AGENCY RULES WITH THE FORCE OF LAW: THE ORIGINAL CONVENTION

INTRODUCTION ................................................................................. 470 I. WHY THE MEANING OF RULEMAKING GRANTS MATTERS ................................................. 476 II. TEXTUALIST INTERPRETATION OF FACIALLY AMBIGUOUS RULEMAKING GRANTS ...... 481 A. Language ................................................................... 482 B. Structure ................................................................... 483 C. Canons of Interpretation ................................................................... 487 i. The Rule of Lenity ................................................................................. 487


INTRODUCTION
W hether Congress has delegated to particular agencies the authority to make rules with the force of law has been, until very recently, a question of little interest in the administrative law community. Controversies simmered in the ig6os and 1970s over whether the Federal Trade Commission (FTC), the Food and Drug Administration (FDA), and the National Labor Relations Board (NLRB) had power to engage in legislative rulemaking. But these debates are now largely forgotten. An unarticulated assumption took hold sometime after the 1970s that virtually every agency is free to make policy in any mode it chooses, including legislative rules, interpretive rules, policy statements, or adjudication. Administrative lawyers have come to believe that an agency's intent in promulgating a rule, not Congress's intent in delegating power to the agency, determines whether an agency's action has the force of law.
This assumption may soon change. The Supreme Court's recent decisions in Christensen v. Harris County' and United States v. Mead Corp. 2 thrust the question whether agencies have been delegated authority to act with the force of law to the forefront of the most contested issue in administrative law: the scope of the Chevron doctrine, which directs courts to accept reasonable agency interpretations of ambiguous statutes. 3 Christensen and Mead hold that Chevron's high level of deference applies only to agency interpretations that have the "force of law." '4 Moreover, Mead makes clear that agencies act with the force of law only if Congress intended to delegate authority to them to so act. 5 To decide whether Chevron deference is appropriate, reviewing courts will therefore have to focus on the intended scope of Congress's authorization.
Unfortunately, Mead provides incomplete guidance about how courts should undertake this inquiry. One problem is that both Christensen and Mead involved agency action considerably more informal than either legislative rulemaking or formal adjudication, both of which unquestionably have the force of law. 6 Christensen involved an opinion letter written by the head of the Wage and Hour Division of the Department of Labor. Mead considered a letter ruling issued by the Customs Service indicating what tariff would be applied to a particular importation of goods. The Court held in both contexts that these actions did not have the force of law, and hence that Chevron did not apply. 7 But the factors that Mead, the more fully considered decision, discussed as being relevant to whether tariff classification letters have the force of law are not likely to provide much help in the rulemaking or formal adjudication context.
With respect to rulemaking, which is the focus of this Article, the central difficulty going forward is that nearly all agency rulemaking grants are facially ambiguous concerning whether the agency is authorized to make rules with the force of law. Statutes typically give agencies the power "to make, amend, and rescind such rules and regulations as may be necessary to carry out the provisions of this title,",, or "to make such rules and regulations ... as may be necessary in the administration of this Act." 9 The phrase "rules and regulations" in these statutes could refer to legislative rulesthat is, rules that have legally binding effect on the general publicor it could refer to interpretive rules that do not have such binding effect. 10 If "rules and regulations" refers to the former type of rule, then Congress has delegated power to the agency to act with the force of law, and under Mead the agency would be eligible for Chevron deference. However, if "rules and regulations" refers to the latter type of rule, then Mead's threshold inquiry presumably would not be satisfied, and the agency would not be eligible for Chevron deference." 1 Although the language of most rulemaking grants is facially ambiguous, we argue in this Article that these grants were not ambiguous during the formative years of the modern administrative stateup to and beyond the enactment of the Administrative Procedure Act (APA) in 1946. Throughout the Progressive and New Deal eras, Congress followed a drafting convention that signaled to agencies whether particular rulemaking grants conferred authority to make rules with the force of law as opposed to mere housekeeping rules. That convention was simple and easy to apply in most cases: If Congress specified in the statute that a violation of agency rules would subject the offending party to some sanctionfor example, a civil or criminal penalty; loss of a permit, license, or benefits; or other adverse legal consequencesthen the grant conferred power to make rules with the force of law. Conversely, if Congress made no provision for sanctions for rule violations, the grant authorized only procedural or interpretive rules. Congress followed this convention from the second decade of the twentieth century through the enactment of the APA, and it can be discerned in statutes enacted as recently as 1967.12 The most remarkable aspect of this drafting convention is that modern administrative lawyers are not aware of its existence. How could a convention that Congress consistently followed during the formative years of the administrative state simply disappear from legal consciousness? The explanation, we suggest, lies in the fact that during the time the convention was developed and followed by Congress, no appellate court rendered a decision that required it to determine whether Congress had conferred authority on an agency to make rules with the force of law. In administrative law, as in other areas of American law, legal knowledge is transmitted through the study of appellate opinions. With no opinion to flag the issue, questions about the meaning of ambiguous rulemaking grants were ignored in post-World War II treatises and instructional materials devoted to administrative law. As a result, knowledge of the convention died out. When, in subsequent years, the Supreme Court occasionally encountered cases that implicated the meaning of such rulemaking grants, none of the parties alerted the Court to the existence of the convention, even if it would have been in their interests to do sopresumably because their lawyers did not know about it.
The collective amnesia about the drafting convention eventually had important consequences. In the i96os, courts and commentators began to urge an expanded use of rulemaking by agencies and a reduced emphasis on adjudication. Eventually, two influential federal appellate judges who strongly favored greater use of rulemaking -Judges J. Skelly Wright of the D.C. Circuit and Henry Friendly of the Second Circuitauthored important opinions construing facially ambiguous rulemaking grants to the FTC and FDA as authorizing legislative rulemaking.1 3 These holdings were inconsistent with what Congress had intended, as measured by the convention. Although some commentators expressed unease about allowing the FTC and FDA to engage in legislative rulemaking under their general rulemaking grants, 1 4 neither Congress nor the Supreme Court attempted to reverse these decisions.
Soon, the assumption took hold that facially ambiguous rulemaking grants always include the authority to adopt rules having the force of law. In i99i, for example, the Supreme Court upheld a legislative rule promulgated by the NLRB pursuant to its general rulemaking grant under the National Labor Relations Act.' 5 Although the case involved the first broad-scale exercise of legislative rulemaking by the NLRB since its creation in 1935, 16 no Justice questioned whether the agency had the authority to promulgate such a rule.' 7 Similarly, the Supreme Court's Chevron decision treated as legally binding a rule adopted by the Environmental Protection Agency (EPA) pursuant to its general rulemaking powers under the Clean Air Act.' 8 The Chevron Court never questioned whether the Act's facially ambiguous rulemaking grant authorized the agency to promulgate a legislative rule. 19 This history is highly relevant to whether agencies have authority to act with the force of law. Christensen and Mead appear to contemplate that courts will engage in a statute-by-statute determination of whether agencies can exercise such authority. For any statute enacted before the revisionist Wright-Friendly view took hold, the convention in most cases provides the key to unlocking Congress's delegatory intent. The convention is obviously less reliable as a guide to congressional intent for statutes enacted after Wright and Friendly wrote their opinions, since the background understanding against which Congress acts has arguably shifted. Nevertheless, courts may wish to consider adopting the convention as a general canon for determining the presumptive meaning of facially ambiguous rulemaking grants. The ultimate objective of Chevron, as interpreted in Mead, appears to be to develop a set of signals by which Congress can indicate when agencies, rather than courts, are to serve as the primary interpreters of federal statutes. From this perspective, the original convention for distinguishing between legislative and housekeeping grantswhether Congress prescribed some sanction for rule violationsnot only has the imprimatur of history, but would also serve as a clear rule for Congress, agencies, courts, and regulated entities to follow in determining whether the critical delegation occurred. Perhaps most importantly, such a canonby identifying an unambiguous signal from Congress of its intent to delegate power to act with the force of lawwould reinforce an important nondelegation principle: executive branch agents have no inherent authority to act with the force of law and instead possess such power only when it has been deliberately given to them by the people's representatives in Congress. 20 This Article is divided into seven Parts. Part I begins by describing two reasons why it is important to know whether facially ambiguous rulemaking grants confer power to adopt rules with the force of law. First, legislative rules are subject to special procedural obligations, 2 ' known as notice-and-comment procedures. To determine whether an agency has promulgated a legislative rule and hence must follow these procedures, we need to know whether Congress has delegated authority to the agency to issue rules that have the force of law. Second, and more urgently, the Supreme Court in its recent Christensen and Mead decisions has confined the scope of the Chevron doctrine to agency interpretations that have the force of law. To implement the Mead doctrine, it is necessary to determine when particular rulemaking grants confer authority to make rules with the force of law.
Anticipating that some interpreters will want to know whether it is possible to ascertain Congress's delegatory intent without recourse to legislative history, