Printer Friendly


In resolving questions of statutory meaning, the lion's share of Roberts Court opinions considers and applies at least one interpretive canon, whether the rule against surplusage or the presumption against state law preemption. This is part of a decades-long turn toward textualist statutory interpretation in the Supreme Court. Commentators have debated how to justify canons, since they are judicially created rules that reside outside the statutory text. Earlier studies have cast substantial doubt on whether these canons can be justified as capturing congressional practices or preferences; commentators have accordingly turned toward second-order justifications, arguing that canons usefully make interpretation constrained and predictable, supplying Congress with a stable interpretive background. Based on an extensive study tracking the use of over 30 interpretive canons in the first 10 years of the Roberts Court, this Article attempts to contribute evidence to the debate over canons.

The data raise substantial questions regarding stability and predictability. Despite a long tradition of use, some canons have essentially disappeared; meanwhile, the Court has created others out of whole cloth. In addition, application is erratic. The Roberts Court Justices have declined to apply even the most widely engaged canons 20-30% or more of the time, often for difficult-to-anticipate reasons; some well-known canons, such as the rule of lenity and the presumption against preemption, were applied roughly at a 50-50 rate. The story is worse in the many cases in which multiple canons are considered. Based on these and other findings, this Article accordingly argues that predictability and stability arguments cannot supply a firm foundation for canon use. The study also reveals troubling mismatches between canons actually in use and congressional staff acceptance of canons. The Article concludes by suggesting some future directions for investigation and reform.



      A. Categories of Canons

          1. Textual Canons
          2. Substantive Canons

      B. The Challenges of Justifying Interpretive Canon Use


      A. Methodology

         1. Textual Canons
         2. Substantive Canons

      B. Caveats


      A. The Supreme Court Uses Canons Often--Especially Canons
         of Consistency and Deliberateness

      B. Some Canons Are Strong; Some Are Weak, Yielding Readily
         to Other Considerations; the Vast Majority of Canons Were
         Not Applied At Least 25% of the Time They Were

      C. When Opinions Use Canons to Resolve a Statutory Issue,
         They Typically Use More Than One Canon..

      D. Canons Were Not Applied for a Wide Variety of

      E. The Court Has Birthed Multiple New Canons, Modified Old
         Ones, and Killed Some Off Altogether

           1. New "No Elephants in Mouseholes" Canon
           2. New Location of Codification Canon
           3. New Veterans' Benefits Canon
           4. New Jurisdictional Rules Canon
           5. Rule of Lenity Modifications
           6. Expansion of Expressio Unius Beyond the List Conte
           7. Whole Session Laws Rule
           8. Continued Expansion of Federalism Canons
           9. Fledgling Canons

IV.   Implications for Stability and Predictability

      A. The Stable Interpretive Background Justification
      B. A First Cut at the Content of Individual Canons in Use




The lion's share of Roberts Court majority opinions engages at least one interpretive canon in resolving a question of statutory meaning. (1) Growing canon use is a component of a distinct rise in so-called textualist methods of statutory interpretation. (2)

Justice Kagan's suggestion that judges are "staring at the words on the page" (3) captures a current consensus within the federal judiciary--a victory for the textualists--that statutory text comes first. When text straightforwardly suffices to answer a question, no further investigation is needed, and evidence about congressional purpose will not override it. (4) Even ambiguous or unclear text can bound the range of permissible interpretations that interpretive strategies such as legislative purpose analysis might otherwise open up. (5)

Yet statutory text is often inadequate to the interpretive task. The text may be silent, indeterminate, ambiguous, or even conflicting on contested legal issues. As Justice Breyer has explained, the judge is then compelled to go beyond the words in carrying out the legislature's will, "for the words have simply ceased to provide univocal guidance to decide the case at hand." (6)

For decades, the Court turned to legislative history as evidence of congressional intent, but this drew blistering criticism from textualists, who argued that legislative history is "not the law" and is subject to manipulation by both judges and legislators. (7) This prompted a significant decline in, though not abandonment of, legislative history use. (8)

Perhaps to supplement a depleted arsenal, the Court has deployed a stockpile of canons to aid the interpretive endeavor. (9) Interpretive canon use, along with dictionary use, is now seen as intrinsic to textualist interpretive modes. The Roberts Court has invoked so-called textual canons, including grammatical rules and canons of deliberateness, such as the rule against surplus-age. (10) The Court has also applied substantive canons, such as the presumption against preemption of state law and canons that call for consultation of the common law, agency interpretations, and other legal sources. (11) Canon use now seems deeply embedded in the Court's interpretive practices. Indeed, every justice in the Roberts Court engaged at least one canon in the majority of statutory interpretation opinions he or she authored in the Roberts Court's first decade, this Article's study period. (12)

But canons, like legislative history, also reside outside the enacted statutory text, and they are judicial creations, raising further concerns about their consistency with legislative supremacy. Canon defenders have argued that interpretive canons are nonetheless acceptable because they approximate Congress's drafting practices and likely preferences. (13) Canon critics have questioned whether this is plausible for many canons. Professors Abbe Gluck and Lisa Bressman raised a substantial challenge to that defense in an empirical study of congressional staff; they found staff were often unfamiliar with canons or rejected their premises outright. (14) Although they found limited evidence for some canons, Gluck and Bressman concluded that their findings undermine any universal justification of canons on the ground that they approximate congressional preferences. (15)

Canon defenders have also advanced second-order reasons. In essence, the second-order defense of canon use is that canons, whatever their content, represent clear interpretive rules that can coordinate and constrain judicial decisionmaking and render interpretation more predictable. (16) Interpretive stability and judicial constraint are independently valuable, so the argument goes.

Justice Gorsuch recently raised such an optimistic defense: "[W]hen judges pull from the same toolbox ... we confine the range of possible outcomes and provide a remarkably stable and predictable set of rules...." (17) The late Justice Scalia's 2012 book with Bryan Garner, Reading Law, catalogs and makes similar express claims for canons, arguing that they "will narrow the range of acceptable judicial decision-making and.... will curb--even reverse--the tendency of judges to imbue authoritative texts with their own policy preferences." (18) Justice Kagan commented in November 2015 that the book serves as a regular reference on the Court. (19)

These claims do not directly address the concern that the application of canons, if they do not approximate congressional practices or preferences, potentially undermines legislative supremacy. Here, canon defenders respond that canons supply a more stable interpretive background for Congress. If Congress can anticipate canon application and predict interpretive outcomes, so the argument goes, it is empowered to draft around undesirable ones, indirectly supporting legislative supremacy. (20)

Judicial reliance on canons has, however, lately drawn its own share of blistering criticism as capricious and potentially manipulative. Abbe Gluck has, for example, deemed the current iteration of the canon-heavy "formalist [interpretive] project" an outright "failure." (21) She joins retired Judge Richard Posner in criticizing judicial canon use as unpredictable, lacking any ranking or precedential approach. (22) The critique is that the sheer number and variety of canons end up actually widening judicial discretion to rule in accord with judicial policy preferences, undermining any promise of interpretive stability.

This project attempts to contribute evidence to the debate over whether canon use contributes to interpretive stability and predictability. It reports findings from a substantial empirical project on interpretive canon use in the Roberts Court. The project assesses such use in the first ten years of the Roberts Court, from October Term 2005 through October Term 2014. It is the most significant assessment of statutory interpretation in the Court to date, tracking thirty-six separate textual and substantive canons across 838 majority, concurring, and dissenting opinions that resolved contested statutory issues. The study identified instances in which a canon was applied to help resolve a contested issue, as well as instances in which the justices considered, but declined to apply, a particular canon. This project adds to the substantial analysis already conducted by Professor Anita Krishnakumar and the significant work of Professors James Brudney and Corey Ditslear in workplace and tax cases, though it differs in important respects from these studies, both by tracking more canons and in certain key findings, as discussed below. (23)

The principal findings are these. First, the Court relies extensively on canons, both textual and substantive. Second, rather than stability, it is change that characterizes the Roberts Court's current collection of interpretive canons. (24) Some traditionally recognized canons, such as the punctuation canon and the remedial purposes canon, are used rarely and may soon evaporate altogether. (25) Meanwhile, the Court continues to create new canons. This Article discusses several new textual and substantive canons, including a location of codification canon, a jurisdictional rules canon, and a veterans' benefits canon, as well as new modifications to existing canons. (26)

Second, the data suggest that canon critics are right to worry about judges using canons unpredictably. Karl Llewellyn's famous point was that for every canon, there was an "opposing" canon that could overcome it. (27) Posner and Gluck also have argued that the Court appears to deploy no hierarchy for canon use, enabling judges freely to choose among them. (28) This study's findings confirm Llewellyn's implication that one indeed cannot predict which of two conflicting canons the Court will apply.

But this Article documents two additional ways in which canon use goes beyond the unpredictability Llewellyn anticipated. First, when canons are discussed, the justices very often decline to apply them, even if there is no other canon present. The most frequently considered and applied canons are still not applied at least 20-30% of the time they are discussed--and numerous canons are applied even less reliably. (29) The common presence of multiple potentially applicable canons simply makes canon use even less predictable. Second, and at least as critically, the reasons the justices have invoked for not using canons are a slippery group, extraordinarily difficult to anticipate or even to define.

In other words, although this study is not (and could not be) a controlled study, the findings strongly suggest that canon use is of dubious value to interpretive predictability and in turn to judicial constraint or a stable interpretive background for Congress. As currently used, then, canons seem hard to justify on these "second-order" grounds. Considerable reform would be required for canon use to positively contribute to interpretive predictability.

Finally, returning to the first-order arguments that canons could approximate congressional practices, this Article's findings reveal a striking mismatch between the top canons in current use in the Roberts Court and Gluck and Bressman's findings. Three, and possibly four, of the five most frequently applied canons in the Roberts Court were identified by Gluck and Bressman as canons outright "rejected" by congressional staff--and as to the fifth, the evidence of congressional staff acceptance is equivocal. (30) In short, the most used canons seem the least defensible on first-order grounds. Meanwhile, the canons seemingly most accepted by congressional staff were used by the Roberts Court only fitfully.

Part I overviews the justifications offered for both textual and substantive canons and lays out the necessary conditions for canon use to supply Congress with a stable interpretive backdrop. Part II discusses the study's methodology and some caveats, and Part III presents key findings. Part IV assesses the implications of those results for the justifiability of canon use in the Roberts Court, both at a general level and for selected individual canons. The conclusion identifies future directions for investigation, including suggesting the consideration of a joint legislative/judicial approach to interpretive rules, perhaps similar to the process that generated the Federal Rules of Evidence.


Interpretive canon use is associated with textualist approaches to statutory interpretation, in which legislative supremacy and the text of the enacted statute are central. In textualist approaches, the judge realizes her role as faithful agent of the legislature by ascertaining the statute's meaning--in contrast, perhaps, to attempting to implement what Congress intended by enacting the statute. (31) This Part overviews the two major types of interpretive canons in current use--textual, or syntax, canons, and substantive canons. It then takes stock of the prevailing justifications for these judicially developed interpretive tools.

A. Categories of Canons

1. Textual Canons

Textual canons are often described as policy-neutral tools to decode the meaning of Congress's language. (32) They include grammatical and punctuation rules, as well as rules that assume internal consistency and nonredundancy in textual drafting. The former rules are often characterized as ordinary English usage rules, assisting courts to discern statutory meaning by reference to what ordinary English speakers mean when they use or read particular words and sentences. (33) They include the rule of the last antecedent, the "and/or" rule, and the "may/shall" rule. (34)

If such textual canons are in fact based on realistic views of how English speakers communicate or how Congress, as a special category of English speakers, legislates, they might not be understood as in tension with legislative supremacy. Judges might be merely recognizing preexisting rules of syntax and semantics, much like recognizing that federal statutes are drafted in American English. Such rules would seem to be less "law." Of course, if these canons merely encapsulate the ordinary rules of communication, one might wonder why they are needed. (35)

Even these purportedly neutral canons do embed some policy commitments. For example, they can represent a systematic judicial choice to focus on one interpretive community over another--a choice to pay attention to the understanding of legislative drafters, rather than readerly understanding, or vice versa. (36) Grammatical canons could be understood as aimed at readers, more likely to be ordinary English speakers. Such canons could implicitly privilege notice concerns by enabling ordinary people to better understand statutes, were they to open up the statute books, compared with determining and honoring Congress's meaning. (37) For example, in Lockhart v. United States, the majority opinion, written by Justice Sotomayor, advocated for application of the rule of the last antecedent, while the dissent, written by Justice Kagan, advocated for the application of the so-called series-qualifier canon; both defended their interpretive rule choices in terms of what a reader would "intuitively" do. (38)

Despite being grouped together with grammatical canons, (39) the remaining textual canons are quite different. They instead embody assumptions that Congress drafts legislation deliberately, coherently, consistently, and without redundancy. For example, the whole act rule calls for courts to interpret a statutory term by assuming that a term used in a statute means the same thing wherever it appears, and that different words mean different things. (40) The whole code rule extends the same set of assumptions throughout the entire body of enacted statutes codified in the U.S. Code. (41)

Judges applying the rule against surplusage interpret statutory language to avoid making text redundant or meaningless, essentially assuming that Congress does not enact words without meaningful application, including because their function is duplicated elsewhere. (42) Judges applying the expressio unius canon assume that omissions are deliberate (43)--when a statute restricts, say, trafficking in guns, drugs, and money, it must be understood to permit trafficking in everything else.

Although some attempt to defend consistency and deliberateness canons as consistent with language conventions, (44) they can differ significantly from ordinary English practices. For example, variety and redundancy in communication both can be par for the course, even desirable. Teachers urge writers to strive for diversity in word choice and sentence structure; redundancy can add clarity and emphasis. And parents know that the statement "Alex, do not hit or kick your sibling," simply provides specific examples of the correct behavioral standard, rather than--as the expressio unius canon might suggest--impliedly authorizing pinching or worse. (45)

Meanwhile, the whole act and whole code rules, which assume consistency in legislative drafting, (46) downgrade the importance of context, critical to ordinary English speakers. (47) An ordinary reader will likely assess a word's meaning only in its immediate context, rather than scour an entire statute or the U.S. Code to find other uses of the word. These rules accordingly also devalue notice, since a reader cannot readily reach a conclusion about a particular text's meaning without comparing it to numerous others. (48) The rule against surplusage similarly devalues notice, since ordinary readers might well not understand redundancy as something to be avoided in a statute, but simply as "belt and suspenders" that clarify or confirm a statute's coverage.

2. Substantive Canons

Substantive canons represent a judicial thumb on the scale in favor of a particular norm. Clear statutory language or another distinct indication of meaning is usually required to overcome the canon's default result. Substantive canons include the rule of lenity--in which ambiguous criminal statutes are to be interpreted in favor of the defendant--and presumptions against waiving sovereign immunity, preempting state law, and derogating common law. (49) Some substantive canon defaults are defended as enforcing constitutional values. The rule of lenity, for example, is understood to effectuate the constitutional due process norm of notice, (50) though the historical justification sounds more in leniency, ensuring that conduct is not criminalized unless the legislature has clearly so stated. (51) The presumption against preemption and the so-called federalism canons are understood to protect core attributes of state sovereignty and an appropriate balance of power between the federal and state governments. (52) Substantive canons of this sort, however, typically enforce the norm more than the independent legal source itself does, without recognizing compromises or countervailing factors. (53) This has prompted the criticism that they represent an unjustified penumbra of protection for the chosen value. (54)

As with textual canons, substantive canon defaults are also sometimes defended as approximating congressional preferences, as analyzed in greater detail below. (55) Finally, some commentators have offered other justifications for certain substantive canons, including the presumption against preemption of state law, the rule of lenity, and the major question canon exception to deference to agency interpretations. The idea is that tipping the scales through interpretation--in favor of the underrepresented or against broad federal powers--can enhance congressional deliberation by increasing the prospects that these interests will receive consideration. (56) These are typically not the justifications offered by textualist interpreters, perhaps because they place the judiciary in such an activist role relative to legislative process. (57) They thus are beyond the immediate scope of this project.

B. The Challenges of Justifying Interpretive Canon Use

Whatever their content, canons purport to be rules. But these rules are external to the particular statutory text being interpreted, as well as to the congressional process. That raises the difficult question of reconciling them with legislative supremacy. To set the stage for the empirical findings, this Section overviews the struggle to justify canons and the major arguments made in their defense.

The first-order justification for canons--fairly consistent with legislative supremacy if true--is that they simply represent a rule-like approach to "trying to ascertain what the enacting legislature probably meant" in a particular statute. (58) Textual canons, then, could be justified as an efficient way--one that minimizes the sum of error and decision costs--to approximate how Congress uses language, including both English usage and drafting practices, (59) and substantive canons could be justified as approximating Congress's substantive policy preferences. (60)

The elephant in the room here, however, is that courts have not systematically investigated actual congressional practices or preferences at any point in time, let alone assessed their evolution over time or across Congresses. (61) Meanwhile, Gluck and Bressman's important study has raised serious questions about this justification. They sharply question, in particular, whether congressional drafting embodies--or is even capable of--the consistency and deliberateness assumptions implicit in textual canons. (62) Some textual rules, such as the rule against surplusage and the whole act rule, were affirmatively rejected by the many staff members Gluck and Bressman interviewed. (63)

Gluck and Bressman's work does suggest some greater staff awareness of substantive canons. They found that many congressional staffers they interviewed expected the courts to adhere to the constitutional avoidance canon, presumption against preemption, and Chevron deference rule, though staffers did not necessarily know canons by name or were sometimes confused about a canon's default result. (64) Indeed, some staffers believed that the presumption against preemption was actually a presumption in favor of preempting state law. (65) In short, although that study was limited to interviewing congressional staff, not members, and only in a single year, (66) Gluck and Bressman's work substantially undermines a general defense of canons as "approximating" congressional preferences or practices.

Perhaps recognizing the potential weaknesses of Congress-approximating justifications, but nonetheless seeking an across-the-board justification for canon use, canon defenders have relied increasingly heavily on second-order reasons. The second-order argument essentially is that canons are rule-like, and a more rule-like interpretive approach could make statutory interpretation more constrained and predictable. (67) If courts interpreting statutes use, say, similar approaches to text or a particular substantive issue, judicial decisionmaking could be better coordinated and the range of available potential outcomes reduced. Citizens could better anticipate how judges will apply statutes. (68) A related benefit would be, in theory, narrowing the range of interpretive choices open to judges, ultimately reducing decisions influenced by the judge's own policy preferences, though this claim depends on comparing a canon-based interpretive regime with other interpretive regimes.

But then what of legislative supremacy? The argument here: canons can facilitate legislative supremacy (and minimize error, though only indirectly) not by aiming to capture likely congressional practices and preferences in the first instance, but by letting Congress have the last word. Congress could, in theory, better predict interpretive rules applied and interpretive outcomes produced and more clearly draft its legislation if it seeks a different outcome. (69) It is worth underscoring that these second-order justifications are independent of content; in theory, they could be advanced for any canon whatsoever. (70)

The particular argument that canon use can supply a stable legislative background is the point of textualist arguments that canons are validated by tradition, even antiquity. (71) Even as he initially opposed "dice-loading" substantive canons as beyond judicial authority, (72) Scalia claimed that a long history of use could legitimate a particular canon: "Once [substantive canons] have been long indulged, they acquire a sort of prescriptive validity, since the legislature presumably has them in mind when it chooses its language...." (73) Scalia and Garner's claim for legitimacy for canons similarly turns significantly on their age: "Most of the canons of interpretation [in our book] are so venerable that many of them continue to bear their Latin names." (74) Then-Professor Amy Coney Barrett discussed six canons for which she located historical foundations and argued that their "long pedigree" was validating. (75)

A canon's pedigree is essential to the "stable background" justification, since Congress obviously cannot anticipate a not-yet-developed canon; so is the need for a stable and settled group of canons in use. Professor John Manning has acknowledged the necessary implication of the "stable background" argument: "textualists [logically] ... must largely accept the world as they find it, treating the existing set of background conventions as a closed set." (76)

Even Manning's "closed set" of canons can usefully coordinate judicial decisionmaking and supply a stable background for Congress only if several additional conditions are satisfied. First, the canon must actually be rule-like. As Professor Frederick Schauer points out, a rule's contribution to a stable system comes in part from clarity over the rule's boundaries. (77) Following Schauer's example, the stability and certainty--the "ruleness"--of a rule like "no dogs in a restaurant" depends not only on knowing the result that follows from the rule's application but also on everyone knowing and agreeing on what qualifies as a "dog." (78)

Similarly, there must be shared knowledge and agreement around each canon's content--the conditions that trigger a canon's application and the interpretive consequences that follow. (79) In other words, both readers and legislators must know what circumstance will cue a judge to apply a canon: for example, "The criminal statutory language is ambiguous"; or "The ambiguous word is one of a listed series." Both readers and legislators must also anticipate what will follow, such as "The statute will be interpreted in favor of the defendant"; or "The word will be interpreted to share the characteristic common to the others in the list."

Second, judges must deploy canons consistently, when prompted by the relevant cue. Unpredictable canon use, blurry canon contours, or a general atmosphere of changeability will undermine a canon's rule-like quality, and, in turn, the stability it contributes to the interpretive background. A general atmosphere of changeability, moreover, likely will reduce congressional drafters' incentives to maintain overall awareness of canon use.

Third, Congress must be able to easily predict the result when a canon is applied to particular statutory language and also to readily draft around that result if necessary. Even if canons are fully rule-like and thus more predictable, the prospects here may be dim. Christiansen and Eskridge have documented the obstacles to and relative rarity of overrides, (80) underscoring that Congress cannot be assumed to readily respond to a new canon. (81) Even older, more consistently used canons could raise significant concerns. As Gluck and Bressman's findings on congressional staff knowledge of canons suggest, it may be unfair to presume widespread congressional awareness of canons. (82)

In this regard, the complexity of legislative bargains also may make it unfair to impute congressional acceptance of a judicially devised canon to a statute's enactment. Textualists have pointed to the difficulty of identifying the content of a legislative compromise as a central reason why courts cannot infer a clear congressional purpose apart from the enacted text itself. (83) By the same token, a variety of demands may impair Congress's ability to freely draft around canons, including the need to satisfy pivotal legislators to get legislation through "vetogates" or to direct instructions to implementing or complying institutions. For example, notwithstanding the rule against surplusage, drafters regularly include redundant language in legislation to emphasize its coverage or to placate interest groups or a key legislator. (84) The "opacity, complexity, and path dependency" of legislative bargain making means one can assume neither that drafters have written with perfect clarity nor that Congress, in enacting legislation, has implicitly accepted any particular judicial canon or assumption about drafting. (85)


A. Methodology

The project's primary goal was to systematically document canon use patterns in the Roberts Court's first ten years. Fourteen excellent research assistants and I read all statutory interpretation decisions from the first ten years of the Roberts Court--October Terms 2005 through 2014. We evaluated any opinion resolving a contested question of federal statutory meaning, excluding, for example, constitutional, regulatory, and nonfederal legal issues. (86) For each such contested federal statutory issue, we tracked the use of multiple listed textual and substantive interpretive canons, based on characterizations of the canons that we specified in advance. Specifications are excerpted below. (87) We coded the type of opinion (majority, concurring, etc.), the opinion's author and the identities of the signing justices, and, generally, the subject matter. In short, we analyzed 838 majority, plurality, concurring, and dissenting opinions, found in 460 separate decisions. Each contained analysis of at least one contested statutory issue, for a total of 931 contested statutory issues over the 838 opinions. We coded over 1,400 separate canon observations.

We coded for the following thirty-two textual and substantive canons, all of which appear in two important canon sources: a classic seven-volume treatise, Sutherland's Statutes and Statutory Construction, which includes an extensive listing of canons, (88) and Scalia and Garner's widely referenced newer book, Reading Law, also a lengthy canon catalogue. (89) For clarity, included below are the particular definitions we used.

1. Textual Canons

Expressio/inclusio unius est exclusio alterius (hereinafter referred to as expressio unius). The expression of one thing implies the exclusion of another. (90)

Whole act rule. Identical terms are presumed to have consistent meaning throughout a statute; different words are presumed to have different meanings. (91)

Whole code rule. Identical terms are presumed to have consistent meaning throughout the U.S. Code; different words are presumed to have different meanings. (92)

Rule against surplusage. If possible, every word or phrase should be given effect. (93)

Harmonious reading canon. Close cousin of rule against surplusage. A statute should not be interpreted to nullify a section or render it contradictory with another section. (94) We coded it separately because it is so described by Scalia and Garner.

Noscitur a sociis. Associated words in a list bear on one another's meaning and should be interpreted to share common characteristics. (95)

Ejusdem generis. A general word following an enumeration should be interpreted to share the characteristics of the preceding words. (96)

Rule of last antecedent. "A pronoun, relative pronoun, or demonstrative adjective generally refers to the nearest reasonable antecedent." (97)

In pari materia. Statutes addressing the same subject matter should be "interpreted together, as though they were one law"; ambiguities or lack of detail in one statute can be resolved or filled in by reference to another statute on the same subject. (98)

And/or rule. "And" joins a conjunctive list; "or" joins a disjunctive list. (99)

May/shall rule. "May" is permissive and grants discretion; "shall" is mandatory and imposes a duty. (100)

Punctuation canon. "Punctuation is a permissible indicator of meaning." (101) Specific rules govern over general ones. (102)

Titles canon. "The title and headings are permissible indicators of meaning." (103)

2. Substantive Canons

Presumption against implied repeal. Where a newly enacted statute is silent regarding a previous existing one, the legislature should be presumed not to have impliedly repealed the existing one. (104)

Rule against absurdity. Statutes should be applied to avoid absurd results. (105)

Constitutional avoidance canon. A statute should be interpreted to avoid "even rais[ing] serious questions of constitutionality." (106)

Presumption against preemption of state law. (107)

Federalism canon (other than the presumption against preemption). (108)

Presumption against extraterritorial application. (109)

Presumption against retroactive application. (110)

Rule of lenity. "Ambiguity in a statute defining a crime or imposing a penalty should be resolved in the defendant's favor." (111)

Native American/Indian canon. Statutory ambiguities are to be resolved in the affected tribes' favor. (112)

Statutes in derogation of the common law shall be strictly construed (hereinafter referred to as the "common law canon"). We tracked this together with its close "cousin," (113) "A statute that uses a common-law term, without defining it, adopts its common-law meaning." (114)

Remedial statutes shall be liberally construed. (115)

Presumption against waiver of sovereign immunity. (116)

Agency deference canons (Chevron/Skidmore). (117) These included deference to an agency's reasonable interpretation of an ambiguous statute (Chevron) and deference to agency interpretation in non-Chevron settings (Skidmore). Owing to an already extensive literature analyzing application of these canons, we did not differentiate further among them. (118)

Legislative acquiescence. Congress is deemed to acquiesce in an interpretation when Congress has enacted the legislation but not modified the relevant provision, or when Congress has not acted in response.

We also tracked four new canons that we had identified during the pilot phase of the study, for a total of thirty-six canons:

Whole session laws rule. Word usage is presumed to be consistent within statutes enacted during the same congressional session. (119)

No elephants in mouseholes. Congress is presumed not to legislate on significant issues through the use of cryptic language. (120)

Major question canon. Absent clear language, Congress is presumed not to delegate "major questions" to agencies; this is an exception to the usual rules of deference to agency interpretations. (121)

Jurisdictional rules canon. Absent clear language, statutory limitations are presumed not to be jurisdictional in nature. (122)

The canon list is not exhaustive. With the notable exceptions of the Native American/Indian canon and the rule of lenity, for example, we did not attempt to track subject-specific canons. (123) But the list is aimed at capturing the most commonly discussed canons of construction. (124)

For each contested statutory issue, we coded which canons were considered--or discussed--in the opinion. To avoid the limitations of searching for particular words in a digital database, we coded the use of a canon even if it was not named explicitly. For example, suppose an opinion were to elect an interpretation to avoid the prospect that a particular word would duplicate the meaning of another. This is application of the rule against surplus-age even if the opinion does not use the phrase.

We also coded whether a particular canon, once discussed, was actually applied with approval to resolve the statutory issue, or whether, instead, the opinion ultimately did not apply the canon. For example, an opinion might discuss a canon's implications but find the conditions for the canon unmet or its application overcome by other factors, such as context or another canon. (125) Identifying instances when a canon was discussed but the opinion ultimately declined to apply it allows a glimpse into a canon's relative strength or weakness, into uncertainty around canon boundaries, and into the range of reasons an opinion might deploy for declining to apply a canon. It also helps us identify the arguments the justices feel they must engage, answer, or distinguish for their opinions to be perceived as legitimate--to address the expectations of the relevant interpretive community, including their colleagues, the parties, or attorneys in general. Finally, we noted when an opinion expressed general disapproval of a particular canon.

When a canon was actually applied, we did not attempt to grade the intensity of reliance on a canon or distinguish the case where a canon was relied on centrally, largely because of the high level of subjectivity in distinguishing among reasons for decision. (126) Thus, all judicial applications of a canon with approval to resolve a particular statutory issue were coded the same way.

Besides tracking canon use, we tracked whether the opinion cited dictionaries (legal or lay) or legislative history, and whether Congress's statutory purpose (or intent, goals, or design) figured in the opinion's interpretation. In an open-ended portion of the survey, we also attempted to identify canons other than those listed, as well as canon-like interpretive moves. In other words, we noted whether the opinion articulated a general principle or rule (127) to help resolve a disputed question of statutory interpretation. Finally, we attempted to track context-specific assumptions about Congress used to resolve interpretive questions. For example, we tracked statements like the following: "We do not believe Congress would have intended [this section defining certain aggravated felonies for purposes of deportation] to apply in so limited and so haphazard a manner." (128) We distinguished these sorts of assumptions from conclusions based on evidence about congressional intent, including legislative history. The data collected in these portions of the study will be the basis for future analyses.

Each Supreme Court decision's use of interpretive tools was entered into a Qualtrics survey instrument, and each reader entered an excerpt of opinion text supporting every canon coding decision, enabling my ready review of individual coding decisions for accuracy. At the outset of the study, I simultaneously read the initial set of opinions each coder evaluated to provide feedback and ensure the accuracy of each coder's work. I also have personally reviewed all of the over 1,400 individual canon coding decisions and have independently read roughly one third of the opinions as part of reviewing the accuracy of canon coding.

Canon-related findings appear in Part III.

B. Caveats

First, this is, of course, not a controlled study. We could only observe explicit engagement with canons. We could not see how cases might have been resolved in the absence of canon use, and we could not observe unwritten decisions not to apply a particular canon. For example, definitively proving whether canons improve interpretive predictability or constrain judges relative to other interpretive methods might require a controlled study, perhaps one in which alternative interpretive methods are used to resolve the identical dispute with external assessment of the ultimate results. (129) On the other hand, such a study would require decisionmakers to resolve hypothetical disputes with awareness that they are study participants, which itself might raise reliability concerns.

Second, this is a Supreme Court study. Supreme Court cases present more difficult and contested legal questions compared with those in the lower courts. If one thought that comparative stability and predictability gains from canon use were largely in easy cases, for example, then a Supreme Court study would understate those gains. On the other hand, canon defenders justify their use in part to constrain judicial willfulness. (130) Difficult cases would seem to present the greatest potential for both interpretive uncertainty and judicial willfulness--and correspondingly, the largest potential payoff for canons, making it appropriate to assess canon use in the Supreme Court.

Third, as one might expect with an observational study, there is a substantial denominator difficulty. Not every disputed statutory issue that reaches the Supreme Court presents an opportunity to use every canon, let alone every interpretive tool. For example, the major question canon, an exception to the usual agency deference rules, was discussed in 0.3% of disputed statutory issues, (131) but its application is limited to the cases in which deference is claimed for an agency interpretation of a statute. As a percentage of those cases, it rises to 2.7%, (132) a more substantial number. Meanwhile, the "and/or" rule can apply only if the statutory language contains an "and" or an "or"; the rule of lenity is generally limited to criminal statutes; (133) and so on. Ideally, one would know which canons could have plausible application; that list could then be compared with the canons actually invoked. We considered, though ultimately rejected as both unreliable and infeasible, filed briefs as a potential source of this information. The sheer volume would have made it impossible to examine all briefing in all cases. Moreover, in the instances we did attempt to examine, we found that parties regularly briefed canons that went unmentioned by the Court, (134) and the Court also regularly relied upon canons that went unmentioned in the parties' briefs. (135) We also ruled out an independent assessment of potential canon applicability in each case as extraordinarily challenging, to say the least. Instead, much more simply, this analysis attempts to compensate for this difficulty by identifying canons that could be expected, in general, to have broader or narrower application.

As noted, we also coded an opinion's engaging or discussing a canon, even if the opinion ultimately declined to apply the canon. An opinion's considering a particular canon, even without applying it, tended to show that the canon was part of the Court's regular interpretive discourse and was important to assessing the canon's overall stability. This data can help us gain a greater understanding of which canons belong in the Court's current "canon of canons," if you will. (136)

Some might suggest, however, that looking at an opinion's consideration of a canon when it did not ultimately apply it could overstate a canon's level of importance. For example, if a brief--or another opinion in the case--cites a canon, perhaps an authoring justice might feel compelled to address it as a matter of meticulousness or courtesy even if the canon obviously has no application. But as noted, parties often cited canons in their briefs that went unmentioned in any opinion. Meanwhile, Krishnakumar has identified a distinct lack of what she terms "dueling" in Supreme Court opinions, in which the majority and dissenting opinions engage and attempt to apply the same canon. (137) Finally, opinions that decline to apply a particular canon typically supply detailed reasons, rather than dispense cursorily with the canon, suggesting that discussion of the canon is not a mere formality. (138)

Some also might suggest that if all (or most) questions regarding whether a canon applies are easy ones, it might make sense to exclude examples when opinions found canon conditions unmet, since mention (without application) of the canon might then count for little. (139) But questions of a canon's applicability are often difficult ones. For example, in United States v. Atlantic Research Corp., the majority declined to apply the rule against surplusage, stating that tolerating some surplusage in the particular setting was "appropriate." (140) That could be characterized as concluding that the statute did not contain sufficient surplusage to trigger the canon's application, or alternatively that the particular language at issue was "clarifying," so that the canon was overcome by the strength of contextual inferences. (141) Moreover, as discussed in greater detail below, one of this study's important findings is that opinions declined to apply a canon for many reasons other than the canon's obvious nonapplicability. Thus, an opinion's discussion of, but ultimate decision not to apply, a canon can tell us something about stability (or the lack thereof) around canon use and the extent to which the canon functions as a well-defined "rule."

Finally, although numerous steps were taken to ensure that canon use was accurately coded, my review was more likely to miss coding errors of omission (instances when opinions considered canons but were not coded at all) compared with errors of commission (incorrect coding of canon use). Thus, this Article could potentially underreport the overall use of canons. No particular canon would seem especially vulnerable to coding omissions, however, so such errors should not bias relative results.


A. The Supreme Court Uses Canons Often--Especially Canons of Consistency and Deliberateness.

Majority opinions considered (though did not necessarily apply) at least one canon in roughly 70% of contested statutory issues. (142) Opinions overall considered at least one canon in about two-thirds of contested issues. (143) Both substantive and textual canons were regularly engaged. Opinions overall considered at least one textual canon in 48% of contested statutory issues and applied at least one textual canon with approval to help resolve 40% of contested statutory issues. (144) Opinions overall considered substantive canons in 37% of issues--applying at least one with approval to help resolve the issue in 27% of issues, not counting agency deference canons. (145) Canon use outstripped that of dictionaries, though dictionary use was greater than any individual canon. Including agency deference canons, opinions overall considered substantive canons in 45% of issues and applied at least one with approval in 32% of issues. (146) Opinions overall cited at least one dictionary in 20.3% of contested statutory issues; majority opinions cited dictionaries in 24.2% of contested statutory issues. (147)

Justice Roberts was the top user of canons in this project, discussing at least one canon in 80% of the statutory interpretation issues in opinions he wrote, and using at least one canon with approval to help resolve 70% of the interpretation issues. Justice Scalia and Justice Alito were close behind, discussing canons in 76% and 72% of contested interpretive issues in opinions each wrote, respectively; both applied at least one canon with approval to help resolve 64% of statutory issues in opinions each wrote. (148) Justices Ginsburg and Kagan, in the opinions each wrote, considered canons at the lowest rate, discussing at least one canon in 59% of the statutory interpretation issues and applying at least one canon to help resolve the issue in 54% and 57% of issues, respectively. But despite considering canons in 64% of issues in opinions he authored, Justice Thomas applied canons with approval less often than any other Justice. He declined to apply any canons with approval in 50% of opinions he authored. (149) In short, though, canons are central to statutory interpretation in the Court. Every justice regularly engaged them in resolving contested interpretive issues.

Krishnakumar found a much lower rate of reliance overall on substantive canons (14.4% across all opinions in the 2005-2011 October Terms). (150) But her lower findings are likely explained by the fact that she did not code as substantive canons two that we found among the most important--the rule against absurdity and the common law canon. (151)

Not all canons were regularly considered. Ranked by the percentage of contested statutory issues in which they were discussed (whether named explicitly or not), the top canon scorers, both textual and substantive, were the following:

Most Frequently             Percentage of Contested
Appearing Canons            Statutory Issues Where
                            Canon Was Discussed
                            (though not necessarily

Expressio unius                      18.6%

Whole act rule                       15.2%

Rule against surplusage              13.2%

Agency deference canons              11.9%
(Chevron, Skidmore)

Harmonious reading canon             11.8%

Statutes in derogation of            11.6%
the common law are to be
strictly construed or
"common law canon"

Whole code rule                      11.5%

Rule against absurdity               7.4%

Textual canons were, in general, considered more frequently than substantive canons. (152) Of the tracked canons, the least engaged included the following: Native American/Indian canon (discussed in 0.2% of issues), punctuation canon (0.3%), remedial statutes shall be liberally construed (0.6%), and the and/or rule (1.1%). (153) The data for all tracked canons appears in the Appendix.

Even if the data can only be impressionistic for the reasons earlier discussed, we can still learn something. For example, several canons could be expected to have relatively general application, including, at a minimum, the rule against surplusage, the whole act and whole code rules, expressio unius, the harmonious reading canon, and the punctuation canon. Also likely to have potentially broad application: canons that seek to incorporate content from other sources, including the agency deference canons, the in pari materia canon (which calls for same-subject statutes to be interpreted as a coherent whole), and the common law canon, the canon that statutes in derogation of the common law are to be strictly construed. A judge likely could apply any of these canons across a wide range of statutes and subject areas.

Thus, it seems useful to note the top scorers among the generally applicable canons. The most frequently considered canons were all textual--the surplusage canon, expressio unius, and the whole act and whole code canons. (154) None of these textual canons involved grammar or syntax rules. Meanwhile, some of the grammatical and punctuation canons were among the least frequently engaged.

The two most frequently considered substantive canons were also both general--the agency deference canons and the common law canon. (155) Surprisingly, given the sharp scholarly criticism leveled against it because of the amount of discretion it vests in judges to determine what is absurd, (156) the rule of absurdity came in third, discussed in 7.4 % of issues. (157) The common law canon's strong showing was also unexpected, given Justice Scalia's prominent questioning, in the mid-1990s, "whether the attitude of the common-law judge ... is appropriate for [working with] statutory law." (158) Among the least popular was the remedial statutes canon, and in pari materia was considered in fewer than 5% of issues. (159)

B. Some Canons Are Strong; Some Are Weak, Yielding Readily to Other Considerations; the Vast Majority of Canons Were Not Applied At Least 25% of the Time They Were Discussed.

The following table lists the approval rates for particularly strong and weak canons. The approval rate refers specifically to the percentage of times a discussed canon was actually applied to help resolve a contested statutory issue. The table also includes the percentage of disputed statutory issues in the database in which the canon's application was discussed. The data for all tracked canons appears in the Appendix. (160)

Subject to the caveats already discussed, the data can tell us something about both the relative and absolute strength of particular canons. At the comparative level, the data yields a Top Five list of textual canons, consisting of expressio unius, the rule against surplusage, the whole act rule, the whole code rule, and the harmonious reading canon. (161) These were the most frequently considered canons. Moreover, all were applied with approval to help resolve statutory issues 70-76% of the time that an opinion discussed them. (162)

Not far behind the "Top Five" textual canons were, somewhat surprisingly given criticism of them, the common law canon (statutes in derogation of the common law are to be strictly construed) and the rule against absurdity. Both were considered in a large number of cases and had rates of application with approval comparable to the Top Five canons. (163) These seven canons were the overall strongest.

Other canons, including well-known canons, were applied more rarely, more erratically, or both. Although the agency deference canons and the constitutional avoidance canon were each discussed relatively frequently, their approval rates were well under 60%. (164) The rule of lenity was considered in only about 3% of statutory issues discussed and applied with approval only in 55% of those issues. (165) In majority opinions specifically, the rule of lenity was applied with approval only 14% of the time it was considered. (166)

And some canons were outright unimportant. Of canons where one might expect more general application, the remedial statutes canon was almost never invoked--it was discussed in fewer than 1% of statutory issues and applied with approval 50% of the time it was discussed. (167) By contrast, the new "no elephants in mouseholes" canon, discussed below, could be on a strong trajectory; although it appeared relatively rarely, it was applied with approval 89% of the time. (168)

These data suggest some strong canons, but also could signify that the Court is on the way to abandoning others. Alternatively, low rates of application could indicate that a particular canon is simply quite low on some sort of hierarchy that the Court has developed but has not (so far) announced.

C. When Opinions Use Canons to Resolve a Statutory Issue, They Typically Use More Than One Canon.

In 70.1% of contested statutory issues in majority opinions, the opinion considered at least one of the tracked canons. (169) In 66.9% of the issues in that group of canon-using majority opinions, the opinion considered two or more canons, and in 37.9% of that group, the opinion considered three or more canons. (170)

Does considering multiple canons increase interpretive uncertainty? If most or all the canons point in the same interpretive direction, then multiple canon use might be entirely consistent with ensuring predictable judicial decisionmaking.

But that does not seem to be the case. Consistent with the findings on rate of application of individual canons, the data show that opinions engaging multiple canons in connection with a statutory issue typically do not apply all of them with approval. For example, for all opinions resolving a contested statutory issue that considered one canon, that canon was applied with approval 75% of the time; analyses considering two canons applied, on average, 1.38 of them with approval; analyses considering three canons applied, on average, 1.99 of them; analyses considering four canons applied, on average, 2.95 of them. (171) In short, when multiple canons were considered in connection with a disputed statutory issue, most often opinions declined to apply at least some of them.

D. Canons Were Not Applied for a Wide Variety of Reasons.

The data show that an individual canon frequently was not applied even when it was considered. Moreover, when an opinion considered multiple canons, it declined to apply at least some. This might be unsurprising (and consistent with a canon's claimed rule-like characteristics) if an opinion's reason for not applying a canon were that the canon's conditions are unsatisfied. So, for example, if a majority of the court were to find a criminal statute to be unambiguous, that would eliminate any scope for the rule of lenity.

But one of this study's striking findings is that opinions in the Roberts Court offered a far wider array of reasons for not applying canons. For example, consider the variety of reasons given for declining to apply the top-scoring expressio unius canon, which requires drawing significance from a textual omission: the omission was of "scant practical significance"; (172) the canon yielded to "text and structure"; (173) the alleged omission should simply be understood as statutory "silen[ce]," with no particular implication; (174) the argument that meaning must be drawn from the omission was simply "unpersuasive"; (175) the canon yielded to the presumption of reviewability (176) or to consistency with interpretation of similar statutory language elsewhere; (177) or the canon yielded to the context of a norm of preexisting authority or perhaps a presumption against implied repeal, specifically: "the [statute's] acknowledgment of ... certain pre-existing authority ... in no way implies a repeal of other pre-existing authority." (178) One final opinion declining to apply the canon is worth highlighting: Justice Thomas, for a majority of the Court, seemed to transform expressio unius right back into an interpretive standard for discerning statutory meaning, rather than a rule: a guideline that can be readily overcome by "contrary indications that [Congress's] adopting a particular rule or statute was probably not meant to signal any exclusion." (179)

Reasons for not applying other canons were similarly variable. Consider the common law canon. Opinions declining to apply that canon gave a wide range of reasons: the lack of statutory ambiguity; (180) the presence of another applicable canon, such as the rule against surplusage; (181) linguistic context; (182) the prospect that the statute was not meant to cover an area formerly governed by the common law; (183) the Court's view (without extensive explanation) that the statute simply changed the common law; (184) and the involvement of Native American tribes. (185)

Reasons for declining to apply the rule against surplusage included that it was possible, even tolerable, for statutes to contain surplusage; (186) that excess language could be "clarifying"; (187) that purported surplusage had the "hallmarks of a compromise" among varying congressional factions; (188) the purported surplus language was "of no moment"; (189) that statutory surplusage is particularly tolerable when it concerns a presumption, burden of persuasion, or proof standard; (190) that a surplusage-avoiding reading still was accompanied by some surplusage elsewhere; (191) that a "small amount of additional work for the words to do" could justify ignoring broader implications of the canon; (192) that surplus language could represent a "safety valve" in case of future legislation; (193) or that surplus language could be appropriately present to "remove any doubt" on the issue. (194)

In a single 2016 decision, Lockhart v. United States, Justice Sotomayor, for a majority, catalogued multiple reasons why the rule of the last antecedent might not apply, including context (in general), the lack of a good reason not to apply the modifying clause to the whole list, and having to swallow "unlikely premises," if the rule did apply. (195) Though the majority did not accept her interpretive conclusion, Justice Kagan, in dissent, offered two more: that the rule of the last antecedent might not apply when the modifying clause appears "at the end of a single, integrated list," and that the rule could be trumped (as she concluded it was) by a different canon, the series qualifier canon. (196)

Finally, consider reasons that majority opinions offered to decline application of the presumption against preemption of state law: Federal language was sufficiently clear to preempt state law, leaving no ambiguity for the canon to resolve; (197) the state law did "not conflict with the federal scheme"; (198) the "presumption carries less force ... than in other contexts because [the federal statute] ... simply denies plaintiffs the right to use the class-action device to vindicate certain claims"; (199) and the presumption carries less force in elections cases. (200)

Of course, interpretive canons can have exceptions. But for canons to function as rules, we must know those exceptions. In these cases, the justices seem to elbow canons aside with slippery, unpredictable justifications.

E. The Court Has Birthed Multiple New Canons, Modified Old Ones, and Killed Some Off Altogether.

Despite its historic pedigree, including a mention in Blackstone's Commentaries, (201) the canon "Remedial statutes shall be liberally construed" seems to have little, if any, meaningful application nowadays. Justice Kennedy, writing for himself and six other justices in 2014, specifically disapproved the Court of Appeals' application of the canon: "After all, almost every statute might be described as remedial in the sense that all statutes are designed to remedy some problem ... [and] 'no legislation pursues its purposes at all costs.'" (202) The last recorded application of the canon in this study (203) was in 2011, in the majority opinion in Kasten v. Saint-Gobain Performance Plastics Corp., interpreting the Fair Labor Standards Act broadly. (204) Despite their supposed pedigrees, the and/or canon and the punctuation canon also appear to be on the way out. (205)

But that does not mean that the Court is shrinking its interpretive canon list. Instead, it has created a significant number of new ones. Abbe Gluck has already pointed out that two now well-established canons (or presumptions) were "invented" by the Court in the twentieth century: the presumption against preemption of state law and the Chevron doctrine of deference to agency interpretations. (206)

But this study documents several more. Some of these new canons or modifications of existing canons may predate the Roberts Court, but all are recently devised--within the last few decades at the earliest. None is listed in Scalia and Garner's catalogue; research has not located them in the seven-volume Sutherland treatise on statutory construction either. To make this list, the purported canon had to take a rule-like form--to be articulated as an interpretive principle applicable across a range of statutory settings--and had to have been applied repeatedly.

1. New "No Elephants in Mouseholes" Canon

Others have also noted this new canon, appealingly named by Justice Scalia. (207) It presumes that Congress does not bury important concepts in obscure provisions or phrases, instructing courts to interpret those phrases narrowly to exclude the important concept. In its current form, this canon originated in the Supreme Court's 2001 decision in Whitman v. American Trucking Associations, Inc., when the Court declined to interpret the phrase "requisite to protect the public health" in the Clean Air Act national ambient air quality standards provisions to authorize the Environmental Protection Agency to consider cost in setting those standards. (208)

While it was not one of the top scorers in the data, the canon appears to have staying power. The canon was discussed in 2% of statutory issues over the ten-year period, more frequently than either ejusdem generis or the rule of the last antecedent. (209) Meanwhile, opinions that considered the canon applied it to resolve statutory issues 89% of the time. (210) Because the canon is a relatively new addition to canonical discourse, opinions not applying it might simply ignore it rather than supply reasons for not applying it. If so, we may see the approval rate drop over time. But it is also worth noting that the canon was engaged during every October Term analyzed but one. (211)

2. New Location of Codification Canon

The justices have several times in majority as well as other opinions applied this canon drawing inferences from a provision's codified location, either within the statute or within the U.S. Code. (212) For example, in Yates v. United States, Justice Ginsburg, for a plurality, reasoned that a statute prohibiting destruction of objects was codified "together with specialized provisions expressly aimed at corporate fraud and financial audits" and consequently should be interpreted not to cover the destruction of fish. (213) Justice Stevens reasoned similarly in a concurring opinion in a patent case, arguing that Congress's placement of the relevant provision in the portion of the U.S. Code title governing protection of patent rights, rather than the portion governing patentability, indicated that the provision did not amend limitations on patent-eligible subject matter. (214) Of course, as with the remainder of the canons, it has not been applied in all opinions in which it has been considered, (215) and, unusual for interpretive canons, its existence provoked an objection from Justice Kagan, dissenting in Yates. (216)

This canon may be especially difficult to defend as approximating congressional drafting practices, because at least some final decisions around codifying legislation into the U.S. Code, including "improv[ing]" the law's organizational structure and clarifying and correcting the legislation, are made after enactment by the U.S. House's Office of Law Revision Counsel. (217)

3. New Veterans' Benefits Canon

In unanimously holding that a 120-day deadline on filing appeals to Veterans Court was not jurisdictional, the Court stated in Henderson v. Shinseki in 2011, "[w]e have long applied 'the canon that provisions for benefits to members of the Armed Services are to be construed in the beneficiaries' favor.'" (218) Although this canon has been used infrequently, the Henderson Court nonetheless termed it a canon and referenced earlier case law applying it. The canon appears to date from the World War II era, though the cases cited for the existence of the veterans' benefit canon do not describe it in that fashion. In Fishgold v. Sullivan Drydock & Repair Corp., a 1946 case, for example, the Court made the following comment regarding the Soldiers' and Sailors' Relief Act of 1940: "This legislation is to be liberally construed for the benefit of those who left private life to serve their country in its hour of great need." (219) The reference to "this legislation" suggests that the Court might have been engaged in purposive interpretation or even applying the remedial purposes canon. (220) By Henderson v. Shinseki, however, the Court had developed the statement into a specifically veteran-focused canon. (221) It does not appear to be mentioned in the Scalia and Garner book or in the Sutherland treatise beyond a single citation to a 1994 case that the treatise includes in a listing of cases on "remedial legislation." (222)

4. New Jurisdictional Rules Canon

Beginning in 2006, the Court developed the following "clear statement" principle regarding whether statutory limitations should be construed as jurisdictional: "[A] rule is jurisdictional '[i]f the Legislature clearly states that a threshold limitation on a statute's scope shall count as jurisdictional.'" (223) Justice Ginsburg explained: "But when Congress does not [clearly] rank a statutory limitation on coverage as jurisdictional, courts should treat the restriction as nonjurisdictional in character." (224) She then applied that "readily administrable bright line" to the case. (225) Since 2006, the jurisdictional rules canon has been applied repeatedly, including in bankruptcy, veterans' benefits, discrimination, and Federal Tort Claims Act cases. (226) We did not locate any opinions that discussed but did not apply the canon. Nonetheless, as with the no elephants in mouseholes canon, the approval rate may conceivably drop over time as the canon ages. (227)

Research could not locate instances of this canon prior to 2006, and it is mentioned neither in the Sutherland treatise nor in the Scalia and Garner book. (228)

The Court did observe in a 2004 decision that courts have been "less than meticulous" in distinguishing prescriptions that "delineat[e] the classes of cases (subject matter jurisdiction) and the persons (personal jurisdiction) falling within a court's adjudicatory authority" from "claim-processing rules," calling them all jurisdictional. (229) This may have prompted the development of the clear statement rule. Notably, the rule requires clear indications from Congress that a particular statutory provision is meant to have jurisdictional consequences, but the Court nonetheless applied it to statutes enacted prior to the canon's development, when Congress could not have anticipated the canon's application.

5. Rule of Lenity Modifications

The Roberts Court appears to be modifying the conditions that trigger the rule of lenity. The Scalia and Garner book describes the rule of lenity this way: "Ambiguity in a statute defining a crime or imposing a penalty should be resolved in the defendant's favor." (230) Roberts Court majority opinions during the study period did indeed sometimes apply that version of the rule, refusing to use the canon when the statute contained "no ambiguity." (231) But in other opinions, majorities declined to apply the rule of lenity because the statute was not "grievous[ly] ambigu[ous]." (232) In addition, some opinions suggested that ambiguity (to which lenity would then be applied) was to be resolved with respect to "ordinary, accepted meaning." (233) But other majority court opinions held that ambiguity was to be determined only after other canons had been applied. (234) If the Court solidifies its jurisprudence, requiring "grievous ambiguity," determined only after use of other canons are applied, it will, of course, limit the rule of lenity's scope. (235) The point, however, is that the Roberts Court seems to be shifting the rule's boundaries.

On the other hand, the Court appears to have expanded the rule to the immigration setting. In Moncrieffe v. Holder, the Court stated, "[A]mbiguity in criminal statutes referenced by the [Immigration and Nationality Act] must be construed in the noncitizen's favor."236 In that case, it was addressing whether a Georgia conviction for possession of marijuana with intent to distribute qualified as an "aggravated felony" under the INA, making the noncitizen both deportable and ineligible for discretionary relief. (237) Using this apparently modified rule of lenity, it interpreted the federal statute in favor of the alien. (238) The Court has not so far explained the expansion, but perhaps this approach conceptualizes deportation as akin to a criminal punishment; the rule of lenity applies not only to "the substantive ambit of criminal prohibitions, but also to the penalties they impose." (239)

6. Expansion of Expressio Unius Beyond the List Context

Professor Richard Pildes wrote in his student note that after decades of disuse, the Court revived the "ancient maxim" of expressio unius in 1974. (240) "Decades of disuse" followed by a canon reappearance alone might suffice to undermine the consistency needed for a canon to form part of a stable interpretive background for Congress. But the Court has also expanded its application.

The Sutherland treatise describes expressio unius as instructing that "where a statute designates a form of conduct, the manner of its performance and operation, and the persons and things to which it refers, courts should infer that all omissions were intentional exclusions." (241) The core application of expressio unius had been to statutory lists "when the items expressed are members of an 'associated group or series.'" (242) But the Court has expanded it in numerous cases by drawing an inference of deliberate omission from the inclusion of particular language in one part of a statute and its nonappearance elsewhere. (243) The Court has even expanded it to compare the presence and absence of particular language across the U.S. Code. (244) For example, in Carr v. United States, the Court compared the use of the present tense for sex offender registration requirements with "numerous federal statutes us[ing] the past-perfect tense ... when coverage of preenactment events is intended.... The absence of similar phrasing here provides powerful evidence that [the statute] targets only postenactment travel." (245)

Finally, during the Roberts Court, the Court applied expressio unius to a comparison of statutory language with common law norms, rather than other statutory language. In Bruesewitz v. Wyeth LLC, in considering the relationship of federal vaccine regulation to state tort law, the Court addressed a limitation on manufacturer liability dependent on the following condition: "the injury or death resulted from side effects that were unavoidable even though the vaccine was properly prepared and accompanied by proper directions and warnings." (246) The majority compared the statutory language to state tort law: "Products-liability law establishes a classic and well known triumvirate of grounds for liability: defective manufacture, inadequate directions or warnings, and defective design." (247) Ultimately, it reasoned that the liability-protecting clause's lack of mention of "design-defect liability" was by choice, not inadvertence, stating: "Expressio unius, exclusio alterius." (248)

7. Whole Session Laws Rule

We found that the opinions in the Court considered the whole code rule in over 11% of analyses of disputed statutory issues, and applied it with approval 74% of the time. (249) Scalia and Garner describe the whole code and whole act rules, collectively, as embodying a "presumption of consistent usage" by the legislature. (250) Regardless of whether this assumption is realistic for a single statute, presuming consistent usage certainly seems a stretch for the entire U.S. Code, which includes both year-old and centuries-old legislation on an enormous range of topics and within the jurisdiction of a wide array of House and Senate committees.

Perhaps recognizing these limitations, though without taking on the whole code rule outright, a number of opinions in the Roberts Court reasoned that a presumption of consistent usage was most appropriate when statutory provisions were enacted during the same congressional session. We termed this the whole session laws rule. For example, in the 2012 decision in Mohamad v. Palestinian Authority, the Court compared the Torture Victim Protection Act's (TVPA's) imposition of liability for torture on an "individual" to other statutes distinguishing between an "individual" and an "entity," concluding that the Act imposed liability only on natural persons. (251) But key to the Court's analysis was its point that "the very same Congress that enacted the TVPA also established a cause of action for" terrorism-related injuries against any "individual or entity." (252) And in the 2008 decision in Gomez-Perez v. Potter, (253) a majority of the Court rejected the application of the whole code canon on the ground that the relevant laws were not enacted closely in time. The majority interpreted the federal sector provisions of the Age Discrimination in Employment Act to authorize retaliation claims, refusing to interpret silence as a deliberate omission even though retaliation claims were specifically authorized by the private sector provisions. (254) The opinion commented that such negative implications were " 'strongest' in those instances in which the relevant statutory provisions were 'considered simultaneously when the language raising the implication was inserted,'" but that, in this case, the relevant statutes were enacted seven years apart. (255)

Even with the development of a whole session laws rule, the whole code rule continues to be applied. For example, in the 2006 majority opinion in Buckeye Check Cashing, Inc. v. Cardegna, Justice Scalia confirmed an interpretation of the word "contract" in the 1947 Federal Arbitration Act to include putative agreements, whether or not legal, by referring to the use of the term in the 1890 Sherman Antitrust Act, commenting: "Our more natural reading is confirmed by the use of the word 'contract' elsewhere in the United States Code...." (256)

8. Continued Expansion of Federalism Canons

Professors William Eskridge and Philip Frickey in particular have noted the Court's creation of federalism canons extending beyond the presumption against state law preemption. (257) In the early 1990s, they noted the clear statement requirement for conditions on federal grants (258) and what they termed a "super-strong" rule against waiver of state sovereign immunity from federal suit and against federal regulation of core state functions, (259) among others.

The Roberts Court has continued to broaden the federalism canons, announcing that federal statutes will be narrowly interpreted if they intrude via overlapping jurisdiction into areas of "traditional" state regulatory authority. Pre-Roberts Court cases had mentioned judicial reluctance to assume that a statute changed the relation between federal and state criminal jurisdiction. (260) The Roberts Court expanded the list of state prerogatives deemed deserving of this canonical thumb on the scale to include land use regulation, (261) the development of contract and tort law, (262) the "independent judgment of state prosecutors," (263) and the "administration of a State's taxation scheme." (264)

9. Fledgling Canons

Two more examples potentially fit the canon mold, since they, too, could constitute rules applicable across a variety of statutes, but they have not yet been articulated sufficiently clearly or applied sufficiently frequently to qualify as a canon.

Presumption in Favor of Arbitrability (In development). In CompuCredit Corp. v. Greenwood, Justice Sotomayor, in a concurring opinion joined by Justice Kagan, noted that although the Credit Repair Organizations Act authorized a "right to sue," which could be understood as a right to sue in court, valid arbitration agreements remained enforceable. (265) She articulated her reasoning in canonical form: "[W]e resolve doubts [regarding whether Congress disallowed arbitration] in favor of arbitration." (266) Sotomayor cited a 1991 case, Gilmer v. Interstate/Johnson Lane Corp., which focused not on any explicit textual requirements, but on the federal policy favoring arbitration embodied in the Federal Arbitration Act. (267) Other cases have drawn from Gilmer to apply arbitration-favoring interpretive rules to contracts and to find state law preempted notwithstanding the presumption against preemption. (268)

Canons of Continuity/Antichange (In development). Several opinions have reasoned that courts should disfavor interpreting statutes to work changes from the status quo. This sort of interpretive move is closely linked to the common law canon, also effectively status-quo preserving. But in these opinions, statutes or even administrative practice define the content of the status quo. In a 2011 patent case, for example, Board of Trustees of the Leland Stanford Junior University v. Roche Molecular Systems, Inc., the Court ruled on a complex patent ownership dispute, rejecting a particular reading in part because it "assumes that Congress subtly set aside two centuries of patent law in a statutory definition." (269) In a 2015 case, Michigan v. EPA, the Court declined to read general statutory language as authorizing the agency not to consider cost in deciding whether to regulate power plant mercury emissions. (270) Among the Court's arguments for rejecting deference to the agency's interpretation was that "[a]gencies have long treated cost as a centrally relevant factor when deciding whether to regulate." (271) The majority found EPA's statutory interpretation unreasonable "[a]gainst the backdrop of this established administrative practice." (272) And in a 2012 case also involving agency deference, Scialabba v. Cuellar de Osorio, the Court upheld an immigration agency interpretation as consistent with longstanding practices, notwithstanding a later enactment that amended the relevant statutory language. (273)

If it evolves and becomes embedded, this canon--a presumption that generally phrased statutory language does not disrupt a long-settled legal norm--could be understood as linked not only to the common law canon, but to an interpretive move that had been associated with legislative history use. This was the notion that if legislation resolved an important issue, one would expect to see discussion of it in legislative deliberations. Silence would imply no intended change. This was the "dog that did not bark" canon. (274) For example, in Zuni Public School District No. 89 v. Department of Education, a 2007 case, Justice Breyer relied on silence in the legislative history as support for an agency interpretation of new school-funding legislation as working no change in longstanding practices. (275) He noted that the agency had prepared the draft legislation, Congress had amended the language without comment, and "[n]o one at the time ... expressed the view that this statutory language ... was intended to require ... the Secretary to change the Department's system of calculation [which had been] followed for nearly 20 years." (276) That interpretive move was repeatedly criticized, particularly by Justice Scalia, who dissented in Zuni and commented colorfully elsewhere: "We are here to apply the statute, not legislative history, and certainly not the absence of legislative history. Statutes are the law though sleeping dogs lie." (277) (Justice Scalia nonetheless joined a concurrence in Scialabba, agreeing with the majority that the agency interpretation at hand was reasonable as "consistent with ... established meaning" notwithstanding statutory change.) (278) In any event, the point here is not to criticize or defend the interpretive move but to note its connection to a possibly developing canon of continuity.


A. The Stable Interpretive Background Justification

Even if impressionistic, this Article's findings nonetheless raise significant questions about whether canons improve interpretive predictability, constrain judicial discretion, or supply a stable interpretive background for Congress. First, interpretive canons, as a group, are in flux. The Roberts Court has added to the list of canons in use, modified existing ones, and apparently withdrawn others. (279) There is no "closed set" of canons.

Conceivably, the pace of change among canons might not be so rapid as to undermine stability altogether. Indeed, if canons were justified as "entrenched generalizations" about congressional intent, (280) one might imagine that the Court would have to modify the list in use as congressional practices evolve. But as discussed, the congress-approximating justification is hard to sustain, and the Court generally does not investigate actual congressional practices. (281) If anything, apparent evolution in the group of canons in use suggests ad hoc canon development rather than stability.

Also in sharp tension with the claim that canons are rule-like: the unpredictable and seemingly limitless catalogue of reasons not to apply canons. Opinions declined to apply canons because their conditions were not met, but also owing to context, other canons, and an assortment of other reasons that are hard to wrestle to the ground. Indeed, this study could also understate the range of grounds available to defeat a particular canon, since we cannot observe the reasons an opinion declined even to discuss a particular canon. (282) Opinions were also vague regarding how strong such a countervailing reason must be to defeat a canon. For example, sometimes expressio unius's implications were sufficient to defeat the rule against surplusage and sometimes not. Cases came out both ways. (283)

Moreover, some canons lack clear criteria for application, making them inherently more standard-like than rule-like. Manning has made this point about the rule against absurdity, arguing that textualists should discard the canon in part because it is "too broad and unintelligible to give either legislators or the public a realistic basis on which to evaluate the specific outcomes." (284) Nonetheless, that canon has made the "most applied" list for the Roberts Court. (285) The newly developed major question canon similarly calls on courts to identify whether a question meets the vague standard of "economic and political significance," (286) while the no elephants in mouseholes canon requires a judge to decide whether an issue is sufficiently elephantine. Even superficially rule-like canons lose their rule-like quality when the conditions that trigger them are indistinct: how ambiguous a statute must be to trigger the rule of lenity; how clear a statute must be to overcome a clear statement rule; and so on.

All these factors may explain the patchwork of uneven canon application observed in this study. As noted, the "Top Five" textual canons and the common law canon were applied to resolve the statutory issues 70-76% of the time they were discussed. (287) But others were applied to resolve issues significantly less often.

Moreover, these less frequently applied canons included some that we might have expected as part of a "stable interpretive background" for Congress, either because of apparent congressional awareness or because of their historical pedigree. These included the rule of lenity, agency deference canons, and the presumption against preemption of state law. These latter three were applied 48-57% of the time an opinion considered them. (288)

So, can we consider any of this objectively "predictable"? Certainly not a purported rule with poorly specified conditions for application--or a canon with an approval rate hovering between 45% and 60%. A 75% approval rate might strike some as predictable--but we should deem it predictable only if such a canon is the sole potential interpretive tool in a case and if the reason for not applying it was that its conditions were unmet, a prospect that the data calls into question.

Cases with multiple canons fare even worse. Scholars have already sharply criticized the lack of explicit hierarchy among the canons. (289) And interpretive canon use in the Roberts Court has given us few, if any, hierarchy hints. For example, in some majority opinions considering both the federalism canon and rule against surplusage, the federalism canon was applied with approval and the rule against surplusage was disregarded, (290) and in other opinions, it was the opposite. (291) As noted, sometimes the rule against surplus-age was applied in preference to expressio unius, and sometimes the reverse. (292) Even so-called clear statement rules, which one might expect to be trumped only by clear statutory language (and not, for example, by other canons), were not obviously at the top of the canon heap. (293) For example, in an analysis of whether a statute was clear enough to avoid the presumption against sovereign immunity waiver, the Court first applied textual canons. The Court explained: "The sovereign immunity canon is just that--a canon of construction.... a tool for interpreting the law, and we have never held that it displaces the other traditional tools of statutory construction." (294) Similarly, in applying the Chevron doctrine, the Court now more often applies textual and at least some substantive canons in determining whether Congress has directly answered the precise interpretive question, making a finding of ambiguity, under which Chevron requires deference to a reasonable agency interpretation, less likely. (295) And it is unclear whether the rule of lenity is applied ahead of or behind other textual canons. (296)

This project's data confirms not only the presence of the hierarchy concern but its significance, because in the clear majority of statutory issues in which any canon was considered, the opinion considered multiple canons. And the vast majority of Roberts Court opinions that considered multiple canons did not apply at least one of those canons. (297)

To illustrate the potential cost to predictability from multiple canons with varying implications for a statute's interpretation, consider a case in which both the relatively high-scoring rule against surplusage and expressio unius canon could apply. The rule against surplusage was applied with approval to resolve a statutory issue 70% of the time it was discussed in our study, and expressio unius applied 75% of the time it was discussed. (298) If the two canons are not related to one other, (299) simple probability theory tells us that the odds that both will be applied are roughly 52.5%; the odds that the rule against surplusage alone will be applied are 17.5%; the odds that expressio unius alone will be applied are 22.5%; and the odds that neither will be applied are 7.5%. If the canons are known to have clearly opposite implications for the case's outcome (thus ruling out the scenario where both canons could apply), then the odds that the rule against surplusage alone will be applied are roughly 37%; the odds that expressio unius alone will be applied are roughly 47%; and the odds that neither will apply are 16%. (300) None of this is consistent with any reasonable notion of "predictable." With three canons, the story only worsens.

Moreover, the hierarchy problem apparently extends to all interpretive tools, not just canon versus canon. Conceivably excepting plain statutory language or previous statutory precedent, (301) the Roberts Court has articulated no organized approach at all to interpretive tools. We have no clear sense of when the Court will turn to canons (302) or where they rank relative to dictionaries, statutory structure and function, more general linguistic context, or legislative context.

Consider the ordering difficulties in the Court's unanimous 2012 opinion in Freeman v. Quicken Loans, Inc. (303) That case involved a rather poorly written statutory provision stating, "No person shall give and no person shall accept any portion, split, or percentage of any charge made or received for the rendering of a [specified] real estate settlement service ... other than for services actually performed." (304) The question was whether a loan servicing company violated the statute by charging fees to consumers for nonexistent services--or whether the statute's reference to "portion, split or percentage" limited its application to an unearned fee only if the company had divided the fee with others, perhaps in exchange for a referral. (305) Meanwhile, the neighboring statutory subsection specifically prohibited any "fee, kickback, or thing of value" in connection with an agreement to refer business connected with a real estate settlement, (306) and the entire section was titled, "Prohibition against kickbacks and unearned fees." (307) Finally (simplifying slightly), the implementing agency had, in 2001, issued a policy statement interpreting the section to prohibit fees imposed by a single provider, the situation at issue in the case. (308)

Interpreting the statute not to cover the unearned fees the consumers paid in the case, the Court had to rationalize a surplusage difficulty, since interpreting the first provision to apply only to a side arrangement might make it surplusage, given the second provision's application to referrals. (309) (The Court was able to identify hypothetical, though likely rare, cases that each provision would reach uniquely. (310)) And despite the title canon, the Court declined to accept the natural implication of "Prohibition of kickbacks and unearned fees," which it did by narrowing the title's "unearned fees" language to refer only to a subset of "unearned fees," those shared with others. (311) The court refused deference to the agency interpretation as contrary to the statute's language. (312) These were only a few of the interpretive rules engaged in the opinion. The Court also applied the noscitur a sociis canon to interpret "portion" to share the meaning of "split" in the list of three nouns, in the process rejecting the consumers' argument that "portion" could encompass an entire payment. (313) This required disregarding a surplusage violation, since "portion" would then add nothing to "split." (314) The Court also considered multiple potential applications of the rule against absurdity, accepting some and rejecting others. (315) Ultimately, Justice Scalia, for the unanimous Court, concluded that the weight of the interpretive clues required the statute to be read to cover "only a settlement-service provider's splitting of a fee with one or more other persons; it cannot be understood to reach a single provider's retention of an unearned fee." (316)

This case might be seen as a difficult one on the language, which is why it reached the Supreme Court, or possibly a relatively easy one, since the Court's decision was unanimous. But given the range of potentially applicable canons pointing in different directions, invoking additional interpretive tools seems hardly to have increased predictability.

Heavy use of canons, including of canons that correspond only poorly to Congress's preferences and practices, raises yet one further concern. That is the prospect that judicial drawing from an increasingly lengthy canon list might detract from the weight placed on other, possibly better, interpretive sources.

Opinions reviewed for this study, including in decisions such as Freeman v. Quicken Loans, Inc., sometimes read as if they were considering a jumble of interpretive clues of unspecified weight, including statutory language, structure, context, canons, dictionaries, purpose, and other tools. Finding four clues pointing in one direction and perhaps three in another, the Court might go in the first direction. Canons add significantly to the pile. Less defensible canons could distract from or even crowd out interpretive clues that arguably could better help a court determine what Congress was communicating in particular statutory language--whether those are textual or statutory context, genuinely reliable canons, stated purpose, statutory structure or function, or conceivably legislative history evidence. (317)

In short, the data raise serious questions regarding whether interpretive canons amount to a "closed set," "coordinate" judicial decisionmaking, or constrain it. Further, although this question is beyond the scope of this study, the data raise concern regarding whether interpretive canons can be justified as more constraining or predictable in their use compared with other interpretive sources, such as legislative history.

Current patterns of canon application could be fodder for arguments that canons simply amount to window dressing. At the same time that textualists like Justice Scalia have claimed stability and predictability advantages for canons, they have acknowledged that canons may not help predict results, stating: "[A]s in any good mystery, different clues often point in different directions. It is a rare case in which each side does not appeal to a different canon to suggest its desired outcome. The skill of sound construction lies in ... deciding where the balance lies." (318) The data suggest, though, that such comments do not simply acknowledge that interpretive canons are imperfect--they give away the game.

If canons were to provide any semblance of interpretive stability, they would have to constitute not only a closed set, but a spartan one. Courts would have to articulate canons more clearly and deploy them more consistently, using an explicit hierarchy. (319) And even if we had a genuinely stable interpretive regime composed of a very small number of canons, the objection would remain that, given the niceties of the legislative process, we could not fairly impute to a statutory enactment acceptance by Congress of a particular judicially devised canon.

If canons cannot be systematically justified on these second-order grounds, as rule-like methods that increase interpretive predictability or constraint, what then? Is there any hope of redeeming canons on other grounds?

Conceivably, an individual canon deployed in a streamlined regime could still be justified on first-order grounds--as connected to congressional preferences or another defensible value. Perhaps a canon defender still could meet the burden of demonstrating that a particular canon is indeed an "entrenched generalization" (320) of congressional preferences or actual drafting practices, that it supplies an assurance that readers will have notice of legal requirements, or that it serves some other appropriate goal. (321) Or if the defense is that a canon serves as a "higher-order rule[]" of some sort, (322) such as bringing an extrastatutory value to bear in interpretation, further work is needed on this as well. (323) Of course, the canon must also function in a genuinely rule-like fashion; it must be well-defined, including limiting the conditions under which the canon will not apply and specifying its relationship to other interpretive tools. And it must be consistently applied.

That raises a further question of which issues warrant canons. (324) Why, for example, a rule of lenity or a presumption against preemption of state law, but not a presumption in favor of federal supremacy (given the Supremacy Clause), endangered species, or disabled individuals (perhaps given the likely disadvantages the latter two would face in the legislative process)? Why a punctuation canon, but not a verb tense canon? Implicit in the choice of canonical treatment for a particular issue is the notion that rule-based treatment of the statutory issue is worthwhile--that it will minimize error and decision costs as a court ascertains the meaning of particular statutory language. But that raises further questions: Would purported gains from predictability, judicial constraint, and decision-cost savings (all higher if a canon is consistently applied) outweigh the prospect that individual cases might be wrongly or unjustly decided? Of course, if a canon fails to approximate congressional preferences, that cost is substantially higher. For a particular issue, a more standards-based approach could be preferable. All these questions in turn raise yet-to-be-resolved issues about appropriate interpretive goals. (325)

B. A First Cut at the Content of Individual Canons in Use

From the standpoint of whether canons in use can be justified as approximating congressional practices or preferences, this study raises another significant concern. The data reveal a gulf between canons actually in use in the Roberts Court and Gluck and Bressman's findings regarding congressional staff knowledge and acceptance of canons. Three of the top five textual canons in this study, including the rule against surplusage, the whole act rule, and the whole code rule, were found by Gluck and Bressman to be rejected outright by congressional staff, because they are disconnected from drafting practices and because redundancy is a regular feature of drafting. (326) These are listed in the upper right-hand quadrant of the table below.

Congressional staff in that study were not asked about the harmonious reading canon, also in the top five in this study. That canon assumes that a statutory section should not be read to be redundant or contradictory. That assumption has greater intuitive plausibility than the rule against surplusage's assumption that Congress would not enact any redundant language whatsoever, but it is closely linked to the surplusage canon, which was rejected by staff. Evidence is also at best incomplete regarding expressio unius. Gluck and Bressman characterized expressio unius as received more positively by congressional staff. (327) But Gluck and Bressman's framing in their congressional staff study of expressio unius focused solely on statutory lists. (328) Meanwhile, the Roberts Court's use of that canon has been far broader, using it in contrasting one section of a statute with another, one statute with another, or even common law with statutory law. Finally, Gluck and Bressman apparently did not ask staff about either the common law canon or the rule against absurdity. These relatively heavily used canons, for which evidence in the Gluck and Bressman study is absent or ambiguous, are listed in the lower left-hand quadrant of the table.

On the other side, the most promising canons from the congressional perspective were only erratically applied by the Roberts Court. Many congressional staff told Gluck and Bressman they were aware of the federalism canons, the presumption against preemption, and agency deference canons, and they tried to draft with those canons in mind. (329) Gluck and Bressman accordingly suggested that the hypothesized "feedback" between the courts and Congress might be present in these settings. (330) Nonetheless, the Roberts Court applied these canons approvingly far more erratically than other interpretive canons--ranging from 48% of the time it was discussed for the presumption against preemption, worse than a coin-flip, to 62% of the time the canon was discussed for the federalism canons, and 53% for the agency deference canons. (331) These are listed in the lower right hand quadrant of the table below.

Only the two canons listed in the upper left-hand quadrant in the table seem to represent a good match between the canons in use in the Roberts Court and those that Gluck and Bressman identified as accepted by congressional staff. These were the two specifically list-focused canons of noscitur a sociis and ejusdem generis, applied with approval 79% and 73%, respectively, of the time they were discussed (in 3% and 1.6% of issues overall). (332) Gluck and Bressman found that congressional staff mostly agreed generally that "terms in a statutory list always or often relate to one another." (333) Some caution is still warranted, however, since staff did not identify the canons by name and were not asked about the precise contours of the rules they embody. (334)

The Court could, of course, justify a canon with reasons other than likely congressional preferences. But except for the infrequently used grammatical and syntax canons, described as required by ordinary English conventions, the Court has rarely offered any other clear rationales for particular canons. (336) More often, the Court seems to simply announce or apply the canon. The erratically applied Chevron doctrine might be a counterexample, because the Court announced its reasons for imputing to Congress the implicit delegation of interpretive authority to agencies. (337) And in applying the new jurisdictional rules canon, the Court explained that interpreting statutes as jurisdictional "alters the normal operation of our adversarial system." (338) By contrast, however, the Court's declaration of the "no elephants in mouseholes" canon was just that: "Congress, we have held, does not alter the fundamental details of a regulatory scheme in vague terms or ancillary provisions--it does not, one might say, hide elephants in mouseholes." (339)


The Court has been devising new interpretive canons, abandoning old ones, and deciding whether to deploy them based on reasons that appear difficult to pin down. Meanwhile, the abundance of canons undermines the prospect of a reliable interpretive hierarchy. It seems hard to conclude that canons usefully coordinate judicial decisionmaking or render statutory interpretation more predictable for Congress or others; as presently conducted, canon use does not seem a useful solution to the bugaboo of judicial discretion in interpretation. A bounded, stripped-down set of canons with clearly defined conditions, consistently applied, could be an improvement, but even then, their use might conflict with legislative supremacy because we cannot readily impute congressional acceptance of these judicially devised canons.

Instead of relying on these generalized, second-order justifications for canons, we might be better off tackling the choice and design of these interpretive tools in a more fine-grained fashion. A more promising option might be to use a particular canon only if its content can be justified on first-order grounds--that is, if it actually approximates congressional preferences or serves an important external value. On these criteria, some of the most popular canons--and here I nominate the whole act, whole code, and surplusage canons--should be abandoned outright. (340) Gluck and Bressman's findings have cast substantial doubt on whether they approximate congressional preferences, and given how demanding these canons are for ordinary readers to apply, they surely do not support fair notice either. The same may be true at least of the newly expanded uses of expressio unius.

Meanwhile, other canons, including the agency deference canons, noscitur a sociis, ejusdem generis, the constitutional avoidance canon, and the new jurisdictional rules canon, might be individually defensible on these first-order criteria. But that may not be sufficient to legitimate canon use. Pronounced unevenness in use, especially when it comes to agency deference and constitutional avoidance, undercuts their value as rules, including any predictability they might contribute to interpretation.

All this raises an important related question: Which institutions should assess and devise, or else discourage the use of, particular canons? Gluck as well as Baude and Sachs have suggested that canons can be understood as a form of federal common law. (341) If so, stare decisis effect might be given to approved canons. (342)

Such a view of canons also potentially implies that legislatures could readily supersede them. Nicholas Rosenkranz has specifically argued that judicial power over interpretive methodology is "common lawmaking power, which may be trumped by Congress." (343) Gluck has pointed out that Congress does occasionally legislate rules of interpretation, though these are mostly limited to savings or preemption clauses that address whether a statute impliedly repeals another federal statute or preempts state law. (344) Congress has not, so far, attempted to enact a full-blown set of interpretive rules. The Dictionary Act, with default definitions of "person," "company," and so forth, for the entire U.S. Code, is an exception, though Congress has provided that the definitions can easily be overcome if "the context indicates otherwise." (345)

Meanwhile, as Gluck's research also documents, state legislative attempts to create interpretive "codes" have been ignored or outright resisted by the supreme courts in those states, (346) an outcome one would not expect if interpretive rules were widely understood to be ordinary common law. This may be because canons, though resembling common law in some ways, also seem closely linked the core function of "say[ing] what the law is." (347) Figuring out--or deciding--how to approach statutory language seems related to inherent judicial functions. (348) That could in part explain Gluck's findings.

We should consider an alternate understanding of canons. Even as they resemble common law, and even as they must be reconciled with legislative supremacy, they also guide interpretation by functioning as source-selecting, burden-allocating rules. For example, canons might prompt an interpreting court to consider and to weigh agency interpretations, word usage outside the statute at hand, values external to that statute, drafting practices (real or assumed), or even legislative history. To that extent, canons resemble evidentiary or procedural rules guiding adjudication. For example, a court's assessment of the implications of word use elsewhere in the U.S. Code has some resemblance to a fact-finding court's assessment of how much to weigh tangentially relevant information.

If canons are seen this way, the power to devise interpretive rules might be seen as integrally linked to the judiciary's expertise, obligation, and responsibility to determine statutory meaning as judges apply the law to a particular dispute. (349) This suggests that we should not assume that Congress can trump canons if it so chooses. At some point, legislative dictation of interpretive approaches may interfere with the independence and inherent functions of the Article III judiciary. (350)

At a more practical level, devising a canon's content--much like constructing an evidentiary or procedural rule--necessitates considering issues of value and policy extending far beyond the individual case or statute. Key issues include whether creation of a "rule" is justified at all; which interpretive rules, as a general matter, can usefully assist an interpreting court in resolving a dispute over statutory meaning; whether a particular canon can be justified as effectuating congressional preferences or specified outside norms; and in what order particular interpretive rules should be applied. It is not obvious that adjudication and the adversary process, focused as they are on individual cases and particular statutes, provide the best way to resolve these issues. We may need more systematic deliberation, coupled with a deeper investigation of particular interpretive rules. This could take place within the judiciary, but legislative deliberation also could be valuable. Legislative deliberation also could have the salutary effect of raising legislative awareness of canon content. (351)

Full development of such a proposal is beyond this Article's scope, but as we consider where issues around canons might best be resolved, we might also put on the table a compromise congressional-judicial approach resembling our process for procedural or evidentiary rules. (352)

In short, canon defenders have not yet succeeded in justifying judicial deployment of interpretive canons, collectively or individually, as consistent with legislative supremacy. Wherever further deliberation on the appropriate use of canons takes place, however, heavy canon use on the Court makes one thing clear: it is vitally needed.


                                                   Approval Rate
                             Percentage            of times canon
                            of Issues in         discussed in which
                            Which Canon          it was ultimately
                           Was Discussed      applied with approval to
Canon                    (any opinion type)   resolve statutory issue)

More Frequently Considered Canons

Expressio Unius                18.6%                    75%

Whole Act Rule                 15.2%                    76%

Rule Against                   13.2%                    70%

Whole Code Rule                11.5%                    75%

Agency                         11.9%                    53%
Deference Canons

Harmonious Readng              11.8%                    75%

Statutes In                    11.6%                    72%
Derogation of the
Common Law Shall Be
Strictly Construed
or "Common Law Canon"

Rule Against                    7.4%                    67%

Constitutional                  6.8%                    57%
Avoidance Canon

Legislative                     5.9%                    71%

Less Frequently Considered Canons

Federalism Canon                5.1%                    62%

In Pari Materia                 4.6%                    63%

Rule of Lenity                  3.3%                    55%

Noscitur a Sociis               3.0%                    79%

May/Shall Rule                  2.9%                    85%

Presumption Against             2.5%                    48%

Tiles Canon                     2.3%                    67%

Specific Riles                  2.3%                    67%
Govern over
General Ones

No Elephants in                 1.9%                    89%

Presumption Against             1.7%                    69%
Implied Repeal

Ejusdem Generis                 1.6%                    73%

Presumption Against             1.3%                    33%

Presumption Against             1.2%                    45%
Waiver of Sovereign

And/Or Rile                     1.1%                    90%

Rule of Last                    1.0%                    56%

Jurisdictional                  0.9%                    100%
Rules Canon

Presumption Against             0.8%                    71%

Remedial Statutes               0.6%                    50%
Shall Be Liberally

Whole Session                   0.5%                    80%
Laws Rule

Punctuation Canon               0.3%                    33%

Major Question Canon            0.3%                    100%

Native                          0.2%                    100%
American/Indian Canon


                                  Percentage          Percentage
                                 of Contested        of Contested
                               Statutory Issues,   Statutory Issues,
Overall Canon Use Data           All Opinions      Majority Opinions

At least one textual                 48.2%               55.9%
canon discussed

At least one textual canon           39.5%               45.7%
applied with approval

At least one substantive             44.9%               44.5%
canon discussed

At least one substantive             32.2%               31.7%
canon applied with approval

At least one substantive             36.8%               37.8%
canon discussed, excluding
agency deference canons

At least one substantive             27.3%               27.4%
canon applied with approval,
excluding agency deference

At least one canon                   67.1%               70.1%
of any sort discussed

At least one canon of any            57.9%               61.4%
sort applied with approval


Justice     Percentage of        Percentage of
            Issues in Authored   Issues in Authored
            Opinions with at     Opinions with at
            Least One Canon      Least One Canon
            Discussed            Applied with

Alito              72%                  64%
Breyer             62%                  52%
Ginsburg           59%                  54%
Kaqan              59%                  57%
Kennedy            70%                  58%
O'Connor           50%                  50%
Roberts            80%                  70%
Scalia             76%                  64%
Sotomayor          61%                  61%
Souter             71%                  60%
Stevens            63%                  52%
Thomas             64%                  50%

Justice     Percentage of        Percentage of
            Issues in Authored   Issues in Authored
            Majority Opinions    Majority Opinions
            with at Least One    with at Least One
            Canon Discussed      Canon Applied with

Alito              72%                  64%
Breyer             60%                  57%
Ginsburg           67%                  62%
Kaqan              63%                  60%
Kennedy            67%                  52%
O'Connor           50%                  50%
Roberts            80%                  69%
Scalia             78%                  67%
Sotomayor          72%                  72%
Souter             83%                  70%
Stevens            67%                  57%
Thomas             69%                  56%


                              Statutory Issues
                            in Majority Opinions

Number of             Number of Issues   Percentage of
Canons Discussed                         Total Issues
(including both
textual and
substantive canons)

0                           152              29.9%
1                           118              23.2%
2                           103              20.3%
3                            66              13.0%
4                            37              7.3%
5                            20              3.9%
6                            9               1.8%
7                            1               0.2%
8                            2               0.4%
9                            0               0.0%
Total Statutory             508              100%

                               Statutory Issues
                               in All Opinions

Number of             Number of Issues   Percentage of
Canons Discussed                         Total Issues
(including both
textual and
substantive canons)

0                           306              32.9%
1                           252              27.1%
2                           151              16.2%
3                           105              11.3%
4                            64              6.9%
5                            30              3.2%
6                            15              1.6%
7                            4               0.4%
8                            3               0.3%
9                            1               0.1%
Total Statutory             931              100%


                         Statutory Issues in Majority Opinions

Number of                Mean Number of    Percentage Appied
Canons Discussed         Canons Applied    (Mean Number of
(Both Textual/Syntax     with Approval     Canons Applied with
Canons and Substantive                     Approval divided by
Canons)                                    Number of Canons

0                             N/A
1                             0.76                76.3%
2                             1.36                68.0%
3                             2.03                67.7%
4                             2.92                73.0%
5                             3.15                63.0%
6                             3.89                64.8%
7                             7.00               100.0%
8                             3.00                37.5%
9                             N/A

                         Statutory Issues in All Opinions

Number of                Mean Number of   Percentage Applied
Canons Discussed         Canons Applied   (Mean Number of
(Both Textual/Syntax     with Approval    Canons Applied with
Canons and Substantive                    Approval divided by
Canons)                                   Number of Canons

0                             N/A
1                             0.75               75.0%
2                             1.38               69.2%
3                             1.99               66.3%
4                             2.95               73.8%
5                             3.13               62.7%
6                             4.07               67.8%
7                             4.75               67.9%
8                             4.00               50.0%
9                             6.00               66.7%

Nina A. Mendelson, Joseph L. Sax Collegiate Professor of Law, University of Michigan Law School. I appreciate the very helpful comments and feedback from Maureen Carroll, Ed Cooper, the late John DiNardo, Abbe Gluck, Don Herzog, Sharon Jacobs, Riyaz Kanji, Anita Krishnakumar, Jessica Litman, John Manning, Martha Minow, Julian Mortenson, Nicholson Price, Intisar Rabb, Daphna Renan, Margo Schlanger, Jonathan Siegel, Peter Strauss, and participants in the University of Michigan Law School Governance Lunch, the Harvard Law School Public Law Workshop, and the University of Colorado Law School Faculty Colloquium. Thanks are due to the terrific research assistants and programmers who devoted so many hours to this project, and whose work and comments were essential to it: Brian Apel, Lauren Babst, Michael Bloem, Erin Chapman, Joe Celentino, Lauren Dayton, Mariel Eben, Jarrett Gross, Simmon Kim, Claire Lally, J. Parker Lee, Collin Metcalf, Emma Notis-McConarty, Julie Siegel, and Anne Yearwood. Thanks to the Cook Fund at the University of Michigan Law School for generous research support. This work is licensed under the Creative Commons Attribution-NonCommercial 4.0 International License. To view a copy of this license, visit by-nc/4.0/[].

(1.) Approximately 70% of statutory issues addressed in majority opinions were resolved after considering one or more canons; roughly 67% of statutory issues addressed in all opinions were resolved after considering one or more interpretive canons. See infra Section III.A. All data cited is on file with author unless otherwise noted.

(2.) See, e.g., James J. Brudney & Corey Ditslear, Canons of Construction and the Elusive Quest for Neutral Reasoning, 58 VAND. L. REV. 1, 34-35 (2005) ("[A] majority opinion written in the Rehnquist Court era [on workplace issues] was twice as likely to rely on language canons as one authored in the Burger Court years."); id. at 30 tbl.l (also tracking a very substantial increase in use in substantive canons).

(3.) Harvard Law School, The Scalia Lecture | A Dialogue with Justice Elena Kagan on the Reading of Statutes, YOUTUBE 9:26-9:48 (Nov. 25, 2015), watch?v=dpEtszFT0Tg8rt=553s (on file with the Michigan Law Review).

(4.) E.g., Abbe R. Gluck, Congress, Statutory Interpretation, and the Failure of Formalism: The CBO Canon and Other Ways That Courts Can Improve on What They Are Already Trying to Do, 84 U. CHI. L. REV. 177, 191 (2017) ("There is now virtually unanimous agreement among federal judges that text always comes first."); Abbe R. Gluck & Richard A. Posner, Statutory Interpretation on the Bench: A Survey of Forty-Two Judges on the Federal Courts of Appeals, 131 HARV. L. REV. 1298, 1311 (2018) ("Even those [appellate] judges who emphasized the importance of purpose as an interpretive tool made clear they still would not use purpose to push a statute's interpretation beyond the limits of its text."); Frederick Schauer, The Practice and Problems of Plain Meaning: A Response to Aleinikoff and Shaw, 45 VAND. L. REV. 715, 720-21 (1992) (arguing that plain meaning has generally trumped purpose); Richard Pildes, Note, Intent, Clear Statements, and the Common Law: Statutory Interpretation in the Supreme Court, 95 HARV. L. REV. 892, 894 (1982) (arguing that intent determined through "literalist reading of statutory terms [has served] as a surrogate for actual legislative intent"). Even with this consensus, the rules of absurdity and scrivener's error provide an escape valve from text.

(5.) John F. Manning, Essay in Honor of Justice Ruth Bader Ginsburg, Justice Ginsburg and the New Legal Process, 127 HARV. L. REV. 455,456 (2013) ("[T]he semantic meaning of the enacted text, when clear, now sets a hard cap on the judge's discretion [to consider other evidence regarding a statute's objectives]."). Manning seems to limit his claim to "clear" text, but even ambiguous text still might place some interpretations out of range. See Jonathan T. Molot, The Rise and Fall of Textualism, 106 COLUM. L. REV. 1, 34 (2006).

(6.) Stephen Breyer, The 1991 Justice Lester W. Roth Lecture, On the Uses of Legislative History in Interpreting Statutes, 65 S. CAL. L. REV. 845, 863 (1992); see Caleb Nelson, What is Textualism?, 91 VA. L. REV. 347,394 (2005) ("[interpreters ... need some way to finish the job and to pick from among the possible meanings that their primary interpretive tools have identified.").

(7.) E.g., ANTONIN SCALIA & BRYAN A. GARNER, READING LAW: THE INTERPRETATION OF LEGAL TEXTS 369-90 (2012); John F. Manning, Textualism As a Nondelegation Doctrine, 97 Colum. L. Rev. 673, 679 (1997) (critiquing legislative history use as arrogating power to individual legislators though noting that the Court had used legislative history "[f]or more than a century"); see Nelson, supra note 6, at 377.

(8.) Legislative history use became routine in the mid-twentieth century but has declined substantially since the mid-1980s. Victoria F. Nourse, The Constitution and Legislative History, 17 U. PA. J. CONST. L. 313, 314 (2014) (noting that legislative history use in courts was not questioned until the late 1980s); Nicholas R. Parrillo, Leviathan and Interpretive Revolution: The Administrative State, the Judiciary, and the Rise of Legislative History, 1890-1950, 123 YALE L.J. 266, 270 (2013) ("The proportion of U.S. Supreme Court opinions citing legislative history in statutory cases has fallen by more than half since the 1980s."); see also Breyer, supra note 6, at 845-46 ("[T]he Supreme Court's actual use of legislative history is in decline.").

(9.) See John F. Manning, Continuity and the Legislative Design, 79 NOTRE DAME L. REV. 1863, 1864-65 (2004) ("Because modern formalists (qua textualists) doubt that intent or purpose gleaned from the legislative history offers a reliable way to resolve statutory indefiniteness, they want clear and predictable background rules to help legislators and interpreters decode textual cues." (footnote omitted)); Parrillo, supra note 8, at 391. Nelson makes the point a slightly different way: "[T]his reluctance [to make ad hoc judgments about intent] need not stem from distinctive views about the relevance of intended meaning.... [S]omeone ... might simply believe that judges will reach more accurate assessments of that intent if they accept the discipline of rules." Nelson, supra note 6, at 376.

(10.) See infra Section III.A.

(11.) See infra Section III.B.

(12.) This excludes Justice Gorsuch, who joined the Court after the study period, though he is a well-known defender of canon use. See infra note 17. Justice O'Connor, who retired a few months into the Roberts Court, wrote only two relevant opinions during the study period. One used canons, Schaffer ex rel. Schajfer v. Weast, 546 U.S. 49, 56-58 (2005); one did not, Lockhart v. United States, 546 U.S. 142 (2005). See infra text accompanying note 149 (summarizing data by Justice). Gluck and Posner report that of the forty-two appellate judges they interviewed, despite some disparagement of canons, "all use[d] them." Gluck & Posner, supra note 4, at 1334.

(13.) E.g., Nelson, supra note 6, at 390 ("Many of the canons used by textualists reflect observations about Congress's own habits."). Scalia and Garner's defense of the canons is slightly more obscure, but they concede that congressional awareness is an important justification: "It might be said that rules like these, so deeply ingrained, must be known to both drafter and reader alike so that they can be considered inseparable from the meaning of the text." SCALIA & GARNER, supra note 7, at 31.

(14.) See Lisa Schultz Bressman & Abbe R. Gluck, Statutory Interpretation from the Inside--An Empirical Study of Congressional Drafting, Delegation, and the Canons: Part II, 66 STAN. L. REV. 725, 732 (2014) [hereinafter Bressman & Gluck, Statutory Interpretation, Part II]; Abbe R. Gluck 8c Lisa Schultz Bressman, Statutory Interpretation from the Inside--An Empirical Study of Congressional Drafting, Delegation, and the Canons: Part I, 65 Stan. L. Rev. 901 (2013) [hereinafter Gluck & Bressman, Statutory Interpretation, Part I], Important earlier work was done by Jane Schacter and Victoria Nourse. Victoria F. Nourse & Jane S. Schacter, The Politics of Legislative Drafting: A Congressional Case Study, 77 N.Y.U. L. REV. 575, 601 (2002) (Senate Judiciary Committee staffers evinced awareness of canons but viewed them as "not important because such rules paled in comparison to what they were trying to do").

(15.) Bressman & Gluck, Statutory Interpretation, Part II, supra_note 14, at 731-32..

(16.) E.g., id. at 730.

(17.) Neil M. Gorsuch, Lecture, Of Lions and Bears, Judges and Legislators, and the Legacy of Justice Scalia, 66 CASE W. RES. L. REV. 905, 917 (2016) ("[W]e can make our decisions based on a comparative assessment of the various legal clues--choosing whether the rule of the last antecedent or one of its exceptions best fits the case in light of the particular language at hand.").

(18.) See SCALIA & GARNER, supra note 7, at xxviii; see also id. at 61 (noting endorsements of canons by Roscoe Pound and Felix Frankfurter).

(19.) See Harvard Law School, supra note 3, at 34:26-34:42. As of July 9, 2018, the book itself has been cited in 22 Supreme Court decisions; in 130 decisions of federal appeals and district courts; and in 167 state court decisions. To determine this, I searched the phrase "scalia /2 garner" in the Westlaw Supreme Court database; the combined Westlaw Federal Courts of Appeals and Federal District Courts databases; and the combined Westlaw State Courts database, respectively. Data is on file with author.

(20.) See infra notes 69-70 and accompanying text.

(21.) Gluck, supra note 4, at 178 ("The formalist project ... has been a failure.").

(22.) Id. at 178-79 (arguing that the failure to establish a clear hierarchy of canon use makes outcomes unpredictable); see RICHARD A. POSNER, REFLECTIONS ON JUDGING 178-235 (2013) (arguing that there is no principled hierarchy of canons of construction); see also James J. Brudney, Festschrift, Canon Shortfalls and the Virtues of Political Branch Interpretive Assets, 98 CALIF. L. REV. 1199, 1229 (2010) ("[T]he Court has not developed an architecture of authoritative priorities for the canons."); Karl N. Llewellyn, Remarks on the Theory of Appellate Decision and the Rules or Canons About How Statutes Are to Be Construed, 3 VAND. L. REV. 395, 401 (1950).

(23.) James Brudney and Corey Ditslear have done important related work in the specific settings of workplace law and tax law; their work runs through 2008. Brudney & Ditslear, supra note 2; James J. Brudney 8c Corey Ditslear, The Warp and Woof of Statutory Interpretation: Comparing Supreme Court Approaches in Tax Law and Workplace Law, 58 DUKE L.J. 1231 (2009) [hereinafter Brudney 8c Ditslear, Warp and Woof] (tracking subject-specific tax canon use in cases from 1969-2008). Professor Anita Krishnakumar is also conducting a large study of interpretive tools in the Roberts Court, but her study (thus far) covers a shorter time period, tracks the use of fewer canons, excluding significant substantive canons such as the common law canon, and does not consider judicial decisions that engage, but ultimately decline to apply, particular canons. E.g., Anita S. Krishnakumar, Reconsidering Substantive Canons, 84 U. Chi. L. Rev. 825, 844 (2017) [hereinafter Krishnakumar, Reconsidering] ("When an opinion mentioned a substantive canon but rejected it as inapplicable or not controlling, I did not count that as a substantive canon reference." (footnote omitted)). See generally Anita S. Krishnakumar, Statutory Interpretation in the Roberts Court's First Era: An Empirical and Doctrinal Analysis, 62 Hastings L.J. 221 (2010) [hereinafter Krishnakumar, Statutory Interpretation in the Roberts Court], David S. Law and David Zaring reported on uses of legislative history in Law Versus Ideology: The Supreme Court and the Use of Legislative History, 51 WM. & MARY L. REV. 1653 (2010). Christopher Walker has recently assessed canon use in agency statutory interpretation. See Christopher J. Walker, Inside Agency Statutory Interpretation, 67 Stan. L. Rev. 999 (2015).

(24.) Cf. William N. Eskridge, Jr. & Philip P. Frickey, Quasi-Constitutional Law: Clear Statement Rules as Constitutional Lawmaking, 45 VAND. L. REV. 593, 597 (1992) ("[T]he current Court emphasizes a different array of clear statement rules than did the Court in the 1970s.").

(25.) See infra text accompanying notes 201-205.

(26.) See infra Section III.E.

(27.) Llewellyn, supra note 22, at 401.

(28.) See supra note 22 and accompanying text.

(29.) See Appendix Table 1.

(30.) See infra Tables 3,4.

(31.) E.g., Nelson, supra note 6, at 352 ("[T]extualists suggest that interpretation should focus 'upon what the text would reasonably be understood to mean, rather than upon what it was intended to mean.'" (quoting ANTONIN SCALIA, A MATTER OF INTERPRETATION 144 (1997))); see also John F. Manning, The Absurdity Doctrine, 116 HARV. L. REV. 2387, 2389 (2003) ("[T]he enacted text is generally considered the best evidence of [congressional] intent...."); Nelson, supra note 6, at 353-54 (arguing that textualists and intentionalists both "seek to identify and enforce the legal directives that an appropriately informed interpreter would conclude the enacting legislature had meant to establish").

(32.) See Amy Coney Barrett, Substantive Canons and Faithful Agency, 90 B.U. L. REV. 109, 117 (2010) ("Linguistic canons pose no challenge to the principle of legislative supremacy because their very purpose is to decipher the legislature's intent."); Manning, supra note 31, at 2465 n.285 ("[Such canons] serve only as rules of thumb ... that help users of legal language discern meaning.").

(33.) See Brudney, supra note 22, at 1202 ("Language canons are generally justified as ... embodying] conventional usage...."); Nelson, supra note 6, at 383 ("Some canons simply reflect broader conventions of language use...."); see also William Baude & Stephen E. Sachs, The Law of Interpretation, 130 HARV. L. REV. 1079, 1121 (2017) ("Linguistic canons.... are just attempts to read whatever the authors wrote, according to the appropriate theory of reading....").

(34.) See infra text accompanying notes 97, 99-101 (describing usage-based canons).

(35.) E.g., FREDERICK SCHAUER, PLAYING BY THE RULES 141 (1991) ("Where the world is regular ... the necessity of rules to harness that regularity will be minimal"). For example, in Can v. United States, 560 U.S. 438, 447-48 (2010), the court paid extensive attention to the verb tense used in the statute despite the lack of a verb tense canon.

(36.) Cf. Michael Robertson, Picking Positivism Apart: Stanley Fish on Epistemology and Law, 8 S. CAL. INTERDISC. L.J. 401, 408 (1999) (discussing Fish's view that "by upbringing and training we have become embedded in a particular community's set of beliefs, values, categories of thought, paradigms, practices, and purposes").

(37.) See LAWRENCE M. SOLAN, THE LANGUAGE OF STATUTES 122 (2010) (arguing that fair notice could prompt decision makers to "undermine what [they] know to be the intent of the legislature in order to promote those competing values").

(38.) 136 S. Ct. 958, 962-63 (2016) (Sotomayor, J., for the majority); see id. at 969-70 (Kagan, J., dissenting) (arguing that her approach better reflected "ordinary understanding of how English works").

(39.) See Baude & Sachs, supra note 33, at 1121 (seemingly assuming that all linguistic canons are "features of how some group of actual people actually speak").

(40.) E.g., Law v. Siegel, 134 S. Ct. 1188,1195 (2014) (applying the " 'normal rule of statutory construction' that words repeated in different parts of the same statute generally have the same meaning" (quoting Dep't of Revenue v. ACF Indus., Inc., 510 U.S. 332, 342 (1994))).

(41.) SCALIA & GARNER, supra note 7, at 168,172.

(42.) E.g., United States v. Spelar, 338 U.S. 217, 225 (1949) (Jackson, J., concurring) ("[W]e should pay Congress the respect of not assuming lightly that it indulges in inconsistencies of speech which make the English language almost meaningless."); United States v. Butler, 297 U.S. 1, 65 (1936) ("These words cannot be meaningless, else they would not have been used."), cited in SCALIA & GARNER, supra note 7, at 174.

(43.) SCALIA & GARNER, supra note 7, at 107.

(44.) E.g., Nelson, supra note 6, at 383 (classifying noscitur and expressio unius as "reflect[ing] ordinary principles that laymen as well as lawyers use to interpret communications").

(45.) William N. Eskridge, Jr., Norms, Empiricism, and Canons in Statutory Interpretation, 66 U. CHI. L. REV. 671, 676 (1999) (discussing "Stop kicking and pushing your brother," as a stand in for a "more general principle (no hurting)").

(46.) See David L. Shapiro, Continuity and Change in Statutory Interpretation, 67 N.Y.U. L. REV. 921, 943 (1992). Another defense of these canons is that they impose "coherence on the U.S. Code," though again, this goal is at the expense of ordinary notice. Gluck & Bressman, Statutory Interpretation, Part I, supra note 14, at 905.

(47.) E.g., SCALIA & GARNER, supra note 7, at xxvii (noting importance of "conventions and contexts" in communication); cf. Frank H. Easterbrook, The Absence of Method in Statutory Interpretation, 84 U CHI. L. REV. 81, 83 (2017) ("[E]very canon implicitly begins or ends with the statement 'unless the context indicates otherwise,' which potentially leaves so much room for maneuver that the canon isn't doing much work.").

(48.) But see Gluck, supra note 4, at 186-87 (suggesting that assuming statutory consistency across the U.S. Code enhances "notice to the public").

(49.) See infra notes 104-122 and accompanying text.

(50.) United States v. Castleman, 134 S. Ct. 1405, 1416 (2014) (declining to apply the rule of lenity since the statute was not sufficiently ambiguous); Crandon v. United States, 494 U.S. 152,158 (1990) (noting that the rule of lenity ensures "fair warning of the boundaries of criminal conduct and that legislatures, not courts, define criminal liability"); Shapiro, supra note 46, at 935 (discussing the value of the rule of lenity as "helping to ensure that people have fair warning of what has been prohibited").

(51.) See SCALIA & GARNER, supra note 7, at 296 (arguing that the rule of lenity is based on the "interpretive reality that a just legislature will not decree punishment without making clear what conduct incurs the punishment and what the extent of the punishment will be; or at least on the judge-made public policy that a legislature ought not to do so"). See generally Intisar A. Rabb, Response, The Appellate Rule of Lenity, 131 HARV. L. REV. F. 179,193-94 (2018), https:// [] (discussing constitutional values of legislative supremacy and fair warning as underlying rule of lenity).

(52.) E.g., Nina A. Mendelson, Chevron and Preemption, 102 MICH. L. REV. 737, 755 (2004).

(53.) Manning argues that clear statement rules go astray by enforcing values in the abstract, without recognizing that the Constitution protects values "in specified contexts through specified means." John F. Manning, Essay, Clear Statement Rules and the Constitution, 110 Colum. L. Rev. 399,438-39 (2010); id. at 445 (critiquing clear statement rules as "sacrific[ing] the idea that constitutional values ... are intelligible only in [a framework] that defines their reach and limits"); id. at 433 (arguing that the Constitution does not endorse "federalism" as such but instead reconciles "competing values about the appropriate sphere of state authority"). David Shapiro defends this as useful anyway. See Shapiro, supra note 46, at 936-37 (arguing that some substantive canons implicitly "recogniz[e] the value of minimal disruption of existing arrangements").

(54.) Manning, supra note 53, at 427 ("[C]lear statement rules ... enforce constitutional values as abstracted from the specific provisions that implement those values."). We also included canons that require deference to agencies, though since some classify those canons as distinct "extrinsic source rules," Eskridge, supra note 45, at 674, we also broke out the data separately. These include the Chevron and Skidmore doctrines requiring judicial deference to certain agency statutory interpretations. Chevron U.S.A. Inc. v. Nat. Res. Def. Council, Inc., 467 U.S. 837,865-66 (1984); Skidmore v. Swift & Co., 323 U.S. 134 (1944).

(55.) E.g., Nelson, supra note 6, at 394 & n. 140; see infra text accompanying notes 60-66.

(56.) E.g., EINER ELHAUGE, STATUTORY DEFAULT RULES 168-69 (2008) (explaining that the rule of lenity can be justified as "forc[ing] legislatures to define just how anti-criminal they wish to be"). Elhauge and Hills have separately defended particular interpretive defaults as improving legislative deliberations by requiring clear language to reach a particular outcome, perhaps one supported by a particularly well-represented interest group or opposed by a relatively poorly represented one. See Roderick M. Hills, Jr., Against Preemption: How Federalism Can Improve the National Legislative Process, 82 N.Y.U. L. REV. 1 (2007); see also Elhauge, supra, at 153, 190 (2008) (arguing that under certain conditions, canons can serve as effective means of eliciting legislative preferences, rather than estimating them). See generally Jane S. Schacter, Metademocracy: The Changing Structure of Legitimacy in Statutory Interpretation, 108 HARV. L. REV. 593,621 (1995).

(57.) But see SCALIA & GARNER, supra note 7, at 299 ("When [statutes] are not clear, the consequences should be visited on the party more able to avoid and correct the effects of shoddy legislative drafting--namely, the federal Department of Justice or its state equivalent.").

(58.) Nelson, supra note 6, at 383.

(59.) E.g., Eskridge, supra note 45, at 675 (articulating argument that "canons supply details ... that the legislature would probably have provided if its members had written a more detailed statute"); Nelson, supra note 6, at 386 (arguing that textualists "instinctively evaluate most canons with reference to the likely intent of the enacting legislature"); id. at 389 (suggesting that specialized canons "can be seen as entrenched generalizations about the likely intent of the enacting legislature"). A rules-based approach might be comparatively efficient if it minimizes the sum of error and decision costs. Adrian Vermeule, Interpretive Choice, 75 N.Y.U. L. REV. 74, 88-89 (2000) (discussing efficiency approach to interpretive choice and aiming for "best mix of error costs and decision costs"); id. at 91 (noting that "[r]ules economize on information"). Nelson offers the most detailed explanation of these types of justifications. See generally Nelson, supra note 6.

(60.) E.g., Barrett, supra note 32, at 120 ("Avoidance, for example, has been justified on the ground that Congress does not usually intend for its statutes to provoke serious constitutional questions...."). But see Nelson, supra note 6, at 397 n.149 ("But insofar as interpreters are putting [constitutional] values in opposition to the intended meaning of statutes, this response sits uneasily with textualist theory.").

(61.) See William N. Eskridge, Jr., The New Textualism and Normative Canons, 113 COLUM. L. REV. 531, 569 (2013) (reviewing SCALIA & GARNER, supra note 7) (noting that Scalia and Garner provide "zero evidence" in support of their canons). Gluck notes that the Court has cited legislative counsel drafting manuals, but only three times in recent decades. Gluck, supra note 4, at 210 & n.145; see also Richard A. Posner, Statutory Interpretation--In the Classroom and in the Courtroom, 50 U. CHI. L. REV. 800, 806 (1983) ("We should demand evidence that statutory draftsmen follow the code before we erect a method of interpreting statutes on the improbable assumption that they do.").

(62.) Gluck & Bressman, Statutory Interpretation, Part I, supra note 14, at 954.

(63.) Id. at 954-55.

(64.) Id. at 1002-04 (agency deference); id. at 942-44 (federalism canons); id. at 948 (constitutional avoidance canon unknown but assumed); see also id. at 946-47 (rule of lenity unfamiliar).

(65.) Id. at 944 (6% of respondents predicted that courts would construe ambiguities against preemption; 12% predicted that the "presumption would run in the opposite [incorrect] direction").

(66.) Id. at 919-24 (describing methodology).

(67.) Not everyone accepts these as central interpretive goals. Eskridge, supra note 45, at 681 (noting "normative disagreement about how highly to value predictability and even objectivity," given the need for "equity" and "justice"). For purposes of discussion, I assume the validity of the goals of stability and predictability.

(68.) Id. at 678-79 ("The canons of statutory interpretation can be defended if they generate greater objectivity and predictability in statutory interpretation.").

(69.) Lockhart v. United States, 546 U.S. 142, 148 (2005) (Scalia, J., concurring) (stating that certain canons "may function as background canons of interpretation of which Congress is presumptively aware"); SCALIA & GARNER, supra note 7, at 51; Gluck, supra note 4, at 185 (suggesting that canons still might "promote legislative supremacy by giving Congress the tools to draw effective lines of inclusion and exclusion" (quoting John F. Manning, Inside Congress's Mind, 115 COLUM. L. REV. 1911, 1942 (2015))); Manning, supra note 31, at 2467 ("These background conventions, if sufficiently firmly established, may be considered part of the interpretive environment in which Congress acts."); id. at 2473 (regarding "established conventions" such as "equitable tolling," canonical application offers "an intelligible basis for legislators and the public to identify and evaluate the legislative bargains struck"); Nelson, supra note 6, at 391 (stating that canons have the advantage of "relative predictability"--they "help courts discern Congress's likely intent ... simply because members of Congress know that the courts use them.... [which] enables members of Congress to convey their intended meaning in a way that the courts will understand"); id. at 386 ("[S]ome textualists seem attracted to the idea that a canon can form part of the backdrop for legislation even if there is little reason to think ... Congress acted in accordance with it."); cf. Morrison v. Nat'l Austl. Bank Ltd., 561 U.S. 247, 261 (2010) (arguing for presumption against extraterritoriality instead of case-by-case analysis of likely congressional intent to "preserv[e] a stable background against which Congress can legislate with predictable effects").

(70.) See Nelson, supra note 6, at 393 (arguing that justification does not "help[] textualists identify the kinds of canons they should support" (emphasis added)).

(71.) Barrett, supra note 32, at 111 (explaining that, however canons began, "courts have used them for so long that they are now part of the way that lawyers think about language").

(72.) SCALIA, supra note 31, at 28-29.

(73.) Antonin Scalia, Essay, Assorted Canards of Contemporary Legal Analysis, 40 CASE W. RES. L. REV. 581, 583 (1990); see Nelson, supra note 6, at 392 ("Justice Scalia suggested] that having clear canons of statutory interpretation--even if they initially seem too unrefined to match actual congressional intent--will ultimately help minimize the gap between ... interpretations ... and the meanings intended by members of Congress.").

(74.) SCALIA & GARNER, supra note 7, at 51.

(75.) Barrett, supra note 32, at 128. Barrett also argues that the "long pedigree" makes it difficult to dismiss the use of substantive canons overall as inconsistent with constitutional limits on judicial power, questions I do not reach in these pages. Id. at 111 (asserting that if federal courts "once possessed the power to develop substantive canons, there is no reason to believe that they lost that power"); see also Baude & Sachs, supra note 33, at 1127 ("We ... would ask whether the canons were rules of law at the Founding or have validly become law since, pursuant to rules of legal change that were themselves valid in this way."); Manning, supra note 31, at 2467 (arguing that canons, "if sufficiently firmly established, may be considered part of the interpretive environment in which Congress acts").

(76.) Manning, supra note 31, at 2474. But see SCALIA & GARNER, supra note 7, at 341-96 (regarding "thirteen falsities," arguing for abandonment of remedial purposes canon, use of legislative history, and use of "spirit" in statutory interpretation, despite historical use of all three).

(77.) See SCHAUER, supra note 35.

(78.) See id. at 24.

(79.) See, e.g., id. at 139 ("Reliance thus requires that a rule encompass a category whose membership ... is substantially non-controversially shared among the addressees of the rule and those who enforce it."); Gluck & Bressman, Statutory Interpretation, Part I, supra note 14, at 917 (recognizing argument that canons might serve a background function if consistently applied).

(80.) See Matthew R. Christiansen & William N. Eskridge, Jr., Congressional Overrides of Supreme Court Statutory Interpretation Decisions, 1967-2011, 92 TEX. L. REV. 1317 (2014). Eskridge and Christiansen, of course, cannot provide a baseline of expected overrides, and their findings could be disregarded if we simply believed that all decisions not overridden were correct. This seems unlikely, however.

(81.) See generally JERRY L. MASHAW, GREED, CHAOS, AND GOVERNANCE ch. 4 (1997) (arguing that because interpretation "rearrange[s] the status quo," "even if the legislature acts to 'correct' an interpretation ... it will almost never end up with its original policy reinstated" (emphasis omitted)).

(82.) See generally Gluck & Bressman, Statutory Interpretation, Part I, supra note 14.

(83.) See John F. Manning, Exchange, What Divides Textualists from Purposivists?, 106 COLUM. L. REV. 70, 102 (2006) ("Given these inherent complexities in the legislative process, purposivism cannot be justified as a demonstrably more reliable measure of what a constitutionally sufficient legislative majority actually would have wanted....").

(84.) See infra note 326 and accompanying text.

(85.) Manning, supra note 83, at 74. Nelson also has expressed substantial reservations regarding this justification; see Nelson, supra note 6, at 388-89 ("[A] canon that runs counter to ordinary understandings of language ... might impede rather than assist the transmission of intended meaning; courts might well find it hard to tell whether members of Congress were legislating in light of the canon or [not]."); cf. Tex. Dep't of Hous. & Cmty. Affairs v. Inclusive Cmtys. Project, Inc., 135 S. Ct. 2507, 2540-41 (2015) (Alito, J., dissenting) (arguing, for four justices, that purported surplusage has the "hallmarks of a compromise" among varying congressional factions). Both Baude and Sachs and David Shapiro have argued for canon use on the ground that it stabilizes the development of law, seemingly independent of whether Congress can be expected to know the canon. Baude & Sachs, supra note 33, at 1123 ("If anything, the lack of knowledge about a canon reinforces the strength of that canon: what legislators were unaware of, they're unlikely to have displaced."); Shapiro, supra note 46, at 925 ("[C]lose questions of construction should be resolved in favor of continuity and against change."); id. at 936-37 (arguing for sympathetic view of common law canon). Both these arguments, however, have the most force in connection with a narrow subset of canons, perhaps the regularly cited, well-settled common law rules such as mens rea requirements in the criminal law or equitable tolling of statutes of limitation. See Baude & Sachs, supra note 33, at 1123 (generalizing from the fact that "[m]any legal canons are common law default rules"). And Shapiro, in particular, notes the need for courts to moderate the risks that this position will mistake congressional meaning by making a "sincere and sympathetic effort to discern legislative purpose." Shapiro, supra note 46, at 926.

(86.) We excluded the interpretation of constitutional provisions, international agreements, private contracts, state statutes, and federal regulations.

(87.) An opinion's characterization of a canon was rarely different from ours; where they differed, we used the opinion's express identification of the canon.


(89.) SCALIA & GARNER, supra note 7.

(90.) Id. at 107.

(91.) Id. at 168.

(92.) Id. at 172.

(93.) Id. at 174.

(94.) Id. at 180-81. We located no named uses of this canon. Our coding relied on a very conservative definition of this canon, finding it applied only when the court considered whether a proposed interpretation of one statutory provision might nullify, effectively nullify, or directly contradict another, rather than simply being in tension with it. (In contrast to the definition we used, a broader understanding of this canon could be understood to encompass arguments regarding the coherence of statutory structure.) We did code as employing the "harmonious reading canon" opinions that avoided reading a statutory section to be superfluous, but also "unnaturally cramped" or "insignificant." E.g., Salazar v. Ramah Navajo Chapter, 567 U.S. 182, 204 (2012) (Roberts, C.J., dissenting) ("insignificant" (quoting TRW Inc. v. Andrews, 534 U.S. 19, 31 (2001))); Cullen v. Pinholster, 563 U.S. 170, 217 (2011) (Sotomayor, J., dissenting) ("unnaturally cramped"). Again, this rule is closely tied, though not identical, to the rule against surplusage. E.g., TRW Inc. v. Andrews, 534 U.S. 19, 31 (2001) (declining to read any part of a statute to be "superfluous, void, or insignificant" (quoting Duncan v. Walker, 533 U.S. 167, 174 (2001))).

(95.) SCALIA & GARNER, supra note 7, at 195.

(96.) Id. at 199-201.

(97.) Id. at 144.

(98.) Id. at 252-53. Where a statute's words were interpreted consistently with identical statutory language located elsewhere in the U.S. Code, we coded it as use of the Whole Code Rule.

(99.) Id. at 116-25.

(100.) Id. at 112.

(101.) Id. at 161.

(102.) Id. at 183.

(103.) Id. at 221.

(104.) 1A SUTHERLAND STATUTORY CONSTRUCTION, supra note 88, [section]23:10, at 479 ("The presumption against implied repeals is founded upon the doctrine that the legislature is presumed to envision the whole body of the law when it enacts new legislation.").

(105.) We also coded this rule as applied when the court considered whether a statute should be construed to avoid potential "irrational," "ludicrous," "outlandish," "perverse," "strange," or significantly "anomalous" results. Although it was a close call, and although such interpretive moves raise legitimacy issues at least as significant as the absurdity canon, we excluded milder expressions regarding difficult-to-accept interpretive results, such as the presence of "odd results," "something arbitrary," or "disharmony." E.g., Barber v. Thomas, 560 U.S. 474, 491 (2010) (Breyer, J., for the majority) ("odd results"); United States v. Rodriquez, 553 U.S. 377, 396 (2008) (Souter, J., dissenting) ("something arbitrary"); BP Am. Prod. Co. v. Burton, 549 U.S. 84, 99 (2006) (Alito, J., for the majority) ("disharmony").

(106.) SCALIA & GARNER, supra note 7, at 248.

(107.) Id. at 290-94.

(108.) We coded for the presence of this canon whenever the court applied a presumption (other than the presumption against preemption of state law) of noninterference with some flavor of state government autonomy.

(109.) SCALIA & GARNER, supra note 7, at 268-72.

(110.) Id. at 261-65.

(111.) Id. at 296.

(112.) CONFERENCE OF W. ATTORNEYS GEN., AMERICAN INDIAN LAW DESKBOOK [section] 1.6 (2017 ed.) (collecting cases); Eskridge & Frickey, supra note 24, at 602 n.29 (quoting McClanahan v. Ariz. State Tax Comm'n, 411 U.S. 164,174 (1973)).

(113.) 3 SUTHERLAND STATUTORY CONSTRUCTION, supra note 88, [section] 61:1, at 314; Abbe R. Gluck, The Federal Common Law of Statutory Interpretation: Erie for the Age of Statutes, 54 Wm. & Mary L. Rev. 753, 769 (2013) ("cousin"). Scalia and Garner rename the canon "Presumption Against Change in Common Law," but similarly would presume that a statute "will not be interpreted as changing the common law unless [it] effect[s] the change with clarity." SCALIA & GARNER, supra note 7, at 318. In tracking this canon, we attempted to exclude cases where the Court reasoned that Congress specifically intended to incorporate the common law.

(114.) SCALIA & GARNER, supra note 7, at 320; e.g., United States v. Castleman, 134 S. Ct. 1405, 1418 (2014) (Scalia, J., concurring) (noting the "settled principle of interpretation" that "Congress intends to incorporate the well-settled meaning of the common-law terms it uses." (quoting Sekhar v. United States, 570 U.S. 729, 732 (2013))).

(115.) 3 SUTHERLAND STATUTORY CONSTRUCTION, supra note 88, [section] 60:1, at 250-51.

(116.) Scalia & Garner, supra note 7, at 281-89.

(117.) Chevron U.S.A. Inc. v. Nat. Res. Def. Council, 467 U.S. 837, 865-66 (1984); Skidmore v. Swift & Co., 323 U.S. 134 (1944). We classified these canons as substantive, though since some classify those canons as distinct "extrinsic source rules," Eskridge, supra note 45, at 674, we also broke out the data on the use of agency deference doctrines separately.

(118.) E.g., William N. Eskridge, Jr. & Lauren E. Baer, The Continuum of Deference: Supreme Court Treatment of Agency Statutory Interpretations from Chevron to Hamdan, 96 Geo. L.J. 1083 (2008).

(119.) See infra Section III.E.7.

(120.) See infra Section IH.E.l.

(121.) E.g., Dialogue: The Future of Administrative Law, 47 ENVTL. L. REP. NEWS & ANALYSIS 10186, 10188 (2017) (comments of Professor Jody Freeman); see also Michael Coenen & Seth Davis, Minor Courts, Major Questions, 70 Vand. L. Rev. Ill, 785-96 (2017) (discussing potential evolution in major question canon).

(122.) See Arbaugh v. Y & H Corp., 546 U.S. 500, 516 (2006); infra Section III.E.4.

(123.) Cf. Brudney & Ditslear, Warp and Woof, supra note 23 (tracking subject-specific tax canon use in cases from 1969-2008).

(124.) But see also Kevin Stack, The Stated Purposes Canon (unpublished draft 2018) (on file with author) (arguing for the recognition of a "stated purposes principle" or canon).

(125.) In the relatively rare circumstance in which an opinion considered the same canon multiple times in resolving a single statutory issue, we coded the canon's use (applied with approval or not) based on the first time it appeared in the analysis.

(126.) I part ways here with Anita Krishnakumar, who has attempted to distinguish "primary" reliance on a canon from "passing" reliance. Krishnakumar, Reconsidering, supra note 23, at 845.

(127.) See Frederick Schauer, Giving Reasons, 47 STAN. L. REV. 633, 635 (1995) ("The act of giving a reason ... is an exercise in generalization.").

(128.) Nijhawan v. Holder, 557 U.S. 29, 40 (2009).

(129.) E.g., Eskridge, supra note 45, at 680.

(130.) See supra notes 67-68 and accompanying text.

(131.) See Appendix Table 1.

(132.) Our study identified 3 occurrences of the major question canon and 111 occurrences of agency deference canons, in 11.9% of cases. Data on file with the author.

(133.) But see infra text accompanying notes 236-239 (noting new trend of applying rule of lenity in immigration cases).

(134.) For example, the parties' briefs (excluding amicus briefs, which may get less attention from the justices) discussed the major question canon in eleven cases during the study period. Meanwhile, the canon was applied only in two majority opinions and one dissent and not otherwise mentioned. I identified briefs through an overly inclusive search in the Westlaw Supreme Court briefs database: adv: (brown or tele!) /p (extraordinary or hesitate or significan!) /p (agency delegat!) and DATE(aft 2003) and DATE(bef 2016). The search was meant to capture citations to the leading canon references in either FDA v. Brown & Williamson Tobacco Corp., 529 U.S. 120 (2000) or MCI Telecommunications Corp. v. AT&T Co., 512 U.S. 218 (1994), together with a mention of a significant agency action or statutory delegation. The search resulted in over 100 briefs, each of which I reviewed individually. All data on file with author.

For another example, party briefs argued for application of the Native American/Indian canon in six cases, identified through multiple Westlaw searches in the Supreme Court briefs database for "be construed liberally in favor of the Indians" and "ambigu! /s stat! /s native" and "ambigu! Is stat! Is indian". Meanwhile, the canon made it into only one majority opinion and a dissent in a different case. It garnered one very oblique reference in a second majority opinion. All data on file with author.

These findings are consistent with Eskridge and Baer's analysis of the Court's reliance on the Chevron doctrine during an earlier period than this study, between 1984 and 2006. They identified over one thousand cases in which an agency statutory interpretation was at issue. Eskridge & Baer, supra note 118, at 1094. Although it seems likely that the Chevron or another deference doctrine was briefed in those cases, Eskridge and Baer found that Chevron (or a Chevron precedent) was applied in only 8.3% of cases during this period, and in a majority of the cases, the Court applied no deference doctrine at all. Id. at 1090.

(135.) For example, we used query "specific Is general" to search all briefing in the Westlaw Supreme Court briefs database, in the 18 decisions we coded as considering the "specific rules govern over general ones" canon. In 28% of these decisions, including multiple decisions in which majority opinions applied the canon, the canon was discussed in none of the briefs filed with the Court. All data on file with author. But see Gluck & Posner, supra note 4, at 1332 ("[O]ur random study of opinions from the [appellate] judges we interviewed revealed that many canons used in the opinions were introduced by the briefs.").

(136.) Cf. Schacter, supra note 56, at 621 (describing prodeliberation canons as within the "republican canon of canons").

(137.) Anita S. Krishnakumar, Dueling Canons, 65 DUKE L.J. 909, 914 (2016) (finding that "the overall rate of dueling canon or interpretive tool use in the Roberts Court's first five terms was ... at or below 25.0 percent for most tools").

(138.) See, e.g., infra Section III.D (giving examples of reasons why canons might not ap ply).

(139.) For example, Krishnakumar draws no inference from a court's engagement with, but ultimate decision not to apply, a canon in a particular setting. Krishnakumar, Reconsidering, supra note 23, at 844 ("When an opinion mentioned a substantive canon but rejected it as inapplicable or not controlling, I did not count that as a substantive canon reference." (footnote omitted)); Krishnakumar, supra note 137, at 925 ("[Disagreements over applicability do not necessarily reflect an underlying looseness ... [in] the canon ... [or] susceptibility to judicial massaging...."). Brudney and Ditslear employ a similar methodology to this project. Brudney & Ditslear, Warp and Woof, supra note 23, at 1249.

(140.) 551 U.S. 128,137 (2007).

(141.) Id. at 137.

(142.) See Appendix Table 2.

(143.) See Appendix Table 2.

(144.) See Appendix Table 2. Krishnakumar's findings for what she calls "combined language canons" are comparable, given the differences in the number of canons tracked and the difference in time period. She found that those canons to be applied with approval in 32.3 % of all opinions and in 67.9 % of majority opinions. Krishnakumar, Reconsidering, supra note 23, at 894.1 found textual canons overall to be used with approval in 46.7 % of majority opinions.

(145.) See Appendix Table 2.

(146.) See Appendix Table 2.

(147.) We recorded citations to congressional legislative history documents in 29.3% of contested statutory issues in all opinions, but these findings are more difficult to interpret, given frequent, but difficult-to-code, disclaimers and other signals that reliance on legislative history were simply for "those who care" about legislative history. E.g. Yates v. United States, 135 S. Ct. 1074, 1093 (2015) ("[Legislative history, for those who care about it, puts extra icing on a cake already frosted.") (Kagan, J., dissenting for four justices); United States v. Hayes, 555 U.S. 415, 418 n.*, 426-29 (2009) (noting that Justice Thomas declined, without opinion, to join Section III of the majority opinion, which analyzed legislative history).

(148.) See Appendix Table 3.

(149.) See Appendix Table 3. With respect to individual canons, our data also showed that Justice Kennedy made use of the constitutional avoidance canon at a higher rate than any other justice in opinions he authored (considering the canon in 15% of issues in opinions he authored; applying it with approval to help resolve the issue 73% of the time the canon was considered), followed by Chief Justice Roberts (considering in 14% of issues in opinions he authored; applying it with approval 56% of the time the canon was considered). Justices Souter and Scalia discussed the rule against absurdity at a higher rate than the other Justices (considering in 14% and 12% of issues; applying with 60% and 69% approval rate, respectively), though Justice Alito applied the rule of absurdity with the highest rate of approval (considering in 11% of issues; applying with 82% approval rate). Meanwhile, Justice Sotomayor used the expressio unius canon at the highest rate of the justices in the study, applying it with a high level of approval (considering it in 28% of issues in opinions she authored; applying with approval to help resolve the issue 86% of the time the canon was considered). Justice Kennedy discussed agency deference canons at the highest rate among the justices in the study (considering in 18% of issues; applying with approval 62% of the time the canon was considered), while Justices Souter and Ginsburg applied those canons with the highest rate of approval (75% approval rate for both; considered in 11 and 12% of issues, respectively). Justice Sotomayor and Justice Thomas were least likely to discuss and apply agency deference canons in opinions each authored (considering in 5% of issues for Sotomayor, with 25% approval rate, and in 6% of issues for Thomas, applied with approval 29% of the time those canons were considered). The justicespecific data on individual canons is more difficult to interpret compared with the aggregate canon data, however, not only because of the nonrandomness of other justices who may have joined the opinion, but because of the relatively small numbers involved and the nonrandomness of opinion assignments.

(150.) Krishnakumar, Reconsidering, supra note 23, at 850 tb1.2.

(151.) See id. (overall statistics); id. at 856 (stating that the top six "substantive" canons tracked were the avoidance canon, the rule of lenity, the presumption against preemption of state law, the federalism clear statement rule, the presumption against waiver of sovereign immunity, and the presumption against retroactivity); id. app. (appendix showing all substantive canons tracked).

(152.) See Appendix Table 1.

(153.) See Appendix Table 1.

(154.) See Appendix Table 1.

(155.) See Appendix Table 1.

(156.) E.g., Manning, supra note 31, at 2390.

(157.) See Appendix Table 1.

(158.) Scalia, supra note 31, at 13 (emphasis omitted). One question that will be taken up in future work is whether the Roberts Court more often has sought statutory coherence with common law norms than with norms drawn from statutes. Cf. Krishnakumar, Statutory Interpretation in the Roberts Court, supra note 23, at 240 ("[W]hen a Justice uses common law precedent as an interpretive aid, he or she is harmonizing the individual statute at issue with the existing legal backdrop.").

(159.) See Appendix Table 1.

(160.) See Appendix Table 1.

(161.) See Appendix Table 1.

(162.) See Appendix Table 1.

(163.) See Appendix Table 1.

(164.) See Appendix Table 1.

(165.) See Appendix Table 1.

(166.) In other words, majority opinions applied the rule of lenity only twice in the fourteen statutory issues in which the canon was discussed. The rule was, however, applied with approval in both plurality opinions that discussed it over this period, all of the five concurring opinions that discussed it, and eight of the ten dissenting opinions that discussed it; that led to an overall application with approval percentage of roughly 55%. Although it seems unlikely that the justices would be more inclined to discuss one inapplicable canon compared with another, it is conceivable that the justices feel a greater obligation to explain a decision not to apply the rule of lenity because of the special status of criminal defendants.

(167.) See Appendix Table 1.

(168.) See infra Section III.E.1; Appendix Table 1.

(169.) In other words, majority opinions used at least one tracked canon to resolve 356 of 508 contested statutory issues in those opinions. See Appendix Table 4. Krishnakumar argues based on her own empirical study that precedent and pragmatic considerations remain important interpretive tools. Krishnakumar, Reconsidering, supra note 23, at 830; Krishnakumar, Statutory Interpretation in the Roberts Court, supra note 23, at 235-36, 236 tbl.1 (showing very significant reliance on Supreme Court precedent in first three years of Roberts Court statutory interpretation opinions). See generally Lawrence M. Solan, Precedent in Statutory Interpretation, 94 N.C. L. REV. 1165 (2016).

(170.) In other words, of the 356 contested statutory issues resolved in majority opinions by considering at least one tracked canon, majority opinions resolved 238 of those issues by considering at least two canons and considered at least three canons in resolving 135 statutory issues. See Appendix Table 4. The corresponding statistics for statutory issues addressed in all opinions were slightly lower, but similar: At least one canon was considered in approximately 67% of contested statutory issues resolved in any opinion (in 625 issues out of 931 total). Within that subset, two or more canons were considered to resolve 60% of the issues (in 373 issues), and three or more canons were considered to resolve 36% of the cases (in 222 issues). See Appendix Table 4.

(171.) Opinions resolving a contested issue that considered 5 tracked canons applied, on average, 3.13 of them with approval; those considering 6 canons applied 4.07 of them with ap proval; those considering 7, 8, and 9 canons applied 4.75, 4.00, and 6.00 of them, respectively, with approval. See Appendix Table 5.

(172.) Wachovia Bank, Nat'l Ass'n v. Schmidt, 546 U.S. 303, 317 n.9 (2006) (Ginsburg, J., for the unanimous Court).

(173.) Winkelman ex rel. Winkelman v. Parma City Sch. Dist., 550 U.S. 516, 531 (2007) (Kennedy, J., for the majority). Justice Scalia's concurring opinion, joined by Justice Thomas, would have relied on expressio unius. See id. at 539 (Scalia, J., dissenting); see also Ala. Dep't of Revenue v. CSX Transp., Inc., 135 S. Ct. 1136, 1146 (2015) (Thomas, J., dissenting) (declining, along with Justice Ginsburg, to apply expressio unius, because a residual clause "need not use the same language as the clauses it follows to derive meaning from those clauses").

(174.) Norfolk S. Ry. v. Sorrell, 549 U.S. 158,171 (2007) (Roberts, C.J., for the majority).

(175.) Richlin Sec. Serv. Co. v. Chertoff, 553 U.S. 571, 577-78 (2008) (Alito, J., for an opinion unanimous on this point) (calling "unpersuasive" the "Government's fractured interpretation of the statute" that the omission of paralegals from a statutory provision governing fee awards to "attorneys, agents, and expert witnesses" indicated a different treatment for the paralegals and instead suggesting that paralegals are relatively analogous to these professionals).

(176.) Sackett v. EPA, 566 U.S 120, 129 (2012) (Scalia, J., for the unanimous Court) (finding an omission of judicial review for agency action in one section of a statute, even though statute expressly provided it elsewhere, insufficient to overcome presumption of reviewability).

(177.) Univ. of Tex. Sw. Med. Ctr. v. Nassar, 570 U.S. 338, 377 (2013) (Ginsburg, J., dissenting) (rejecting expressio unius arguments because the courts should read the phrases at hand "informed by this Court's consistent holdings" regarding phrasal meaning).

(178.) Setser v. United States, 566 U.S. 231, 238 (2012) (Scalia, J., for the majority).

(179.) Marx v. Gen. Revenue Corp., 568 U.S. 371, 381 (2013) (Thomas, J., for the majority) (citation omitted) (quoting United States v. Vonn, 535 U.S. 55, 65 (2002)); id. ("We have long held that the expressio unius canon does not apply 'unless it is fair to suppose that Congress considered the unnamed possibility and meant to say no to it,'....") (quoting Barnhart v. Pea body Coal Co., 537 U.S. 149, 168 (2003))). Justice Sotomayor, joined by Justice Kagan, would have applied the canon. Id at 392; see also Dan's City Used Cars, Inc. v. Pelkey, 569 U.S. 251, 264 (2013) (Ginsburg, J., for the unanimous Court) (declining to apply expressio unius and commenting, "[exceptions to a general rule, while sometimes a helpful interpretive guide, do not in themselves delineate the scope of the rule"); United States v. Vonn, 535 U.S. 55, 65 (2002) (Souter, J., for the majority) (canon is "only a guide").

(180.) Sebelius v. Cloer, 569 U.S. 369, 381 (2013) (Sotomayor, J., for the majority).

(181.) Rapanos v. United States, 547 U.S. 715, 731 n.3 (2006) (Scalia J., for a plurality) (stating that petitioners' interpretation "would preserve the traditional import of the qualifier 'navigable' in the defined term 'navigable waters,' at the cost of depriving the qualifier 'of the United States' in the definition of all meaning" (emphasis omitted)); see also Sossamon v. Texas, 563 U.S. 277, 291 n.8 (2011) (noting that common law canon yielded to presumption against waiver of sovereign immunity).

(182.) E.g., Bridge v. Phx. Bond & Indem. Co., 553 U.S. 639, 652 (2008) ("[T]he key term ... is a defined term, and Congress defined the predicate act not as fraud simpliciter, but mail fraud--a statutory offense unknown to the common law. In these circumstances, the presumption that Congress intends to adopt the settled meaning of common-law terms [fraud, in particular] has little pull." (emphasis omitted)).

(183.) E.g., Samantar v. Yousuf, 560 U.S. 305, 320 (2010) (Stevens, J., for the majority) ("The canon of construction that statutes should be interpreted consistently with the common law helps us interpret a statute that clearly covers a field formerly governed by the common law. But the canon does not help us decide the antecedent question whether ... Congress intended the statute to govern a particular field...." (footnote omitted)).

(184.) Jesinoski v. Countrywide Home Loans, Inc., 135 S. Ct. 790, 793 (2015) (Scalia, J., for the unanimous Court) ("Nothing in our jurisprudence, and no tool of statutory interpretation, requires that a congressional Act must be construed as implementing its closest common-law analogue.... To the extent [section] 1635(b) alters the traditional process for unwinding such a unilaterally rescinded transaction, this is simply a case in which statutory law modifies commonlaw practice.").

(185.) Compare United States v. Jicarilla Apache Nation, 564 U.S. 162,177 (2011) (Alito, J., for the majority) (declining to utilize common law trust principles in Indian law setting: "The Government assumes Indian trust responsibilities only to the extent it expressly accepts those responsibilities by statute."), with id. at 195, 204 (Sotomayor, J., dissenting) (arguing that "the statutory regime governing the United States' obligations with regard to Indian trust funds 'bears the hallmarks of a conventional fiduciary relationship,' " and thus "application of common-law trust principles" is appropriate (quoting United States v. Navajo Nation, 556 U.S. 287, 301 (2009))).

(186.) Corley v. United States, 556 U.S. 303, 325 (2009) (Alito, J., dissenting for four justices) ("There are times when Congress enacts provisions that are superfluous, and this may be such an instance."); United States v. Atl. Research Corp., 551 U.S. 128, 137 (2007) (Thomas, J., for the unanimous Court) (stating that the rule against surplusage does not require "avoid[ing] surplusage at all costs. It is appropriate to tolerate a degree of surplusage"); Arlington Cent. Sch. Dist. Bd. of Educ. v. Murphy, 548 U.S. 291, 299 n.l (2006) (Alito, J., for the majority) (refusing to find that statute clearly notified school districts not in compliance with IDEA that they might be liable for parents' expert witness fees; "[T]he reference [to costs, in addition to expenses] may be surplusage. While it is generally presumed that statutes do not contain surplusage, instances of surplusage are not unknown.").

(187.) Atl. Research, 551 U.S. at 137 (arguing that surplus language performed significant "clarifying" function regarding the statute's coverage).

(188.) Tex. Dep't of Hous. 8t Cmty. Affairs v. Inclusive Cmtys. Project, Inc., 135 S. Ct. 2507,2540-41 (2015) (Alito, J., dissenting for four justices).

(189.) Kawasaki Risen Kaisha Ltd. v. Regal-Beloit Corp., 561 U.S. 89, 122 n.6 (2010) (Sotomayor, J., dissenting).

(190.) Microsoft Corp. v. i4i Ltd. P'ship, 564 U.S. 91,107 (2011) (Sotomayor, J., for the majority) ("[T]he kind of excess language that Microsoft identifies [in the statute] is hardly unusual.").

(191.) Id. at 106-07 ("[N]o interpretation ... avoids excess language.").

(192.) See Scheidler v. Nat'l Org. for Women, Inc., 547 U.S. 9, 21-22 (2006) (Breyer, J., for the unanimous Court) (holding that the Hobbs Act prohibited physical violence only in fur therance of a plan to engage in robbery or extortion, not all physical violence, commenting that purportedly superfluous words prohibiting "violence and threats of violence" either addressed the "small" issues identified or were enacted "to emphasize the problem of violence").

(193.) Abbott v. United States, 562 U.S. 8, 27 (2010) (Ginsburg, J., for the unanimous Court) (quoting United States v. Studifin, 240 F.3d 415, 423 (4th Cir. 2001)) ("We do not regard this allowance [for future enactments] as 'implausible.' ").

(194.) Kawashima v. Holder, 565 U.S. 478,487 (2012) (Thomas, J., for the majority) (refusing to apply rule against surplusage to limit "plain language" elsewhere; "We think it more likely that Congress specifically included tax evasion offenses under [the other clause] to remove any doubt that tax evasion qualifies as an aggravated felony").

(195.) 136 S. Ct. 958, 965 (2016) (citing multiple cases dating from 1920 through 2005). Lockhart was decided in October Term 2015 and was thus beyond the scope of the study.

(196.) Lockhart, 136 S. Ct. at 970 (quoting Jama v. Immigration & Customs Enft, 543 U.S. 335, 344 n.4 (2005), and Series Qualifier Canon, BLACK'S LAW DICTIONARY (10th ed. 2014)). Scalia and Garner list the "series qualifier" canon, though they caution that it is "highly sensitive to context." Scalia & Garner, supra note 7, at 147, 150. It was not among the canons tracked in this study.

(197.) Hillman v. Maretta, 569 U.S. 483 (2013).

(198.) Chamber of Commerce v. Whiting, 563 U.S. 582,608 (2011).

(199.) Merrill Lynch, Pierce, Fenner & Smith Inc. v. Dabit, 547 U.S. 71, 87 (2006).

(200.) Arizona v. Inter Tribal Council of Ariz., Inc., 570 U.S. 1, 13-14 (2013). In addition, the justices have split over whether the presumption against preemption should be applied to statutory language expressly preempting statute law. Altria Grp., Inc. v. Good, 555 U.S. 70, 99 (2008) (Thomas, J., dissenting) ("Since Cipollone, the Court's reliance on the presumption against pre-emption has waned in the express pre-emption context."). In CTS Corp. v. Waldburger, three justices would have applied the presumption against preemption to express preemption language, 134 S. Ct. 2175, 2188-89 (2014) (Kennedy, J., joined by Sotomayor and Kagan, JJ.), while four justices claimed that applying the presumption to express pre-emption provisions was novel and inappropriate, id. at 2189 (Roberts, C.J., joined by Scalia, Thomas, and Alito, JJ.).

(201.) E.g., 1 WILLIAM BLACKSTONE, COMMENTARIES *87 (with respect to the "construction of all remedial statutes ... it is the business of the judges so to construe the act, as to suppress the mischief and advance the remedy"); see also Galveston R.R. v. Cowdrey, 78 U.S. (11 Wall.) 459, 475 (1870) (applying canon to statute governing sale of railroad bonds: "It is a remedial law for the benefit of creditors, and should be liberally construed"). Perhaps in recognition of its current status, however, Bressman and Gluck did not ask congressional staff about this canon. Gluck & Bressman, Statutory Interpretation, Part I, supra note 14, at 940 (listing substantive canons included in survey).

(202.) Waldburger, 134 S. Ct. at 2185 (quoting Rodriguez v. United States, 480 U.S. 522, 525-26 (1987) (per curiam)) (joined by Roberts, C.J., Scalia, Thomas, Alito, Sotomayor, and Kagan, JJ.); see also SCALIA & GARNER, supra note 7, at 364-65 (criticizing this canon as "false notion" and observing, rhetorically, "[i]s any statute not remedial?").

(203.) Of the four majority opinions interpreting remedial statutes broadly, two of them seem better characterized as applying statute-specific instructions. E.g., Boyle v. United States, 556 U.S. 938, 944 (2009) (applying RICO-specific instruction to interpret statute broadly); Burwell v. Hobby Lobby Stores, Inc., 134 S. Ct. 2751, 2772-73 (2014) (applying RLUIPA instruction to construe statute "in favor of a broad protection of religious exercise" (quoting 42 U.S.C. [section] 2000cc-3(g) (2012))). The other two opinions are Kasten v. Saint-Gobain Performance Plastics Corp., 563 U.S. 1, 12-14 (2011) and Atlantic Sounding Co. v. Townsend, 557 U.S. 404, 415-17 (2009). Excluding Boyle and Burwell from the numbers above, we could say that the remedial purposes canon was considered in 0.7% of issues and applied with approval in 0.3% of issues. (3 of 904 issues analyzed in any opinion; 2 majority opinions and 1 concurrence.). See Appendix Tables 1,2.

(204.) 563 U.S. 1, 13 (2011) (Breyer, J., for the majority) (quoting Tenn. Coal, Iron & R.R. v. Muscoda Local No. 123, 321 U.S. 590, 597 (1944)) (interpreting Fair Labor Standards Act broadly in light of its "remedial and humanitarian ... purpose").

(205.) Some also view the Native American/Indian canon as dying on the ground that the Court does not mention it in cases when one might expect it to appear. E.g. Carcieri v. Salazar, 555 U.S. 379,413-14 (2009) (Stevens, J., dissenting) ("[T]he Court ignores the 'principle deeply rooted in [our] Indian jurisprudence' that 'statutes are to be construed liberally in favor of the Indians.' " (quoting County of Yakima v. Confederated Tribes & Bands of the Yakima Indian Nation, 502 U.S. 251, 269 (1992)). It appeared in only one majority opinion and one dissent over the study period, both times applied with approval. See also supra note 134 (discussing briefing of this canon).

(206.) Gluck, supra note 113, at 765. Krishnakumar, in her reading of opinions from 20062012, also identified four new canons or new modifications of existing ones. Krishnakumar, Reconsidering, supra note 23, at 855-56 & 856 n.140 (identifying four canons, including equal state sovereignty principle; clear statement for jurisdictional rules; background norm denying inventors patent rights if public has paid for research; clear statement for military commission jurisdiction).

(207.) E.g., Jacob Loshin & Aaron Nielson, Hiding Nondelegation in Mouseholes, 62 Admin. L. Rev. 19 (2010). Bruhl has argued that this canon could be understood to have antecedents in earlier cases. See Aaron-Andrew P. Bruhl, Communicating the Canons: How Lower Courts React When the Supreme Court Changes the Rules of Statutory Interpretation, 100 MINN. L. REV. 481, 543 (2015) ("[Scalia] did not invent the sensible idea that one should hesitate before finding a serious change in the law ... hiding in an unassuming, easy-to-miss provision.").

(208.) 531 U.S. 457, 468-69 (2001) (Scalia, J., for eight justices). The opinion cited MCI Telecommunications Corp. v. American Telephone & Telegraph Co., 512 U.S. 218, 231 (1994), a case applying the major question canon. As noted, the "no elephants in mouseholes" canon can be seen as linked to the major question canon. See Loshin & Nielson, supra note 207, at 61 n.255.

(209.) See Appendix Table 1.

(210.) See Appendix Table 1.

(211.) Data on file with author.

(212.) E.g., Kellogg Brown & Root Servs., Inc. v. United States ex rel. Carter, 135 S. Ct. 1970, 1977 (2015) ("[I]t is significant that Congress chose to place the WSLA in Title 18."); United States v. Kwai Fun Wong, 135 S. Ct. 1625, 1633 (2015) ("Whereas [section] 2401(b) houses the FTCA's time limitations, a different section of Title 28 confers power on federal district courts to hear FTCA claims. Nothing conditions the jurisdictional grant on the limitations periods, or otherwise links those separate provisions. Treating [section]2401(b)'s time bars as jurisdictional would thus disregard the structural divide built into the statute." (citations omitted)); Adoptive Couple v. Baby Girl, 570 U.S. 637, 652 (2013) ("Our interpretation of (d) is also confirmed by the provision's placement next to [section][section] 1912(e) and (f), both of which condition the outcome of proceedings on the merits of an Indian child's "continued custody' with his parent."); United States v. O'Brien, 560 U.S. 218, 234 (2010) (Kennedy, J., for eight justices) ("Had Congress intended to treat firearm type as a sentencing factor, it likely would have listed firearm types as clauses (iv) and (v) of subparagraph (A) [proximate to other sentencing factors]. By listing firearm type in stand-alone subparagraph (B), Congress set it apart ... ; this is consistent with preserving firearm type as an element of a separate offense."); Nken v. Holder, 556 U.S. 418, 431 (2009) ("Although the dissent 'would not read too much into Congresses] decision to locate such a provision in one subsection rather than in another,' the Court frequently takes Congress's structural choices into consideration when interpreting statutory provisions." (citation omitted) (quoting id. at 446 (Alito, J., dissenting))); see also Lockhart v. United States, 136 S. Ct. 958, 964 (2016) (suggesting Congress might have used section headings in a US Code chapter as a template for a list of crimes that would lead to sentence enhancement; "Congress followed the federal template") (Sotomayor, J., for the majority); Nat'l Fed'n of Indep. Bus. v. Sebelius, 567 U.S. 519, 668 (2012) (Scalia, Kennedy, Thomas, and Alito, JJ., dissenting) ("And the nail in the coffin is that the mandate and penalty are located in Tide I of the [Affordable Care] Act, its operative core, rather than where a tax would be found--in Title IX, containing the Act's 'Revenue Provisions.' In sum, 'the terms of [the] act rende[r] it unavoidable' that Congress imposed a regulatory penalty, not a tax." (quoting Parsons v. Bedford, 28 U.S. (3 Pet.) 433, 448 (1830)). Lockhart was decided after the study period.

(213.) 135 S. Ct. 1074,1084 (2015) (plurality opinion).

(214.) Bilski v. Kappos, 561 U.S. 593, 645 (2010) (Stevens, J., concurring).

(215.) E.g., Gonzalez v. Thaler, 565 U.S. 134, 146-47 (2012) ("[T]he United States argues that the placement of [section]2253(c)(3) in a section containing jurisdictional provisions signals that it too is jurisdictional. In characterizing certain requirements as nonjurisdictional, we have on occasion observed their 'separation]' from jurisdictional provisions.... The converse, however, is not necessarily true: Mere proximity will not turn a rule that speaks in nonjurisdictional terms into a jurisdictional hurdle. In fact, [section]2253(c)(3)'s proximity to [section][section]2253(a), (b), and (c)(1) highlights the absence of clear jurisdictional terms in [section]2253(c)(3)." (quoting Reed Elsevier, Inc. v. Muchnick, 559 U.S. 154, 162 (2010))); United States v. O'Brien, 560 U.S. 218, 235 (2010) ("These points [regarding the positioning of the machine gun provision] are overcome, however, by the substantial weight of the other Castillo factors and the principle that Congress would not enact so significant a change without a clear indication of its purpose to do so.").

(216.) Kagan objected in part based on the omission of the canon by Scalia and Garner. See Yates, 135 S. Ct. at 1095 (Kagan, J., dissenting) ("As far as I can tell, this Court has never once suggested that the section number assigned to a law bears upon its meaning. Cf. Scalia [& Garner], at xi-xvi (listing more than 50 interpretive principles and canons without mentioning the plurality's new number-in-the-Code theory).").

(217.) Positive Law Codification, OFF. L. REVISION COUNS.: U.S. Code http:// []; see also Gluck, supra note 4, at 208-09 (describing this office's involvement as an "underbelly of Congress").

(218.) 562 U.S. 428, 441 (2011) (quoting King v. St. Vincent's Hosp., 502 U.S. 215, 220 n.9 (1991)); see Brown v. Gardner, 513 U.S. 115,118 (1994); see also Krishnakumar, Reconsidering, supra note 23, at 835 (listing canon).

(219.) 328 U.S. 275, 285 (1946). The other cases the Court cited for the proposition were King v. St. Vincent's Hospital, 502 U.S. 215, 220-21, 220 n.9 (1991) (interpreting the Veterans' Reemployment Rights Act), and Coffy v. Republic Steel Corp., 447 U.S. 191, 196 (1980) (interpreting the Veterans' Readjustment Assistance Act of 1974). Le Maistre v. Leffers, 333 U.S. 1 (1948), states a similar principle, interpreting the Soldiers' and Sailors' Civil Relief Act of 1940 liberally in favor of veterans. Id. at 4, 6 ("[T]he Act must be read with an eye friendly to those who dropped their affairs to answer their country's call."); see also Boone v. Lightner, 319 U.S. 561,575 (1943) (similar comment regarding same statute, but ruling against veteran). Supreme Court cases prior to the 1940s appear to have rejected any such canon. E.g., United States v. Towery, 306 U.S. 324, 329 (1939) (ruling in favor of repose and against veteran in interpreting time limitation on suit on war risk policy, overruling appellate decision to construe statute "to comport with liberal policy of Congress towards veterans").

(220.) See generally Linda D. Jellum, Heads I Win, Tails You Lose: Reconciling Brown v. Gardner's Presumption that Interpretive Doubt Be Resolved in Veterans' Favor with Chevron, 61 Am. U. L. Rev. 59,65-70 (2011) (arguing that canon began life as a variant of remedial purposes canon).

(221.) The only explicit reasons given for the modern form of the veterans' benefit canon appear to have come from Justice Stevens, in dissent. See Hooper v. Bernalillo Cty. Assessor, 472 U.S. 612, 626 & n.3 (1985) (Stevens, J., dissenting) (arguing that use of canon "express[es] gratitude" for sacrifices made by veterans and supports their reentry into civilian society).

(222.) See 3 SUTHERLAND STATUTORY CONSTRUCTION, supra note 88, [section] 60.2, at 286 n.60 (citing Smith v. Brown, 35 F.3d 1516 (Fed. Cir. 1994)).

(223.) Gonzalez v. Thaler, 565 U.S. 134, 141-42 (2012) (quoting Arbaugh v. Y & H Corp., 546 U.S. 500,515 (2006)).

(224.) Arbaugh v. Y & H Corp., 546 U.S. 500,516 (2006).

(225.) Id.

(226.) Hamer v. Neighborhood Hous. Servs. of Chi., 138 S. Ct. 13, 20 n.9 (2017) (noting application of "clear-statement rule"); United States v. Kwai Fun Wong, 135 S. Ct. 1625, 1632 (2015) ("[T]he Government must clear a high bar to establish that a statute of limitations is Jurisdictional ... cabin[ing] a court's power only if Congress has 'clearly state[d]' as much." (quoting Sebelius v. Auburn Reg'l Med. Ctr., 568 U.S. 145, 153 (2013))); Stern v. Marshall, 564 U.S. 462, 479-80 (2011) ("Because '[branding a rule as going to a court's subject-matter jurisdiction alters the normal operation of our adversarial system' ... we are not inclined to interpret statutes as creating a jurisdictional bar when they are not framed as such." (quoting Henderson v. Shinseki, 562 U.S. 428, 434 (2011))); Henderson v. Shinseki, 562 U.S. 428, 435-36 (2011) ("[W]e look to see if there is any 'clear' indication that Congress wanted the rule to be 'jurisdictional.'" (quoting Arbaugh, 546 U.S. at 515-16)).

(227.) See supra text accompanying note 211.

(228.) Arbaugh, 546 U.S. at 515-16, which seems to be the first appearance of the canon, cites no authority for it. In addition, the following Westlaw search in the Supreme Court database turned up no appearances of the canon prior to Arbaugh: ("clearly states" or "clearly state" or "clear statement") Is jurisdictional. Scalia and Garner do mention a "false notion that a statute cannot oust courts of jurisdiction unless it does so expressly," SCALIA & GARNER, supra note 7, at 367-68, but this appears to relate to a different issue--interpreting statutes to foreclose judicial review altogether, including of executive action.

(229.) Kontrick v. Ryan, 540 U.S. 443, 454-55 (2004) (Ginsburg, J., for the unanimous Court).

(230.) SCALIA & GARNER, supra note 7, at 296.

(231.) Burgess v. United States, 553 U.S. 124, 136 (2008) (Ginsburg, J, for the unanimous Court) (canon not applied).

(232.) United States v. Hayes, 555 U.S. 415, 429 (2009) (emphasis added) (Ginsburg, J., for the majority) (quoting Huddleston v. United States, 415 U.S. 814, 831 (1974)); Dean v. United States, 556 U.S. 568, 577 (2009) (finding the statute was not "grievously ambiguous") (Roberts, C.J., for the majority); see also Lockhart v. United States, 136 S. Ct. 958, 968 (2016) ("[T]he arguable availability of multiple, divergent principles of statutory construction cannot automatically trigger the rule of lenity.").

(233.) Burrage v. United States, 134 S. Ct. 881,891 (2014).

(234.) Compare id. at 891 (Scalia, J., for the majority) (seeming to apply canon to statute's "ordinary, accepted meaning"), with Robers v. United States, 134 S. Ct. 1854, 1859 (2014) (Breyer, J., for the majority) ("[R]ule of lenity applies only if, after using the usual tools of statutory construction, we are left with a 'grievous ambiguity or uncertainty....'" (quoting Mus carello v. United States, 524 U.S. 125,139 (1998))).

(235.) Scalia and Garner particularly note the uncertainty stemming from the "multiplicity in expressed standards" around the rule of lenity. SCALIA & GARNER, supra note 7, at 298-99.

(236.) 569 U.S. 184,205(2013).

(237.) Moncrieffe, 569 U.S. at 189-90.

(238.) Id. at 205. In another case, the Court acknowledged its practice of "constru[ing] ambiguities in deportation statutes in the alien's favor," though it did not apply the rule of lenity because it found the statute unambiguous. Kawashima v. Holder, 565 U.S. 478, 489 (2012) (acknowledging, but not applying canon, since "[i]t is true that we have, in the past, construed ambiguities in deportation statutes in the alien's favor.... We think the application of the present statute clear enough that resort to the rule of lenity is not warranted").

(239.) Albernaz v. United States, 450 U.S. 333, 342 (1981). Despite this rationale, Scalia and Garner note an apparent incongruity; the rule of lenity has thus far attached neither to tax obligations nor to statutes imposing remedies for fraud. SCALIA & GARNER, supra note 7, at 299300.

(240.) Pildes, supra note 4, at 895 8t n.28 (citing National Railroad Passenger Corp. v. National Association of Railroad Passengers, 414 U.S. 453, 458 (1974) and noting that the court cited an "obscure 1929 case" to justify the canon's use).

(241.) 2A SUTHERLAND STATUTORY CONSTRUCTION, supra note 88, [section]47:23, at 406-13 (footnotes omitted).

(242.) See Barnhart v. Peabody Coal Co., 537 U.S. 149,168 (2003) (quoting United States v. Vonn, 535 U.S. 55, 65 (2002)) ("We do not read the enumeration of one case to exclude another unless it is fair to suppose that Congress considered the unnamed possibility and meant to say no to it. As we have held repeatedly, the canon expressio unius est exclusio alterius does not apply to every statutory listing or grouping; it has force only when the items expressed are members of an 'associated group or series,' justifying the inference that items not mentioned were excluded by deliberate choice, not inadvertence." (citation omitted) (quoting United States v. Vonn, 535 U.S. 55,65 (2002))); see also Eskridge, supra note 45, at 674.

(243.) E.g., Sebelius v. Cloer, 569 U.S. 369, 378 (2013) ("We have long held that '[w]here Congress includes particular language in one section of a statute but omits it in another section of the same Act, it is generally presumed that Congress acts intentionally and purposely in the disparate inclusion or exclusion.'" (quoting Bates v. United States, 522 U.S. 23,29-30 (1997))).

(244.) E.g., Descamps v. United States, 570 U.S. 254, 267-68 (2013) ("If Congress had wanted to increase a sentence based on the facts of a prior offense, it presumably would have said so; other statutes, in other contexts, speak in just that way." (citing case applying immigration statute)).

(245.) 560 U.S. 438, 450 (2010); see also Astrue v. Ratliff, 560 U.S. 586, 595 (2010) ("Given the stark contrast between the SSA's express authorization of direct payments to attorneys and the absence of such language in subsection (d)(1)(A), we are reluctant to interpret the later provision to contain a direct fee requirement.... ").

(246.) 562 U.S. 223,230 (2011) (quoting 42 U.S.C. [section] 300aa-22(b) (1)).

(247.) Bruesewitz v. Wyeth LLC, 562 U.S. at 232.

(248.) Id. at 233.

(249.) See Appendix Table 1.

(250.) See SCALIA & GARNER, supra note 7, at 168,170.

(251.) 566 U.S. 449,454-55 (2012).

(252.) Mohamad, 566 U.S. at 455 (emphasis omitted) (quoting 18 U.S.C. [section] 2331(3) (2012)).

(253.) 553 U.S. 474 (2008).

(254.) Gomez-Perez, 553 U.S. at 477,487-88.

(255.) Id. at 486 (quoting Lindh v. Murphy, 521 U.S. 320, 330 (1997)). This could also be read to impose a requirement of simultaneous enactment, though the prospect that the Court is envisioning omnibus passage of multiple sets of statutory amendments seems an unlikely one.

(256.) 546 U.S. 440, 448 n.3 (2006).

(257.) Eskridge & Frickey, supra note 24, at 619 (noting "the greater enthusiasm the Court brought to the federalism-based conditions" during the 1980s).

(258.) Id. at 621.

(259.) Id. at 622-25 (including discussion of Gregory v. Ashcroft, 501 U.S. 452 (1991)).

(260.) E.g., United States v. Bass, 404 U.S. 336, 349 (1971); see Bond v. United States, 134 S. Ct. 2077, 2087-88 (2014) (interpreting statute narrowly not to interfere with "traditional state criminal jurisdiction"); Scheidler v. Nat'l Org. for Women, Inc., 547 U.S. 9, 20 (2006) (declining to adopt statutory reading that would "federalize much ordinary criminal behavior").

(261.) E.g., Rapanos v. United States, 547 U.S. 715, 738 (2006) ("We ordinarily expect a 'clear and manifest' statement from Congress to authorize an unprecedented intrusion into traditional state authority." (quoting BFP v. Resolution Tr. Corp., 511 U.S. 531, 544 (1994))). This statement expanded upon language in a 2001 case applying the constitutional avoidance canon to avoid a potential Commerce Clause issue. Solid Waste Agency of N. Cook Cty. v. U.S. Army Corps of Eng'rs, 531 U.S. 159,172-73 (2001) (assuming that "Congress does not casually authorize administrative agencies to interpret a statute to push the limit of congressional authority. This concern is heightened where the administrative interpretation ... permit[s] federal encroachment upon a traditional state power" (citation omitted)).

(262.) See, e.g., Mac's Shell Serv., Inc., v. Shell Oil Prods. Co., 559 U.S. 175,186 (2010) (declining to interpret Petroleum Marketing Practices Act provisions on franchise relationships as covering effective termination of service station franchises, and calling for "clearer indication that Congress intended to federalize such a broad swath of the law governing petroleum franchise agreements"); see also Stoneridge Inv. Partners, LLC, v. Scientific-Atlanta, Inc., 552 U.S. 148, 161 (2008) (refusing to extend implied right of action under securities law where "there would be a risk that the federal power would be used to invite litigation ... in areas already governed by functioning and effective state-law guarantees").

(263.) Carachuri-Rosendo v. Holder, 560 U.S. 563,579-80 (2010).

(264.) Fla. Dep't. of Revenue v. Piccadilly Cafeterias, Inc., 554 U.S. 33, 48, 50 (2008) (Thomas, J., for the majority).

(265.) 565 U.S. 95, 108-09 (2012) (Sotomayor, J., concurring) (quoting 15 U.S.C. [section] 1679c(a) (2012)).

(266.) CompuCredit, 565 U.S. at 109.

(267.) 500 U.S. 20, 25 (1991) ("[The FAA's] provisions manifest a 'liberal federal policy favoring arbitration agreements.'" (quoting Moses H. Cone Mem'l Hosp. v. Mercury Constr. Corp., 460 U.S. 1, 24 (1983))), cited with approval in CompuCredit, 565 U.S. at 108 (Sotomayor, J., concurring); see also Shearson/Am. Express Inc. v. McMahon, 482 U.S. 220, 227 (1987) ("The burden is on the party opposing arbitration ... to show that Congress intended to preclude a waiver of judicial remedies for the statutory rights at issue.").

(268.) These cases address the interpretation of arbitration contracts, however, rather than statutes, and therefore do not directly suggest the existence of an interpretive canon. E.g., Mastrobuono v. Shearson Lehman Hutton, Inc., 514 U.S. 52, 62 (1995) (interpreting arbitration clause broadly in light of "federal policy favoring arbitration" (quoting Volt Info. Scis., Inc. v. Bd. of Trs. of Leland Stanford Junior Univ., 489 U.S. 468, 476 (1989))); Moses H. Cone Mem'l Hosp. v. Mercury Constr. Corp., 460 U.S. 1, 24-25 (1983) ("The Arbitration Act establishes that, as a matter of federal law, any doubts concerning the scope of arbitrable issues should be resolved in favor of arbitration, whether the problem at hand is the construction of the contract language itself or an allegation of waiver, delay, or a like defense to arbitrability."); see also AT&T Mobility LLC v. Concepcion, 563 U.S. 333, 339 (2011) (interpreting Federal Arbitration Act to preempt California state law in view of the "liberal federal policy favoring arbitration"; no consideration of presumption against preemption (quoting Moses H. Cone Mem'l Hosp., 460 U.S. at 24)).

(269.) 563 U.S. 776, 787 (2011). Later in the opinion, the Court included a cf. citation to Whitman v. American Trucking Associations, Inc., 531 U.S. 457, 468 (2001), the source of the no elephants in mouseholes canon. Bd. of Trs., 563 U.S. at 792.

(270.) 135 S. Ct. 2699 (2015).

(271.) Michigan v. EPA, 135 S. Ct. at 2707.

(272.) Id. at 2708.

(273.) "That understanding of [section] 1153(h)(3)'s 'automatic conversion' language matches the exclusive way immigration law used the term when Congress enacted the [Child Status Protection Act]." 134 S. Ct. 2191, 2204 (2014) (Kagan, J., for the majority). Although the dissent disagreed on the outcome, it quoted this language in the majority and engaged it on its own terms: "But immigration law has long allowed petitions to be converted from one category to another in contexts where doing so requires changing the sponsor's identity." Scialabba, 134 S. Ct. at 2226 (Sotomayor, J., dissenting).

(274.) E.g., Anita S. Krishnakumar, The Sherlock Holmes Canon, 84 GEO. WASH. L. REV. 1 (2016).

(275.) 550 U.S. 81 (2007).

(276.) Zuni, 550 U.S. at 90-91.

(277.) Chisom v. Roemer, 501 U.S. 380, 406 (1991) (Scalia, J., dissenting). In Zuni, Scalia expressed himself similarly: "Nor do I see any significance in the fact that no legislator in 1994 expressed the view that [the statute] was designed to upend the Secretary's [approach]." 550 U.S. at 120 (Scalia, J., dissenting).

(278.) Scialabba, 134 S. Ct. at 2215 (Roberts, C.J., concurring).

