, 85 N.Y.U. L. Rev. 738, 777–89 (2010) (same); see also Tobias A. Dorsey, Remarks, Sense and Severability, 46 U. Rich. L. Rev. 877, 891–92 (2012) (arguing that courts should always find unconstitutional provisions severable, without exception); Rachel J. Ezzell, Note, Statutory Interdependence in Severability Analysis, 111 Mich. L. Rev. 1481, 1481 (2013) (arguing for a “qualified clear statement rule,” where statutes would be severable unless (1) there is an inseverability clause or (2) the severed law could not logically be enforced). Others argue that the courts should rarely sever. 11Michael C. Dorf, Facial Challenges to State and Federal Statutes, 46 Stan. L. Rev. 235 (1994) (arguing for a default presumption of inseverability); David H. Gans, Severability as Judicial Lawmaking, 76 Geo. Wash. L. Rev. 639 (2008) (advocating inseverability if severing a provision would substantially change the statute). Still others argue that courts should always (or should never) treat severability (or inseverability) clauses as controlling. 12See, e.g., Israel E. Friedman, Comment, Inseverability Clauses in Statutes, 64 U. Chi. L. Rev. 903, 904 (1997) (arguing that inseverability clauses should always be respected, but that severability clauses can sometimes be ignored); see also Michael C. Dorf, Fallback Law, 107 Colum. L. Rev. 303, 339–42 (2007) (defining limits to fallback law, including severability and inseverability clauses, namely that they cannot be impermissibly coercive to courts); Dorsey, supra note 10, at 892 (arguing that inseverability clauses are nullities that cannot be enforced, and that severability clauses are redundant); Fred Kameny, Are Inseverability Clauses Constitutional?, 68 Alb. L. Rev. 997, 1001 (2005) (arguing that inseverability clauses are unconstitutional when they attempt to coerce the judiciary or are intended as a weapon to eliminate a law); Movsesian, supra note 10, at 73–79 (arguing for a purely textualist approach to severability, including strict adherence to severability clauses); Nagle, supra note 10, at 206 (arguing that severability and inseverability clauses should be applied according to their plain meaning). Dean Tom Campbell, a former member of Congress, has taken the position that no part of a statute should ever be deemed severable—that if even a tiny, inconsequential provision is held unconstitutional the whole thing must fall. 13Tom Campbell, Severability of Statutes, 62 Hastings L.J. 1495 (2011). All of these academic commentators defend their preferred approaches by appealing to principles of judicial interpretation, separation of powers concerns, and normative views about the proper role of the judiciary. But none of this work has answered the fundamental question of where judges find this power in the first place.

This Article seeks to answer that question. It does so by articulating three different theories of the federal judicial authority to declare statutes inseverable, rejecting each of these theories, and then showing that a fourth theory is the most plausible account of the inseverability power. Each of the initial three initial theories is, to a greater or lesser extent, implicit in the existing judicial and academic commentary on severability. However they have not heretofore been fully described or explicitly distinguished. The first theory is that federal judges have an equitable remedial power to strike down or edit partly unconstitutional statutes. The second theory (and the theory that best fits the Supreme Court’s current doctrinal tests) is that severability is a form of hypothetical intent-based statutory interpretation, where the reviewing court asks whether all of a statute’s remaining provisions still further the goals of the legislature, and “interprets” the statute to invalidate those provisions that do not. This second theory is premised on the belief that if the legislature had known the one provision was unconstitutional it would have preferred that the others fall as well. The third theory is that legislative deals are enforceable as contracts, and that judges can make statutes inseverable to guarantee legislators the benefit of their bargain. The final theory, which this Article defends, is that inseverability is the product of a legislative decision to make one part of a statute conditional on another part of a statute. A legislature can create such conditionality explicitly through an inseverability clause. 14See, e.g., Social Security Amendments of 1977, Pub. L. No. 95-216, § 334(g)(3), 91 Stat. 1509, 1547 (“If any provision of this subsection . . . is held invalid . . . the application of this subsection to any other persons or circumstances shall also be considered invalid.”). But it can also do so implicitly by writing a statutory provision so that its text is nonsensical without the unconstitutional language, so that it cannot have legal effect without the unconstitutional language, or so that it serves no plausible purpose without the unconstitutional language.

As this Article will show, only the last theory—legislative conditionality—is both plausible as an account of severability and compatible with the federal judiciary’s limited powers under Article III. The main benefit of the conditionality theory is that it limits the incidence of judges reasoning like legislatures. The line between “legislative” and “judicial” reasoning, admittedly, is often slippery, and much ink has been spilled debating whether judicial review can ever be truly policy neutral. But current severability doctrine goes far beyond the “policymaking” that inheres in normal constitutional review, since judges finding statutes inseverable are not limited by any interpretation of the constitutional text, but instead are able to invalidate provisions merely on the grounds that they are bad policy or that the legislature hypothetically would not have wanted them. Such judicial repeal of statutory language is hard to distinguish from legislative repeal. Legislative conditionality solves this problem because it turns severability into a legitimate statutory interpretation issue, not an inquiry into the policy consequences of severing a provision. It also, in the great majority of cases, lets Congress determine the proper remedy by letting the political process continue, rather than rewinding that process by deleting an entire law. The major downside of the conditionality theory is that it does not permit judges to avoid the perverse consequences of judicial review, which include leaving behind an illogical law that Congress never would have enacted. But this harm is mitigated by the possibility of a legislative solution such as an amendment, a repeal, or an explicit inseverability clause. The conditionality theory also has two important implications for severability doctrine. First, a federal court cannot strike down a provision as inseverable unless it has a party with standing to challenge that provision. Second, there is no reason to constrain severability analysis to only a single bill or a single act—any statutory provision can be made conditional on another provision anywhere else in the legislative code, as long as Congress so provides (explicitly or implicitly).

The argument of this Article proceeds in six parts. Part I tracks how the Supreme Court’s current ad hoc approach to severability has evolved over time, describes the three severability tests that the Court applies, and shows that the Court inevitably makes policy judgments when it applies these tests. Part II introduces the basic problem: that judges must either act like legislators or else enforce a statutory regime that Congress never meant to enact. Part III shows that the proper framing question when debating theories of severability is to ask how courts can find statutory language inseverable, rather than how courts can find statutes severable. Part IV describes the first three theories of judges’ power to declare statutes inseverable (that it is an equitable remedial power, that it is a matter of hypothetical legislative intent, and that it is a form of contract remedy), showing the flaws in each. Part V presents the legislative conditionality theory and lays out its consequences, showing that conditionality can be found by implication, requires standing, and is not limited to provisions within a single bill or act. Part VI considers the practical implications of the conditionality theory. It demonstrates that any harm from legislative conditionality is mitigated by legislative overrides and the greater use of inseverability clauses, that state and not federal judges should determine whether state statutes are inseverable, that state judges might legitimately exercise a broader inseverability power than can federal judges, and that adopting the conditionality theory would not require a major break from the Supreme Court’s current severability precedents (although the theory is likely inconsistent with parts of the holdings in two recent Supreme Court cases).

I. The Supreme Court’s Conflicting Approaches to Severability

Severability doctrine is confusing. The Supreme Court currently applies three very different tests. The first is to ask whether the remaining provisions of the statute are capable of functioning independently of the unconstitutional provision (the independence test). The second is to ask whether Congress would have passed the legislation without the severed provision—or, as it is sometimes framed, whether Congress would have wanted the whole enactment to fall, or just the unconstitutional part (the hypothetical passage test). The third is to ask whether the legislation operates in the way Congress intended if the unconstitutional piece is severed (the intent test). At the oral argument in NFIB, some of the justices expressed confusion about which of these tests they were supposed to apply. 15For example, Justice Scalia and Paul Clement engaged in the following colloquy:JUSTICE SCALIA: . . . . Why do we look to the—are you sure we look to the intent of the Congress? I thought that, you know, sometimes Congress says that these provisions will—all the provisions of this Act will be severable, and we ignore that when the Act really won’t work, when the remaining provisions just won’t work. Now, how can you square that reality with the proposition that what we’re looking for here is what would this Congress have wanted?MR. CLEMENT: Well, two responses, Justice Scalia: We can look at this Court’s cases on severability, and they all formulate the test a little bit differently.JUSTICE SCALIA: Yes, they sure do.Transcript of Oral Argument, Day 2, at 9, NFIB, 132 S. Ct. 2566 (2012) (Nos. 11-393 & 11-400), available at http://www.supremecourt.gov/oral_arguments/argument_transcripts/11-393.pdf. Similarly, Justice Kennedy asked,If you—suppose you had party A wants proposal number 1; party B wants proposal number 2. Completely unrelated. One is airline rates; the other is milk regulation. And we—and they decide them together. The procedural rules are these have to be voted on as one. They are both passed. Then one is declared unconstitutional. The other can operate completely independently. Now, we know that Congress would not have intended to pass one without the other. Is that the end of it, or is there some different test? Because we don’t want to go into legislative history, that’s intrusive, so we ask whether or not an objective—as an objective rational matter, one could function without—I still don’t know what the test is that we’re supposed to apply. And this is the same question as Justice Scalia asked. Could you give me some help on that?Id. at 17–18. The dissenters who would have struck down the entire ACA recited the intent test and the hypothetical passage test, while the controlling opinion by Chief Justice Roberts relied only on the hypothetical passage test in finding the Act’s Medicaid expansion severable. 16NFIB, 132 S. Ct. at 2607 (“The question here is whether Congress would have wanted the rest of the Act to stand, had it known that States would have a genuine choice whether to participate in the new Medicaid expansion.”); id. at 2668–69 (Scalia, Kennedy, Thomas, & Alito, JJ., dissenting) (“First, if the Court holds a statutory provision unconstitutional, it then determines whether the now truncated statute will operate in the manner Congress intended. . . . Second, even if the remaining provisions can operate as Congress designed them to operate, the Court must determine if Congress would have enacted them standing alone and without the unconstitutional portion.”). And they asked severability, “What is thy name?” And severability answered, “My name is Legion: for we are many.” This Part will show how the doctrine has evolved into this confusing multitude. It will also show that the Court essentially makes policy judgments when it finds statutes inseverable. 17For more on the history of severability doctrine, see Kenneth A. Klukowski, Severability Doctrine: How Much of a Statute Should Federal Courts Invalidate?, 16 Tex. Rev. L. & Pol. 1, 10–30 (2011); Stern, supra note 9, at 79–82.

The Supreme Court’s first severability decision was Marbury v. Madison, in which the Court struck down a provision of the Judiciary Act of 1789 that extended the Court’s original jurisdiction beyond the limit set by the Constitution. 185 U.S. (1 Cranch) 137, 175–76 (1803). Chief Justice John Marshall’s opinion did not discuss severability, but it implicitly found that the entire Judiciary Act did not have to fall along with the problematic provision. He could hardly have held otherwise without destroying the federal judiciary. In an 1829 case, Bank of Hamilton v. Lessee of Dudley, Chief Justice Marshall first acknowledged the principle that when a provision is struck down, the rest of the statute may remain in force. 1927 U.S. (2 Pet.) 492, 526 (1829). The first court to discuss the rationale for inseverability, and to find a statute inseverable, was the Massachusetts Supreme Judicial Court in the 1854 case of Warren v. Mayor of Charlestown. 2068 Mass. (2 Gray) 84 (1854). The court in that case articulated both the independence test and the hypothetical passage test:

[T]he parts, so held respectively constitutional and unconstitutional, must be wholly independent of each other. But if they are so mutually connected with and dependent on each other, as conditions, considerations or compensations for each other, as to warrant a belief that the legislature intended them as a whole, and that, if all could not be carried into effect, the legislature would not pass the residue independently, and some parts are unconstitutional, all the provisions which are thus dependent, conditional or connected, must fall with them. 21Id. at 99.

This two-part approach was adopted by other state supreme courts, 22See Klukowski, supra note 17, at 8; Stern, supra note 9, at 80. and ultimately by the United States Supreme Court in the 1880 case Allen v. Louisiana. 23103 U.S. 80, 83–84 (1881). It was applied during the mid- to late-1800s, though with a healthy degree of caution. Courts were generally reluctant, in that period, to damn the whole statute for the sins of the part. 24See Stern, supra note 9, at 81 (“During this early period, the courts were very reluctant to invalidate an entire law on grounds of inseparability. Doubts were resolved in favor of the severance of legislation; indeed, in a number of cases, in both state and federal courts, it was apparently assumed that laws were intended to be sustained to the extent that they could possibly be held valid. Although the language of presumption was not employed, it is clear that the courts at that time presumed that laws were intended to be severable, rather than the contrary.” (footnotes omitted)).

This favoring of severance ended in the early 1900s, when the Supreme Court began adopting a more aggressive approach to judicial review. Prior to 1910 the Court had found state statutes inseverable in only two cases and a federal statute inseverable in only one. 25Allen, 103 U.S. 80; see also Pollock v. Farmers’ Loan & Trust Co., 158 U.S. 601 (1895); Poindexter v. Greenhow, 114 U.S. 270 (1884); Stern, supra note 9, at 107–08 n.138. Note that this list includes only those cases in which the Court held that the unconstitutionality of one provision required striking down one or more entirely separate provisions. It does not include cases where the Court struck down a provision with some constitutional applications and some unconstitutional applications (e.g., a provision regulating both intra- and interstate commerce). See Great N. Ry. Co. v. United States, 208 U.S. 452 (1908). For lists of the latter type of case from this period, see Alfred Hayes, Jr., Partial Unconstitutionality with Special Reference to the Corporation Tax, 11 Colum. L. Rev. 120, 124 n.8 (1911); Stern, supra note 9, at 90 nn.64–65. This list also does not include cases where the Court struck down an entire statute because it contained an unconstitutional exception, because that situation does not create a severability question. Rather, it presents the Court with a choice of whether to remedy the constitutional violation by striking down the rule or the exception to the rule. See infra note 71. From 1910 to 1937, it would find state statutes inseverable in seven cases and federal statutes inseverable in five. 26The seven state law cases are as follows: Williams v. Standard Oil Co. of La., 278 U.S. 235, 241–42 (1929); Lemke v. Farmers Grain Co. of Embden, N.D., 258 U.S. 50, 60 (1922); Looney v. Crane Co., 245 U.S. 178, 190–91 (1917); McFarland v. Am. Sugar Ref. Co., 241 U.S. 79, 87 (1916); Myers v. Anderson, 238 U.S. 368, 370 (1915); Harrison v. St. Louis & S.F. R.R., 232 U.S. 318, 332–34 (1914); Int’l Textbook Co. v. Pigg, 217 U.S. 91, 114 (1910). The five federal law cases are as follows: Carter v. Carter Coal Co., 298 U.S. 238, 316 (1936); R.R. Ret. Bd. v. Alton R.R., 295 U.S. 330, 361–62 (1935); Hill v. Wallace, 259 U.S. 44, 70–71 (1922); Muskrat v. United States, 219 U.S. 346, 362–63 (1911); Weems v. United States, 217 U.S. 349, 381–82 (1910). For more on inseverability of statutes, see Stern, supra note 9, at 107 n.138. The qualifications about severable applications and striking down the exception versus the rule, both mentioned supra note 25 and infra note 71, also apply to this list. In this period the Court frequently used findings of inseverability to strike down laws that interfered with the free market, such as taxes, industry regulations, price controls, and so forth. This new approach occasioned a default rule presuming statutes to be inseverable. The Court announced in Williams v. Standard Oil Co. of Louisiana that it presumed “that the legislature intends an act to be effective as an entirety,” and that any unconstitutional provision would thus be held inseverable unless it were clearly shown to be independent from the rest of the statute. 27278 U.S. at 241–42. The Court reversed this presumption of inseverability if the statute contained an explicit severability clause. 28Id. at 242 (“The effect of the statutory declaration is to create in the place of the presumption just stated the opposite one of separability. That is to say, we begin, in the light of the declaration, with the presumption that the legislature intended the act to be divisible; and this presumption must be overcome by considerations which make evident the inseparability of its provisions or the clear probability that the invalid part being eliminated the legislature would not have been satisfied with what remains.”). However, in many cases during this period, the Court found statutes inseverable despite such clauses. 29See Carter, 298 U.S. at 312–13; R.R. Ret. Bd., 295 U.S. at 361–62; Hill, 259 U.S. at 71. While the Court in these years was more willing to strike down statutes in toto, the tests for determining severability remained essentially the same. The Court still looked to the independence test and the hypothetical passage test that had been elaborated in Warren and Allen. For example it declared in Champlin Refining Co. v. Corp. Commission that “[u]nless it is evident that the legislature would not have enacted those provisions which are within its power, independently of that which is not, the invalid part may be dropped if what is left is fully operative as a law.” 30286 U.S. 210, 234 (1932); see also Dorchy v. Kansas, 264 U.S. 286, 290 (1924) (“But a provision, inherently unobjectionable, cannot be deemed separable unless it appears both that, standing alone, legal effect can be given to it and that the legislature intended the provision to stand, in case others included in the act and held bad should fall.”); Stern, supra note 9, at 76 (“The inquisitive legislator seeking light on this problem will find that the Supreme Court, the state courts, and secondary authorities all appear to agree that the invalidity of part of a law or of some of its applications will not affect the remainder (1) if the valid provisions or applications are capable of being given legal effect standing alone, and (2) if the legislature would have intended them to stand with the invalid provisions stricken out.”).

Since the end of the Lochner era, the Supreme Court has been more reluctant to find statutes inseverable. Since 1940 it has found a federal statute inseverable in only one case, United States v. Booker, 31543 U.S. 220 (2005). I am not counting Northern Pipeline Construction Co. v. Marathon Pipe Line Co., 458 U.S. 50 (1982), as a severability decision, because that case concerned potentially severable applications of the same statutory language (a provision granting bankruptcy courts jurisdiction over cases) rather than the severability of different statutory language. and has found an executive order inseverable in one other case, Minnesota v. Mille Lacs Band of Chippewa Indians. 32526 U.S. 172, 191 (1999). When reviewing state laws, the Court has usually either found them severable or, per Erie Railroad Co. v. Tompkins, 33304 U.S. 64 (1938). avoided the question so that it could be decided as a matter of state law. 34See, e.g., Ayotte v. Planned Parenthood of N. New Eng., 546 U.S. 320, 331 (2006); Leavitt v. Jane L., 518 U.S. 137, 139–40 (1996) (per curiam); Hooper v. Bernalillo Cnty. Assessor, 472 U.S. 612, 624 (1985); Zobel v. Williams, 457 U.S. 55, 65 (1982); Harrison v. NAACP, 360 U.S. 167, 178 (1959); Watson v. Buck, 313 U.S. 387, 396 (1941). For a more complete list of such cases, see Ryan Scoville, The New General Common Law of Severability, 91 Tex. L. Rev. 543, 564 n.139 (2013). Professor Scoville argues that Ayotte portends that the Supreme Court will begin applying a federal common law of severability to state statutes. Id. at 569–74. However this is probably an overreading of part of the opinion, which elsewhere explicitly refers to New Hampshire law. See Kevin C. Walsh, There Is No General Common Law of Severability, 91 Tex. L. Rev. See Also 49 (2012). It has declared state laws inseverable in only a handful of cases. 35After extensive searching I have located only six post-1940 cases in which the Supreme Court declared a state statute inseverable, but it is possible that some have eluded my grasp. See Randall v. Sorrell, 548 U.S. 230, 262 (2006); Thornburgh v. Am. Coll. of Obstetricians & Gynecologists, 476 U.S. 747, 764–65 (1986); City of Akron v. Akron Ctr. for Reprod. Health, Inc., 462 U.S. 416, 445 n.37 (1983); Planned Parenthood of Cent. Mo. v. Danforth, 428 U.S. 52, 83 (1976); Meek v. Pittenger, 421 U.S. 349, 371 n.21 (1975); Sloan v. Lemon, 413 U.S. 825, 833–35 (1973). Keeping with this trend, the Court in 1983 reversed the presumption it had announced in Williams, establishing that “the presumption is in favor of severability” even if the statute lacks a severability clause. 36Regan v. Time, Inc., 468 U.S. 641, 653 (1983) (plurality opinion); see also Ayotte, 546 U.S. at 328–29; Alaska Airlines, Inc. v. Brock, 480 U.S. 678, 686 (1987) (“In the absence of a severability clause, however, Congress’ silence is just that—silence—and does not raise a presumption against severability.”). This declining willingness to strike down statutes in their entirety, or to determine that some provisions must fall alongside others, has coincided with a massive increase in the length and complexity of federal legislation. 37See W. David Slawson, Legislative History and the Need to Bring Statutory Interpretation Under the Rule of Law, 44 Stan. L. Rev. 383, 402–03 (1992) (“The United States Code consisted of 27,308 pages in 1988, . . . compared to 9797 pages in 1964, the last year in which it was subjected to major revision before the great expansion of federal law that started in the sixties. . . . Federal statutes have also become more complicated. The number of subjects they treat has increased and so has the technological difficulty of many of the subjects.” (footnote omitted)); Christopher Beam, Paper Weight: The Health Care Bill is More than 1,000 Pages. Is that a Lot?, Slate (Aug. 20, 2009, 6:12 PM), http://www.slate.com/articles/news_and_politics/explainer/2009/08/paper_weight.html (noting several examples of recent laws that were over 1,000 pages, that “the total number of pages of legislation has gone up from slightly more than 2,000 pages in 1948 to more than 7,000 pages in 2006,” and that “[t]he average bill length increased over the same period from 2.5 pages to 15.2 pages”). Many of the Court’s major recent severability decisions have involved federal laws that are hundreds or thousands of pages in length—for example Booker, NFIB, 38See 132 S. Ct. 2566, 2675 (2012) (Scalia, Kennedy, Thomas, & Alito, JJ., dissenting).Alaska Airlines, Inc. v. Brock, 39480 U.S. at 684. and Free Enterprise Fund v. Public Company Accounting Oversight Board. 40561 U.S. 477 (2010) (striking down for-cause removal limitations for the Public Company Accounting Oversight Board, and finding them severable them from the rest of Sarbanes-Oxley). The modern trend thus might be explained in part by the Court’s unwillingness to search every inch of a legislative giant for connected pieces, or to bring such a giant down by its toe. 41Cf. Transcript, supra note 15, at 38 (“JUSTICE SCALIA: Mr. Kneedler, what happened to the Eighth Amendment? You really want us to go through these 2,700 pages? (Laughter.) JUSTICE SCALIA: And do you really expect the Court to do that? Or do you expect us to give this function to our law clerks? (Laughter.) JUSTICE SCALIA: Is this not totally unrealistic, that we’re going to go through this enormous bill item by item and decide each one?”). However this relative lack of inseverability holdings may not persist into the future. The fact that four Supreme Court justices in NFIB were willing to strike down the entire ACA because of one unconstitutional provision, and that none of the other justices voiced disagreement on this point, could portend more aggressive exercise of this power.

Notwithstanding this increased reluctance to declare laws inseverable, the Supreme Court in the 1980s introduced a new test that seems to make it harder to slice them apart, and that renders the doctrine more confusing. In Alaska Airlines, Inc. v. Brock, after reciting the classic independence and hypothetical passage tests, the Court declared that “[t]he more relevant inquiry in evaluating severability is whether the statute will function in a manner consistent with the intent of Congress.” 42480 U.S. at 685. It thus introduced a third test, the intent test, which is easier to satisfy than the other two.

Indeed, it is not clear why the first two tests are still needed. If the remaining provisions of a statute cannot function independently of the unconstitutional part, or if the legislature would not have passed it without the unconstitutional part, then a fortiori it would seem that the statute will not function as the legislature intended. The three tests also lack a unifying principle. Indeed they seem rather inconsistent. Should a court be focused on the language of the statute, as per the independence test, or on the intentions of the legislature, as per the other two? And if the latter, should the court look to what the legislature would have done had it known its law was partly invalid, or does the court instead decide on its own whether the severed law fulfills the legislature’s original goals? It seems strange to say both, much less all three. Yet that is where we are.

This confusing doctrinal framework has failed to produce neutral principles that courts can apply in severability cases. Indeed, the Supreme Court’s decisions seem largely determined by its views on the substantive merits of the partially invalidated laws. In cases where it considers severability questions at length, the Court’s deliberations resemble those of a legislative body. It considers what the law will look like without the unconstitutional part and determines whether the statute remains a good policy idea. It proceeded in this fashion well before the intent test announced in Alaska Airlines. For example, in two cases from the early 1920s—Hill v. Wallace and Board of Trade of Chicago v. Olsen—the Court considered whether two enforcement mechanisms for substantially identical laws regulating grain futures, respectively a penalty tax and the exclusion of transactions from interstate commerce, could be severed from the larger laws. 43Bd. of Trade of Chi. v. Olsen, 262 U.S. 1, 42 (1923); Hill v. Wallace, 259 U.S. 44, 70 (1922). The Court found that the tax could not be severed while the exclusion could, and it is difficult to find a principled ground for this distinction, aside from the fact that the Court approved of one law and not the other. 44See Stern, supra note 9, at 112 (“The two decisions on separability, written within eleven months of each other by the same judge, can be reconciled only on the ground that the Court approved of the substance of one statute and not of the other.”); see also id. at 109–14 (providing this and other examples of the Court making arbitrary, essentially legislative severability decisions in the early part of the twentieth century). Politicized decisionmaking of this sort is a feature of severability doctrine, not a bug. The Court inevitably puts itself in the place of the legislature and makes a policy judgment. To illustrate, consider three major, semi-recent opinions that addressed severability questions.

In Buckley v. Valeo, the Supreme Court employed policy-driven analysis in deciding to preserve portions of a major campaign finance law. The Court struck down several provisions of the Federal Election Campaign Act on First Amendment grounds, including limits on independent expenditures and on candidates’ expenditures from their personal funds. In doing so, the Court severed the law’s public financing provisions and also implicitly severed the law’s campaign contribution limits. 45424 U.S. 1, 108–09 (1976) (per curiam). It recited the independence and hypothetical passage tests and concluded that the statute was severable because “[o]ur discussion of ‘what is left’ leaves no doubt that the value of public financing is not dependent on the existence of a generally applicable expenditure limit.” 46Id. at 109. The Court thus relied on its approving analysis of the merits of the public financing provisions in finding them severable from the expenditure limits, exemplified by its statement that “[i]t cannot be gainsaid that public financing as a means of eliminating the improper influence of large private contributions furthers a significant governmental interest.” 47Id. at 96. See generally id. at 90–109 (upholding the public financing provisions against various constitutional challenges). The crux of the analysis in Buckley was whether, in the Court’s judgment, the remainder of the law still worked well as a policy matter.

In Booker, the Court engaged in a detailed policy debate over how to rewrite federal sentencing law. After the Court found the federal system of mandatory sentencing guidelines unconstitutional for violating the right to a jury trial, it then had to decide on the proper remedy. 48United States v. Booker, 543 U.S. 220, 245 (2005). Justice Breyer’s majority opinion determined that the best course was to strike down Section 3553(b)(1) of the Sentencing Reform Act (SRA), which had made the Federal Sentencing Guidelines mandatory, with the result that the Guidelines are now advisory. 49Id. Justice Stevens’s dissent argued instead that all facts leading to a higher sentence should need to be proven to a jury. Justice Breyer’s opinion argued that its remedial excisions could be severed from the rest of the statute, but that the dissent’s could not. 50Id. at 249 (“Several considerations convince us that, were the Court’s constitutional requirement added onto the Sentencing Act as currently written, the requirement would so transform the scheme that Congress created that Congress likely would not have intended the Act as so modified to stand.”); id. at 258 (“Although, as we have explained . . . we believe that Congress would have preferred the total invalidation of the statute to the dissenters’ remedial approach, we nevertheless do not believe that the entire statute must be invalidated.”). In so arguing, the majority relied on a number of policy justifications: that the dissent’s remedy would increase sentencing disparities for similarly situated defendants, weaken the link between the sentence length and the offender’s real conduct, require more complex charging documents, make jury decisionmaking more difficult, render the plea bargaining process more arbitrary and disuniform, increase prosecutorial power, and leave sentences easier to shorten than to lengthen. 51Id. at 249–58. By contrast, the majority opinion defended its own remedy as essentially consistent with congressional intent because it left the basic structure of the SRA in place, and Congress likely would have preferred that to eliminating the Guidelines altogether. 52Id. at 265 (“In our view, it is more consistent with Congress’ likely intent in enacting the Sentencing Reform Act (1) to preserve important elements of that system while severing and excising two provisions (§§ 3553(b)(1) and 3742(e)) than (2) to maintain all provisions of the Act and engraft today’s constitutional requirement onto that statutory scheme.”). As in Buckley, it is clear from the opinion that the majority in Booker reasoned as a legislature, looking primarily to the policy consequences of each remedy when arguing about severability. 53See Walsh, supra note 10, at 752 (“It should come as no surprise that the Court’s conclusion in Booker about what Congress would have wanted lines up closely with what five Justices think Congress should have wanted.”).

NFIB presents the most striking example of the Court basing severability decisions on policy reasoning. The majority opinion struck down a provision of the ACA conditioning state Medicaid funding on an expansion of Medicaid, and found that provision severable. 54NFIB, 132 S. Ct. 2566, 2607 (2012). The dissent, by contrast, found both the Medicaid expansion and the individual mandate unconstitutional, and determined that the entire ACA must fall along with them. 55Id. at 2668 (Scalia, Kennedy, Thomas, & Alito, JJ., dissenting). The dissent’s analysis relied largely on policy arguments. It noted that “distinguished economists” had informed the Court that the $700 billion insurers would take in from the Medicaid expansion and individual mandate were necessary to offset $700 billion in new expenses that the ACA would impose on them, and that without this new revenue insurance premiums would rise. 56Id. at 2672. The dissenters also determined that hospitals would increase prices to make up for lost funds, and that government subsidies for the ACA’s insurance exchanges would have to increase if there were no individual mandate or Medicaid expansion. 57Id. at 2672–74. Indeed, the dissent sometimes reads more like a policy brief than a judicial opinion, and the oral argument on severability often resembled a legislative committee hearing on insurance regulation. 58See, e.g., Transcript, supra note 15, at 48–65. The dissent would also have struck down a number of totally ancillary provisions of the ACA—regulations for nursing mothers and tanning booths, a law requiring the display of nutritional information at restaurants, the extension of Medicare to people exposed to asbestos in Montana, and others—on the grounds that “[t]here is no reason to believe that Congress would have enacted them independently.” 59NFIB, 132 S. Ct. at 2675 (Scalia, Kennedy, Thomas, & Alito, JJ., dissenting). This reverses the Court’s longstanding default rule—rather than presuming these items severable, the Court presumed them inseverable and required proof that they would have been passed independently. 60One possible reason for this reversal is that the dissenters could not feasibly have gone through the ACA line by line to determine which of its many minor provisions could be preserved. See supra note 41. The dissenters in NFIB thus engaged in legislative-style decisionmaking, striking down functionally independent provisions and relying on economists for the proposition that a severed ACA would have adverse policy consequences. 61NFIB, 132 S. Ct. at 2672 (Scalia, Kennedy, Thomas, & Alito, JJ., dissenting) (“The Court has been informed by distinguished economists that the Act’s Individual Mandate and Medicaid Expansion would each increase revenues to the insurance industry by about $350 billion over 10 years; that this combined figure of $700 billion is necessary to offset the approximately $700 billion in new costs to the insurance industry imposed by the Act’s insurance regulations and taxes; and that the new $700–billion burden would otherwise dwarf the industry’s current profit margin.”).

The judicial legislating that these three decisions highlight is not just an occasional pathology of severability analysis—it is the mode of reasoning demanded by the official doctrine. The intent test announced in Alaska Airlines is the most transparently policy-driven of the three tests, because it asks courts to interpret Congress’s goals and use them as a basis for decision. 62Alaska Airlines, Inc. v. Brock, 480 U.S. 678 (1987). Yet Congress’s “intent” in passing a law can often be characterized however a court chooses. For example, the NFIB dissenters characterized Congress’s intent as creating a system of “shared responsibility” and justified striking down the entire Act on the grounds that severing the two invalid provisions would disrupt this “sharing.” 63NFIB, 132 S. Ct. at 2671 (Scalia, Kennedy, Thomas, & Alito, JJ., dissenting). But one could just as easily characterize the ACA as serving any number of purposes—expanding health coverage, making health care less expensive, reining in abusive insurance company practices, and so on—and make severability decisions in light of the chosen end. And further, even where Congress’s intent is objectively clear, whether the severed law still furthers that intent is ultimately a legislative judgment. The hypothetical passage test at least gestures at external evidence, namely whether the legislature indicated that it would have passed the severed law. But such evidence is rarely if ever available, and so the hypothetical passage test collapses into an inquiry about the policy effects of severance—presumably Congress would not have passed the severed law if it had sufficiently bad consequences. The independence test is more amenable to neutral judicial standards of decision. One might determine whether the remaining provisions still have legal effect without making a policy judgment. But, in part because it is so narrow, the independence test is rarely the crux of the Court’s analysis, and it has been explicitly deemphasized in recent cases. 64See id. at 2668 (characterizing the independence test as merely a subsidiary part of the intent test); see also Alaska Airlines, 480 U.S. at 685 (“The independent operation of a statute in the absence of a legislative veto provision thus could be said to indicate little about the intent of Congress regarding severability of the veto.”). The decisions where severability analysis is dominated by policy concerns are not flukes or bastardizations of the case law. The Court acts as a legislature because the doctrine demands it.

II. The Dilemma: Statutory Distortion or Judicial Legislating

Inseverability responds to a peculiar kind of problem. A court that finds part of a law unconstitutional but keeps the rest in place will often leave behind a statutory scheme very different from what the legislature created. And much like removing lines of code from a computer program, eliminating parts of a statute can mutate the whole in undesirable ways. As the NFIB dissenters noted, severance “imposes on the Nation, by the Court’s decree, its own new statutory regime, consisting of policies, risks, and duties that Congress did not enact.” 65NFIB, 132 S. Ct. at 2668 (Scalia, Kennedy, Thomas, & Alito, JJ., dissenting). The search for a theory of inseverability is motivated by a desire to give judges the power to alleviate this distortion problem. However, as just illustrated, the downside of having judges wield such power is that it causes them to invalidate statutory language that is perfectly constitutional, and to do so for purely policy reasons (as opposed to interpretive reasons, although that distinction is admittedly sometimes fuzzy). The challenge for this Article is thus to find a theory of inseverability that will allow judges to prevent statutory distortion while staying within their power to interpret the law.

Many severability decisions do not distort the statutory scheme because the severed provision is basically autonomous from the rest of the law. In other words, the severed portion has no effect on anything else in the statute. These cases are generally easy to decide in favor of severance. In more difficult cases, however, removing part of a law will fundamentally change how that law functions. Consider again the opinions in Buckley, Booker, and NFIB. In Buckley, the Supreme Court struck down a ceiling on candidates’ spending on their own political campaigns but upheld a limit on third-party campaign donations. 66Buckley v. Valeo, 424 U.S. 1 (1976) (per curiam). It thus created an asymmetric system that Congress never intended, where wealthy candidates can self-finance without limit, while poorer candidates face significant legal barriers to raising funds. 67This aspect of Buckley has been heavily criticized, including by Chief Justice Burger in his partial concurrence, and is cited as evidence that courts sever too frequently. See id. at 254–55 (Burger, C.J., concurring in part and dissenting in part) (“All candidates can now spend freely; affluent candidates, after today, can spend their own money without limit; yet, contributions for the ordinary candidate are severely restricted in amount—and small contributors are deterred. I cannot believe that Congress would have enacted a statutory scheme containing such incongruous and inequitable provisions.”); see also Transcript, supra note 15, at 84 (“[T]his Court in Buckley created a halfway house, and it took Congress 40 years to try to deal with the situation, when contrary to any time of their intent, they had to try to figure out what are we going to do when we’re stuck with this ban on contributions, but we can’t get at expenditures because the Court told us we couldn’t. . . . Why make them do that in health care?”); Campbell, supra note 13, at 1522 (“What no one proposed was a rule that would allow individuals who were wealthy to be able to spend their own money without limit, while individuals who were not wealthy were restricted, as a practical matter, to spending what they could raise, with limits on how much they could raise from any other person. Yet that is the system we now have, after the Supreme Court struck down the limit on individual expenditure, but left intact the rest of the statute regarding limits on fundraising.”). In Booker, the Court severed the provision that made the Federal Sentencing Guidelines mandatory. 68United States v. Booker, 543 U.S. 220 (2005). It thus transformed the Guidelines from a rule-bound code that judges were compelled to follow into a discretionary system of suggested sentences. While the Guidelines remain in place, they now function in a manner Congress never intended. 69Indeed, Congress had in prior years rejected other versions of the SRA that would have created more flexible sentencing systems. See id. at 293 (Stevens, J., dissenting in part) (“Congress explicitly rejected as a model for reform the various proposals for advisory guidelines that had been introduced in past Congresses.”); Kate Stith & Steve Y. Koh, The Politics of Sentencing Reform: The Legislative History of the Federal Sentencing Guidelines, 28 Wake Forest L. Rev. 223, 238 (1993). In NFIB, if a majority of the Court had found the individual mandate unconstitutional and severed it from the rest of the ACA, this would likely have caused a big rise in insurance premiums. Health insurers would have been compelled by the ACA to offer their products to consumers regardless of preexisting medical conditions, and consumers could thus have waited for an illness to strike before buying insurance. Congress contemplated and sought to avoid this adverse selection problem by creating the individual mandate, and the Obama administration even asked for the Supreme Court to hold the mandates on insurance companies inseverable, risking destruction of the heart of the ACA to avoid an ACA sans individual mandate. 70See 42 U.S.C. § 18091(2)(I) (2012) (“[I]f there were no requirement, many individuals would wait to purchase health insurance until they needed care. By significantly increasing health insurance coverage, the requirement, together with the other provisions of this Act, will minimize this adverse selection and broaden the health insurance risk pool to include healthy individuals, which will lower health insurance premiums.”); Brief for Respondents (Severability) at 13–14, 26, NFIB, 132 S. Ct. 2566 (2012) (Nos. 11-393 & 11-400), 2012 WL 273133. These three cases illustrate the basic problem of statutory distortion. When a law contains multiple parts that operate in concert, like an animal with four legs, severing one will fundamentally change how the law operates. 71One might argue that a worse type of distortion occurs where a legislature enacts a rule with an exception, and the exception is held unconstitutional but the rule remains, so that the rule applies to cases the legislature never extended it to. For instance if the legislature passes a tax on “all persons who are not from Minnesota,” and the exclusion of Minnesotans is found unconstitutional, the legislature must then determine whether strike down the tax or the exception. However this is not, properly conceived, a problem of severability—it is, rather, a choice of how the court exercises its power of constitutional review. The court can remedy the constitutional violation by striking down either the exception or the rule. See Welsh v. United States, 398 U.S. 333, 361 (1970) (Harlan, J., concurring in result) (“Where a statute is defective because of underinclusion there exist two remedial alternatives: a court may either declare it a nullity and order that its benefits not extend to the class that the legislature intended to benefit, or it may extend the coverage of the statute to include those who are aggrieved by exclusion.”).

The purpose of severability doctrine is to lessen the harms of statutory distortion by giving courts the power to fix a partially invalidated law, either by removing additional parts or by striking the law down entirely. The problem is that, in exercising this power to declare statutes inseverable, a court finds itself reasoning and acting like a legislature. It does so in two senses.

First, when a court finds a provision inseverable, it is invalidating statutory language. Such invalidation is a fundamentally legislative function. A court generally cannot erase part of a statute unless it is resolving a conflict between that statute and the higher law of the Constitution. Furthermore, inseverability decisions sometimes skirt the Article III case or controversy requirement—several Supreme Court justices have adopted the view that current doctrine lets them declare a statute inseverable even if there is no party with standing to challenge the entire law. 72See NFIB, 132 S. Ct. at 2671 (Scalia, Thomas, Kennedy, & Alito, JJ., dissenting) (“To be sure, an argument can be made that those portions of the Act that none of the parties has standing to challenge cannot be held nonseverable. The response to this argument is that our cases do not support it.”); see also, e.g., Williams v. Standard Oil Co. of La., 278 U.S. 235, 242–44 (1929) (holding nonseverable statutory provisions that did not burden the parties). The main task for a theory of inseverability, then, is to explain how the judiciary can exercise these legislative powers consistent with Article III of the Constitution. If it is to justify current practice, such a theory must show that a finding of partial unconstitutionality allows a court to rummage around in the rest of a law and strike out other parts too, and that the court may do so even if no party has standing to challenge those other parts.

Second, even if we solve this theoretical problem of Article III power, courts in severability cases still often act like legislatures in a more normative sense. This is because judges rely on policy arguments (i.e. that a remaining provision is a bad idea without the invalidated provision, or that the legislature would not have wanted it to remain), as opposed to interpretive arguments (i.e. that a provision violates the Constitution), when they decide severability questions. This line between “policy” arguments and “interpretive” arguments is admittedly somewhat slippery in practice—a certain degree of purposive, policy-based reasoning may be inevitable in many interpretive contexts. But when judges decide severability questions under the current approach, there is nothing much to interpret—it is policy all the way down. As a former counsel to the House of Representatives put it, “[t]he Supreme Court decides what needs to be stricken for constitutional reasons, and then asks what else should be stricken for political reasons.” 73Dorsey, supra note 10, at 880–81. The prior Part illustrated how two of the Supreme Court’s main doctrinal tests—the intent test and the hypothetical passage test—require a reviewing court to determine severability on the basis of policy justifications. If courts are going to proceed in such a fashion, they may as well conduct legislative committee-style hearings to fully consider the consequences of severing a law. This would be unseemly, of course, because judges are not elected. Thus another desideratum of a theory of severability is that it should allow courts to depend on interpretive principles as opposed to pure policy arguments. This is not as crucial as finding an account of severability that is consistent with Article III, but it is surely important if we wish to preserve the legitimacy of the judicial function.

The current approach to severability presents a basic dilemma: judges must either distort the statutes they review, or they must exceed their Article III powers and reason like legislatures in fixing those statutes. An ideal theory of severability would let judges duck this unfortunate choice by achieving several things. It would allow judges to avoid statutory distortion, at least in some cases, by giving them the power to strike down perfectly constitutional portions of a statute. It would also account for how judges can exercise such a power consistent with Article III. Finally, it would provide judges with a way to base their reasoning on interpretive arguments rather than pure policy arguments.

III. What is the Fundamental Unit of Legislation?

Before analyzing the various theories of severability, it is important to resolve an issue of metatheory: what is the fundamental unit of legislation for judicial review purposes? Or, more precisely, when a court conducts constitutional review of a law, what is the smallest unit of statutory language it can strike down—a word, a sentence, an entire statute, or something else? This Part will show that the correct answer is “a word.” 74Here I am only discussing the remedy of invalidating words entirely. The courts may also use other constitutional remedies—such as interpreting a statute to avoid a constitutional problem or striking down unconstitutional applications of a statute—that leave the words of a statute intact but change their meaning. Such remedies are not relevant here. That is, a court can remedy a violation of the Constitution by striking down a single word or a group of words, but it need not strike down the larger legislative unit (be it a section, statute, chapter, or title) that contains those words. Thus the normal framing of the severability question—where we ask if an unconstitutional provision “can be severed” from the rest of a statute—is exactly backwards. 75See, e.g., Gans, supra note 11, at 653 (“For better or worse, courts need to have some power to save partially invalid legislation by severing the invalid portions and thereby reconstructing the statute.”). Here, I am only discussing how to frame the meta-question of courts’ power to sever statutes. It is of course possible that a particular doctrinal test will impose a presumption of severability and then require courts to ask how a provision can be “saved.” But that is distinct from assuming that any time a court engages in judicial review it initially strikes down the entire statute, and that it then “saves” or “reconstructs” parts of the statute by severing the unconstitutional part. The question is not “can we dislodge part of a statute without destroying the whole?” Rather, it is “once we have decided to remove one or more words of a statute to fix a constitutional problem, which other words can we also remove?” Two implications flow from this insight. First, any theory of inseverability must account for how the invalidation of some set of words permits a court to also invalidate other words. Second, any theory of inseverability must account for the larger unit of legislation within which severability analysis occurs—whether it be a “bill” that Congress passes at a single moment, or an “act” that is codified at a specific place in the legislative code and that has been amended over time.

Dean Tom Campbell argues that the fundamental unit of legislation is a bill. Thus when a court strikes down part of a law, it automatically “renders void the entire bill of which that unconstitutional provision was a part.” 76Campbell, supra note 13, at 1496. Campbell argues from this premise that courts should never engage in severability analysis. 77Id. at 1497–98. And his reasoning is quite valid. If we accept that the fundamental unit of legislation is a single bill enacted by Congress at one point in time, and that a court is unable to invalidate any smaller unit, then the concept of “severability” is nonsensical. A court cannot strike down an entire law and then reenact parts of it. If striking down part of a law entails striking down the whole, it follows that statutes cannot be severed and must be accepted or rejected in toto.

The main problem with Campbell’s approach is that it forces us to accept some untenable results. Any invalidation of even the smallest part of a law would require a court to destroy the entire law. 78One might imagine a variant of Campbell’s approach in which bills are treated as inseverable units unless they contain a severability clause, and those that do have such a clause are treated as severable. This would let Congress reverse the default rule of inseverability. The likely effect of such an approach would be that Congress would include pro forma severability clauses in every statute, as it does in many statutes now. See Yule Kim, Cong. Research Serv., No. 97–589, Statutory Interpretation: General Principles and Recent Trends 38 (2008), available at http://fas.org/sgp/crs/misc/97-589.pdf. Thus, the courts would end up treating most laws as severable, and the problems discussed in this Part would arise only if Congress forgets to include a severability clause (or consciously decides not to include one). Thus in Marbury v. Madison the Supreme Court would have been forced to invalidate the Judiciary Act of 1789. In Free Enterprise Fund v. Public Company Accounting Oversight Board, 79561 U.S. 477 (2010). the Court would have had to destroy the entire Sarbanes-Oxley Act. The decision in INS v. Chadha would have required the total invalidation of 196 different federal laws that contained legislative veto provisions. 80462 U.S. 919, 944 (1983) (“Since 1932, when the first veto provision was enacted into law, 295 congressional veto-type procedures have been inserted in 196 different statutes . . . .” (quoting James Abourezk, The Congressional Veto: A Contemporary Response to Executive Encroachment on Legislative Prerogatives, 52 Ind. L.J. 323, 324 (1977) (internal quotation mark omitted)). And consider Booker. The Sentencing Reform Act was enacted in 1984 as part of an omnibus bill, the “Act of October 12, 1984.” 81Act of Oct. 12, 1984, Pub. L. No. 98-473, tit. II, ch. II, 98 Stat. 1837, 1987–2040. This bill enacted an estimated 450 pages of statutory text (comprising roughly 25% of all the statutory text enacted by Congress in 1984), 82Dorsey, supra note 10, at 882 (explaining that in addition to the text of the Act reported in the United States Statutes at Large, the Act also enacted dozens of provisions by reference). and it did dozens of different things, including funding most of the federal government, 83Id. at 883–84 (citing § 101, 98 Stat. at 1837, 1844, 1867–77, 1884, 1904). authorizing federal agencies to contract with local governments for firefighting services, requiring a federal statistics office to combine St. Louis into a single statistical area, providing grants to strengthen state and local law enforcement, eliminating parole, making it harder to establish an insanity defense, and so on. 84See id. at 881–83. Under Campbell’s approach Booker’s invalidation of Section 3553(b)(1) of the Sentencing Reform Act would require undoing everything in this massive bill, including billions of dollars in appropriations that were spent decades ago. It would overburden Congress to force it to periodically reenact such massive laws, and given the realities of partisanship and legislative inertia, such reenactments could take a long time or even not happen at all. If judges were forced to choose in this way between leaving small constitutional violations unremedied and destroying entire bills, they would likely uphold statutes as a matter of course simply to avoid such major disruptions. Perversely, the Supreme Court would exercise even more power under such a theory, because it could use the certiorari process to decide whether or not to hear a constitutional challenge to a small part of a statute (and, consequently, destroy the entire statute). And no law would ever be safe—for example any successful challenge to a single tiny provision of the Affordable Care Act, whether brought today or in several decades, would fundamentally transform the American health care system. Campbell’s approach would thus turn judicial review from a scalpel into a wrecking ball.

One might respond that these are merely pragmatic concerns and that Campbell’s approach is required for formalist reasons. In other words legislation is simply indivisible, consequences be damned. This assertion runs against centuries of practice in the United States—the Supreme Court has treated laws as severable since at least Marbury, and Congress often explicitly makes laws severable through severability clauses. 85See Kim, supra note 78, at 38. Canada, by contrast, treated statutes as inseverable until the 1980s. See Kent Roach, Constitutional Remedies in Canada 14-1 (2014) (“Historically, legislation in Canada that violated the federal division of powers was struck down in its entirety.”). However Canada has in the last few decades begun finding statutes severable, largely due to their increased complexity and the resulting problems from invalidating them in toto. Thus we would have to believe that bills are indivisible even though Congress and the courts do not believe or intend them to be. This could only be achieved by constructing a constitutional metaphysics of statutes in which a bill is an atom of law that cannot be divided. One could do so by interpreting bicameralism and presentment to make bills into unbreakable units, but the features of the legislative process do not compel that interpretation. 86In defending his formalism, Campbell draws an analogy to the line-item veto: the President cannot choose which parts of a bill he vetoes, and so the courts cannot choose which parts of a bill they strike down. Campbell, supra note 13, at 1498–99. But this analogy does not hold. The line-item veto is invalid because the Presentment Clause requires that the President accept or reject a bill in toto during its passage, not because bills are inherently indivisible once Congress approves them. See Clinton v. City of New York, 524 U.S. 417, 439–41 (1998). Judges can strike down part of an enacted bill without affecting the rest of it, just as Congress can repeal part of an enacted bill while leaving the rest in place. Certainly Congress can repeal parts of an enacted bill without needing to repeal and replace the entire bill. Why, then, must a court that strikes down part of a bill also strike down the entire thing? Laws are human artifacts, and it is transcendental nonsense to argue that they are a priori indivisible even though Congress and the courts understand them to be divisible, and even though making them indivisible leads to perverse consequences. 87See Felix S. Cohen, Transcendental Nonsense and the Functional Approach, 35 Colum. L. Rev. 809 (1935).

So we must reject the view that a bill is the fundamental unit of legislation. Most of the other potential units can be dismissed with little discussion. It cannot be the act, chapter, or title that contains the invalid provision, nor can it be the entire legislative code—those options would mandate similarly problematic results. And to say that a “section” of a law is the fundamental unit of legislation is to again rely on a false formalism. Congress commonly does many independent things in the same section of a law, and the internal structure of legislation is often just a product of arbitrary organizational decisions or path dependence (e.g. the permutation of amendments and the order in which they were adopted). 88See Stern, supra note 9, at 110 (“[A]s far back as 1855 it had been wisely recognized that ‘the point is not whether they are contained in the same section, for the distribution into sections is purely artificial; but whether they are essentially and inseparably connected in substance.’” (quoting Commonwealth v. Hitchings, 71 Mass. (5 Gray) 482, 486 (1855))). No, the fundamental unit of legislation must be a unit of natural language. Letting a court strike down a single morpheme would blur the line between deleting language and rewriting it—surely a court could not remove the “fe” from “female” or the “re” from “resentencing.” However there are cases where a court can strike down a single word without rewriting the law. This could happen if a law contains a list of items, and only one of those items is unconstitutional. In the sentence “a person shall not be eligible to join the military if they are under eighteen years of age, nearsighted, female, or overweight,” the word “female” could be taken out without changing the meaning of what remains. 89See, e.g., Noble v. Mitchell, 164 U.S. 367, 369, 372 (1896) (treating as binding the Alabama Supreme Court’s severance of the words “association or partnership” from the statutory sentence “[t]he term ‘insurance company,’ as used in this article, includes every company, corporation, association or partnership organized for the purpose of transacting the business of insurance”). Of course this is an unusual example—in most cases the court will have to strike out more than just one word in order to avoid rewriting the statute or making nonsense of it. For instance, a court could not strike out the word “not” or the word “military” in the above example, at least not without making further excisions from the statute. Thus the fundamental unit of legislation for judicial review purposes is the word, but in most cases a court will have to strike down a group of words—a phrase, a sentence, or some larger unit—in order to avoid rewriting or making nonsense of the statute. 90In some cases, the Supreme Court has struck down a unit smaller than a sentence. See Stern, supra note 9, at 110 n.152.

The main task for a theory of inseverability, then, is to explain why striking down one group of words means that the court can also strike down another. Whenever a court finds a statute unconstitutional, it must determine which words it will invalidate in order to bring that statute into line with the Constitution. Often the choice will be obvious, but sometimes the court has some discretion—for example where it can strike down either the exception or the rule (as in an otherwise constitutional tax that is not applied to females), 91See supra note 71. or in complex cases where several different excisions could remedy the constitutional problem (as in Booker). 92Some scholars have argued that declaring a statute partly unconstitutional does not involve any act of excision by the reviewing court, but that the court is merely recognizing that the Constitution automatically “displaced” the problematic language. See Tribe, supra note 10, at 25; Walsh, supra note 10, at 778–84. This theory supposedly shows that there is no judicial power to find a statute inseverable because a judge that strikes down an unconstitutional provision has not actually changed anything: the statute was already partially invalid the moment it was enacted because the unconstitutional part was never real law, and the judge is merely recognizing that fact. This theory fails for two reasons. First, the Constitution does not strike down statutes—judges do. Making a statute consistent with the Constitution inevitably requires judges to exercise discretion when they excise language. Sometimes a judge has to choose between striking down an exception or a rule, or in complex cases like Booker, has to choose between two different excisions that would each solve the constitutional problem. In other kinds of cases, the judge has to choose between striking down the provision altogether or merely interpreting it to exclude an application. And even in cases where it is clear which statutory language must go, courts still exercise other sorts of remedial discretion, such as deciding whether to delay the remedy or to make it retroactive. See, e.g., N. Pipeline Constr. Co. v. Marathon Pipe Line Co., 458 U.S. 50, 88 (1982) (finding federal bankruptcy courts unconstitutional, but staying the judgment). Thus a reviewing court does not just passively recognize that the Constitution “displaced” part of a statute the moment the statute was passed. The court actively edits the statute to craft a solution to the constitutional problem. Second, even if we assume that the Constitution “displaces” the invalid parts of a statute such that they were never valid in the first place, inseverability questions still arise. It does not matter whether the language was removed through automatic “displacement” or through judicially managed excision—the statute is distorted either way. To make the case that there is no power to declare statutes inseverable one must grapple with the theories presented in Parts IV and V of this Article, each of which provides a possible account for why a judge that has found one provision unconstitutional can then find other provisions inseverable. The claim that judicial review is a passive activity does not refute any of these theories—judges could still have a remedial or interpretive power to fix statutory distortion after they recognize a constitutional “displacement.” In this first phase, the court cannot strike down more words than necessary to render the statute constitutional. 93The Supreme Court has recognized this principle. See Regan v. Time, Inc., 468 U.S. 641, 652 (1984) (plurality opinion) (“A ruling of unconstitutionality frustrates the intent of the elected representatives of the people. Therefore, a court should refrain from invalidating more of the statute than is necessary. As this Court has observed, ‘whenever an act of Congress contains unobjectionable provisions separable from those found to be unconstitutional, it is the duty of this court to so declare, and to maintain the act in so far as it is valid.’” (quoting El Paso & Ne. Ry. Co. v. Gutierrez, 215 U.S. 87, 96 (1909))). In the second phase, after the constitutional violation has been remedied, the court takes a second look at the statute and determines what word or group of words, if any, to strike out on inseverability grounds. A theory of inseverability must explain how judges can go from phase one to phase two—that is, it must explain how after remedying the constitutional violation by excising some words, the court can then go on to excise further words in order to limit statutory distortion.

A theory of inseverability must also account for the larger statutory unit within which words’ fates can be tied together. That is, it must address the question “where else in the legislative code should a court look to find inseverable language?” There are several possible answers. 94See Dorf, supra note 12, at 370. One would simply let a court look anywhere in the code for legislative language that is inseverable from the stricken words (the Code approach). A second would let a court look only at other provisions that were enacted as part of the same bill (the Bill approach). A third would let a court look only at other provisions that are part of the same uninterrupted sequence of language in the code, which is designated as a single “act” and has been amended over time (the Act approach). To see the difference between the Bill approach and the Act approach, consider the Voting Rights Act (VRA). 9552 U.S.C.A. §§ 10101–10702 (West 2014). The VRA is codified in Title 52 of the United States Code in Sections 10101 through 10702. The first VRA bill was enacted in 1965, and among other provisions it contained Section 2, which prohibited discriminatory voting laws, and Section 5, which required several states to obtain permission before changing any election procedures. 96Voting Rights Act of 1965, Pub. L. No. 89-110, §§ 2, 5, 79 Stat. 437, 437, 439. The second VRA bill was enacted in 1970, and it reenacted Section 5, as well as adding a number of other provisions, including one that lowered the national voting age to eighteen. 97See Voting Rights Act Amendments of 1970, Pub. L. No. 91-285, §§ 5, 6, 84 Stat. 314, 315–19; see also Eric S. Fish, Note, The Twenty-Sixth Amendment Enforcement Power, 121 Yale L.J. 1168, 1183 (2012). The Supreme Court then struck down this voting age provision in Oregon v. Mitchell. 98400 U.S. 112, 117–18, 130–31 (1970). So the question is, what provisions might the Court have found inseverable in Oregon v. Mitchell? Under the Bill approach, it could only have looked at other parts of the bill enacting the 1970 VRA amendments (including any non-VRA provisions, had it been an omnibus bill). Under the Act approach, it could have looked at any provisions of the VRA as codified in Title 52, including those that had been enacted in the 1965 bill. Under the Code approach, it could have looked anywhere in the legislative code. None of these three approaches is necessarily correct a priori—to determine what assumptions make each valid, we must flesh out the various theories of inseverability and explore their logics.

IV. Three Theories of Severability

Judges do not have a freestanding authority to delete language from statutes. To justify the practice of inseverability, we must somehow explain why the power of constitutional review carries with it the power to excise further language. This Part will explore three possible explanations: (1) That judges have a remedial power in equity to edit the legislative code in order to correct for the statutory distortion caused by judicial review, (2) That inseverability is an exercise in hypothetical intent-based statutory interpretation, where judges vindicate the legislature’s goals by deleting parts of the statute’s text, and (3) That laws are akin to contracts between legislators, and inseverability is a way for judges to enforce the terms of a legislative bargain. As shall be seen, none of these three theories is a plausible account of inseverability for federal courts.

A. Equitable Remedy

In A Common Law for the Age of Statutes, Guido Calabresi argues that judges should have a general power to update laws that have become obsolete. 99Guido Calabresi, A Common Law for the Age of Statutes 2 (1982). Calabresi reasons that since legislative inertia prevents many laws from keeping up with the times, courts ought to step in and either eliminate anachronistic laws or change their meanings. 100Calabresi equivocates, however, on the question whether such a power currently exists. See id. at 116–17. One possible theory of inseverability would follow a similar logic. In order to remedy statutory distortion, which arises when judges strike down part of a statute, judges would be granted a remedial power to delete other legislative language to correct the problem. 101David Gans has advocated such a theory, arguing that inseverability is a matter of judicial remedial discretion. See Gans, supra note 11, at 656–62. This remedial deletion power might stem from judges’ inherent common law authority, or it might be delegated by the legislature. The power would also be discretionary, much like equitable remedies in civil cases. Thus, for example, after striking down the individual mandate in NFIB, the Court would have had the discretion to rummage through the remainder of the ACA and determine which other provisions ought to be struck down, and which preserved, based on the Court’s vision of the best version of the statute sans mandate. This theory would take the power to invalidate language in order to make a statute constitutional, and add to it a power to invalidate further language in order to make a statute sensible as a policy matter.

It is difficult to see limits to a remedial inseverability power. When courts exercised it they would be acting as essentially legislative bodies, combing through a statute and deciding which provisions to strike down for policy reasons. The Supreme Court could articulate standards to govern these cases, or elaborate a set of tests to create a common law of severability remedies. The most natural touchstone for such tests would be the purposes of the statute as defined by the legislature. But that approach would only do so much to limit judicial discretion; since each severability case is very different, the Court could modify or reinterpret the doctrine as new situations arose, and Congress’s goals are often subject to multiple characterizations. It is also difficult to see any potential limit to the unit of legislation that a court could strike down. The logic of this theory does not point to either the Bill approach or the Act approach, and unless a limit is borrowed from some other theory, the proper unit of analysis would seem to be the entire code. Thus, for example, a court operating under this theory might cure the damage caused by striking down the individual mandate in NFIB by also striking down the age limit on Medicare, solving the adverse selection problem by creating a single payer health care system. Further, it is not clear why courts would have to stop at striking down other provisions. If courts have a remedial power to delete additional language for policy reasons, why could they not also add language to salvage a partly invalid law? 102See Glenn Chatmas Smith, From Unnecessary Surgery to Plastic Surgery: A New Approach to the Legislative Veto Severability Cases, 24 Harv. J. on Legis. 397, 463–66 (1987) (arguing that courts in severability cases have the power to reformulate the language of partly unconstitutional laws, and not just to strike language down); cf. Weinberger v. Wiesenfeld, 420 U.S. 636 (1975) (exemplifying a case in which the Supreme Court remedied a constitutionally underinclusive statute by effectively adding language to the statute). This intuitively seems like a more extreme exercise of legislative power by courts. But in a world where courts have the general authority to delete statutory language for policy reasons, it is difficult to articulate a principled reason for denying them this further power.

The equitable remedy theory is not a plausible account of inseverability, given the limited nature of federal judicial power. Federal judges cannot act without an explicit grant of jurisdiction; 103E.g., Steel Co. v. Citizens for a Better Env’t, 523 U.S. 83, 94–95 (1998). they cannot answer political questions that are ill-suited to judicial resolution or are more properly decided by a coordinate branch of government; 104E.g., Baker v. Carr, 369 U.S. 186, 217 (1962). they cannot do anything without a properly presented case or controversy; 105E.g., Lujan v. Defenders of Wildlife, 504 U.S. 555, 559–60 (1992). and they are supposed to exercise restraint in constitutional review cases and defer to the elected branches of government. 106E.g., United States v. Morrison, 529 U.S. 598, 607 (2000); INS v. Chadha, 462 U.S. 919, 944 (1983) (“We begin, of course, with the presumption that the challenged statute is valid.”). These limitations cannot be squared with a general power to invalidate legislative language for policy reasons, even if that power is only exercised to avoid statutory distortion. The equitable remedy theory would further pose a counter-majoritarian problem—federal judges lack the democratic legitimacy to reshape laws in such a fashion. And unlike in the context of judicial review, findings of inseverability do not vindicate constitutional rights. 107See Alexander M. Bickel, The Least Dangerous Branch: The Supreme Court at the Bar of Politics 16 (2d ed. 1986). When a court strikes part of a law down, it does not become a legislature. Such a power has no basis in federal courts’ inherent common law authority, because it would involve changing statutes rather than changing judicially created doctrines. It is also difficult to defend the equitable remedy theory as an implied delegation—the ability to delete statutory language as a remedy is a massive amount of power to grant federal judges without mentioning it, and Congress’s frequent pro forma use of severability clauses suggests some antipathy to declarations of inseverability. 108Kim, supra note 78, at 38 (“Congress frequently includes a pro forma severability clause in a statute . . . .”). The theory might be more plausible if Congress were to enact a law explicitly granting federal courts the power to strike down additional statutory language so as to avert the negative consequences of partial unconstitutionality. 109For example Kentucky has enacted a statute instructing courts how to resolve severability questions, although that statute frames the issue as one of legislative intent rather than remedial power. See Ky. Rev. Stat. Ann. § 446.090 (West 2006). But such a grant would raise nondelegation issues, and it would be difficult for courts to exercise this delegated power within the limits of Article III. 110Cf. Clinton v. City of New York, 524 U.S. 417, 448 (1998) (striking down a law empowering the president to issue line-item vetoes); Calabresi, supra note 99, at 114–16 (arguing that his theory of a common law for statutes could be achieved through legislative delegation to courts). It would be especially difficult to exercise this broad remedial power within the constraints of the case and controversy requirement, since the court would need a party with standing to request whatever remedy the court found appropriate. In any event, Congress has not granted the courts such a power, and it cannot be inferred from the general power of constitutional review.

B. Hypothetical Legislative Intent

A second theory would treat severability as a statutory interpretation problem. Under this theory, a court would decide a severability case by determining which provisions the legislature would hypothetically have wanted invalidated had it known that part of the statute was unconstitutional. The court would then strike down those provisions. It would do so on the theory that every statute without an explicit severability or inseverability clause contains implied instructions with respect to severability, and that the court can determine what those instructions are by looking to the legislature’s goals. Thus severability questions are treated as interpretive lacunae, akin to vague phrases or other ambiguities, and are filled in by looking at legislative intent. This approach allows courts to avoid statutory distortion by doing what the legislature would have preferred in light of its goals. 111In contrast, some scholars call for a “textualist approach” to severability. See, e.g., Movsesian, supra note 10, at 73–82. But textualism cannot provide an account of why some statutory language might be inseverable from other language, aside from the obvious case of an explicit inseverability clause. Thus to take a strict “textualist” approach to severability questions would basically mean finding everything severable. I do not explore a “textualist” theory here because the premises of textualism are consistent with the conditionality theory presented in Part V. Thus, for example, in NFIB the Supreme Court might have determined that striking down the individual mandate but leaving the rest of the ACA in place would be inconsistent with Congress’s goals, as it would lead to higher premiums or potentially drive insurance companies out of business. And under this theory the Court would not be making that decision of its own authority, it would simply be interpreting Congress’s handiwork in light of the enactors’ intentions.

This is the theory most compatible with the Supreme Court’s current severability doctrine. Both the hypothetical passage test and the intent test follow naturally from it. Indeed, if one adopts this theory, those two doctrinal tests become basically two versions of the same question. In asking whether Congress would have passed the law with the unconstitutional part severed, one is also asking whether the law still functions in the manner that Congress intended. The hypothetical intent theory also seems to follow from the Supreme Court’s statements about severability. The Court has declared that legislative intent is the key to its severability decisions, noting in the NFIB majority opinion that “[w]e seek to determine what Congress would have intended in light of the Court’s constitutional holding” 112NFIB, 132 S. Ct. 2566, 2607 (2012) (alteration in original) (quoting United States v. Booker, 543 U.S. 220, 246 (2005)) (internal quotation marks omitted). and that “[o]ur ‘touchstone for any decision about remedy is legislative intent, for a court cannot use its remedial powers to circumvent the intent of the legislature.’” 113Id. (quoting Ayotte v. Planned Parenthood of N. New Eng., 546 U.S. 320, 330 (2006)). While the Court employs a default presumption that Congress intends provisions to be severable, this default can be overcome where severance would distort the statute in ways that Congress would not have wanted.

The major problem with this theory is that, in most cases, a court cannot determine Congress’s intent with respect to severability. Except in the rare case where Congress includes explicit inseverability instructions in the statute (i.e. an inseverability clause) or makes its intentions clear in the legislative history, a court will necessarily be putting itself in Congress’s place and imagining whether Congress would have wanted to keep the remainder of the statute. The hypothetical intent theory thus devolves into imaginative reconstruction. 114See Walsh, supra note 10, at 752–53 (“The hypothetical legislative intent test gets around this absence of any actual legislative intent to discern but does so by posing a question whose answer often calls for rank speculation. The inquiry can perhaps be given some content by reference to legislative purpose, or to the ‘enterprise’ to which the statute belongs. But it can easily deteriorate into a question of what ought to happen, in which case ‘legislative intent’ adds nothing.” (citing and quoting Emily L. Sherwin, Rules and Judicial Review, 6 Legal Theory 299, 305 (2000)) (some internal quotation marks omitted)). Rather than looking to the specific intentions of the legislature, which are not generally available, courts must look to the general purposes of the statute and decide whether, given those purposes, Congress would have preferred other parts of the statute to fall alongside the unconstitutional provision. This leads to situations where a judge must go through a statute provision by provision and decide what stays or goes based on the legislature’s assumed preferences. And most complex statutes have many conflicting purposes, so that a judge can emphasize one or the other when deciding what to invalidate. This “picking your friends out of a crowd” problem is minimal when a purposivist judge uses a statute’s goals to resolve interpretive ambiguities, because the judge is restrained by the statute’s text, and the purpose only works as a kind of tiebreaker between plausible interpretations. 115The argument I am making here is not intended as a general broadside against intentionalist or purposivist methods of statutory interpretation, but only against the use of hypothetical intent and statutory purpose to declare provisions inseverable. But the problem is much worse when a judge uses the statute’s purpose to override and delete parts of the text. Indeed there is a crucial difference between using the legislative purpose to resolve an ambiguity and using the legislative purpose to slash and burn your way through a statute. The hypothetical intent approach effectively makes a court a kind of replacement Congress, telling judges to get into the heads of legislators and decide what they would have wanted to repeal.

Consider the Affordable Care Act—it had many different goals, including reducing the cost of health care, expanding coverage, reining in insurance company abuses, and more. As Chief Justice Roberts noted at the oral argument in NFIB:

[S]traight from the title, we have two complementary purposes, patient protection and affordable care. And you can’t look at something and say this promotes affordable care; therefore, it’s consistent with Congress’s intent. Because Congress had a balanced intent. You can’t look at another provision and say this promotes patient protection without asking if it’s affordable.

So, it seems to me if you ask what is going to promote Congress’s purpose, that’s just an inquiry that you can’t carry out. 116Transcript, supra note 15, at 51–52.

The existence of so many competing purposes means that a court can justify any severability result it wishes by construing Congress’s goals a particular way. Nor does this problem disappear when Congress has one clear overriding purpose—the court must still determine for itself whether the excision so undermines that purpose as to make the remainder of the law counterproductive. Consider the Federal Election Campaign Act, which had a fairly clear purpose—fighting corruption in politics by putting limits on fundraising and expenditures. If invalidating part of the law makes the goal harder to achieve (as in Buckley), the court still has to decide whether the remainder of the law is desirable in light of the goal. And the court has to balance the policy considerations on its own, because there is no way of knowing what Congress would hypothetically have done had it known that part of its bill was unconstitutional. It could have passed another version of the bill with a replacement provision, the bill minus the unconstitutional provision, no bill at all, or a totally different bill. The legislative history will not reveal which of these Congress would have done—even if Congress considered and rejected a version of the bill without the unconstitutional part (as happened with prior versions of the Sentencing Reform Act), 117See supra note 69. there is no way of knowing whether Congress would have passed that version had it known about the constitutional problem. A court cannot know what Congress’s decision would have been in a counterfactual world. Thus a court is forced to put itself in Congress’s place and do the policy weighing itself.

A further problem is that this theory of inseverability cannot be treated as merely an exercise in statutory interpretation. If it is to be justified at all, it must be by invoking some form of inherent judicial remedial power, akin to the equitable remedy theory discussed in section A. When a court finds a statute inseverable under the intent theory it is not interpreting a vague phrase in the statute or filling in any sort of gap. It is enacting judicially created fallback law by taking Congress’s purposes and using them to trump the actual text of the statute. 118See Dorf, supra note 12 (discussing legislatively created fallback law). This is amending the text, not interpreting the text. To justify such an inseverability power, one would have to embrace a theory akin to the English principle of “equity of the statute”—that judges have an inherent authority to extend a statute beyond its text, or to create exceptions not present in the text, when doing so would advance the statute’s purpose. 119See John F. Manning, Textualism and the Equity of the Statute, 101 Colum. L. Rev. 1, 22 (2001) (“The most important alternative justification for atextual and purposive interpretation relates to an ancient common law doctrine: the equity of the statute. As discussed below, when applied, that doctrine authorized courts to extend a clear statute to reach omitted cases that fell within its ratio or purpose, and conversely, to imply exceptions to such a statute when the text would inflict harsh results that did not serve the statutory purpose.”). This would transform the hypothetical intent theory from an interpretive theory into a remedial theory. And much as with the equitable remedy theory, it means that we must posit an inherent judicial authority to change the content of statutes in order to remedy statutory distortion.

The hypothetical intent theory of inseverability also, if taken on its face, entails further judicial powers that are not plausible. If courts have a power to interpret Congress’s general purposes as creating implied inseverability clauses, there is no principled way to limit this power to provisions within a single legislative bill or act. If Congress passes a law in time A, then passes a number of separate laws that assume or rely on that first law in time B, and then the first law is held unconstitutional in time C, the court could strike down or rewrite the laws passed in time B. Take the War Powers Resolution. Ten years elapsed between the enactment of the WPR and the decision in INS v. Chadha, 120462 U.S. 919 (1983). which had the effect of invalidating the WPR’s legislative veto provision. In those ten years, Congress enacted numerous laws financing and regulating the armed forces, providing for new weapons systems, and affecting our relations with other nations, all under the assumption that if the President sent troops into foreign combat then Congress would be able to withdraw them. After INS v. Chadha, the hypothetical intent theory would let a court rummage through each of those defense laws and delete the parts of them it thinks Congress would have intended to delete had Congress known about its diminished war powers. There is also no good reason to limit such a power to constitutional review cases. Why can’t judges invalidate legislative language whenever any event undermines Congress’s intended purpose? If the courts can delete parts of a statute in order to preserve Congress’s intentions after a partial invalidation, they could well do the same when dealing with any other intervening events that cause a statute to no longer serve its intended function. 121This would bring us quite close to Calabresi’s theory in A Common Law for the Age of Statutes—courts would be able to invalidate statutes when events have made them anachronistic. See Calabresi, supra note 99. And why stop at deletion? The English principle of equity of the statute permitted courts to actually add to statutes, extending them to new circumstances or creating exceptions when doing so would serve the law’s goals. 122See Manning, supra note 119, at 22. If a court can use intentionalist reasoning to delete language from a statute, it is hard to see why it could not create judicial amendments in the same fashion. This is all to say that while the hypothetical intent theory of severability may seem modest at first, its premises entail granting the judiciary very broad powers to reshape a statute’s text so as to better achieve Congress’s ends.

On examination, then, the hypothetical intent theory is indistinguishable from the inherent judicial power theory. They start from different premises but end up looking the same in practice. Both theories must posit a power to delete statutory language once part of a law is declared unconstitutional, and both theories require judges to rely on non-interpretive policy arguments when exercising such power. Invoking the hypothetical intentions of the legislature to delete parts of its text cannot be passed off as merely an act of interpretation. If it is to be justified at all, it must be as a judicially created remedy to the problem of statutory distortion. And Congress’s general intentions with regard to a partially unconstitutional statute will not give a court much guidance concerning what else Congress would have deleted. Thus, when deciding what to make inseverable, courts must produce their own policy decisions. And this theory cannot be limited to severability cases, but would give courts the power to declare statutory provisions invalid whenever an intervening event thwarts Congress’s intentions. The hypothetical intent theory is therefore implausible for the same reasons as the inherent judicial power theory. Both are inconsistent with the limited nature of judicial authority in our constitutional tradition.

C. Legislative Contract Remedy

A third theory would draw an analogy to contract law. 123See, e.g., Movsesian, supra note 10 (noting and arguing against the treatment of statutes as contracts for severability purposes). Enforcing private contracts is a classic judicial power, and legislation might be seen as a form of contract. Legislators debate and compromise with one another when determining the language of a statute, and the judiciary can play a role in preserving these compromises. On this account, judges would only exercise a remedial power to declare statutes inseverable when doing so preserves the terms of the legislative bargain. If legislators trade provision A for provision B in a single bill, and a court subsequently finds A unconstitutional, it can also invalidate B on the grounds that leaving B in place would unfairly burden the legislators who negotiated for A. The court would do so on analogy to the doctrine that contracts can be rescinded if part of the contract is void on public policy grounds. 124See Restatement (Second) of Contracts § 184 cmt. a (1981). Judges would thus provide legislators with a guarantee that the deals they make in the legislative process will not be later undone through constitutional review. The NFIB dissenters seemed to endorse a version of this theory when they declared that the dozens of minor provisions in the ACA, many of which were included as “quid pro quos” to induce legislators to support the law, should be found inseverable. 125See 132 S. Ct. 2566, 2675–76 (2012) (Scalia, Kennedy, Thomas, & Alito, JJ., dissenting) (“Often, 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. . . . Some provisions, such as requiring chain restaurants to display nutritional content, appear likely to operate as Congress intended, but they fail the second test for severability. There is no reason to believe that Congress would have enacted them independently.”).

The contract remedy account of inseverability relies on a common intuition about legislation—that it is enacted as a bundle of deals, not as many individual parts, and that much like a private contract it cannot be partially undone without changing the nature of the whole. This theory calls for the Bill approach, since the relevant unit of legislation would necessarily be a single enactment that Congress agrees to at one point in time. 126It would be absurd to analogize an act to a contract, since acts are modified by different Congresses that have different compositions. For instance, the Voting Rights Act was enacted in 1965 and amended in 1970, 1975, 1982, 1992, and 2006. George P. Laney, Cong. Research Serv., No. 95-896, The Voting Rights Act of 1965, As Amended: Its History and Current Issues (2008), available at http://fpc.state.gov/documents/organization/109556.pdf. The current VRA, which reflects all of these successive amendments, cannot meaningfully be treated as a single legislative agreement. A closer question would arise if two bills—one establishing an “Act” and the second amending the same “Act”—were passed in the same session of Congress, or alternatively if two separate bills were passed simultaneously as part of a single legislative agreement. But this is not a terribly important point to the legislative contract theory, and could be resolved either way. If courts adopted either possible rule (either that different laws enacted in the same Congress can be part of the same agreement, or that they cannot), then legislators would be able to plan their legislative bargains accordingly. Of course, two provisions of the same bill might be totally unrelated and grouped together merely for convenience (as in omnibus legislation). But such cases are easy to deal with—a court would simply find the independent parts of such bills severable absent some indication that they were traded for one another. It is also not difficult to limit the contract theory to the text of the bill in question, excluding side deals between legislators. This could be done by analogy to the parol evidence rule: if legislators are unwilling to put their agreements in the language of the statute, then they cannot use contract remedies in the courts. 127A potentially more interesting case would arise if the President made a guarantee to a crucial group of legislators that he would enact certain regulations if they voted for a law. For example, during the passage of the ACA a group of pro-life Democrats in the House of Representatives agreed to support the law only if President Obama promised to issue an executive order that would limit the use of federal funds for abortion. See Monica Davey, Under Fire for Abortion Deal, Stupak to Retire, N.Y. Times, Apr. 10, 2010, at A16, available at http://www.nytimes.com/2010/04/10/us/politics/10stupak.html. Given the fact that any important regulations could have been included in the statute itself, it would be sensible to avoid making such negotiated-for regulations part of the judicially preserved legislative contract. Including executive action would blur the distinction between laws and regulations, creating an odd situation where if a court finds a regulation invalid it may also have to invalidate a statute. This both simplifies the task of determining what is in the legislative contract and limits the enforcement of corrupt legislative deals by the courts (by excluding those that legislators are unwilling to memorialize in a statute). The Bill approach does create the curious result that subsequent amendments to an act cannot be declared inseverable—thus, for example, if the ACA had been amended after its initial passage, the Court would be unable to strike down the amendments as inseverable from the individual mandate. But that problem is not fatal, it just limits the scope of the inseverability power in potentially odd ways. The contract theory also carries the considerable benefit that it avoids the problem of judicial legislating. It does not call on courts to make judgments of policy, but instead to analyze a statute’s text and enactment history in order to determine the contours of any deals made between the legislators. Such agreements may sometimes be difficult to divine from the legislative history, 128See Movsesian, supra note 10, at 71–72 (noting that it may be difficult to find evidence of a legislative contract in cases where legislators do not express their disagreements or engage in the key negotiations on the record). but that is simply a problem of implementation. Courts could apply a default rule favoring severability, and only let that default be overcome if the legislative history reveals that an unconstitutional provision was traded for one or more other provisions in the bargaining process.

This theory of legislative contract remedies has much to recommend it. It facilitates the legislative process by guaranteeing to legislators that the terms of their deals will be enforced. Further, it removes any incentive for legislators to make compromises with the hope that the federal courts will bail them out of their concessions by declaring them unconstitutional. Unfortunately, however, it is not a convincing account of the inseverability power. This is so for two reasons.

First, legislative contract remedies do not solve the problem of statutory distortion because the operational parts of a law are not generally traded for one another. Rather, they are crafted to work together to advance Congress’s policy goals. Many compromises certainly happen while determining what those goals are and how the statute should pursue them, but it does not follow that the main provisions of a statute are consideration for one another. Consider a few familiar examples. In the campaign finance statute at issue in Buckley, the limits on fundraising and the limits on expenditures were not parallel concessions—they were complementary regulations that Congress packaged together. The same is true of the ACA—the individual mandate was not traded for the other main provisions of the ACA, such as the requirement that health insurers cover everybody that applies; rather, they were all enacted as a comprehensive system with multiple parts. This is also true of the Sentencing Reform Act—the provision making the Guidelines mandatory was not exchanged for some other provision, it was part of the enactors’ overall plan for an administrative sentencing system. While private contracts involve each party promising to perform some act or omission in exchange for an act or omission by the counterparty (e.g. “Party A agrees to satisfy all the lumber needs of Party B in exchange for one thousand dollars per month”), the core components of legislative agreements do not formalize an interparty exchange of one benefit for another. Rather, legislative agreements create legal frameworks that confer rights and obligations on third parties, who are the objects of the legislation. A final bill is certainly the result of dozens of microbargains that happen throughout the committee process and floor debates, during which alternative statutory designs are considered and rejected, and plenty of horse trading takes place. 129See Dorsey, supra note 10, at 879 (“A statute is not the result of a legislative bargain. A statute is the product of a convergence of microbargains, between and among 100 senators, 435 representatives, and the White House. Tradeoffs are made within a single legislative provision, sure, but tradeoffs are also made across the various legislative provisions in a single bill. And tradeoffs are also made between this legislative provision and that funding provision, between this provision and that decision to do something or to refrain from doing something else, between this bill and that bill, between this bill and that hearing, and between this bill and that nomination. There are people who feel confident they can pore over the statute or the legislative history and identify the legislative bargain. They may also be confident they can pore over a sonar image from the depths of Loch Ness and identify the plesiosaur. I agree there are dark shapes in the water but that’s as far as I will go.”). But these bargains are akin to negotiations between two co-authors over what their article will look like—there are debates and creative compromises, but only infrequently is there an explicit quid pro quo over the basic argument or structure of the article. Contract rescission does not give us the conceptual tools to deal with classic inseverability problems because a group of legislators rarely receives a windfall or loses the benefit of their bargain when an integral part of a statute is removed.

Second, to the extent that there are quid pro quo deals in the legislative process, these generally take the form of adding provisions to a bill in order to gain votes. This means that the contract remedies approach would create a strange version of severability. The key argument in any severability decision would not be “provisions X and Y are necessary for one another to function properly,” but rather “a crucial group of legislators would not have supported this bill without provision X.” Thus the only relevant test would be a strong version of the hypothetical passage test, and courts would decide severability cases by essentially doing a whip count. Rather than considering whether excising a provision distorts the operation of a statute, courts would have to look at whether the excision would have prevented passage of the bill. For example, during the debate over the ACA, Senator Ben Nelson traded his crucial support for roughly $100 million in Medicaid funding for Nebraska—the so-called “cornhusker kickback.” 130See Chris Frates, Payoffs for States Get Harry Reid to 60 Votes, Politico (Dec. 19, 2009, 7:56 PM EST), http://www.politico.com/news/stories/1209/30815.html. So if this provision were struck down on constitutional grounds, presumably the entire ACA would have to fall alongside it. 131Justice Scalia made this precise point at oral argument in Sebelius. See Transcript, supra note 15, at 10 (“JUSTICE SCALIA: All right. The consequence of your proposition, would Congress have enacted it without this provision, okay, that’s the consequence. That would mean that if we struck down nothing in this legislation but the—what is it called—the ‘Corn Husker kickback,’ okay, we find that to violate the constitutional proscription of venality, okay? (Laughter.) JUSTICE SCALIA: When we strike that down, it’s clear that Congress would not have passed it without that. It was—it was the means of getting the last necessary vote in the Senate. And you’re telling us that the whole statute would fall because the Cornhusker kickback is bad. That can’t be right.”). And there were dozens of other provisions that were likely added to the ACA to secure votes for its passage—it would be strange indeed if the entire law had to fall because any one of these was struck down. 132See id. at 27 (“CHIEF JUSTICE ROBERTS: The reality of the passage—I mean, this was a piece of legislation which, there was—had to be a concerted effort to gather enough votes so that it could be passed. And I suspect with a lot of these miscellaneous provisions that Justice Breyer was talking about, that was the price of a vote: Put in the Indian health care provision and I will vote for the other 2700 pages. Put in the black lung provision, and I’ll go along with it. That’s why all—many of these provisions, I think, were put in, not because they were unobjectionable. So, presumably, what Congress would have done is they wouldn’t have been able to put together, cobble together the votes to get it through.”). One could try and avoid this outcome by imposing a substantiality requirement, such that only “major” provisions lead to legislative contract rescission. But this would be a difficult distinction to draw, and would not change the basic problem. Rather than tracking the substantive logic of the statute, severability doctrine would depend on finding agreements over specific provisions. It also follows that the only available remedy is to invalidate the entire bill, since provisions like the cornhusker kickback generate legislative support for the overall package and are not tied to any other specific provision. Thus inseverability would function as a wrecking ball rather than a scalpel—courts would not be able to find one single provision inseverable from another single provision, but would have to choose between severing a provision and striking down the entire bill. The contract remedy theory is thus not a plausible account of inseverability, because it tracks the logic of legislative dealmaking rather than the substantive internal logic of a statute. 133It is possible that bargained-for treaties, which in a sense are both laws and contracts, might sometimes be treated as contracts for severability purposes. See generally Curtis J. Mahoney, Note, Treaties as Contracts: Textualism, Contract Theory, and the Interpretation of Treaties, 116 Yale L.J. 824 (2007) (arguing that relational contract theory should govern treaty interpretation).

We might also imagine a variant of this contract remedies theory. Rather than focusing on the existence of specific quid pro quos during the legislative bargaining process, we could treat severability analysis as a form of contractual gap filling. Just as courts in contract cases look to the goals of the drafters when filling in incomplete contracts, courts in severability cases could look to the goals of the drafters when determining how to respond to an event that the drafters did not anticipate (such as partial invalidation of the law). 134See Gillian K. Hadfield, Incomplete Contracts and Statutes, 12 Int’l Rev. L. & Econ. 257, 257 (1992). And by analogy to contract law, this might be done by disaggregating the intentions of the different actors who maintain legislative veto-gates and trying to determine what fix each of them would have consented to had they been informed during the drafting process that the bill was partly unconstitutional. 135See McNollgast, Positive Canons: The Role of Legislative Bargains in Statutory Interpretation, 80 Geo. L.J. 705 (1992) (arguing for such an approach to statutory interpretation). This variant contract remedies theory need not be analyzed here, because it is indistinguishable from the hypothetical intent theory addressed in the prior section. 136See supra Part IV.B. Just like this contract theory variant, the hypothetical intent theory relies on judicial recognition of Congress’s hypothetical intentions to motivate reshaping a statute that has been partially invalidated. 137Unfortunately, the analogy to contract law does not provide much help in locating congressional intent. See Hadfield, supra note 134, at 257–58 (“We will be disappointed, therefore, if we look to incomplete contracting principles in hope of avoiding the morass of ‘intent.’ Indeed, if we imported the standard gap filling analysis from contracting into the statutory interpretation arena, we would still run smack into Arrow’s theorem: in the legislative case there is simply no answer to the gap-filling counterfactual, ‘what would the parties have voted for?’”).

V. Severability as Legislative Conditionality

A. The Theory

Having rejected the three theories that might justify an expansive approach to inseverability, it is time to consider a narrower one: legislative conditionality. The legislative conditionality theory would make inseverability a simple problem of statutory interpretation. It would instruct courts to decide inseverability issues by looking at whether the legislature made part of a statute conditional on the continued validity of another part. Generally legislatures have the power to enact conditional legislation, making a provision operative or inoperative once a triggering event occurs. 138Michael Dorf has argued that such conditional legislation could be unconstitutional if it is used to coerce the courts into not exercising their constitutional review powers. For example, if Congress created a fallback law that eliminated school lunches if the Supreme Court struck down the Defense of Marriage Act, he argues that that would be unconstitutionally coercive. See Dorf, supra note 12, at 327–29, 333–36; see also Kameny, supra note 12; cf. Bruce K. Miller, Constitutional Remedies for Underinclusive Statutes: A Critical Appraisal of Heckler v. Mathews, 20 Harv. C.R.-C.L. L. Rev. 79 (1985). This argument is appealing but ultimately not convincing. Congress has the power to enact punitive legislation if it likes, so long as it stays within the limits set by the Constitution. Congress cannot cut judicial pay or limit judicial tenure, but it can certainly enact bad laws to punish the judiciary, and it can design those laws to take effect only after a triggering event. Nothing in the Constitution protects any of the three branches from bullying. If Congress wants to steal lunch money because the judiciary struck down a law, that is Congress’s prerogative. As a practical matter, though, it seems unlikely that a legislature would actually create such a punitive fallback law. And if it did, it would almost certainly repeal the law once the triggering condition was satisfied in order to avoid an electoral backlash. The public would rightly blame Congress, not the courts, for their kids going hungry. It is also worth noting that policy blackmail is not limited to Congress. See Calabresi, supra note 99, at 269–70 (describing Reid v. Covert, 354 U.S. 1 (1957), as a failed attempt at judicial blackmail of Congress). Thus a legislature could create explicit fallback law through an inseverability clause—a clause establishing that if provision A is held unconstitutional, then provision B is no longer valid law. But legislatures can also create such fallback law implicitly. 139A number of scholars have adopted the view that courts should find everything severable unless there is an explicit inseverability clause. See supra note 10. However none of these have explored the possibility of implicit inseverability. They can do so in three ways: by writing a statute so that its language no longer makes sense without the excised part, by writing a statute so that one part cannot be enforced if another part is excised, or by enacting a provision that serves no purpose without another provision. 140Rachel Ezzell has made a similar proposal, arguing that a statute should be held inseverable only (1) when there is an explicit inseverability clause or (2) when the severed law could not logically be enforced. See Ezzell, supra note 10. The legislative conditionality approach would add to these (3) when statutory language is inseverable as a consequence of its syntactic logic and (4) when one provision serves no purpose without another. For example, the sentence “the Sentencing Commission shall have exclusive authority to enact sentencing guidelines” is implicitly conditional on the existence of a body called the “Sentencing Commission.” If a court strikes down the provision creating that body, 141Cf. Mistretta v. United States, 488 U.S. 361, 413 (1989) (Scalia, J., dissenting) (concluding that the Sentencing Commission is unconstitutional on nondelegation grounds). then this statutory language becomes nonsense. Similarly, if Congress enacts a law creating a crime, and then a separate law providing the sentence for that crime, the sentencing provision is implicitly conditional on the crime provision. If the crime is invalidated, then the sentencing provision cannot have independent effect even though its syntax still makes sense. Finally, even if a provision is textually coherent and independently enforceable without the excised provision, it can still be implicitly conditional if it no longer serves any plausible purpose. For example if a statute provides for regulation of gasoline prices and also for the collection of data solely to assist that regulation, the data collection could be found conditional on the ability to regulate. 142See Williams v. Standard Oil Co. of La., 278 U.S. 235 (1929) (striking down a law fixing gas prices and finding inseverable provisions creating a state agency to administer the price fixing).

When a court finds a provision implicitly conditional, it engages in a type of statutory interpretation. Since it cannot find an explicit inseverability clause in the text of the statute, it looks instead to the logical relationship between the statute’s provisions to see if one is necessarily implied. This is quite similar to two other interpretive doctrines: implied repeal and conflict preemption. When a court finds a law repealed by implication, it concludes that the legislature repealed the law without saying so. A court can make such a finding only in situations where there is a “plain repugnancy” between the two provisions, such that they cannot both be given full effect without creating a contradiction. 143Gordon v. NYSE, 422 U.S. 659, 682 (1975) (finding that the Securities Exchange Act impliedly repeals the antitrust laws insofar as they regulate securities transactions). Similarly, when a court finds a state law preempted by a federal law through the doctrine of implied conflict preemption, it concludes that Congress preempted the state law without acknowledging having done so. A court finds such preemption in situations where it is “impossible for a private party to comply with both state and federal requirements.” 144Geier v. Am. Honda Motor Co., 529 U.S. 861, 899 (2000) (Stevens, J., dissenting) (quoting Freightliner Corp. v. Myrick, 514 U.S. 280, 287 (1995)) (internal quotation marks omitted). Under both of these doctrines, members of Congress might not have known that by enacting a law they were repealing or preempting anything, but the law is still properly interpreted as including the repeal or preemption because that is the only sensible reading. Implied conditionality follows the same logic. Even if Congress did not enact an inseverability clause, courts can infer one where removing legislative language will render a statute meaningless or incoherent.

The legislative conditionality theory is similar to the hypothetical intent theory discussed above, since both treat severability as a feature of statutory interpretation. However, there is an important distinction between them. The legislative conditionality theory calls for interpretation of a statute, rather than imaginative reconstruction of a statute. It looks at whether the statutory remainder remains logically coherent with the unconstitutional provision severed, rather than at what Congress might hypothetically have wanted if that provision were invalidated. When a court decides whether two provisions are implicitly conditional on one another, it looks at the texts and purposes of the respective provisions and determines whether they depend on each other within the architectural logic of the statute. It asks questions concerning the statute’s internal design—does provision X make syntactic sense without provision Y? Can X be enforced without Y? Does X still serve any purpose without Y? However the court does not, as in the hypothetical intent theory, ask whether severing Y but leaving X in place will better serve Congress’s legislative goals. The court thus engages in a narrower, more “interpretive” inquiry that is focused on the internal architecture of the statute and how its pieces fit together, and not on the policy consequences of partial invalidation. Similar to the implied repeal doctrine, the court only concludes that two provisions are implicitly conditional when it would be either impossible or plainly illogical to preserve one provision without the other. It does not find implied conditionality merely to avoid negative consequences where the provisions are functionally independent.

Legislative conditionality thus has one major downside: it gives courts limited tools to deal with the problem of statutory distortion. This is because, unlike with the hypothetical intent theory, courts cannot strike down legislative language for non-interpretive policy reasons. If the Supreme Court had struck down the individual mandate in NFIB and applied the conditionality theory, the severability inquiry would have been a simple one. All of the other operative provisions of the Affordable Care Act were both syntactically coherent and capable of being enforced without the individual mandate, and they still served a purpose without the mandate. 145See Patient Protection and Affordable Care Act, 42 U.S.C. §§ 18001–18121 (2012). Thus the rest of the ACA would have remained in place, notwithstanding the facts that insurance premiums would have gone up or that health insurers may have gone bankrupt. Similarly, the restrictions on campaign donations in the Federal Election Campaign Act were neither explicitly nor implicitly conditional on the expenditure restrictions struck down in Buckley. 146See Federal Election Campaign Act of 1971, Pub. L. No. 92-225, 86 Stat. 3. Thus, under the conditionality theory, the Buckley Court got the severability question right. It is not the Court’s role to strike down additional provisions for pure policy reasons. If such broken statutes are going to be fixed, Congress must do the mending.

However, legislative conditionality is the only theory of inseverability that is ultimately defensible in the federal judiciary. Unlike the hypothetical intent and equitable remedy theories, it does not require positing an implausibly broad power to strike down statutory language. And unlike the contract remedies theory, it does not turn legislative deals into agreements enforceable in court. Instead it infers that Congress made certain statutory provisions inseverable from others, and shows that the courts’ job is simply to interpret statutes correctly so as to find out which language is conditioned on which other language. This theory also avoids making courts approach severability questions like legislatures. Rather than considering the policy justifications for deleting additional language, courts look to the language and internal logic of a statute and identify implicit or explicit inseverability instructions. And while legislative conditionality cannot fix all instances where severing language creates undesirable policy outcomes, it still lets courts avoid situations where severing a provision makes other statutory language ineffectual or incoherent. It is thus at least better than the contract remedy theory, which only lets courts enforce the terms of legislative deals. While legislative conditionality is certainly not a perfect theory, it is the most plausible account of the inseverability power in the federal system. Its details and implications will be explored further in the following sections.

B. Finding Conditionality

Legislative conditionality is found through the normal process of statutory interpretation. Thus, when judges search for conditionality they explore the kinds of evidence that determine a statute’s meaning—text, legislative history, purpose, and so forth. Here we will consider the different circumstances under which a judge could find legislative conditionality. We will consider seven kinds of cases: (1) where there is an explicit inseverability clause; (2) where there is no inseverability clause, but the legislative history shows an intention to make part of the statue conditional; (3) where severing language would leave the text incoherent; (4) where severing language would leave part of the statute incapable of being given effect; (5) where part of the statute serves no purpose once another part is severed; (6) where severing part of the statute changes the statute’s overall effect in an undesirable way; and (7) where there is implicit conditionality, but the statute contains an express severability clause.

The easiest case is a statute containing a clear inseverability clause, a clause that instructs judges to strike down provision Y if provision X is held unconstitutional. 147See, e.g., Social Security Amendments of 1977, Pub. L. No. 95-216, § 334(g)(3), 91 Stat. 1509, 1547 (“If any provision of this subsection . . . is held invalid . . . the application of this subsection to any other persons or circumstances shall also be considered invalid.”); Zobel v. Williams, 457 U.S. 55, 65 (1982) (reviewing an Alaska statute that provided “If any provision enacted in sec. 2 of this Act . . . is held to be invalid by the final judgment, decision or order of a court of competent jurisdiction, then that provision is nonseverable, and all provisions enacted in sec. 2 of this Act are invalid and of no force or effect” (quoting 1980 Alaska Sess. Laws ch. 21, § 4, at 9) (internal quotation mark omitted)). If a law specifically instructs that the invalidation of one provision causes others to lose their force, then judges must follow that instruction and find the provisions inseverable. This follows from the premise that inseverability is a legislative power and not a judicial one. A legislature can enact a law that takes effect or sunsets after some triggering condition, and a judicial finding of unconstitutionality could be one such condition. Some scholars have argued that inseverability clauses are unconstitutional because they tell judges how to construe statutes, which encroaches on the powers of the judiciary. 148See Dorsey, supra note 10, at 892; see also Kameny, supra note 12, at 999–1000, 1009. But this is backwards. Inseverability clauses are not instructions for how to interpret a statute; they are substantive provisions with their own binding force. And making statutes inseverable is not a judicial power; it is a legislative power the exercise of which sometimes requires judicial interpretation.

The situation is somewhat more complicated if the legislature does not write an explicit inseverability clause, but makes it clear in the legislative history that it means for one provision to be conditional on another. For example, imagine that a statute’s text is silent on inseverability, but the main committee report states that “provision A serves no purpose without provision B, and if provision B were invalidated we would like provision A struck down as well.” A judge’s approach in such a case will necessarily reflect that judge’s theory of statutory interpretation. A textualist in the mode of Justice Scalia or Judge Easterbrook would not find the legislative history relevant and so would not count it in favor of inseverability. But a judge with a more intentionalist orientation might use the legislative history to find implied inseverability. It should be noted, however, that an intentionalist judge would need to find evidence that Congress actually intended to make one provision conditional on the other, that is, that it meant one to fall if the other fell. Evidence that Congress meant the two provisions to work in concert, or that it thought that one would not function well without the other, is insufficient by itself to establish such conditionality.

Courts can also find implied conditionality as a function of a statute’s syntax, where severing language would render a provision either meaningless or absurd. This type of conditionality is not a product of the legislature’s conscious intentions—it is a feature of the nature of written language. Parts of a sentence are often necessary to the whole, and so when we speak in sentences we sometimes create unbreakable units of meaning. The sentence “Emily’s puppy will get all the treats” makes no sense if Emily has no puppy. Similarly, the sentence “bankruptcy courts shall have exclusive jurisdiction over such actions” makes no sense if the Supreme Court has made bankruptcy courts unconstitutional. This principle of syntactic conditionality sometimes holds even if the reformed sentence still makes sense after an excision, if it gives the sentence an accidental new meaning. For example, the sentence “there shall never be more than one hundred female students enrolled at the U.S. Naval Academy” is unconstitutional gender discrimination, but a court cannot remedy it by striking out only the word “female.” That would leave us with a totally accidental meaning, limiting the size of the Academy to 100 students, which is absurd in light of the statute’s purpose. Thus the entire sentence is implicitly conditional on the word “female.” 149This principle also helps explain why courts never strike down fragments of words, for example the prefix “fe” in “female.” Words are implicitly indivisible, such that each part of a word is conditional on each other part. See supra Part III (discussing why courts cannot strike down single morphemes); see also Noble v. Mitchell, 164 U.S. 367, 372 (1896). If one strikes down that word, the rest of the sentence must go with it.

This concept of syntactic conditionality helps us make sense of a kind of severability that this Article has not yet dealt with—severable applications. When a statutory provision has many applications, some of them constitutional and some not, a court must decide whether to strike the provision down entirely or sever the unconstitutional applications. For example, in Northern Pipeline v. Marathon Pipe Line the Supreme Court held that bankruptcy courts could not hear state law contract claims, and found the rest of bankruptcy courts’ jurisdiction inseverable in part because all of bankruptcy jurisdiction was provided by a single statutory sentence: 28 U.S.C. § 1471(b), providing jurisdiction for “all civil proceedings arising under title 11.” 15028 U.S.C. § 1471(b) (1982), invalidated by N. Pipeline Constr. Co. v. Marathon Pipe Line Co., 458 U.S. 50 (1982). Under the principle of textual conditionality, the Court’s basis for finding this sentence facially invalid would be that the sentence becomes nonsense if its main applications are taken out. The sentence is implicitly conditional on its main applications, and the result in the case would be different if the bankruptcy code contained two separate textual grants of jurisdiction—one for state common law claims and the other for all other claims. Consider another example. If a law provides that “civil rights will be protected throughout the United States,” but the Supreme Court determines that the statute is invalid as applied to the fifty states and the territories, it might also strike down the statute as applied to American flagged ships. 151See Butts v. Merchs. & Miners Transp. Co., 230 U.S. 126, 138 (1913) (holding that the Civil Rights Act of 1875, which had been invalidated as applied within the states, was also invalid as to United States vessels engaged in coastal trade). In doing so, the Court would be determining that the phrase “throughout the United States” becomes nonsensical if it is only applied to conduct on a few boats. But a separate sentence specifically providing for civil rights on ships would have been severable. Textually implied conditionality thus might help make sense of the facial versus as-applied distinction in at least some cases. 152See generally Dorf, supra note 11, at 251–83 (showing that the Supreme Court often makes statutory applications inseverable from one another, notwithstanding the principle announced in Salerno v. United States, 481 U.S. 739 (1987), that applications are presumptively severable).

A similar but distinct category of implied conditionality consists of cases where one statutory provision cannot be given legal effect without another. This kind of case is captured by the Supreme Court’s independence test—a provision is inseverable unless “standing alone, legal effect can be given to it.” 153Dorchy v. Kansas, 264 U.S. 286, 290 (1924). One example of such a provision would be an appropriation for an unconstitutional enactment. If Congress passes a law to build an unconstitutional structure (say a statue of Moses to be placed in the national mall), and then separately enacts a law appropriating funds for that structure, the funds are conditional on the structure. Money cannot be spent on something unless that thing is permitted to happen, and thus a court can reasonably conclude that the appropriation is conditional on the other enactment. Similarly, if a tax is held unconstitutional, any provisions for the collection of that tax could be struck down as well. And if a law creating a crime is held unconstitutional, a separate sentencing statute cannot remain valid. A similar finding of implied conditionality might be made where a provision can have no legal effect after the larger statutory structure that contains it has been invalidated. So for example, if the Supreme Court were to strike down the Sentencing Reform Act and its entire Guidelines structure, any separate statute creating specific Guidelines rules would also have to go. Those rules could have no independent legal effect without the larger structure, so if they remained they would be free-floating enactments without any context or force. 154See Stern, supra note 9, at 76 n.1 (“In theory, a legislative body has power to enact laws or parts of laws which are incapable of being given legal effect. Accordingly, if a legislative body should for some or no reason desire that an ineffectual or meaningless part of a law stand alone, the legislative intention should prevail. But the fact that valid provisions of a statute are incapable of having legal effect by themselves is ordinarily conclusive proof that the legislature did not intend them to stand by themselves. Legislatures do not pass laws merely to increase the size of the statute books. Inasmuch as the same results are reached under either theory, the difference in approach has no substantial significance and will not be considered further here.”). Even if judges could not find provisions inseverable for lack of independence, such provisions would still not have any legal consequence since they could not meaningfully be enforced. They would essentially be deadwood—filler for the legislative code. But there is no meaningful distinction between a court recognizing a provision as deadwood and the court striking that provision down. 155When a court strikes down statutory language, it does not actually delete that language from the code. Only the legislature can do that. See, e.g., 18 U.S.C. § 700 (2012) (establishing criminal penalties for defacing the American flag), invalidated by United States v. Eichman, 496 U.S. 310 (1992). When a court strikes a provision down, this is only a declaration that it has no legal effect. Hence there is no difference between a judge striking a provision down and a judge finding that a provision is deadwood. In both cases the judge is declaring that the provision will no longer have any legal effect, but is leaving its language in place. And in both cases, if the provision’s unconstitutionality is cured (say by a constitutional amendment), then the provision becomes valid again even if it is not reenacted. Thus, when a court recognizes that a provision cannot have legal effect without the invalidated provision, the court is necessarily declaring those two provisions inseverable and striking them down together.

Implied conditionality might also be found where a remaining provision no longer serves any plausible purpose. Even if a provision makes syntactic sense and can be enforced without the excised language, a court could potentially find it invalid because it no longer has any function. 156See 2 Norman J. Singer, Sutherland Statutes and Statutory Construction § 44:4, at 576 (6th ed. 2001) (“Even where part of an act is independent and valid, other parts which are not themselves substantively invalid but have no separate function to perform independent of the invalid portions of the act are also held invalid.”). Consider again the statue of Moses on the national mall, which would violate the Establishment Clause. If Congress had a separate provision that created a wooden pedestal for the statue (the pedestal would, let’s stipulate, lack any religious symbolism), that provision could be found conditional on the validity of the statue. The pedestal provision might make perfect textual sense, and the pedestal might logically still be built without a statue atop it, but the pedestal is so clearly intended only to support the statue that a court could reasonably conclude Congress intended the one to be conditional on the other. In doing so the court would not be imposing its own policy judgment about whether the pedestal is still desirable, but merely determining whether the pedestal has any independent function in the statute. The Supreme Court case Williams v. Standard Oil Co. of Louisiana involved precisely this kind of inseverability. 157278 U.S. 235 (1929). Tennessee had created an agency to fix gasoline prices, and the Supreme Court determined that this price regulation was unconstitutional. However the agency was also empowered to do other tasks, such as collecting data that were intended solely for the purpose of helping it fix gas prices. The Court found these other functions inseverable because they were “mere adjuncts of the price-fixing provisions of the law or mere aids to their effective execution.” 158Id. at 243. Such provisions are like a barnacle going down with a ship. Finding implied conditionality in such a case does require looking at the purposes of the statute (e.g. to regulate gas prices, or construct a statue), and not just at its text. As a consequence, this kind of implied conditionality seems perilously close to an inquiry into hypothetical congressional intent. To avoid that danger, it must be limited to only clear cases—cases where conditionality can be inferred from the fact that a provision plainly serves no other function. If there is a valid argument that the legislature intended the provision to serve an independent purpose, then a court cannot find it implicitly conditional. By doing so the court would go beyond interpreting the internal logic of the statute, and would instead be looking at Congress’s hypothetical preferences. 159There may be borderline cases. For example, in Williams v. Standard Oil Co. of Louisiana one might argue that the data-collecting agency also serves an academic research function for professors with an interest in oil prices, even though there is probably little evidence of such a purpose in the statute’s text or legislative history (and it seems like an expensive way to generate data for professors). In deciding whether the data collection provision is conditional on the rest of the statute, judges would have to determine whether legislators actually had that independent purpose in mind when they enacted the statute. Such an inquiry would be similar in structure to a Title VII case, where one party claims that they were fired for discriminatory reasons and the other party must show an independent rationale for the firing. Here one party claims that provision A exists only to assist the (now unconstitutional) provision B, and the other party must show an independent justification for A. This inquiry is much more narrow than the hypothetical passage test, which asks whether the legislature (given its substantive policy preferences) would have enacted B without A.

The conditionality theory does not permit courts to find provisions inseverable merely on the grounds that leaving them in place would create adverse policy consequences or that it would frustrate legislative intent. The approach is therefore incompatible with the intent test and the hypothetical passage test, and is generally a weak tool for dealing with statutory distortion. There is no defensible basis for interpreting the ACA to find the mandates on insurance companies (much less the entire Act) conditional on the individual mandate. Congress enacted two clearly separate provisions—a requirement that individuals buy health insurance and a requirement that insurers sell their products without regard for preexisting conditions. These provisions make textual sense without one another, can each be enforced without the other, and would each serve a plausible purpose even if the other were no longer in effect. The mere fact that Congress might not hypothetically have wanted one without the other is insufficient to find them conditional. The same is true of the Federal Election Campaign Act. Eliminating the restrictions on campaign expenditures may create major problems for that statute, and may even make it undesirable as a matter of policy, but that is insufficient evidence for a court to find the remainder of the statute conditional on the excised provisions. Courts may only find laws conditional if they can reasonably be interpreted as containing legislative instructions to eliminate one provision if another falls.

A final issue to consider is what legislative conditionality means for severability clauses. Under current doctrine severability clauses are largely irrelevant because such clauses only establish a presumption of severability, and there is already a default presumption favoring severability in all cases regardless of whether there is such a clause. 160Severability clauses are officially understood to create a presumption favoring severability. See Alaska Airlines, Inc. v. Brock, 480 U.S. 678, 686 (1987). However the absence of such a clause does not reverse the presumption favoring severability. See id. Rather, a presumption of severability operates in every case regardless of whether there is a severability clause. See Regan v. Time, Inc., 468 U.S. 641, 653 (1984) (plurality opinion). It is not clear from current doctrine whether the default presumption is as easily overcome as the presumption created by a severability clause. Under the conditionality approach severability clauses would also ordinarily be surplusage, since this approach finds statutes inseverable only in a very limited set of cases. Thus such clauses would affect a court’s decision in only a few kinds of circumstances. A severability clause would presumably overcome any legislative history suggesting that parts of the statute were conditional on one another, unless a judge believed that legislative history could trump the plain text. A more interesting question is how such clauses interact with implied conditionality. In the first two kinds of cases—where severing leaves the remainder of the statute textually incoherent or incapable of being enforced—it is difficult to see how a severability clause could be followed. Severability clauses are generally not applied to specific statutory text, but communicate a general legislative intention that pieces of the statute be treated as severable. 161See Max Radin, A Short Way with Statutes, 56 Harv. L. Rev. 388, 419 (1942) (“[T]he ‘severability clause’ is generally a formulaic phrase, added almost as a matter of course whenever there is the least question as to the constitutionality of the statute. It is derived from a resentment against the power which the courts of the United States have so long exercised and which the legislature has accepted, although neither fully nor with good grace. The decision to take the severability clause literally, whenever it can be so taken, is an example of comity between coordinate branches of the government, but it is not an example of attempting to discover what was really intended.”). That intention could be defeated in any particular case if severance would leave the remaining language incoherent or unenforceable. And even if courts felt constrained to keep such language in place, it could have no legal consequence. There is no meaningful difference between striking a provision down and leaving it in place but denying it any legal effect. However a severability clause could have legal effect in the third type of implied severability case, where the remaining provision serves no plausible purpose without the invalidated provision. So, for example, if the law creating both the statue of Moses and the pedestal it rests on contained a severability clause, then the court would keep the pedestal in place. In such a case, the court infers conditionality from the uselessness of the remainder of the statute, but such an inference would be easily overcome by explicit textual instructions to sever.

In summary, legislative conditionality can be found in four types of cases: (1) where the legislature instructs it (through an explicit inseverability clause, or possibly through legislative history suggesting conditionality); (2) where a severed statute’s text becomes nonsensical or takes on a totally accidental meaning; (3) where a remaining provision cannot be enforced; and (4) where a remaining provision serves no plausible purpose. Legislative conditionality cannot be found in cases where severing would merely have undesirable policy consequences, or where implicit conditionality is trumped by an explicit severability clause.

C. The Need for Standing

If a court believes that provision A is unconstitutional, and also believes that provisions A and B are inseverable, it will need a party with standing to challenge both A and B if it is to strike them both down. If the only case or controversy in the lawsuit is over A, then the court is powerless to dispose of B. This follows from the conditionality theory, but it is also true regardless of which theory of inseverability one adopts—any of the four approaches discussed in this Article would require a party to show a concrete injury from the challenged provision before it is struck down. 162See, e.g., Raines v. Byrd, 521 U.S. 811, 818–20 (1997). It does not matter whether the ability to declare laws inseverable is a matter of statutory interpretation or of remedial power—it can only be exercised within the limits of Article III, which requires an active case or controversy before a court can provide any sort of remedy. And the fact that a party has standing to challenge one provision of a statute does not automatically grant that party standing to challenge other provisions. 163See Davis v. FEC, 554 U.S. 724, 733–34 (2008) (“The fact that Davis has standing to challenge § 319(b) does not necessarily mean that he also has standing to challenge the scheme of contribution limitations that applies when § 319(a) comes into play. ‘[S]tanding is not dispensed in gross.’ Rather, ‘a plaintiff must demonstrate standing for each claim he seeks to press’ and ‘for each form of relief’ that is sought.” (alteration in original) (quoting, in turn, Lewis v. Casey, 518 U.S. 343, 358 n.6 (1996) and DaimlerChrysler Corp. v. Cuno, 547 U.S. 332, 352 (2006)) (some internal quotation marks omitted)). A party must separately establish standing for each form of relief sought. 164Friends of the Earth, Inc. v. Laidlaw Envtl. Servs. (TOC), Inc., 528 U.S. 167, 185 (2000) (“Laidlaw is right to insist that a plaintiff must demonstrate standing separately for each form of relief sought.”). But see Erik R. Zimmerman, Supplemental Standing for Severability, 109 Nw. L. Rev. (forthcoming 2015) (manuscript at 7–9), available at http://ssrn.com/abstract=2428332 (noting this conflict between the Sebelius dissent and the Laidlaw principle that standing must be established for every claim, but arguing that it can be resolved through “supplemental standing”). The Supreme Court recognized this principle in the severability context in Printz v. United States. 165521 U.S. 898 (1997). In that case, it struck down a provision of the Brady Act requiring state officials to conduct background checks on people purchasing handguns. 166Id. at 933. However the Court declined to consider whether the rest of the Brady Act was severable from this background checks requirement, because the parties before it lacked standing to challenge the law’s other provisions. The Court noted, “These are important questions, but we have no business answering them in these cases. These provisions burden only firearms dealers and purchasers, and no plaintiff in either of those categories is before us here. We decline to speculate regarding the rights and obligations of parties not before the Court.” 167Id. at 935. In sum, there is no inseverability exception to the Article III standing requirement.

The four dissenters in NFIB argued to the contrary. In deciding that the constitutional problems with the individual mandate required striking down the entire ACA, they found no barrier in the fact that the parties before them lacked standing to challenge many of the ACA’s provisions. The dissenters reasoned:

To be sure, an argument can be made that those portions of the Act that none of the parties has standing to challenge cannot be held nonseverable. The response to this argument is that our cases do not support it. It would be particularly destructive of sound government to apply such a rule with regard to a multifaceted piece of legislation like the ACA. It would take years, perhaps decades, for each of its provisions to be adjudicated separately—and for some of them (those simply expending federal funds) no one may have separate standing. The Federal Government, the States, and private parties ought to know at once whether the entire legislation fails. 168NFIB, 132 S. Ct. 2566, 2671 (2012) (Scalia, Kennedy, Thomas, & Alito, JJ., dissenting) (citing Williams v. Standard Oil Co. of La., 278 U.S. 235, 242–44 (1929)).

It may be true that an inseverability exception to the standing requirement would be desirable for judicial economy reasons. But that does not mean such an exception exists. The dissenters’ reasoning would seem to allow for advisory opinions whenever multiple adjudications would be “destructive of sound government.” But such disruption is a necessary cost to our ex post system of judicial review. 169See Brianne J. Gorod, The Collateral Consequences of Ex Post Judicial Review, 88 Wash. L. Rev. 903, 904 (2013). In INS v. Chadha, the Court invalidated a type of provision—the legislative veto—that existed in 196 different statutes. 170462 U.S. 919, 944 (1983). But the Court did not then go on to issue advisory judgments on all of those statutes, much less determine whether each of their legislative veto provisions were severable. 171Id. at 959. Such an exercise would have saved the federal courts considerable time and effort, and the entities affected by those statutes considerable uncertainty. But our system simply does not allow courts to pass judgment on statutes before a case or controversy has ripened. 172The dissent in NFIB also seems to contradict the general principle that a person who is burdened by provision A cannot, except under limited circumstances, sue to invalidate a statute because provision B is unconstitutional and inseverable from provision A. See Broadrick v. Oklahoma, 413 U.S. 601, 610 (1973) (“Embedded in the traditional rules governing constitutional adjudication is the principle that a person to whom a statute may constitutionally be applied will not be heard to challenge that statute on the ground that it may conceivably be applied unconstitutionally to others . . . .”). If severability determinations do not require standing, it is difficult to see why a party could not bring this kind of challenge.

Further, contra the NFIB dissenters, the administrability consequences of requiring Article III standing are not insurmountable. The Supreme Court embraced a case-by-case approach in Printz, refusing to decide severability questions that were not properly before it. And the resulting uncertainty over whether the rest of the Brady Act remains valid has not created major problems. To the extent that uncertainty is disruptive, embracing the conditionality theory would drastically improve the situation by making severability determinations more predictable. Legislative conditionality does not permit a court to find part of a statute inseverable on policy grounds. Thus it limits the number of situations where the fate of the statutory remainder is left uncertain. Unless the remaining provisions of the statute appear to be textually incoherent, unenforceable, or totally without purpose, they will be severable. Such situations are generally easy to spot and could be pointed out in dicta. The Supreme Court may also, if seeking to avoid uncertainty, grant certiorari only to plaintiffs that have standing to challenge possibly inseverable provisions, or allow plaintiffs with such standing to join a case at a later procedural stage. 173Another potential solution for complex statutes with many provisions would be to articulate the severability ruling broadly. Imagine that a statute has five provisions: A, B, C, D, and E. A is unconstitutional, and the parties before the court only have standing to challenge A and B. The court might, in some cases, articulate its inseverability decision in a way that clearly applies to C, D, and E as well. Of course it could only do so if all of the provisions were inseverable for the same reason. See Richard H. Fallon, Jr., Commentary, As-Applied and Facial Challenges and Third-Party Standing, 113 Harv. L. Rev. 1321 (2000). But creating a sui generis inseverability exception to the standing requirement is not an option.

D. Conditionality Is Not Limited to the Same Bill or Act

A final implication of the conditionality theory is that any provision in a legislative code can be found inseverable from any other provision in the code. Legislative conditionality cannot be confined to a single bill or act because conditionality is ultimately a legislative decision, and the legislature is not bound by such formal categories. If it decides to make a provision in one law conditional on a provision in an entirely separate law, it can do so. This fact does not create a major problem for the conditionality theory in the same way that it would for the hypothetical intent theory or the remedial power theory, because conditionality does not afford judges nearly as much discretion as those other theories. With the hypothetical intent or remedial theories, a court might comb through the entire code looking for legislative language to find inseverable on policy grounds. But the conditionality theory does not permit that, because it does not allow courts to find provisions inseverable simply because severing them creates policy problems. A court would only be able to find different enactments conditional if it could fit them into one of the categories discussed in Part V.B. For example, if Congress enacts one bill providing for construction of an unconstitutional statue, and later appropriates funding for that statue through an entirely separate bill, the funding provision can be found conditional on the provision creating the statue. Similarly, if Congress passes a bill creating an agency, and then many years later passes a separate bill providing certain new powers to that agency, the language granting the powers is conditional on the language creating the agency even though they are in separate bills (and, indeed, even if they are in separate acts and titles).

This is a counterintuitive approach to severability because it abandons the idea that the question “is it severable?” should be confined to discrete enactments. But it is the correct approach, because Congress often intends provisions to work in concert even where it does not enact them as part of the same bill or act. A legislative code is an intricate tapestry, and laws are interwoven in many different ways. They can refer to and rely on one another explicitly, or the legislature can assume the presence of one when enacting another. Consider the Americans with Disabilities Act. Its remedial section simply cross-references Title VII of the Civil Rights Act of 1964 and states that all remedies available under Title VII are also available under the ADA. 17442 U.S.C. § 12117(a) (2012) (“The powers, remedies, and procedures set forth in sections 2000e-4, 2000e-5, 2000e-6, 2000e-8, and 2000e-9 of this title shall be the powers, remedies, and procedures this subchapter provides to the Commission, to the Attorney General, or to any person alleging discrimination on the basis of disability in violation of any provision of this chapter, or regulations promulgated under section 12116 of this title, concerning employment.”). Thus if the Supreme Court were to declare Title VII unconstitutional, the ADA might, under one potential interpretation, no longer provide any remedies. Or consider the Prison Litigation Reform Act 175Prison Litigation Reform Act of 1995, Pub. L. No. 104-131, 110 Stat. 1321, 1321-66 to 132177. and the Antiterrorism and Effective Death Penalty Act. 176Antiterrorism and Effective Death Penalty Act of 1996, Pub. L. No. 104-132, 110 Stat. 1214. These two laws were enacted at the same time to restrict two different types of prisoner litigation—Section 1983 suits and habeas petitions—that often cover similar subject matter. The main provisions of each law could thus logically be understood to assume and rely on one another. If courts limited severability inquiries to formalistic groupings of legislative language such as “bills” or “acts,” they would miss such cross pollination, which is a common feature of modern law.

VI. Legislative Conditionality in Practice

This last Part considers several pragmatic concerns about legislative conditionality. In a sense, one should not care if the theory works in practice, so long as it is the only one consistent with Article III. But if it forces us to accept crazy results, it loses much of its shine. And we are not necessarily stuck with it, even if it is theoretically pristine. Congress could pass a law empowering judges to declare statutes inseverable for policy reasons, 177This might, however, raise nondelegation issues. See supra note 110. or we could embrace one of the other theories despite its problems and muddle through. No system is without flaws, and better to let judges stray from a marked trail than insist they follow it over a cliff. Fortunately, legislative conditionality does not force this choice. It is easily brought into practice, its harms can be mitigated, and its benefits are substantial.

A. Legislative Fixes

The main cost of the conditionality theory is that it gives courts few tools to deal with statutory distortion. Inseverability exists only where a court can credibly interpret a law as making one provision conditional on another. Thus, a court cannot find parts of a law inseverable merely because leaving them in place would have negative consequences or undermine the law’s purpose. This means that the remainders of statutes like the ACA, the Sentencing Reform Act, and the Federal Election Campaign Act must remain in place even though major provisions have been stripped out through constitutional review. The judiciary can take away language to make laws conform to the Constitution, but it cannot take away more language to maintain sensible policy.

That is not the end of the story, however. Congress can amend or repeal a statute that has been partially invalidated. As the Supreme Court noted in Booker, “Ours, of course, is not the last word: The ball now lies in Congress’ court.” 178United States v. Booker, 543 U.S. 220, 222, 265 (2005). Judicial distortion of statutes can be action forcing; it can push Congress to debate and enact a new law that will fix the mess the Court has left behind. 179See J. Mitchell Pickerill, Constitutional Deliberation in Congress: The Impact of Judicial Review in a Separated System 66−67 (2004) (arguing that major Supreme Court cases can prompt and shape debates over new legislation). Judicial severance could thus be seen as part of an ongoing constitutional dialogue between the branches, which generates new legislative ideas that avoid constitutional problems while still achieving Congress’s policy goals. This idea has become influential in Canadian constitutional law, where many scholars view judicial review as an ongoing process of judicial legislative dialogue. 180See, e.g., Peter W. Hogg & Allison A. Bushnell, The Charter Dialogue Between Courts and Legislators (or Perhaps the Charter of Rights Isn’t Such a Bad Thing After All), 35 Osgoode Hall L.J. 75, 79–80 (1997). There is empirical support for such a pattern: recent work by Matthew Christiansen and William Eskridge has shown that Congress sometimes overrides the Supreme Court’s statutory interpretation opinions. From 1967 to 2011, 275 Supreme Court opinions have been overridden by Congress. 181See Matthew R. Christiansen & William N. Eskridge, Jr., Congressional Overrides of Supreme Court Statutory Interpretation Decisions, 1967–2011, 92 Tex. L. Rev. 1317, 1328–29 (2014). These overrides peaked in the 1990s, fell off precipitously after the Clinton impeachment, and are (or so Christiansen and Eskridge predict) likely to make a comeback in the coming years. 182See id. at 1325 n.31. While the Christiansen and Eskridge study does not consider overrides in constitutional review cases, it does show that a significant number of overrides occur when Congress believes that a Supreme Court decision has undermined a statute’s purpose. 183See id. at 1371–73. This type of override has continued even in the relative drought of the 2000s. See id. at 1375 (“[T]he recent era of fewer overrides still has managed to adopt a good many restorative overrides.”). This evidence suggests that overrides will also sometimes occur after constitutional review cases that result in distortion of a statute.

When the Court strikes down part of a law, Congress is capable of repealing or amending the remainder, or of choosing to leave the partially invalid law in place. More expansive approaches to severability would ignore the possibility of a legislative fix and give the judiciary the first crack at revising statutes. But there is no reason for judges to be doing Congress’s job when Congress is able. For example, in NFIB, if the Court had struck down the individual mandate but left the rest of the ACA in place, it is difficult to imagine that Congress would have let insurance costs and premiums rise precipitously, especially given how powerful the insurance lobby is. Congress might have enacted insurance subsidies, repealed further provisions of the ACA, or crafted some other fix, possibly after delaying the law’s implementation. 184See Transcript, supra note 15, at 73 (“JUSTICE SCALIA: . . . You can’t repeal the rest of the Act because you’re not going to get 60 votes in the Senate to repeal the rest. It’s not a matter of enacting a new Act. You got to get 60 votes to repeal it. So, the rest of the Act is going to be the law. So, you’re just put to the choice of, I guess, bankrupting insurance companies and the whole system comes tumbling down, or else enacting a Federal subsidy program to the insurance companies, which is what the insurance companies would like, I’m sure.”); Adam D. Chandler & Luke Norris, How Conservatives Could Revive the Public Option, Slate (Jan. 7, 2011, 1:26 PM), http://www.slate.com/articles/news_and_politics/jurisprudence/2011/01/how_conservatives_could_revive_the_public_option.html. Legislative responses of that sort are a much better solution to statutory distortion, because they give the problem of fixing the statute to the branch with the most democratic legitimacy and expertise in lawmaking.

Overrides are not the only tool available to Congress—it can also decide severability questions ex ante through explicit inseverability clauses. And the conditionality theory gives Congress a greater incentive to make use of such clauses, because it renders the judiciary’s severability decisions much more predictable. Under the status quo doctrine, the legislature can anticipate that if part of a statute is rendered unconstitutional, the courts will make essentially a policy determination about whether the remainder can be severed. But legislative conditionality prevents such judicial policymaking—if Congress wants one part of a law to stand or fall with another, it must say so explicitly. In this sense, legislative conditionality works as a preference-eliciting punitive default rule. 185See Ian Ayres & Robert Gertner, Filling Gaps in Incomplete Contracts: An Economic Theory of Default Rules, 99 Yale L.J. 87, 89, 115 (1989); Einer Elhauge, Preference-Estimating Statutory Default Rules, 102 Colum. L. Rev. 2027 (2002). It establishes a general default that statutory language is severable except in a few narrow circumstances, and forces Congress to reveal any legislative preference it has for inseverability by writing an explicit clause. Thus, if Congress had wished to avoid the problems that would have attended striking down the individual mandate in the ACA, it would have known to make the law’s provisions textually inseverable from each other and not to rely on the courts to solve the problem (as the Obama administration ultimately did by arguing that the law was partially inseverable). 186See Brief for Respondents (Severability), supra note 70. The reverse approach of striking down the entire bill if part of it is unconstitutional, unless there is a severability clause, would also work as a preference eliciting default rule. This strategy would force Congress to specify when it wants a statute made severable, or else lose everything in a bill. However, this is a less effective punitive default because it is easily reversed by Congress through the use of pro forma severability clauses (indeed, that is Congress’s practice now). The conditionality approach, by contrast, compels Congress to specify which clauses it wants made conditional on which other clauses. Presumably, Congress would not start putting pro forma inseverability clauses in every bill. This approach is only possible where Congress successfully predicts a constitutional challenge, but that sometimes happens. For instance, during the debate over the McCain Feingold campaign finance reform bill, 187Bipartisan Campaign Reform Act of 2002, Pub. L. No. 107-155, 116 Stat. 81. Congress considered and rejected a clause that would have made the law’s ban on soft money inseverable from its restrictions on issue ads. 188See Excerpts from Senate Debate on Donations: Skirmishing and Predictions, N.Y. Times, Mar. 30, 2001, at A16, available at http://www.nytimes.com/2001/03/30/us/excerpts-from-senate-debate-on-donations-skirmishing-and-predictions.html [hereinafter Excerpts] (“The Senate is being asked to pass an amendment that would make two provisions of this bill non-severable, one from another.”). And Congress has enacted inseverability clauses in the past. 189See, e.g., 4 U.S.C. § 125 (2012) (“If a court of competent jurisdiction enters a final judgment on the merits that . . . substantially limits or impairs the essential elements of sections 116 through 126 of this title, then sections 116 through 126 of this title are invalid and have no legal effect as of the date of entry of such judgment.”); 25 U.S.C. § 1760 (2012) (“In the event that any provision of section 1753 of this title is held invalid, it is the intent of Congress that the entire subchapter be invalidated.”). The conditionality theory could prompt more such debates, causing the legislative branch to plan for the contingencies of judicial review. 190Presently inseverability clauses are fairly rare, occurring at a rate of roughly one per year as of 2001. See Excerpts, supra note 188 (“[D]uring the last 12 years only 12 bills have been introduced, let alone passed, that contain a nonseverability clause.”).

Now consider the reverse problem. When the Supreme Court finds a law inseverable, not only does it leave Congress’s goals entirely unvindicated, it also renders futile all of the political and deliberative processes that led to the law’s enactment. At least a partial invalidation leaves the law’s supporters with something to build on. Congress cannot easily reenact entire statutes like the ACA, the Sentencing Reform Act, and Federal Election Campaign Act. These were major transformative laws that were the subjects of lengthy and intense debate, were passed in response to specific historical catalysts, and were enacted by political coalitions that are no longer in power. To undo them is essentially to rewind politics, and in some cases to erase the results of one or more national elections. 191Severability also affects less significant statutes, of course, but these are less important when we are considering the practical effects of legislative conditionality. Take the ACA as an example. The ACA could only be enacted after the Democratic Party won a presidential election and, through several successive congressional elections, obtained a majority in the House of Representatives and a sixty-vote supermajority in the Senate. The Democrats spent a year passing various versions of the ACA through each chamber of Congress, encountering intense opposition from the Republican Party. 192See Emily Smith, Timeline of the Health Care Law, CNN (June 28, 2012, 10:24 AM ET), http://www.cnn.com/2012/06/28/politics/supreme-court-health-timeline; see also Carl Hulse & Jeff Zeleny, Democrats Seem Set to Go it Alone on a Health Bill, N.Y. Times, Aug. 19, 2009, at A1, available at http://www.nytimes.com/2009/08/19/health/policy/19repubs.html. The enactment of this law dominated national politics for basically an entire term of Congress and the first two years of a new presidency. And by the time NFIB was decided in 2012, there was no way Congress was going to enact the ACA again—Republicans had taken control of the House of Representatives, and America was in the midst of a presidential election. Thus if the Supreme Court had struck down the entire ACA because part of it was unconstitutional, it would not just have been deciding that a partial law was worse than no law at all. It would have been undoing all the deliberative and political events that formed that law’s history.

This is not respectful of the democratic process. When the Court strikes down part of a law but leaves the rest intact, it may badly distort how the law operates. But it will at least avoid a situation where the law’s entire deliberative and political history comes to nothing. The political coalition that supports the original law will still have parts to rebuild with, broken though they may be. And if a rebuilding operation proves impossible, Congress can repeal the law. But finding a statute inseverable so as to create a clean slate usurps the power of the enacting legislature, especially when a court bases such a finding on policy arguments. It is better to let the elected branches muddle through than to have courts fully erase the fruits of our national debate.

In sum, the conditionality theory has one major cost: it prevents the courts from avoiding statutory distortion because it denies the judiciary the power to declare laws inseverable for policy reasons. This cost is mitigated by legislative overrides, and by the fact that the conditionality theory gives legislators an incentive to use inseverability clauses. The conditionality theory also carries a significant advantage. It prevents the judiciary from reversing national debates over policy matters by invalidating entire statutes. Because it usually forces the judiciary to leave the remainder of a statute in place, the conditionality theory allows the outcome of a prior debate to frame the national discussion going forward. Rather than rewinding politics, it lets politics play on.

B. State Statutes and State Courts

So far this Article has focused primarily on federal statutes, but state statutes also give rise to severability questions. This section considers how both federal and state courts should apply the conditionality theory of severability when they consider state statutes.

When a federal court finds part of a state statute unconstitutional, any severability issues must be decided under state law. Thus the conditionality theory must yield to state interpretive rules. This follows from the logic of Erie 193Erie R.R. Co. v. Tompkins, 304 U.S. 64 (1938). —state laws are governed by state principles of statutory interpretation, not by federal principles. 194See Michael C. Dorf, The Heterogeneity of Rights, 6 Legal Theory 269, 290 (2000) (“[S]everability is in turn a question of statutory construction, and in a challenge to a state law, state rather than federal principles of statutory construction govern.”); Abbe R. Gluck, Intersystemic Statutory Interpretation: Methodology As “Law” and the Erie Doctrine, 120 Yale L.J. 1898, 1990–91 (2011). And all of the states have adopted their own severability formulas either through statutes, judicial rulings, or both. 195Dorf, supra note 11, at 295 (“Every state except Tennessee and Virginia employs a presumption of severability, unless the statute in question includes a nonseverability clause. In general, state courts hold that provisions and applications of statutes are severable.”); Gluck, supra note 194, at 1950 (“[W]ith respect to avoidance’s cousin, severability, forty-eight states have a statute, a judicial decision, or both requiring courts to apply a presumption of severability.”); see also Dorf, supra note 11, at 295–304 (surveying the severability laws of all fifty states). Some states, like Minnesota, employ principles that closely resemble federal law, namely the intent test and the hypothetical passage test. 196Minn. Stat. Ann. § 645.20 (West 1947). Other states, like Wisconsin, flatly declare everything severable without exception. 197Wis. Stat. Ann. § 990.001(11) (West 2007). And two states—Virginia and Tennessee—adopt presumptions against severability. 198See Vollmer v. City of Memphis, 730 S.W.2d 619 (Tenn. 1987); Bd. of Supervisors v. Rowe, 216 S.E.2d 199 (Va. 1975); see also Dorf, supra note 11, at 301–03. A state’s laws are enacted in the context of its established severability rules. Thus, when a federal court strikes down part of a state law, it should either not address inseverability at all, or it should explicitly leave the question to the state courts. This is generally the approach that the Supreme Court has taken, with a few exceptions. 199See supra note 35 (noting that the Supreme Court has declared state statutes inseverable in only six cases, by my count, since 1940); see also Scoville, supra note 34, at 564–69 (surveying severability decisions after Erie and concluding that “the Court generally settled upon the rule that the sovereign whose statute is at issue dictates the severance test”). It has explicitly recognized that “[s]everability is of course a matter of state law.” 200Leavitt v. Jane L., 518 U.S. 137, 139 (1996) (per curiam). And it has deferred to state courts even in the most obvious cases. For instance, in Zobel v. Williams the Court determined that an Alaska statute was partly unconstitutional, and even though the statute contained an explicit inseverability clause, the Court left it to the Alaska courts to decide whether other parts of the statute were inseverable. 201457 U.S. 55, 65 (1982).

If a federal court does succumb to the urge to find a state statute inseverable, it can only do so by applying state severability rules. But this possibility raises another issue. Most state systems have severability doctrines that are similar to the current federal doctrine, letting state courts strike down additional language under the intent test and the hypothetical passage test. 202See Dorf, supra note 11, at 295. But a federal court cannot apply such tests because it is constrained by Article III. Therefore, even if a state court can legitimately exercise its remedial power to find a statute inseverable for policy reasons, a federal court cannot do the same. The federal court would not merely be interpreting state law in such a case. It would be usurping a state court’s remedial power to delete parts of statutes. And as shown above, a federal court cannot exercise such power consistent with Article III. Thus, if a federal court is to declare a state statute inseverable, it can only do so in circumstances where the state law is inseverable due to state principles of statutory construction: that is, where the legislature has made the statute inseverable, either explicitly or implicitly, and the court is merely recognizing that fact through judicial interpretation. It cannot do so when a state’s severability rules call on state judges to exercise their remedial power to strike down further statutory provisions for policy reasons.

A further question is whether state courts should also adopt the conditionality theory. This will depend on the features of a particular state’s legal system. Some states establish their severability rules by statute. For example the legislature of Kentucky has enacted a law instructing judges to apply both the hypothetical passage test and the intent test in severability decisions (essentially the same as in federal doctrine), 203Ky. Rev. Stat. Ann. § 446.090 (West 2006) (“It shall be considered that it is the intent of the General Assembly, in enacting any statute, that if any part of the statute be held unconstitutional the remaining parts shall remain in force, unless the statute provides otherwise, or unless the remaining parts are so essentially and inseparably connected with and dependent upon the unconstitutional part that it is apparent that the General Assembly would not have enacted the remaining parts without the unconstitutional part, or unless the remaining parts, standing alone, are incomplete and incapable of being executed in accordance with the intent of the General Assembly.”). while the legislature of Iowa has enacted a law that only permits judges to find a provision inseverable if it “cannot be given effect” without the invalid provision (similar to the approach advocated in this Article). 204Iowa Code Ann. § 4.12 (West 2008) (“If any provision of an Act or statute or the application thereof to any person or circumstance is held invalid, the invalidity does not affect other provisions or applications of the Act or statute which can be given effect without the invalid provision or application, and to this end the provisions of the Act or statute are severable.”). In states like Kentucky and Iowa that have codified severability rules, those rules serve as a delegation of remedial power—the legislature has empowered judges to declare provisions inseverable and instructed them how and when to do so. But in the absence of an explicit statute, judges must interpret the constitutional system of the state to determine how broad their severability power is. The bulk of this Article has been devoted to showing that the general plan of the federal judiciary—the limits imposed by Article III, the political question doctrine, the narrowness of federal common law, etc.—is inconsistent with a broad power to find statutes inseverable. But these structural arguments have somewhat less force in most state systems. State judges have general jurisdiction, and have much broader authority to create common law than do federal courts. State judges are usually elected, which diminishes the countermajoritarian problem when they engage in political decisionmaking. 205Cf. David E. Pozen, Judicial Elections as Popular Constitutionalism, 110 Colum. L. Rev. 2047 (2010) (considering the impact of judicial elections on the legitimacy of constitutional change). And many states permit third-party lawsuits and advisory opinions because they lack a jurisdictional “case or controversy” requirement. These features of state systems make them seem broadly more compatible with the remedial or hypothetical intent theories of severability, which permit judges to strike down additional portions of a partially invalid statute in order to make it more workable or to preserve the legislature’s intentions.

C. Does Legislative Conditionality Fit the Most Recent Cases?

A final issue is whether the conditionality theory requires a sharp break from the Supreme Court’s current severability jurisprudence. Doctrinal consistency is a special concern in the severability context, since Congress presumably relies on the judiciary’s prior severability practice when drafting new legislation. One might argue that courts should not change the way they decide severability cases, since Congress enacts laws with the existing severability rules in mind. The conditionality theory presented in this Article would, admittedly, require changing the official doctrine substantially. The Court would have to jettison the intent test and the hypothetical passage test, keep the independence test, and add tests for the other two kinds of implied conditionality (a “textual incoherence” test and a “no independent purpose” test). But legislative conditionality fits remarkably well with the Supreme Court’s actual severability decisions, given that the Court has so rarely found a statute inseverable. Since 1940, the Court has only declared part of a federal statute inseverable in one case, Booker, and has declared part of an executive order inseverable in another case, Mille Lacs (it has held a handful of state statutes inseverable, but as previously noted, these cases are governed by state severability rules). 206See supra note 35; see also supra Part VI.B. Since legislative conditionality would compel judges to find statutes severable in all but a narrow set of cases, this modern tendency to make everything severable fits the theory well. Thus we should not expect too much disruption if the Supreme Court were to embrace it.

Legislative conditionality also comports with the Court’s emphasis on legislative intent. In Ayotte v. Planned Parenthood of Northern New England, the Court announced three principles that would govern its severability analysis:

First, we try not to nullify more of a legislature’s work than is necessary, for we know that “[a] ruling of unconstitutionality frustrates the intent of the elected representatives of the people.” . . . Second, mindful that our constitutional mandate and institutional competence are limited, we restrain ourselves from “rewrit[ing] state law to conform it to constitutional requirements” even as we strive to salvage it. . . . Third, the touchstone for any decision about remedy is legislative intent, for a court cannot “use its remedial powers to circumvent the intent of the legislature.” 207546 U.S. 320, 329–30 (2006) (alterations in original) (quoting, in turn, Regan v. Time, Inc., 468 U.S. 641, 652 (1984) (plurality opinion), Virginia v. American Booksellers Association, 484 U.S. 383, 397 (1988), and Califano v. Westcott, 443 U.S. 76, 94 (1979) (Powell, J., concurring in part and dissenting in part)).

The conditionality theory follows these three principles better than any other. It prevents a court from nullifying more of a law than necessary, and it restrains the court from rewriting laws. It also puts the court’s focus exclusively on interpreting the legislature’s intent—not a hypothetical intent gleaned from the legislature’s general policy ideas, but a specific intent concerning what would happen if part of the statute were struck down. The conditionality theory thus seems to fit the thrust and emphasis of current law quite well. All that remains is to consider whether it is consistent with the two cases in which federal laws were found inseverable—Mille Lacs and Booker.

In Mille Lacs, the Chippewa sued for the right to hunt and fish without state regulation on certain lands in Minnesota. At issue was the legality of an 1850 executive order revoking those rights from the Chippewa. The order contained two relevant provisions, the first stating that “The privileges granted temporarily to the Chippewa Indians . . . ‘of hunting, fishing and gathering the wild rice, upon the lands, the rivers and the lakes included in the territory ceded’ . . . are hereby revoked” and the second that “all of the said Indians remaining on the lands ceded as aforesaid, are required to remove to their unceded lands.” 208Minnesota v. Mille Lacs Band of Chippewa Indians, 526 U.S. 172, 179 (1999) (some internal quotation marks omitted). The Court determined that the removal provision was unlawful because it was not authorized by any federal law and that the provision revoking hunting and fishing rights was inseverable from the removal provision. Can the conditionality theory support this decision? The order contained no explicit inseverability clause, and the revocation of fishing rights remained textually coherent, was capable of being enforced, and had a plausible independent purpose (preventing hunting and fishing) without the removal provision. Thus, if the Court’s finding of inseverability is to be justified, it must be because the legislative history of the order shows that the President intended that its parts be mutually conditional. The majority argued precisely this, claiming that “[w]e think it is clear that President Taylor intended the 1850 order to stand or fall as a whole.” 209Id. at 191. However its evidence from the legislative history is fairly weak—it only cites the fact that during enforcement of the order, the Governor of Minnesota and the Minnesota legislature “explicitly tied revocation of the treaty privileges to removal.” 210Id. at 193. In normal circumstances, this would not be sufficient to demonstrate that one part of the order was meant to be conditional on another. 211See id. at 215–16 (Rehnquist, C.J., dissenting) (“[T]he order’s termination of the treaty privileges should be sustained unless the Chippewa are able to clearly demonstrate that President Taylor would not have terminated them without a removal order. But there is no such evidence . . . .”). However, the decision might still be justified by the longstanding doctrine that statutes will be liberally construed to favor the interests of Native American tribes. 212See Montana v. Blackfeet Tribe of Indians, 471 U.S. 759, 766–68 (1985). And even if the result in Mille Lacs cannot be defended under the conditionality theory, the Court at least framed the inquiry properly—the question was whether President Taylor intended the two parts of the order to stand or fall together, that is, whether the President meant for one part to actually be conditional on the other, not merely whether the President thought that one part was a bad policy without the other.

In Booker, the Court struck down Section 3553(b)(1) of the Sentencing Reform Act, which had made the Guidelines mandatory for federal judges. The Court also invalidated Section 3742(e) of the SRA, finding that it was inseverable from Section 3553(b)(1) because it “depends on the Guidelines’ mandatory nature.” 213United States v. Booker, 543 U.S. 220, 245 (2005). Section 3742(e) established de novo appellate review for departures from the Guidelines. It is easy to see why this section would be deemed inseverable under the conditionality theory: Section 3742(e) cannot be enforced without Section 3553(b)(1). First, Section 3742(e) explicitly cross references Section 3553(b) as a basis for invalidating a statute on appeal. 21418 U.S.C. § 3742(e)(3)(B)(ii) (2012), invalidated by Booker, 543 U.S. 220. If Section 3553(b) is struck down, then, that part of Section 3742(e) cannot be enforced. Second, and more broadly, if the Guidelines are not mandatory, then it is not logically possible to provide de novo review for a lower court’s failure to follow them. The Court was thus fully justified in striking down the parts of Section 3742(e) that concerned appellate review of departures from the Guidelines. However, it was not justified in striking down all of Section 3742(e), which contained several clauses that could still be given full effect even if the Guidelines were advisory. The provision also provided for de novo review and reversal if a sentence “was imposed in violation of law,” “was imposed as a result of an incorrect application of the sentencing guidelines,” was based on a departure factor that is “not justified by the facts of the case,” or if the district court “failed to provide the written statement of reasons required by section 3553(c).” All of these clauses are perfectly valid even if Section 3553(b) is struck down and the Guidelines made fully nonmandatory. The Court’s severability holding in Booker is thus only partially supportable under the conditionality theory.

The holding in Mille Lacs is probably inconsistent with the conditionality theory, and the holding in Booker is partly justifiable but goes too far. So if the Court were to embrace legislative conditionality, there would be a small cost to the principle of stare decisis. But this theory is nonetheless more consistent with modern severability practice than is the current doctrine. While federal courts are able to use very broad tests to find a law inseverable—looking at whether the severed statute would hypothetically have been passed by Congress, or whether it is consistent with Congress’s broad intentions—they have done so in only two cases, one of which involved an executive order. And even in those cases they have approached severability questions with a scalpel, severing single provisions rather than destroying the entire law (the dissent in NFIB notwithstanding). This rarity of inseverability findings helps to answer the reliance objection. Findings of inseverability are like lightning bolts: they happen so uncommonly that they cannot really be prepared for. Thus Congress cannot meaningfully be said to rely on current doctrine in enacting its laws, and indeed the conditionality theory fits the courts’ practice better than current doctrine. More than any other theory, legislative conditionality explains and justifies this modern approach. It accounts for why courts should sever in the overwhelming majority of cases and should fail to sever only in a few narrow and easily identified circumstances.

Conclusion

This Article has sought to show that inseverability only exists where the legislature instructs the courts that one part of a statute must stand or fall alongside another. This happens most obviously where the legislature writes an explicit inseverability clause, or makes it clear in the legislative history that two provisions are conditional on one another. But a legislature can also make statutes conditional implicitly by choosing to write them a particular way. If severing part of the statute leaves another part textually incoherent, incapable of being given effect, or utterly pointless, the two are implicitly inseverable. But, contrary to current doctrine, a court cannot find a provision inseverable simply because invalidating it would advance the legislature’s policy goals. The Article has defended this theory, showing that its alternatives are unsound and its consequences sanguine. Along the way, it has also shown that a party must have standing to challenge a provision for that provision to be found inseverable, and that any two provisions in the entire code can be inseverable from each other (meaning a court is not limited to provisions that share the same bill, act, or title). Finally, and perhaps most importantly, this Article has connected the severability issue to deeper questions about the nature of judges’ power in our constitutional system. It has broadened the inquiry from “what should severability doctrine look like?” to “what gives judges the power to make things inseverable in the first place?” Hopefully, in the process, it has brought some clarity to a very confusing legal issue.

Footnotes

Ph.D. candidate in Law, Yale Law School. For helpful feedback on prior drafts I would like to thank Kiel Brennan-Marquez, William Eskridge, Heather Gerken, Emily Gerrick, Abbe Gluck, Daniel Hemel, Mike Knobler, Lisa Ouellette, Alexander Schwab, Gordon Silverstein, Kate Stith, Maggie Wittlin, my Ph.D. classmates, and the participants in the Yale Doctoral Colloquium. I would also like to thank the editors of the Emory Law Journal for their excellent editing.

1132 S. Ct. 2566 (2012).

2Id. at 2668 (Scalia, Kennedy, Thomas, & Alito, JJ., dissenting) (“In our view, both these central provisions of the Act—the Individual Mandate and Medicaid Expansion—are invalid. It follows, as some of the parties urge, that all other provisions of the Act must fall as well.”).

3See id. at 2675–76.

4See id. at 2671.

5See Alaska Airlines, Inc. v. Brock, 480 U.S. 678, 685 (1987) (“The more relevant inquiry in evaluating severability is whether the statute will function in a manner consistent with the intent of Congress.”); see also NFIB, 132 S. Ct. at 2668 (Scalia, Kennedy, Thomas, & Alito, JJ., dissenting) (“First, if the Court holds a statutory provision unconstitutional, it then determines whether the now truncated statute will operate in the manner Congress intended.”); United States v. Booker, 543 U.S. 220, 247 (2005) (“Hence we must decide whether we would deviate less radically from Congress’ intended system (1) by superimposing the constitutional requirement announced today or (2) through elimination of some provisions of the statute.”).

6See Champlin Ref. Co. v. Corp. Comm’n, 286 U.S. 210, 234 (1932) (“Unless it is evident that the legislature would not have enacted those provisions which are within its power, independently of that which is not, the invalid part may be dropped if what is left is fully operative as a law.”); see also Randall v. Sorrell, 548 U.S. 230, 262 (2006) (“Given these difficulties, we believe the Vermont Legislature would have intended us to set aside the statute’s contribution limits, leaving the legislature free to rewrite those provisions in light of the constitutional difficulties we have identified.”); Denver Area Ed. Telecomms. Consortium, Inc. v. FCC, 518 U.S. 727, 767 (1996) (plurality opinion) (“The question is one of legislative intent: Would Congress still ‘have passed’ § 10(a) ‘had it known’ that the ‘remaining provision[s were] invalid’? If so, we need not invalidate all three provisions.” (alteration in original) (quoting Brockett v. Spokane Arcades, Inc., 472 U.S. 491, 506 (1985))); accord NFIB, 132 S. Ct. at 2607; Ayotte v. Planned Parenthood of N. New Eng., 546 U.S. 320, 330 (2006); Minnesota v. Mille Lacs Band of Chippewa Indians, 526 U.S. 172, 191 (1999); Regan v. Time, Inc., 468 U.S. 641, 653 (1984) (plurality opinion); Buckley v. Valeo, 424 U.S. 1, 108 (1976) (per curiam); United States v. Jackson, 390 U.S. 570, 585 (1968).

7See Alaska Airlines, 480 U.S. at 686; Brockett, 472 U.S. at 506; INS v. Chadha, 462 U.S. 919, 931–34 (1983) (“Here, however, we need not embark on that elusive inquiry since Congress itself has provided the answer to the question of severability in § 406 of the Immigration and Nationality Act . . . .”); Champlin Ref. Co., 286 U.S. at 235; see also NFIB, 132 S. Ct. at 2607 (“The chapter of the United States Code that contains § 1396c includes a severability clause confirming that we need go no further.”).

8United States v. Reese, 92 U.S. 214, 220–21 (1876); see also Free Enter. Fund v. Pub. Co. Accounting Oversight Bd., 561 U.S. 477, 509 (2010) (noting that Sarbanes-Oxley “remains fully operative as law” and that “[t]he remaining provisions are not incapable of functioning independently” (quoting, in turn, New York v. United States, 505 U.S. 144, 186 (1992) and Alaska Airlines, 480 U.S. at 684) (internal quotation marks omitted)); Hill v. Wallace, 259 U.S. 44, 70 (1922) (“Section 4 with its penalty to secure compliance with the regulations of Boards of Trade is so interwoven with those regulations that they cannot be separated. None of them can stand.”).

9Robert L. Stern, Separability and Separability Clauses in the Supreme Court, 51 Harv. L. Rev. 76, 111 (1937).

10See Mark L. Movsesian, Severability in Statutes and Contracts, 30 Ga. L. Rev. 41, 79–80 (1995) (arguing for a default presumption of severability); John Copeland Nagle, Severability, 72 N.C. L. Rev. 203, 206 (1993) (same); Michael D. Shumsky, Severability, Inseverability, and the Rule of Law, 41 Harv. J. on Legis. 227, 272–78 (2004) (same); Laurence H. Tribe, The Legislative Veto Decision: A Law by Any Other Name?, 21 Harv. J. on Legis. 1, 21–27 (1984) (arguing that every unconstitutional provision should be held severable unless Congress has provided an explicit inseverability clause); Kevin C. Walsh, Partial Unconstitutionality, 85 N.Y.U. L. Rev. 738, 777–89 (2010) (same); see also Tobias A. Dorsey, Remarks, Sense and Severability, 46 U. Rich. L. Rev. 877, 891–92 (2012) (arguing that courts should always find unconstitutional provisions severable, without exception); Rachel J. Ezzell, Note, Statutory Interdependence in Severability Analysis, 111 Mich. L. Rev. 1481, 1481 (2013) (arguing for a “qualified clear statement rule,” where statutes would be severable unless (1) there is an inseverability clause or (2) the severed law could not logically be enforced).

11Michael C. Dorf, Facial Challenges to State and Federal Statutes, 46 Stan. L. Rev. 235 (1994) (arguing for a default presumption of inseverability); David H. Gans, Severability as Judicial Lawmaking, 76 Geo. Wash. L. Rev. 639 (2008) (advocating inseverability if severing a provision would substantially change the statute).

12See, e.g., Israel E. Friedman, Comment, Inseverability Clauses in Statutes, 64 U. Chi. L. Rev. 903, 904 (1997) (arguing that inseverability clauses should always be respected, but that severability clauses can sometimes be ignored); see also Michael C. Dorf, Fallback Law, 107 Colum. L. Rev. 303, 339–42 (2007) (defining limits to fallback law, including severability and inseverability clauses, namely that they cannot be impermissibly coercive to courts); Dorsey, supra note 10, at 892 (arguing that inseverability clauses are nullities that cannot be enforced, and that severability clauses are redundant); Fred Kameny, Are Inseverability Clauses Constitutional?, 68 Alb. L. Rev. 997, 1001 (2005) (arguing that inseverability clauses are unconstitutional when they attempt to coerce the judiciary or are intended as a weapon to eliminate a law); Movsesian, supra note 10, at 73–79 (arguing for a purely textualist approach to severability, including strict adherence to severability clauses); Nagle, supra note 10, at 206 (arguing that severability and inseverability clauses should be applied according to their plain meaning).

13Tom Campbell, Severability of Statutes, 62 Hastings L.J. 1495 (2011).

14See, e.g., Social Security Amendments of 1977, Pub. L. No. 95-216, § 334(g)(3), 91 Stat. 1509, 1547 (“If any provision of this subsection . . . is held invalid . . . the application of this subsection to any other persons or circumstances shall also be considered invalid.”).

15For example, Justice Scalia and Paul Clement engaged in the following colloquy:JUSTICE SCALIA: . . . . Why do we look to the—are you sure we look to the intent of the Congress? I thought that, you know, sometimes Congress says that these provisions will—all the provisions of this Act will be severable, and we ignore that when the Act really won’t work, when the remaining provisions just won’t work. Now, how can you square that reality with the proposition that what we’re looking for here is what would this Congress have wanted?MR. CLEMENT: Well, two responses, Justice Scalia: We can look at this Court’s cases on severability, and they all formulate the test a little bit differently.JUSTICE SCALIA: Yes, they sure do.Transcript of Oral Argument, Day 2, at 9, NFIB, 132 S. Ct. 2566 (2012) (Nos. 11-393 & 11-400), available at http://www.supremecourt.gov/oral_arguments/argument_transcripts/11-393.pdf. Similarly, Justice Kennedy asked,If you—suppose you had party A wants proposal number 1; party B wants proposal number 2. Completely unrelated. One is airline rates; the other is milk regulation. And we—and they decide them together. The procedural rules are these have to be voted on as one. They are both passed. Then one is declared unconstitutional. The other can operate completely independently. Now, we know that Congress would not have intended to pass one without the other. Is that the end of it, or is there some different test? Because we don’t want to go into legislative history, that’s intrusive, so we ask whether or not an objective—as an objective rational matter, one could function without—I still don’t know what the test is that we’re supposed to apply. And this is the same question as Justice Scalia asked. Could you give me some help on that?Id. at 17–18.

16NFIB, 132 S. Ct. at 2607 (“The question here is whether Congress would have wanted the rest of the Act to stand, had it known that States would have a genuine choice whether to participate in the new Medicaid expansion.”); id. at 2668–69 (Scalia, Kennedy, Thomas, & Alito, JJ., dissenting) (“First, if the Court holds a statutory provision unconstitutional, it then determines whether the now truncated statute will operate in the manner Congress intended. . . . Second, even if the remaining provisions can operate as Congress designed them to operate, the Court must determine if Congress would have enacted them standing alone and without the unconstitutional portion.”).

17For more on the history of severability doctrine, see Kenneth A. Klukowski, Severability Doctrine: How Much of a Statute Should Federal Courts Invalidate?, 16 Tex. Rev. L. & Pol. 1, 10–30 (2011); Stern, supra note 9, at 79–82.

185 U.S. (1 Cranch) 137, 175–76 (1803).

1927 U.S. (2 Pet.) 492, 526 (1829).

2068 Mass. (2 Gray) 84 (1854).

21Id. at 99.

22See Klukowski, supra note 17, at 8; Stern, supra note 9, at 80.

23103 U.S. 80, 83–84 (1881).

24See Stern, supra note 9, at 81 (“During this early period, the courts were very reluctant to invalidate an entire law on grounds of inseparability. Doubts were resolved in favor of the severance of legislation; indeed, in a number of cases, in both state and federal courts, it was apparently assumed that laws were intended to be sustained to the extent that they could possibly be held valid. Although the language of presumption was not employed, it is clear that the courts at that time presumed that laws were intended to be severable, rather than the contrary.” (footnotes omitted)).

25Allen, 103 U.S. 80; see also Pollock v. Farmers’ Loan & Trust Co., 158 U.S. 601 (1895); Poindexter v. Greenhow, 114 U.S. 270 (1884); Stern, supra note 9, at 107–08 n.138. Note that this list includes only those cases in which the Court held that the unconstitutionality of one provision required striking down one or more entirely separate provisions. It does not include cases where the Court struck down a provision with some constitutional applications and some unconstitutional applications (e.g., a provision regulating both intra- and interstate commerce). See Great N. Ry. Co. v. United States, 208 U.S. 452 (1908). For lists of the latter type of case from this period, see Alfred Hayes, Jr., Partial Unconstitutionality with Special Reference to the Corporation Tax, 11 Colum. L. Rev. 120, 124 n.8 (1911); Stern, supra note 9, at 90 nn.64–65. This list also does not include cases where the Court struck down an entire statute because it contained an unconstitutional exception, because that situation does not create a severability question. Rather, it presents the Court with a choice of whether to remedy the constitutional violation by striking down the rule or the exception to the rule. See infra note 71.

26The seven state law cases are as follows: Williams v. Standard Oil Co. of La., 278 U.S. 235, 241–42 (1929); Lemke v. Farmers Grain Co. of Embden, N.D., 258 U.S. 50, 60 (1922); Looney v. Crane Co., 245 U.S. 178, 190–91 (1917); McFarland v. Am. Sugar Ref. Co., 241 U.S. 79, 87 (1916); Myers v. Anderson, 238 U.S. 368, 370 (1915); Harrison v. St. Louis & S.F. R.R., 232 U.S. 318, 332–34 (1914); Int’l Textbook Co. v. Pigg, 217 U.S. 91, 114 (1910). The five federal law cases are as follows: Carter v. Carter Coal Co., 298 U.S. 238, 316 (1936); R.R. Ret. Bd. v. Alton R.R., 295 U.S. 330, 361–62 (1935); Hill v. Wallace, 259 U.S. 44, 70–71 (1922); Muskrat v. United States, 219 U.S. 346, 362–63 (1911); Weems v. United States, 217 U.S. 349, 381–82 (1910). For more on inseverability of statutes, see Stern, supra note 9, at 107 n.138. The qualifications about severable applications and striking down the exception versus the rule, both mentioned supra note 25 and infra note 71, also apply to this list.

27278 U.S. at 241–42.

28Id. at 242 (“The effect of the statutory declaration is to create in the place of the presumption just stated the opposite one of separability. That is to say, we begin, in the light of the declaration, with the presumption that the legislature intended the act to be divisible; and this presumption must be overcome by considerations which make evident the inseparability of its provisions or the clear probability that the invalid part being eliminated the legislature would not have been satisfied with what remains.”).

29See Carter, 298 U.S. at 312–13; R.R. Ret. Bd., 295 U.S. at 361–62; Hill, 259 U.S. at 71.

30286 U.S. 210, 234 (1932); see also Dorchy v. Kansas, 264 U.S. 286, 290 (1924) (“But a provision, inherently unobjectionable, cannot be deemed separable unless it appears both that, standing alone, legal effect can be given to it and that the legislature intended the provision to stand, in case others included in the act and held bad should fall.”); Stern, supra note 9, at 76 (“The inquisitive legislator seeking light on this problem will find that the Supreme Court, the state courts, and secondary authorities all appear to agree that the invalidity of part of a law or of some of its applications will not affect the remainder (1) if the valid provisions or applications are capable of being given legal effect standing alone, and (2) if the legislature would have intended them to stand with the invalid provisions stricken out.”).

31543 U.S. 220 (2005). I am not counting Northern Pipeline Construction Co. v. Marathon Pipe Line Co., 458 U.S. 50 (1982), as a severability decision, because that case concerned potentially severable applications of the same statutory language (a provision granting bankruptcy courts jurisdiction over cases) rather than the severability of different statutory language.

32526 U.S. 172, 191 (1999).

33304 U.S. 64 (1938).

34See, e.g., Ayotte v. Planned Parenthood of N. New Eng., 546 U.S. 320, 331 (2006); Leavitt v. Jane L., 518 U.S. 137, 139–40 (1996) (per curiam); Hooper v. Bernalillo Cnty. Assessor, 472 U.S. 612, 624 (1985); Zobel v. Williams, 457 U.S. 55, 65 (1982); Harrison v. NAACP, 360 U.S. 167, 178 (1959); Watson v. Buck, 313 U.S. 387, 396 (1941). For a more complete list of such cases, see Ryan Scoville, The New General Common Law of Severability, 91 Tex. L. Rev. 543, 564 n.139 (2013). Professor Scoville argues that Ayotte portends that the Supreme Court will begin applying a federal common law of severability to state statutes. Id. at 569–74. However this is probably an overreading of part of the opinion, which elsewhere explicitly refers to New Hampshire law. See Kevin C. Walsh, There Is No General Common Law of Severability, 91 Tex. L. Rev. See Also 49 (2012).

35After extensive searching I have located only six post-1940 cases in which the Supreme Court declared a state statute inseverable, but it is possible that some have eluded my grasp. See Randall v. Sorrell, 548 U.S. 230, 262 (2006); Thornburgh v. Am. Coll. of Obstetricians & Gynecologists, 476 U.S. 747, 764–65 (1986); City of Akron v. Akron Ctr. for Reprod. Health, Inc., 462 U.S. 416, 445 n.37 (1983); Planned Parenthood of Cent. Mo. v. Danforth, 428 U.S. 52, 83 (1976); Meek v. Pittenger, 421 U.S. 349, 371 n.21 (1975); Sloan v. Lemon, 413 U.S. 825, 833–35 (1973).

36Regan v. Time, Inc., 468 U.S. 641, 653 (1983) (plurality opinion); see also Ayotte, 546 U.S. at 328–29; Alaska Airlines, Inc. v. Brock, 480 U.S. 678, 686 (1987) (“In the absence of a severability clause, however, Congress’ silence is just that—silence—and does not raise a presumption against severability.”).

37See W. David Slawson, Legislative History and the Need to Bring Statutory Interpretation Under the Rule of Law, 44 Stan. L. Rev. 383, 402–03 (1992) (“The United States Code consisted of 27,308 pages in 1988, . . . compared to 9797 pages in 1964, the last year in which it was subjected to major revision before the great expansion of federal law that started in the sixties. . . . Federal statutes have also become more complicated. The number of subjects they treat has increased and so has the technological difficulty of many of the subjects.” (footnote omitted)); Christopher Beam, Paper Weight: The Health Care Bill is More than 1,000 Pages. Is that a Lot?, Slate (Aug. 20, 2009, 6:12 PM), http://www.slate.com/articles/news_and_politics/explainer/2009/08/paper_weight.html (noting several examples of recent laws that were over 1,000 pages, that “the total number of pages of legislation has gone up from slightly more than 2,000 pages in 1948 to more than 7,000 pages in 2006,” and that “[t]he average bill length increased over the same period from 2.5 pages to 15.2 pages”).

38See 132 S. Ct. 2566, 2675 (2012) (Scalia, Kennedy, Thomas, & Alito, JJ., dissenting).

39480 U.S. at 684.

40561 U.S. 477 (2010) (striking down for-cause removal limitations for the Public Company Accounting Oversight Board, and finding them severable them from the rest of Sarbanes-Oxley).

41Cf. Transcript, supra note 15, at 38 (“JUSTICE SCALIA: Mr. Kneedler, what happened to the Eighth Amendment? You really want us to go through these 2,700 pages? (Laughter.) JUSTICE SCALIA: And do you really expect the Court to do that? Or do you expect us to give this function to our law clerks? (Laughter.) JUSTICE SCALIA: Is this not totally unrealistic, that we’re going to go through this enormous bill item by item and decide each one?”).

42480 U.S. at 685.

43Bd. of Trade of Chi. v. Olsen, 262 U.S. 1, 42 (1923); Hill v. Wallace, 259 U.S. 44, 70 (1922).

44See Stern, supra note 9, at 112 (“The two decisions on separability, written within eleven months of each other by the same judge, can be reconciled only on the ground that the Court approved of the substance of one statute and not of the other.”); see also id. at 109–14 (providing this and other examples of the Court making arbitrary, essentially legislative severability decisions in the early part of the twentieth century).

45424 U.S. 1, 108–09 (1976) (per curiam).

46Id. at 109.

47Id. at 96. See generally id. at 90–109 (upholding the public financing provisions against various constitutional challenges).

48United States v. Booker, 543 U.S. 220, 245 (2005).

49Id.

50Id. at 249 (“Several considerations convince us that, were the Court’s constitutional requirement added onto the Sentencing Act as currently written, the requirement would so transform the scheme that Congress created that Congress likely would not have intended the Act as so modified to stand.”); id. at 258 (“Although, as we have explained . . . we believe that Congress would have preferred the total invalidation of the statute to the dissenters’ remedial approach, we nevertheless do not believe that the entire statute must be invalidated.”).

51Id. at 249–58.

52Id. at 265 (“In our view, it is more consistent with Congress’ likely intent in enacting the Sentencing Reform Act (1) to preserve important elements of that system while severing and excising two provisions (§§ 3553(b)(1) and 3742(e)) than (2) to maintain all provisions of the Act and engraft today’s constitutional requirement onto that statutory scheme.”).

53See Walsh, supra note 10, at 752 (“It should come as no surprise that the Court’s conclusion in Booker about what Congress would have wanted lines up closely with what five Justices think Congress should have wanted.”).

54NFIB, 132 S. Ct. 2566, 2607 (2012).

55Id. at 2668 (Scalia, Kennedy, Thomas, & Alito, JJ., dissenting).

56Id. at 2672.

57Id. at 2672–74.

58See, e.g., Transcript, supra note 15, at 48–65.

59NFIB, 132 S. Ct. at 2675 (Scalia, Kennedy, Thomas, & Alito, JJ., dissenting).

60One possible reason for this reversal is that the dissenters could not feasibly have gone through the ACA line by line to determine which of its many minor provisions could be preserved. See supra note 41.

61NFIB, 132 S. Ct. at 2672 (Scalia, Kennedy, Thomas, & Alito, JJ., dissenting) (“The Court has been informed by distinguished economists that the Act’s Individual Mandate and Medicaid Expansion would each increase revenues to the insurance industry by about $350 billion over 10 years; that this combined figure of $700 billion is necessary to offset the approximately $700 billion in new costs to the insurance industry imposed by the Act’s insurance regulations and taxes; and that the new $700–billion burden would otherwise dwarf the industry’s current profit margin.”).

62Alaska Airlines, Inc. v. Brock, 480 U.S. 678 (1987).

63NFIB, 132 S. Ct. at 2671 (Scalia, Kennedy, Thomas, & Alito, JJ., dissenting).

64See id. at 2668 (characterizing the independence test as merely a subsidiary part of the intent test); see also Alaska Airlines, 480 U.S. at 685 (“The independent operation of a statute in the absence of a legislative veto provision thus could be said to indicate little about the intent of Congress regarding severability of the veto.”).

65NFIB, 132 S. Ct. at 2668 (Scalia, Kennedy, Thomas, & Alito, JJ., dissenting).

66Buckley v. Valeo, 424 U.S. 1 (1976) (per curiam).

67This aspect of Buckley has been heavily criticized, including by Chief Justice Burger in his partial concurrence, and is cited as evidence that courts sever too frequently. See id. at 254–55 (Burger, C.J., concurring in part and dissenting in part) (“All candidates can now spend freely; affluent candidates, after today, can spend their own money without limit; yet, contributions for the ordinary candidate are severely restricted in amount—and small contributors are deterred. I cannot believe that Congress would have enacted a statutory scheme containing such incongruous and inequitable provisions.”); see also Transcript, supra note 15, at 84 (“[T]his Court in Buckley created a halfway house, and it took Congress 40 years to try to deal with the situation, when contrary to any time of their intent, they had to try to figure out what are we going to do when we’re stuck with this ban on contributions, but we can’t get at expenditures because the Court told us we couldn’t. . . . Why make them do that in health care?”); Campbell, supra note 13, at 1522 (“What no one proposed was a rule that would allow individuals who were wealthy to be able to spend their own money without limit, while individuals who were not wealthy were restricted, as a practical matter, to spending what they could raise, with limits on how much they could raise from any other person. Yet that is the system we now have, after the Supreme Court struck down the limit on individual expenditure, but left intact the rest of the statute regarding limits on fundraising.”).

68United States v. Booker, 543 U.S. 220 (2005).

69Indeed, Congress had in prior years rejected other versions of the SRA that would have created more flexible sentencing systems. See id. at 293 (Stevens, J., dissenting in part) (“Congress explicitly rejected as a model for reform the various proposals for advisory guidelines that had been introduced in past Congresses.”); Kate Stith & Steve Y. Koh, The Politics of Sentencing Reform: The Legislative History of the Federal Sentencing Guidelines, 28 Wake Forest L. Rev. 223, 238 (1993).

70See 42 U.S.C. § 18091(2)(I) (2012) (“[I]f there were no requirement, many individuals would wait to purchase health insurance until they needed care. By significantly increasing health insurance coverage, the requirement, together with the other provisions of this Act, will minimize this adverse selection and broaden the health insurance risk pool to include healthy individuals, which will lower health insurance premiums.”); Brief for Respondents (Severability) at 13–14, 26, NFIB, 132 S. Ct. 2566 (2012) (Nos. 11-393 & 11-400), 2012 WL 273133.

71One might argue that a worse type of distortion occurs where a legislature enacts a rule with an exception, and the exception is held unconstitutional but the rule remains, so that the rule applies to cases the legislature never extended it to. For instance if the legislature passes a tax on “all persons who are not from Minnesota,” and the exclusion of Minnesotans is found unconstitutional, the legislature must then determine whether strike down the tax or the exception. However this is not, properly conceived, a problem of severability—it is, rather, a choice of how the court exercises its power of constitutional review. The court can remedy the constitutional violation by striking down either the exception or the rule. See Welsh v. United States, 398 U.S. 333, 361 (1970) (Harlan, J., concurring in result) (“Where a statute is defective because of underinclusion there exist two remedial alternatives: a court may either declare it a nullity and order that its benefits not extend to the class that the legislature intended to benefit, or it may extend the coverage of the statute to include those who are aggrieved by exclusion.”).

72See NFIB, 132 S. Ct. at 2671 (Scalia, Thomas, Kennedy, & Alito, JJ., dissenting) (“To be sure, an argument can be made that those portions of the Act that none of the parties has standing to challenge cannot be held nonseverable. The response to this argument is that our cases do not support it.”); see also, e.g., Williams v. Standard Oil Co. of La., 278 U.S. 235, 242–44 (1929) (holding nonseverable statutory provisions that did not burden the parties).

73Dorsey, supra note 10, at 880–81.

74Here I am only discussing the remedy of invalidating words entirely. The courts may also use other constitutional remedies—such as interpreting a statute to avoid a constitutional problem or striking down unconstitutional applications of a statute—that leave the words of a statute intact but change their meaning. Such remedies are not relevant here.

75See, e.g., Gans, supra note 11, at 653 (“For better or worse, courts need to have some power to save partially invalid legislation by severing the invalid portions and thereby reconstructing the statute.”). Here, I am only discussing how to frame the meta-question of courts’ power to sever statutes. It is of course possible that a particular doctrinal test will impose a presumption of severability and then require courts to ask how a provision can be “saved.” But that is distinct from assuming that any time a court engages in judicial review it initially strikes down the entire statute, and that it then “saves” or “reconstructs” parts of the statute by severing the unconstitutional part.

76Campbell, supra note 13, at 1496.

77Id. at 1497–98.

78One might imagine a variant of Campbell’s approach in which bills are treated as inseverable units unless they contain a severability clause, and those that do have such a clause are treated as severable. This would let Congress reverse the default rule of inseverability. The likely effect of such an approach would be that Congress would include pro forma severability clauses in every statute, as it does in many statutes now. See Yule Kim, Cong. Research Serv., No. 97–589, Statutory Interpretation: General Principles and Recent Trends 38 (2008), available at http://fas.org/sgp/crs/misc/97-589.pdf. Thus, the courts would end up treating most laws as severable, and the problems discussed in this Part would arise only if Congress forgets to include a severability clause (or consciously decides not to include one).

79561 U.S. 477 (2010).

80462 U.S. 919, 944 (1983) (“Since 1932, when the first veto provision was enacted into law, 295 congressional veto-type procedures have been inserted in 196 different statutes . . . .” (quoting James Abourezk, The Congressional Veto: A Contemporary Response to Executive Encroachment on Legislative Prerogatives, 52 Ind. L.J. 323, 324 (1977) (internal quotation mark omitted)).

81Act of Oct. 12, 1984, Pub. L. No. 98-473, tit. II, ch. II, 98 Stat. 1837, 1987–2040.

82Dorsey, supra note 10, at 882 (explaining that in addition to the text of the Act reported in the United States Statutes at Large, the Act also enacted dozens of provisions by reference).

83Id. at 883–84 (citing § 101, 98 Stat. at 1837, 1844, 1867–77, 1884, 1904).

84See id. at 881–83.

85See Kim, supra note 78, at 38. Canada, by contrast, treated statutes as inseverable until the 1980s. See Kent Roach, Constitutional Remedies in Canada 14-1 (2014) (“Historically, legislation in Canada that violated the federal division of powers was struck down in its entirety.”). However Canada has in the last few decades begun finding statutes severable, largely due to their increased complexity and the resulting problems from invalidating them in toto.

86In defending his formalism, Campbell draws an analogy to the line-item veto: the President cannot choose which parts of a bill he vetoes, and so the courts cannot choose which parts of a bill they strike down. Campbell, supra note 13, at 1498–99. But this analogy does not hold. The line-item veto is invalid because the Presentment Clause requires that the President accept or reject a bill in toto during its passage, not because bills are inherently indivisible once Congress approves them. See Clinton v. City of New York, 524 U.S. 417, 439–41 (1998). Judges can strike down part of an enacted bill without affecting the rest of it, just as Congress can repeal part of an enacted bill while leaving the rest in place.

87See Felix S. Cohen, Transcendental Nonsense and the Functional Approach, 35 Colum. L. Rev. 809 (1935).

88See Stern, supra note 9, at 110 (“[A]s far back as 1855 it had been wisely recognized that ‘the point is not whether they are contained in the same section, for the distribution into sections is purely artificial; but whether they are essentially and inseparably connected in substance.’” (quoting Commonwealth v. Hitchings, 71 Mass. (5 Gray) 482, 486 (1855))).

89See, e.g., Noble v. Mitchell, 164 U.S. 367, 369, 372 (1896) (treating as binding the Alabama Supreme Court’s severance of the words “association or partnership” from the statutory sentence “[t]he term ‘insurance company,’ as used in this article, includes every company, corporation, association or partnership organized for the purpose of transacting the business of insurance”).

90In some cases, the Supreme Court has struck down a unit smaller than a sentence. See Stern, supra note 9, at 110 n.152.

91See supra note 71.

92Some scholars have argued that declaring a statute partly unconstitutional does not involve any act of excision by the reviewing court, but that the court is merely recognizing that the Constitution automatically “displaced” the problematic language. See Tribe, supra note 10, at 25; Walsh, supra note 10, at 778–84. This theory supposedly shows that there is no judicial power to find a statute inseverable because a judge that strikes down an unconstitutional provision has not actually changed anything: the statute was already partially invalid the moment it was enacted because the unconstitutional part was never real law, and the judge is merely recognizing that fact. This theory fails for two reasons. First, the Constitution does not strike down statutes—judges do. Making a statute consistent with the Constitution inevitably requires judges to exercise discretion when they excise language. Sometimes a judge has to choose between striking down an exception or a rule, or in complex cases like Booker, has to choose between two different excisions that would each solve the constitutional problem. In other kinds of cases, the judge has to choose between striking down the provision altogether or merely interpreting it to exclude an application. And even in cases where it is clear which statutory language must go, courts still exercise other sorts of remedial discretion, such as deciding whether to delay the remedy or to make it retroactive. See, e.g., N. Pipeline Constr. Co. v. Marathon Pipe Line Co., 458 U.S. 50, 88 (1982) (finding federal bankruptcy courts unconstitutional, but staying the judgment). Thus a reviewing court does not just passively recognize that the Constitution “displaced” part of a statute the moment the statute was passed. The court actively edits the statute to craft a solution to the constitutional problem. Second, even if we assume that the Constitution “displaces” the invalid parts of a statute such that they were never valid in the first place, inseverability questions still arise. It does not matter whether the language was removed through automatic “displacement” or through judicially managed excision—the statute is distorted either way. To make the case that there is no power to declare statutes inseverable one must grapple with the theories presented in Parts IV and V of this Article, each of which provides a possible account for why a judge that has found one provision unconstitutional can then find other provisions inseverable. The claim that judicial review is a passive activity does not refute any of these theories—judges could still have a remedial or interpretive power to fix statutory distortion after they recognize a constitutional “displacement.”

93The Supreme Court has recognized this principle. See Regan v. Time, Inc., 468 U.S. 641, 652 (1984) (plurality opinion) (“A ruling of unconstitutionality frustrates the intent of the elected representatives of the people. Therefore, a court should refrain from invalidating more of the statute than is necessary. As this Court has observed, ‘whenever an act of Congress contains unobjectionable provisions separable from those found to be unconstitutional, it is the duty of this court to so declare, and to maintain the act in so far as it is valid.’” (quoting El Paso & Ne. Ry. Co. v. Gutierrez, 215 U.S. 87, 96 (1909))).

94See Dorf, supra note 12, at 370.

9552 U.S.C.A. §§ 10101–10702 (West 2014).

96Voting Rights Act of 1965, Pub. L. No. 89-110, §§ 2, 5, 79 Stat. 437, 437, 439.

97See Voting Rights Act Amendments of 1970, Pub. L. No. 91-285, §§ 5, 6, 84 Stat. 314, 315–19; see also Eric S. Fish, Note, The Twenty-Sixth Amendment Enforcement Power, 121 Yale L.J. 1168, 1183 (2012).

98400 U.S. 112, 117–18, 130–31 (1970).

99Guido Calabresi, A Common Law for the Age of Statutes 2 (1982).

100Calabresi equivocates, however, on the question whether such a power currently exists. See id. at 116–17.

101David Gans has advocated such a theory, arguing that inseverability is a matter of judicial remedial discretion. See Gans, supra note 11, at 656–62.

102See Glenn Chatmas Smith, From Unnecessary Surgery to Plastic Surgery: A New Approach to the Legislative Veto Severability Cases, 24 Harv. J. on Legis. 397, 463–66 (1987) (arguing that courts in severability cases have the power to reformulate the language of partly unconstitutional laws, and not just to strike language down); cf. Weinberger v. Wiesenfeld, 420 U.S. 636 (1975) (exemplifying a case in which the Supreme Court remedied a constitutionally underinclusive statute by effectively adding language to the statute).

103E.g., Steel Co. v. Citizens for a Better Env’t, 523 U.S. 83, 94–95 (1998).

104E.g., Baker v. Carr, 369 U.S. 186, 217 (1962).

105E.g., Lujan v. Defenders of Wildlife, 504 U.S. 555, 559–60 (1992).

106E.g., United States v. Morrison, 529 U.S. 598, 607 (2000); INS v. Chadha, 462 U.S. 919, 944 (1983) (“We begin, of course, with the presumption that the challenged statute is valid.”).

107See Alexander M. Bickel, The Least Dangerous Branch: The Supreme Court at the Bar of Politics 16 (2d ed. 1986).

108Kim, supra note 78, at 38 (“Congress frequently includes a pro forma severability clause in a statute . . . .”).

109For example Kentucky has enacted a statute instructing courts how to resolve severability questions, although that statute frames the issue as one of legislative intent rather than remedial power. See Ky. Rev. Stat. Ann. § 446.090 (West 2006).

110Cf. Clinton v. City of New York, 524 U.S. 417, 448 (1998) (striking down a law empowering the president to issue line-item vetoes); Calabresi, supra note 99, at 114–16 (arguing that his theory of a common law for statutes could be achieved through legislative delegation to courts). It would be especially difficult to exercise this broad remedial power within the constraints of the case and controversy requirement, since the court would need a party with standing to request whatever remedy the court found appropriate.

111In contrast, some scholars call for a “textualist approach” to severability. See, e.g., Movsesian, supra note 10, at 73–82. But textualism cannot provide an account of why some statutory language might be inseverable from other language, aside from the obvious case of an explicit inseverability clause. Thus to take a strict “textualist” approach to severability questions would basically mean finding everything severable. I do not explore a “textualist” theory here because the premises of textualism are consistent with the conditionality theory presented in Part V.

112NFIB, 132 S. Ct. 2566, 2607 (2012) (alteration in original) (quoting United States v. Booker, 543 U.S. 220, 246 (2005)) (internal quotation marks omitted).

113Id. (quoting Ayotte v. Planned Parenthood of N. New Eng., 546 U.S. 320, 330 (2006)).

114See Walsh, supra note 10, at 752–53 (“The hypothetical legislative intent test gets around this absence of any actual legislative intent to discern but does so by posing a question whose answer often calls for rank speculation. The inquiry can perhaps be given some content by reference to legislative purpose, or to the ‘enterprise’ to which the statute belongs. But it can easily deteriorate into a question of what ought to happen, in which case ‘legislative intent’ adds nothing.” (citing and quoting Emily L. Sherwin, Rules and Judicial Review, 6 Legal Theory 299, 305 (2000)) (some internal quotation marks omitted)).

115The argument I am making here is not intended as a general broadside against intentionalist or purposivist methods of statutory interpretation, but only against the use of hypothetical intent and statutory purpose to declare provisions inseverable.

116Transcript, supra note 15, at 51–52.

117See supra note 69.

118See Dorf, supra note 12 (discussing legislatively created fallback law).

119See John F. Manning, Textualism and the Equity of the Statute, 101 Colum. L. Rev. 1, 22 (2001) (“The most important alternative justification for atextual and purposive interpretation relates to an ancient common law doctrine: the equity of the statute. As discussed below, when applied, that doctrine authorized courts to extend a clear statute to reach omitted cases that fell within its ratio or purpose, and conversely, to imply exceptions to such a statute when the text would inflict harsh results that did not serve the statutory purpose.”).

120462 U.S. 919 (1983).

121This would bring us quite close to Calabresi’s theory in A Common Law for the Age of Statutes—courts would be able to invalidate statutes when events have made them anachronistic. See Calabresi, supra note 99.

122See Manning, supra note 119, at 22.

123See, e.g., Movsesian, supra note 10 (noting and arguing against the treatment of statutes as contracts for severability purposes).

124See Restatement (Second) of Contracts § 184 cmt. a (1981).

125See 132 S. Ct. 2566, 2675–76 (2012) (Scalia, Kennedy, Thomas, & Alito, JJ., dissenting) (“Often, 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. . . . Some provisions, such as requiring chain restaurants to display nutritional content, appear likely to operate as Congress intended, but they fail the second test for severability. There is no reason to believe that Congress would have enacted them independently.”).

126It would be absurd to analogize an act to a contract, since acts are modified by different Congresses that have different compositions. For instance, the Voting Rights Act was enacted in 1965 and amended in 1970, 1975, 1982, 1992, and 2006. George P. Laney, Cong. Research Serv., No. 95-896, The Voting Rights Act of 1965, As Amended: Its History and Current Issues (2008), available at http://fpc.state.gov/documents/organization/109556.pdf. The current VRA, which reflects all of these successive amendments, cannot meaningfully be treated as a single legislative agreement. A closer question would arise if two bills—one establishing an “Act” and the second amending the same “Act”—were passed in the same session of Congress, or alternatively if two separate bills were passed simultaneously as part of a single legislative agreement. But this is not a terribly important point to the legislative contract theory, and could be resolved either way. If courts adopted either possible rule (either that different laws enacted in the same Congress can be part of the same agreement, or that they cannot), then legislators would be able to plan their legislative bargains accordingly.

127A potentially more interesting case would arise if the President made a guarantee to a crucial group of legislators that he would enact certain regulations if they voted for a law. For example, during the passage of the ACA a group of pro-life Democrats in the House of Representatives agreed to support the law only if President Obama promised to issue an executive order that would limit the use of federal funds for abortion. See Monica Davey, Under Fire for Abortion Deal, Stupak to Retire, N.Y. Times, Apr. 10, 2010, at A16, available at http://www.nytimes.com/2010/04/10/us/politics/10stupak.html. Given the fact that any important regulations could have been included in the statute itself, it would be sensible to avoid making such negotiated-for regulations part of the judicially preserved legislative contract. Including executive action would blur the distinction between laws and regulations, creating an odd situation where if a court finds a regulation invalid it may also have to invalidate a statute.

128See Movsesian, supra note 10, at 71–72 (noting that it may be difficult to find evidence of a legislative contract in cases where legislators do not express their disagreements or engage in the key negotiations on the record).

129See Dorsey, supra note 10, at 879 (“A statute is not the result of a legislative bargain. A statute is the product of a convergence of microbargains, between and among 100 senators, 435 representatives, and the White House. Tradeoffs are made within a single legislative provision, sure, but tradeoffs are also made across the various legislative provisions in a single bill. And tradeoffs are also made between this legislative provision and that funding provision, between this provision and that decision to do something or to refrain from doing something else, between this bill and that bill, between this bill and that hearing, and between this bill and that nomination. There are people who feel confident they can pore over the statute or the legislative history and identify the legislative bargain. They may also be confident they can pore over a sonar image from the depths of Loch Ness and identify the plesiosaur. I agree there are dark shapes in the water but that’s as far as I will go.”).

130See Chris Frates, Payoffs for States Get Harry Reid to 60 Votes, Politico (Dec. 19, 2009, 7:56 PM EST), http://www.politico.com/news/stories/1209/30815.html.

131Justice Scalia made this precise point at oral argument in Sebelius. See Transcript, supra note 15, at 10 (“JUSTICE SCALIA: All right. The consequence of your proposition, would Congress have enacted it without this provision, okay, that’s the consequence. That would mean that if we struck down nothing in this legislation but the—what is it called—the ‘Corn Husker kickback,’ okay, we find that to violate the constitutional proscription of venality, okay? (Laughter.) JUSTICE SCALIA: When we strike that down, it’s clear that Congress would not have passed it without that. It was—it was the means of getting the last necessary vote in the Senate. And you’re telling us that the whole statute would fall because the Cornhusker kickback is bad. That can’t be right.”).

132See id. at 27 (“CHIEF JUSTICE ROBERTS: The reality of the passage—I mean, this was a piece of legislation which, there was—had to be a concerted effort to gather enough votes so that it could be passed. And I suspect with a lot of these miscellaneous provisions that Justice Breyer was talking about, that was the price of a vote: Put in the Indian health care provision and I will vote for the other 2700 pages. Put in the black lung provision, and I’ll go along with it. That’s why all—many of these provisions, I think, were put in, not because they were unobjectionable. So, presumably, what Congress would have done is they wouldn’t have been able to put together, cobble together the votes to get it through.”).

133It is possible that bargained-for treaties, which in a sense are both laws and contracts, might sometimes be treated as contracts for severability purposes. See generally Curtis J. Mahoney, Note, Treaties as Contracts: Textualism, Contract Theory, and the Interpretation of Treaties, 116 Yale L.J. 824 (2007) (arguing that relational contract theory should govern treaty interpretation).

134See Gillian K. Hadfield, Incomplete Contracts and Statutes, 12 Int’l Rev. L. & Econ. 257, 257 (1992).

135See McNollgast, Positive Canons: The Role of Legislative Bargains in Statutory Interpretation, 80 Geo. L.J. 705 (1992) (arguing for such an approach to statutory interpretation).

136See supra Part IV.B.

137Unfortunately, the analogy to contract law does not provide much help in locating congressional intent. See Hadfield, supra note 134, at 257–58 (“We will be disappointed, therefore, if we look to incomplete contracting principles in hope of avoiding the morass of ‘intent.’ Indeed, if we imported the standard gap filling analysis from contracting into the statutory interpretation arena, we would still run smack into Arrow’s theorem: in the legislative case there is simply no answer to the gap-filling counterfactual, ‘what would the parties have voted for?’”).

138Michael Dorf has argued that such conditional legislation could be unconstitutional if it is used to coerce the courts into not exercising their constitutional review powers. For example, if Congress created a fallback law that eliminated school lunches if the Supreme Court struck down the Defense of Marriage Act, he argues that that would be unconstitutionally coercive. See Dorf, supra note 12, at 327–29, 333–36; see also Kameny, supra note 12; cf. Bruce K. Miller, Constitutional Remedies for Underinclusive Statutes: A Critical Appraisal of Heckler v. Mathews, 20 Harv. C.R.-C.L. L. Rev. 79 (1985). This argument is appealing but ultimately not convincing. Congress has the power to enact punitive legislation if it likes, so long as it stays within the limits set by the Constitution. Congress cannot cut judicial pay or limit judicial tenure, but it can certainly enact bad laws to punish the judiciary, and it can design those laws to take effect only after a triggering event. Nothing in the Constitution protects any of the three branches from bullying. If Congress wants to steal lunch money because the judiciary struck down a law, that is Congress’s prerogative. As a practical matter, though, it seems unlikely that a legislature would actually create such a punitive fallback law. And if it did, it would almost certainly repeal the law once the triggering condition was satisfied in order to avoid an electoral backlash. The public would rightly blame Congress, not the courts, for their kids going hungry. It is also worth noting that policy blackmail is not limited to Congress. See Calabresi, supra note 99, at 269–70 (describing Reid v. Covert, 354 U.S. 1 (1957), as a failed attempt at judicial blackmail of Congress).

139A number of scholars have adopted the view that courts should find everything severable unless there is an explicit inseverability clause. See supra note 10. However none of these have explored the possibility of implicit inseverability.

140Rachel Ezzell has made a similar proposal, arguing that a statute should be held inseverable only (1) when there is an explicit inseverability clause or (2) when the severed law could not logically be enforced. See Ezzell, supra note 10. The legislative conditionality approach would add to these (3) when statutory language is inseverable as a consequence of its syntactic logic and (4) when one provision serves no purpose without another.

141Cf. Mistretta v. United States, 488 U.S. 361, 413 (1989) (Scalia, J., dissenting) (concluding that the Sentencing Commission is unconstitutional on nondelegation grounds).

142See Williams v. Standard Oil Co. of La., 278 U.S. 235 (1929) (striking down a law fixing gas prices and finding inseverable provisions creating a state agency to administer the price fixing).

143Gordon v. NYSE, 422 U.S. 659, 682 (1975) (finding that the Securities Exchange Act impliedly repeals the antitrust laws insofar as they regulate securities transactions).

144Geier v. Am. Honda Motor Co., 529 U.S. 861, 899 (2000) (Stevens, J., dissenting) (quoting Freightliner Corp. v. Myrick, 514 U.S. 280, 287 (1995)) (internal quotation marks omitted).

145See Patient Protection and Affordable Care Act, 42 U.S.C. §§ 18001–18121 (2012).

146See Federal Election Campaign Act of 1971, Pub. L. No. 92-225, 86 Stat. 3.

147See, e.g., Social Security Amendments of 1977, Pub. L. No. 95-216, § 334(g)(3), 91 Stat. 1509, 1547 (“If any provision of this subsection . . . is held invalid . . . the application of this subsection to any other persons or circumstances shall also be considered invalid.”); Zobel v. Williams, 457 U.S. 55, 65 (1982) (reviewing an Alaska statute that provided “If any provision enacted in sec. 2 of this Act . . . is held to be invalid by the final judgment, decision or order of a court of competent jurisdiction, then that provision is nonseverable, and all provisions enacted in sec. 2 of this Act are invalid and of no force or effect” (quoting 1980 Alaska Sess. Laws ch. 21, § 4, at 9) (internal quotation mark omitted)).

148See Dorsey, supra note 10, at 892; see also Kameny, supra note 12, at 999–1000, 1009.

149This principle also helps explain why courts never strike down fragments of words, for example the prefix “fe” in “female.” Words are implicitly indivisible, such that each part of a word is conditional on each other part. See supra Part III (discussing why courts cannot strike down single morphemes); see also Noble v. Mitchell, 164 U.S. 367, 372 (1896).

15028 U.S.C. § 1471(b) (1982), invalidated by N. Pipeline Constr. Co. v. Marathon Pipe Line Co., 458 U.S. 50 (1982).

151See Butts v. Merchs. & Miners Transp. Co., 230 U.S. 126, 138 (1913) (holding that the Civil Rights Act of 1875, which had been invalidated as applied within the states, was also invalid as to United States vessels engaged in coastal trade).

152See generally Dorf, supra note 11, at 251–83 (showing that the Supreme Court often makes statutory applications inseverable from one another, notwithstanding the principle announced in Salerno v. United States, 481 U.S. 739 (1987), that applications are presumptively severable).

153Dorchy v. Kansas, 264 U.S. 286, 290 (1924).

154See Stern, supra note 9, at 76 n.1 (“In theory, a legislative body has power to enact laws or parts of laws which are incapable of being given legal effect. Accordingly, if a legislative body should for some or no reason desire that an ineffectual or meaningless part of a law stand alone, the legislative intention should prevail. But the fact that valid provisions of a statute are incapable of having legal effect by themselves is ordinarily conclusive proof that the legislature did not intend them to stand by themselves. Legislatures do not pass laws merely to increase the size of the statute books. Inasmuch as the same results are reached under either theory, the difference in approach has no substantial significance and will not be considered further here.”).

155When a court strikes down statutory language, it does not actually delete that language from the code. Only the legislature can do that. See, e.g., 18 U.S.C. § 700 (2012) (establishing criminal penalties for defacing the American flag), invalidated by United States v. Eichman, 496 U.S. 310 (1992). When a court strikes a provision down, this is only a declaration that it has no legal effect. Hence there is no difference between a judge striking a provision down and a judge finding that a provision is deadwood. In both cases the judge is declaring that the provision will no longer have any legal effect, but is leaving its language in place. And in both cases, if the provision’s unconstitutionality is cured (say by a constitutional amendment), then the provision becomes valid again even if it is not reenacted.

156See 2 Norman J. Singer, Sutherland Statutes and Statutory Construction § 44:4, at 576 (6th ed. 2001) (“Even where part of an act is independent and valid, other parts which are not themselves substantively invalid but have no separate function to perform independent of the invalid portions of the act are also held invalid.”).

157278 U.S. 235 (1929).

158Id. at 243.

159There may be borderline cases. For example, in Williams v. Standard Oil Co. of Louisiana one might argue that the data-collecting agency also serves an academic research function for professors with an interest in oil prices, even though there is probably little evidence of such a purpose in the statute’s text or legislative history (and it seems like an expensive way to generate data for professors). In deciding whether the data collection provision is conditional on the rest of the statute, judges would have to determine whether legislators actually had that independent purpose in mind when they enacted the statute. Such an inquiry would be similar in structure to a Title VII case, where one party claims that they were fired for discriminatory reasons and the other party must show an independent rationale for the firing. Here one party claims that provision A exists only to assist the (now unconstitutional) provision B, and the other party must show an independent justification for A. This inquiry is much more narrow than the hypothetical passage test, which asks whether the legislature (given its substantive policy preferences) would have enacted B without A.

160Severability clauses are officially understood to create a presumption favoring severability. See Alaska Airlines, Inc. v. Brock, 480 U.S. 678, 686 (1987). However the absence of such a clause does not reverse the presumption favoring severability. See id. Rather, a presumption of severability operates in every case regardless of whether there is a severability clause. See Regan v. Time, Inc., 468 U.S. 641, 653 (1984) (plurality opinion). It is not clear from current doctrine whether the default presumption is as easily overcome as the presumption created by a severability clause.

161See Max Radin, A Short Way with Statutes, 56 Harv. L. Rev. 388, 419 (1942) (“[T]he ‘severability clause’ is generally a formulaic phrase, added almost as a matter of course whenever there is the least question as to the constitutionality of the statute. It is derived from a resentment against the power which the courts of the United States have so long exercised and which the legislature has accepted, although neither fully nor with good grace. The decision to take the severability clause literally, whenever it can be so taken, is an example of comity between coordinate branches of the government, but it is not an example of attempting to discover what was really intended.”).

162See, e.g., Raines v. Byrd, 521 U.S. 811, 818–20 (1997).

163See Davis v. FEC, 554 U.S. 724, 733–34 (2008) (“The fact that Davis has standing to challenge § 319(b) does not necessarily mean that he also has standing to challenge the scheme of contribution limitations that applies when § 319(a) comes into play. ‘[S]tanding is not dispensed in gross.’ Rather, ‘a plaintiff must demonstrate standing for each claim he seeks to press’ and ‘for each form of relief’ that is sought.” (alteration in original) (quoting, in turn, Lewis v. Casey, 518 U.S. 343, 358 n.6 (1996) and DaimlerChrysler Corp. v. Cuno, 547 U.S. 332, 352 (2006)) (some internal quotation marks omitted)).

164Friends of the Earth, Inc. v. Laidlaw Envtl. Servs. (TOC), Inc., 528 U.S. 167, 185 (2000) (“Laidlaw is right to insist that a plaintiff must demonstrate standing separately for each form of relief sought.”). But see Erik R. Zimmerman, Supplemental Standing for Severability, 109 Nw. L. Rev. (forthcoming 2015) (manuscript at 7–9), available at http://ssrn.com/abstract=2428332 (noting this conflict between the Sebelius dissent and the Laidlaw principle that standing must be established for every claim, but arguing that it can be resolved through “supplemental standing”).

165521 U.S. 898 (1997).

166Id. at 933.

167Id. at 935.

168NFIB, 132 S. Ct. 2566, 2671 (2012) (Scalia, Kennedy, Thomas, & Alito, JJ., dissenting) (citing Williams v. Standard Oil Co. of La., 278 U.S. 235, 242–44 (1929)).

169See Brianne J. Gorod, The Collateral Consequences of Ex Post Judicial Review, 88 Wash. L. Rev. 903, 904 (2013).

170462 U.S. 919, 944 (1983).

171Id. at 959.

172The dissent in NFIB also seems to contradict the general principle that a person who is burdened by provision A cannot, except under limited circumstances, sue to invalidate a statute because provision B is unconstitutional and inseverable from provision A. See Broadrick v. Oklahoma, 413 U.S. 601, 610 (1973) (“Embedded in the traditional rules governing constitutional adjudication is the principle that a person to whom a statute may constitutionally be applied will not be heard to challenge that statute on the ground that it may conceivably be applied unconstitutionally to others . . . .”). If severability determinations do not require standing, it is difficult to see why a party could not bring this kind of challenge.

173Another potential solution for complex statutes with many provisions would be to articulate the severability ruling broadly. Imagine that a statute has five provisions: A, B, C, D, and E. A is unconstitutional, and the parties before the court only have standing to challenge A and B. The court might, in some cases, articulate its inseverability decision in a way that clearly applies to C, D, and E as well. Of course it could only do so if all of the provisions were inseverable for the same reason. See Richard H. Fallon, Jr., Commentary, As-Applied and Facial Challenges and Third-Party Standing, 113 Harv. L. Rev. 1321 (2000).

17442 U.S.C. § 12117(a) (2012) (“The powers, remedies, and procedures set forth in sections 2000e-4, 2000e-5, 2000e-6, 2000e-8, and 2000e-9 of this title shall be the powers, remedies, and procedures this subchapter provides to the Commission, to the Attorney General, or to any person alleging discrimination on the basis of disability in violation of any provision of this chapter, or regulations promulgated under section 12116 of this title, concerning employment.”).

175Prison Litigation Reform Act of 1995, Pub. L. No. 104-131, 110 Stat. 1321, 1321-66 to 132177.

176Antiterrorism and Effective Death Penalty Act of 1996, Pub. L. No. 104-132, 110 Stat. 1214.

177This might, however, raise nondelegation issues. See supra note 110.

178United States v. Booker, 543 U.S. 220, 222, 265 (2005).

179See J. Mitchell Pickerill, Constitutional Deliberation in Congress: The Impact of Judicial Review in a Separated System 66−67 (2004) (arguing that major Supreme Court cases can prompt and shape debates over new legislation).

180See, e.g., Peter W. Hogg & Allison A. Bushnell, The Charter Dialogue Between Courts and Legislators (or Perhaps the Charter of Rights Isn’t Such a Bad Thing After All), 35 Osgoode Hall L.J. 75, 79–80 (1997).

181See Matthew R. Christiansen & William N. Eskridge, Jr., Congressional Overrides of Supreme Court Statutory Interpretation Decisions, 1967–2011, 92 Tex. L. Rev. 1317, 1328–29 (2014).

182See id. at 1325 n.31.

183See id. at 1371–73. This type of override has continued even in the relative drought of the 2000s. See id. at 1375 (“[T]he recent era of fewer overrides still has managed to adopt a good many restorative overrides.”).

184See Transcript, supra note 15, at 73 (“JUSTICE SCALIA: . . . You can’t repeal the rest of the Act because you’re not going to get 60 votes in the Senate to repeal the rest. It’s not a matter of enacting a new Act. You got to get 60 votes to repeal it. So, the rest of the Act is going to be the law. So, you’re just put to the choice of, I guess, bankrupting insurance companies and the whole system comes tumbling down, or else enacting a Federal subsidy program to the insurance companies, which is what the insurance companies would like, I’m sure.”); Adam D. Chandler & Luke Norris, How Conservatives Could Revive the Public Option, Slate (Jan. 7, 2011, 1:26 PM), http://www.slate.com/articles/news_and_politics/jurisprudence/2011/01/how_conservatives_could_revive_the_public_option.html.

185See Ian Ayres & Robert Gertner, Filling Gaps in Incomplete Contracts: An Economic Theory of Default Rules, 99 Yale L.J. 87, 89, 115 (1989); Einer Elhauge, Preference-Estimating Statutory Default Rules, 102 Colum. L. Rev. 2027 (2002).

186See Brief for Respondents (Severability), supra note 70. The reverse approach of striking down the entire bill if part of it is unconstitutional, unless there is a severability clause, would also work as a preference eliciting default rule. This strategy would force Congress to specify when it wants a statute made severable, or else lose everything in a bill. However, this is a less effective punitive default because it is easily reversed by Congress through the use of pro forma severability clauses (indeed, that is Congress’s practice now). The conditionality approach, by contrast, compels Congress to specify which clauses it wants made conditional on which other clauses. Presumably, Congress would not start putting pro forma inseverability clauses in every bill.

187Bipartisan Campaign Reform Act of 2002, Pub. L. No. 107-155, 116 Stat. 81.

188See Excerpts from Senate Debate on Donations: Skirmishing and Predictions, N.Y. Times, Mar. 30, 2001, at A16, available at http://www.nytimes.com/2001/03/30/us/excerpts-from-senate-debate-on-donations-skirmishing-and-predictions.html [hereinafter Excerpts] (“The Senate is being asked to pass an amendment that would make two provisions of this bill non-severable, one from another.”).

189See, e.g., 4 U.S.C. § 125 (2012) (“If a court of competent jurisdiction enters a final judgment on the merits that . . . substantially limits or impairs the essential elements of sections 116 through 126 of this title, then sections 116 through 126 of this title are invalid and have no legal effect as of the date of entry of such judgment.”); 25 U.S.C. § 1760 (2012) (“In the event that any provision of section 1753 of this title is held invalid, it is the intent of Congress that the entire subchapter be invalidated.”).

190Presently inseverability clauses are fairly rare, occurring at a rate of roughly one per year as of 2001. See Excerpts, supra note 188 (“[D]uring the last 12 years only 12 bills have been introduced, let alone passed, that contain a nonseverability clause.”).

191Severability also affects less significant statutes, of course, but these are less important when we are considering the practical effects of legislative conditionality.

192See Emily Smith, Timeline of the Health Care Law, CNN (June 28, 2012, 10:24 AM ET), http://www.cnn.com/2012/06/28/politics/supreme-court-health-timeline; see also Carl Hulse & Jeff Zeleny, Democrats Seem Set to Go it Alone on a Health Bill, N.Y. Times, Aug. 19, 2009, at A1, available at http://www.nytimes.com/2009/08/19/health/policy/19repubs.html.

193Erie R.R. Co. v. Tompkins, 304 U.S. 64 (1938).

194See Michael C. Dorf, The Heterogeneity of Rights, 6 Legal Theory 269, 290 (2000) (“[S]everability is in turn a question of statutory construction, and in a challenge to a state law, state rather than federal principles of statutory construction govern.”); Abbe R. Gluck, Intersystemic Statutory Interpretation: Methodology As “Law” and the Erie Doctrine, 120 Yale L.J. 1898, 1990–91 (2011).

195Dorf, supra note 11, at 295 (“Every state except Tennessee and Virginia employs a presumption of severability, unless the statute in question includes a nonseverability clause. In general, state courts hold that provisions and applications of statutes are severable.”); Gluck, supra note 194, at 1950 (“[W]ith respect to avoidance’s cousin, severability, forty-eight states have a statute, a judicial decision, or both requiring courts to apply a presumption of severability.”); see also Dorf, supra note 11, at 295–304 (surveying the severability laws of all fifty states).

196Minn. Stat. Ann. § 645.20 (West 1947).

197Wis. Stat. Ann. § 990.001(11) (West 2007).

198See Vollmer v. City of Memphis, 730 S.W.2d 619 (Tenn. 1987); Bd. of Supervisors v. Rowe, 216 S.E.2d 199 (Va. 1975); see also Dorf, supra note 11, at 301–03.

199See supra note 35 (noting that the Supreme Court has declared state statutes inseverable in only six cases, by my count, since 1940); see also Scoville, supra note 34, at 564–69 (surveying severability decisions after Erie and concluding that “the Court generally settled upon the rule that the sovereign whose statute is at issue dictates the severance test”).

200Leavitt v. Jane L., 518 U.S. 137, 139 (1996) (per curiam).

201457 U.S. 55, 65 (1982).

202See Dorf, supra note 11, at 295.

203Ky. Rev. Stat. Ann. § 446.090 (West 2006) (“It shall be considered that it is the intent of the General Assembly, in enacting any statute, that if any part of the statute be held unconstitutional the remaining parts shall remain in force, unless the statute provides otherwise, or unless the remaining parts are so essentially and inseparably connected with and dependent upon the unconstitutional part that it is apparent that the General Assembly would not have enacted the remaining parts without the unconstitutional part, or unless the remaining parts, standing alone, are incomplete and incapable of being executed in accordance with the intent of the General Assembly.”).

204Iowa Code Ann. § 4.12 (West 2008) (“If any provision of an Act or statute or the application thereof to any person or circumstance is held invalid, the invalidity does not affect other provisions or applications of the Act or statute which can be given effect without the invalid provision or application, and to this end the provisions of the Act or statute are severable.”).

205Cf. David E. Pozen, Judicial Elections as Popular Constitutionalism, 110 Colum. L. Rev. 2047 (2010) (considering the impact of judicial elections on the legitimacy of constitutional change).

206See supra note 35; see also supra Part VI.B.

207546 U.S. 320, 329–30 (2006) (alterations in original) (quoting, in turn, Regan v. Time, Inc., 468 U.S. 641, 652 (1984) (plurality opinion), Virginia v. American Booksellers Association, 484 U.S. 383, 397 (1988), and Califano v. Westcott, 443 U.S. 76, 94 (1979) (Powell, J., concurring in part and dissenting in part)).

208Minnesota v. Mille Lacs Band of Chippewa Indians, 526 U.S. 172, 179 (1999) (some internal quotation marks omitted).

209Id. at 191.

210Id. at 193.

211See id. at 215–16 (Rehnquist, C.J., dissenting) (“[T]he order’s termination of the treaty privileges should be sustained unless the Chippewa are able to clearly demonstrate that President Taylor would not have terminated them without a removal order. But there is no such evidence . . . .”).

212See Montana v. Blackfeet Tribe of Indians, 471 U.S. 759, 766–68 (1985).

213United States v. Booker, 543 U.S. 220, 245 (2005).

21418 U.S.C. § 3742(e)(3)(B)(ii) (2012), invalidated by Booker, 543 U.S. 220.