(279.) Lower courts have also occasionally created their own canons. E.g., Safe Food & Fertilizer v. EPA, 350 F.3d 1263, 1269 (D.C. Cir. 2003) (applying the " 'reverse ejusdem generis' principle occasionally invoked by this court, under which 'the phrase "A, B, or any other C" indicates that A is a subset of C' " (quoting United States v. Williams-Davis, 90 F.3d 490, 509 (D.C. Cir. 1996))).

(280.) Nelson, supra note 6, at 389.

(281.) See Gluck, supra note 4, at 209-10. Gluck does, however, note three references, in 2004 and 2010 opinions, to Office of Legislative Counsels' drafting manuals. Id. at 210 & n.145.

(282.) See supra notes 134-135 and accompanying text (discussing examination of briefs in cases invoking selected canons.

(283.) Compare Milavetz, Gallop & Milavetz, P.A., v. United States, 559 U.S. 229, 236-37 (2010) (declining to apply expressio unius but applying rule against surplusage), with Arlington Cent. Sch. Dist. Bd. of Educ. v. Murphy, 548 U.S. 291, 299 & n.l (2006) (declining to apply rule against surplusage but applying expressio unius).

(284.) Manning, supra note 31, at 2471.

(285.) See Appendix Table 4.

(286.) E.g., FDA v. Brown & Williamson Tobacco Corp., 529 U.S. 120, 159-60 (2000) ("[W]e are confident that Congress could not have intended to delegate a decision of such economic and political significance to an agency in so cryptic a fashion.").

(287.) See supra note 162 and accompanying text.

(288.) See Appendix Table 1.

(289.) See supra note 22 and accompanying text.

(290.) E.g., Scheidler v. Nat'l Org. for Women, Inc., 547 U.S. 9, 20-22 (2006) (applying federalism canon to Hobbs Act to avoid "federaliz[ing] much ordinary criminal behavior," but rejecting surplusage canon in light of possible need for Congress to "emphasize the problem of violence") (Breyer, J., for the unanimous Court).

(291.) E.g., Loughrin v. United States, 134 S. Ct. 2384, 2390-93 (2014) (applying surplusage canon in rejecting argument that "intent to defraud a bank" was needed for conviction, but rejecting federalism argument for requiring such intent to avoid extending statute into gardenvariety fraud claims, the usual focus of state prosecutors).

(292.) See supra note 283 and accompanying text.

(293.) For example, the presumption against extraterritoriality and the federalism canon, which counsel against reading against a statute to entrench on state prerogatives or apply extraterritoriality, might seem to call for an assessment of a statute's plain language prior to other canons. E.g., Bond v. United States, 134 S. Ct. 2077, 2090 (2014) ("[W]e can insist on a dear indication that Congress meant to reach purely local crimes, before interpreting the statute's expansive language in a way that intrudes on the police power of the States."); Morrison v. Nat'l Austl. Bank Ltd., 561 U.S. 247,255 (2010) ("When a statute gives no clear indication of an extraterritorial application, it has none.").

(294.) Richlin Sec. Serv. Co. v. Chertoff, 553 U.S. 571, 589-90 (2008) (Alito, J., for the majority) (declining to use canon to narrowly read the right to recover paralegal fees under EAJA).

(295.) E.g., STEPHEN G. BREYER ET AL., ADMINISTRATIVE LAW AND REGULATORY POLICY 313 (8th ed. 2017) ("In several cases, the Court has held that canons of construction trump Chevron."); Kenneth A. Bamberger, Normative Canons in the Review of Administrative Policymaking, 118 YALE L.J. 64, 74-77 (2008). The status of some substantive canons within the Chevron framework remains unclear. E.g., Carter v. Welles-Bowen Realty, Inc., 736 F.3d 722, 729-35 (6th Cir. 2013) (Sutton, J., concurring) (discussing, in dicta, an approach to reconciling the rule of lenity with the Chevron doctrine); Mendelson, supra note 52, at 478 n.48. But see Solid Waste Agency of N. Cook Cty. v. United States Army Corps of Eng'rs, 531 U.S. 159 (2001) (constitutional avoidance canon trumped Chevron). See generally Gluck, supra note 4, at 197 (arguing that "a contingent of federal judges ... [has been] strongly signaling their desire to reclaim some of that power [to interpret statutes despite Chevrons implications]").

(296.) See, e.g., Abbott v. United States, 562 U.S. 8, 28 n.9 (2010) (declining to apply rule of lenity because no ambiguity remained after application of other canons); supra note 234 (comparing use of rule of lenity in Burrage v. United States with Robers v. United States).

(297.) See supra notes 170-171 and accompanying text.

(298.) See Appendix Table 1.

(299.) By contrast, two canons that might be considered inversely related are noscitur a sociis and the rule against surplusage, since the more likely a word is to be interpreted to share a character similar to the others on a list, the greater the risk that the word will not retain a "character of its own not to be submerged by its association." Babbitt v. Sweet Home Chapter of Cmtys. for a Great Or., 515 U.S. 687, 702 (1995) (quoting Russell Motor Car Co. v. United States, 261 U.S. 514, 519 (1923)).

(300.) The hypothetical uses these canons because they seem likely to occur in the same setting, but even pairing commonly occurring canons with the strongest approval rates, such as the harmonious reading canon (75%) and the whole act rule (76%), see Appendix Table 1, the numbers are not much more encouraging. The odds of both applying are 57%; the odds of neither applying are 6%; the odds of a particular one applying, and not the other, are 19% (with the whole act rule applying) and 18% (with harmonious reading applying).

(301.) But see Anita S. Krishnakumar, Textualism and Statutory Precedents, 104 VA. L. REV. (forthcoming 2018), [] (suggesting that statutory stare decisis may be yielding to textualist methods).

(302.) Fallon has suggested that a judge's sense of dissonance in plain statutory language will prompt the judge to invoke further interpretive tools and that the judge's values will influence which implications are considered dissonant. Richard H. Fallon, Jr., Three Symmetries Between Textualist and Purposivist Theories of Statutory Interpretation--And the Irreducible Roles of Values and Judgment Within Both, 99 Cornell L. Rev. 685, 693 (2014) ("[T]he role of values is irreducible in triggering apprehensions that statutory language really may not mean what at first blush it might appear to mean----").

(303.) 566 U.S. 624 (2012).

(304.) Freeman, 566 U.S. at 627 (quoting 12 U.S.C. [section] 2607(b) (2012)).

(305.) Id. at 628-29 (quoting 12 U.S.C. [section] 2607(b) (2012)).

(306.) 12 U.S.C. 2607(a) (2012).

(307.) Id. [section] 2607.

(308.) Freeman, 566 U.S. at 629-30.

(309.) Id. at 635-36. Because the reading allegedly deprived an entire section of meaning, Scalia and Garner would characterize it as implicating the harmonious reading canon, see SCALIA & GARNER, supra note 7, at 180, but Justice Scalia's majority opinion characterized it as a surplusage problem. Freeman, 566 U.S. at 635-36.

(310.) Id., 566 U.S. at 636.

(311.) Id. at 635-36 (quoting [section] 2607). One might have thought the Court would mention the headings for the individual subsections: "(a) Business referrals," and "(b) Splitting charges," which would have supported its conclusion, but these garnered nary a mention. [section] 2607.

(312.) Freeman, 566 U.S. at 630-32.

(313.) Id. at 633-34.

   Petitioners rightly point out that under our interpretation
   "portion," "split," and "percentage" all mean the same thing--a
   perhaps regrettable but not uncommon sort of lawyerly iteration ...
   But the canon against surplusage merely favors that interpretation
   which avoids surplusage ... and petitioners' interpretation [also
   has surplusage].... It is impossible to imagine a "portion" ... or
   a "split" that is not also a "percentage."

Id. at 635 (quoting [section] 2607). The Court rejected a second surplusage argument as well. See id. at 637 n.9 (rejecting the government's argument from surplusage: "[T]he phrase does not have the significance attributed to it by the United States").

(315.) For example, the lenders argued that a literal reading of the section would mean that consumers would be liable for paying the unearned fee; the government argued that it would be absurd for a service provider to "charge and keep the entirety of a $1,000 unearned fee, while imposing liability if the provider shares even a nickel of a $10 charge with someone else." Id. at 638. The Court answered that the consumers' reading had its own absurdity problem: "a service provider could avoid liability by providing just a dollar's worth of services" and at least the lenders' reading had the virtue of a "coherent response to that particular problem." Id.

(316.) Id. at 631.

(317.) Cf. Brudney & Ditslear, supra note 2, at 98 (noting that canon use may make it harder for legislative purpose evidence to be deemed "sufficient" for a particular interpretation); Gluck, supra note 4. Though assessing legislative history as an interpretive tool is beyond the scope of this project, commentators have argued that a more sensitive use of those materials is superior to abandoning them altogether. E.g., Victoria F. Nourse, A Decision Theory of Statutory Interpretation: Legislative History by the Rules, 122 YALE L.J. 70 (2012).

(318.) SCALIA & GARNER, supra note 7, at 59-60; e.g., Lockhart v. United States, 136 S. Ct. 958, 963 (2016) ("[A]s with any canon of statutory interpretation, the rule of the last antecedent 'is not an absolute and can assuredly be overcome by other indicia of meaning.'" (quoting Barnhart v. Thomas, 540 U.S. 20,26 (2003))).

(319.) As Gluck has documented, several state supreme courts have undertaken this task, with some seeming success, though they do not appear to have established hierarchies among canons. Abbe R. Gluck, The States as Laboratories of Statutory Interpretation: Methodological Consensus and the New Modified Textualism, 119 YALE L.J. 1750 (2010). It is unclear, from Gluck's work, how important canons are to these hierarchies. Moreover, even a better hierarchy would not address lack of clarity in the canons themselves or the difficulty of assuming that a particular enacting legislature has acquiesced in a canon.

(320.) Nelson, supra note 6, at 389.

(321.) Shapiro, supra note 46 (arguing that certain canons foster "interpretations that do not alter relationships any more than is necessary to achieve the statutory objectives"). See generally Gluck, supra note 113, at 769 (whether certain canons can be justified may depend on how expansively one views the judicial role, including the ability of courts to update obsolete statutes or reinforce constitutional norms, and noting that certain canons simply embody "the way in which courts use interpretive methodology to retain some law-making power for themselves in a changing legal world").

(322.) Baude & Sachs, supra note 33, at 1123.

(323.) E.g., Gluck & Bressman, Statutory Interpretation, Part I, supra note 14 (considering whether particular canons could be justified as "approximating" congressional practices); Manning, supra note 53; Manning, supra note 31.

(324.) Cf. Manning, supra note 53, at 425 (arguing that seeing clear statement rules as resulting from "nondelegation canon" does not relieve court of explaining "why it deploys such a canon in some contexts but not others").

(325.) Cf. Eskridge, supra note 45, at 681 (noting "equity" and "justice" as alternative interpretive goals); Vermeide, supra note 59, at 89 ("[T]he relevant theory of authority [may] specify a minimum of accuracy that permissible interpretive doctrines must achieve----").

(326.) Bressman and Gluck explain, based on their findings, that the consistency rules (whole code and whole act) fail to approximate drafting practices for a variety of reasons, including the scope of committee jurisdiction. Gluck & Bressman, Statutory Interpretation, Part I, supra note 14, at 936-38. Meanwhile, redundancy is a regular feature of drafting, undercut ting the rule against surplusage. Id. at 934-36; e.g., James J. Brudney, Faithful Agency Versus Ordinary Meaning Advocacy, 57 ST. LOUIS U. L.J. 975, 986 (2013) ("[P]resumptions that statutes should be understood as structurally integrated with no surplus phrases or provisions is at odds with the drafting realities that produce Congress's complex statutory schemes such as ERISA [or] Title VII...."). But see Brudney & Ditslear, Warp and Woof, supra note 23, at 1299-1300 (suggesting that for tax statutes, "[t]he Court's tendency to assume thoroughness and cohesion in tax statutory language may in turn be less naive or aspirational given the bipartisan, interbranch, and professionalized drafting process that has long characterized the tax area").

(327.) Gluck & Bressman, Statutory Interpretation, Part I, supra note 14, at 932.

(328.) Gluck and Bressman's substantive question on expressio unius asked for drafters' level of agreement with the statement, "[W]hen terms appear in a statutory list, to what extent are any other terms intended to be excluded from that list?" See Abbe R. Gluck & Lisa Schultz Bressman, Statutory Interpretation from the Inside: Methods Appendix, 65 STAN. L. REV. 901 app. at 33 (2013), [https://]. Gluck and Bressman apparently did not question staff regarding other, more expansive variants of expressio unius in use in the Roberts Court, making comparisons more difficult.

(329.) Gluck & Bressman, Statutory Interpretation, Part I, supra note 14, at 940-42.

(330.) Id. at 958-59.

(331.) See Appendix Table 1.

(332.) See Appendix Table 1.

(333.) Gluck & Bressman, Statutory Interpretation, supra note 14, at 933.

(334.) Id. at 933. Gluck and Bressman's substantive question to drafters on these canons was fairly general: "[T]o what extent do the terms in a statutory list relate to one another?" Gluck & Bressman, supra note 328, at 32. Staff also identified the "major question" canon as one they were comfortable with, but again, the study was unclear on what might make up a "major" question. Gluck & Bressman, Statutory Interpretation, Part I, supra note 14, at 1003; see Gluck & Bressman, supra note 328, at 37.

(335.) Gluck & Bressman, Statutory Interpretation, Part I, supra note 14, at 1016 tbl.3. Quantitative data are derived from the author's research. See Appendix Table 1.

(336.) The Scalia and Garner treatise offers analysis, but it is often unsatisfying. For example, the authors justify the rule against surplusage as making "intuitive sense," in view of expectations for drafting, but then nearly immediately concede that the canon "assumes a perfection of drafting that... is not often achieved," and thus should be seen as "particularly defeasible by context." SCALIA & GARNER, supra note 7, at 170-71. For expressio unius, they suggest, "We encounter the device ... frequently in our daily lives," but nonetheless caution that it "must be applied with great caution, since its application depends so much on context." Id. at 107.

(337.) Chevron U.S.A. Inc. v. Nat. Res. Def. Council, Inc., 467 U.S. 837, 865-866 (1984) (discussing agency expertise and political accountability, commenting, "Perhaps [Congress] consciously desired the Administrator to strike the balance at this level, thinking that those with great expertise and charged with responsibility for administering the provision would be in a better position to do so; perhaps it simply did not consider the question at this level; and perhaps Congress was unable to forge a coalition on either side of the question----[Meanwhile, j Judges are not experts in the field, and are not part of either political branch of the Government").

(338.) Henderson v. Shinseki, 562 U.S. 428, 434 (2011), quoted in Stern v. Marshall, 564 U.S. 462, 479-80 (2011).

(339.) Whitman v. Am. Trucking Ass'ns, 531 U.S. 457, 468 (2001). The Court cited two other cases; both were similarly conclusory in their assumption about Congress. See FDA v. Brown & Williamson Tobacco Corp., 529 U.S. 120, 159-60 (2000) ("[W]e are confident that Congress could not have intended to delegate a decision of such economic and political significance to an agency in so cryptic a fashion."); MCI Telecomms. Corp. v. AT&T Co., 512 U.S. 218, 231 (1994) ("It is highly unlikely that Congress would leave the determination of whether an industry will be entirely, or even substantially, rate-regulated to agency discretion--and even more unlikely that it would achieve that through such a subtle device as permission to 'modify' rate-filing requirements.").

(340.) Gluck has also noted reasons to abandon the rule against surplusage. See Gluck, supra note 4, at 190 (discussing the "rule against superfluities"). Posner expressed doubt about the rule against surplusage as early as 1983. Posner, supra note 61, at 812 ("[A] statute that is the product of compromise may contain redundant language [from] the strains of the negotiating process.").

(341.) Gluck, supra note 113, at 763-68 (also noting that some canons represent policy choices "out of thin air"); Baude & Sachs, supra note 33, at 1122 ("[T]he canons stand on their own authority as a form of common law."); see also Abbe R. Gluck, Intersystemic Statutory Interpretation: Methodology as "Law" and the Erie Doctrine, 120 YALE L.J. 1898 (2011) (considering whether, if canons are common law, federal courts ought be applying state canons as substantive rules of decision under the doctrine of Erie Railroad Co. v. Tompkins, 304 U.S. 64 (1938)).

(342.) See Gluck & Posner, supra note 4, at 1332 (noting that some of the appellate judges told the authors that the lenity and avoidance canons "are 'not canons' but rather are 'substantive law' "). But see Evan J. Criddle & Glen Staszewski, Essay, Against Methodological Stare Decisis, 102 GEO. L.J. 1573 (2014). See generally Sydney Foster, Should Courts Give Stare Decisis Effect to Statutory Interpretation Methodology?, 96 GEO. L.J. 1863 (2008).

Some have asserted that compared with other canons, the Chevron doctrine is more of a "precedent." E.g., Abbe R. Gluck, What 30 Years of Chevron Teach Us About the Rest of Statutory Interpretation, 83 FORDHAM L. REV. 607, 609 (2014). Rabb argues that the rule of lenity essentially receives the same treatment. Rabb, supra note 51, at 184 (arguing that "the Court's approach to lenity corresponds to that of Chevron"). Even if that were true as a formal matter, approval rates for both canons as documented in this study suggest that they are weak precedents at best.

This study did, however, reveal an occasional example of interpretive stare decisis. For example, the Court appears to have decided which interpretive sources must be consulted to determine whether Congress meant statutory criteria in a criminal statute to constitute elements of that crime (which must be proved to a jury beyond a reasonable doubt) or instead sentencing factors. These factors "directed at determining congressional intent [are]: (1) language and structure; (2) tradition; (3) risk of unfairness; (4) severity of the sentence; and (5) legislative history." United States v. O'Brien, 560 U.S. 218, 225-229 (2010) (applying so-called Castillo factors, after Castillo v. United States, 530 U.S. 120, 124-29 (2000)).

(343.) Nicholas Quinn Rosenkranz, Federal Rules of Statutory Interpretation, 115 HARV. L. REV. 2085, 2103 (2002).

(344.) See Gluck, supra note 113, at 803-04 (acknowledging savings and preemption clauses). General interpretive instructions are rare. But see, e.g., Religious Land Use and Institutionalized Persons Act (RLUIPA) of 2000 [section] 5(g), 42 U.S.C. [section] 2000cc-3(g) (2012) (stating that the chapter shall be "construed in favor of a broad protection of religious exercise, to the maximum extent permitted by the terms of this chapter and the Constitution"); Organized Crime Control Act of 1970, Pub. L. No. 91-452, [section] 904(a), 84 Stat. 922, 947 ("The provisions of this title [Racketeer Influenced and Corrupt Organizations Act] shall be liberally construed to effectuate its remedial purposes.").

(345.) See 1 U.S.C. [section][section] 1-8 (2012).

(346.) Gluck, supra note 319, at 1824-29 (2010) ("There is a power struggle going on here."); see also SCALIA & GARNER, supra note 7, at 43-44 ("Some states have enacted a repealer of the rule of lenity, and the courts of some of those states have ignored such repealers." (footnote omitted)); id. at 245 (suggesting that such rules are likely unconstitutional).

(347.) Marbury v. Madison, 5 U.S. (1 Cranch) 137,177 (1803); see Gluck, supra note 319, at 1827 (arguing that the power to say how to interpret may be "inherent in the power to 'say what the law is' ").

(348.) Cf. Chambers v. NASCO, Inc., 501 U.S. 32, 43 (1991) (declining to hold that FED. R. CIV. P. 11 displaces the judiciary's inherent authority to control its proceedings, since those and other implied powers "are necessary to the exercise of all others" (quoting United States v. Hudson, 11 U.S. (7 Cranch) 32, 34 (1812))).

(349.) See Gluck, supra note 4, at 184 (asserting that "judges have a visceral, highly negative reaction" to the prospect of Congress legislating interpretive rules). I part ways here with Nicholas Rosenkranz. Although he, too, recommends a compromise congressional-judicial process for settling on interpretive rules, he ultimately argues that Congress's power over interpretive rules is supreme and that the judiciary would simply add valuable expertise to such a process. Rosenkranz, supra note 343, at 2151 (rule proposal "harnesses the expertise of the judiciary"); id. at 2152 (arguing that "interpretive methodology is ... an incident of the power to legislate").

(350.) See also Jennifer Bandy, Note, Interpretive Freedom: A Necessary Component of Article III Judging, 61 Duke L.J. 651,688-89 (2011) (arguing that methodological independence on statutory interpretation is necessary to judiciary's constitutional function of checking other branches).

(351.) Cf. Vermeule, supra note 59, at 77 ("The empirical questions relevant to interpretive choice frequently strain the limits of judicial information and predictive capacities.").

(352.) These rules are sometimes referred to as a "treaty" between Congress and the courts involving systematic deliberation inside the judiciary and a congressional opportunity to participate, as well as congressional respect for the ultimate results. I am indebted to my colleague Edward Cooper for this analogy and many useful discussions on the subject.

                                              Rate of Application
                                                 with Approval
                                              (percentage of times
                                             a discussed canon was
                                             ultimately applied to
Canon                                       resolve statutory issue)

Relatively Strong Canons

Expressio unius (discussed                            75%
in 18.6% of all statutory issues)

Whole act rule (15.2% of issues)                      76%

Rule against surplusage                               70%
(discussed in 13.2% of issues)

Harmonious reading canon                              75%
(11.8% of issues)

Whole code rule (11.5% of issues)                     75%

Common law canon (11.6% of issues)                    72%

Rule against absurdity (7.4%)                         67%

Legislative acquiescence (5.9%)                       71%

Noscitur a sociis (3.0%)                              79%

Elephants in mouseholes (1.9%)                        89%

Ejusdem generis (1.6%)                                73%

And/or rule (1.1%)                                    90%

Relatively Weak Canons

Agency deference canons                               53%
(discussed in 11.9% of issues)

Constitutional avoidance canon                        57%
(discussed in 6.8% of issues)

Rule of lenity (3.3% of issues)                       55%

Presumption against preemption                        48%
(2.5% of issues)

Remedial statutes canon (0.6% of issues)              50%

Punctuation canon (0.3% of issues)                    33%


The Seven Strongest Canons, Including
the "Top Five" Textual Canons

Canon                      Percentage of      Rate of
                           Contested Issues   Application
                           Where Canon        with Approval

Expressio unius                 18.6%              75%
Whole act rule                  15.2%              76%
Rule against surplusage         13.2%              70%
Harmonious reading canon        11.8%              75%
Whole code rule                 11.5%              75%
Common law canon                11.6%              72%
Rule against absurdity           7.4%              67%


High approval rate canons also       Top-scoring canons in Roberts
identified in Gluck and Bressman     Court (frequently applied, high
findings as "approximating"          approval rate) "rejected" in
congressional preferences (335)      Gluck and Bressman findings

* Ejusdem generis (73% approval      * Rule against surplusage
  rate; discussed in 1.6% of           (70% approval rate; discussed
  issues)                              in 13.2% of issues)

* Noscitur a sotiis                  * Whole act rule (76% approval
  (79% approval rate; discussed        rate; discussed in 15.2% of
  in 3.0% of issues)                   issues)

                                     * Whole code rule (75% approval
                                       rate; discussed in 11.5% of

Top-scoring canons with equivocal    Low approval rate canons
or unknown status in Gluck and       identified as "approximating"
Bressman study                       congressional preferences or
                                     driving congressional drafting
* Expressio unius (75% approval      in Gluck and Bressman findings
  rate; discussed in 18.6% of
  issues) (Approximates
  congressional practices in
  drafting statutory lists;          * Agency deference canons
  unknown in other settings)           (53% approval rate; discussed
                                       in 11.9% of issues)
* Common law canon (72% approval
  rate; discussed in 11.6% of        * Presumption against preemption
  issues) (Unknown status)             (48% approval rate; discussed
                                       in 2.5% of issues)
* Rule against absurdity
  (67% approval rate; discussed      * Federalism canons (62% approval
  in 7.4% of cases)                    rate; discussed in 5.1% of
  (Unknown status)                     issues).

* Harmonious reading canon
  (75% approval rate; discussed
  in 11.8% of issues) (Possibly
  rejected as variant of rule
  against surplusage)
COPYRIGHT 2018 Michigan Law Review Association
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2018 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Mendelson, Nina A.
Publication:Michigan Law Review
Date:Oct 1, 2018

Terms of use | Privacy policy | Copyright © 2021 Farlex, Inc. | Feedback | For webmasters |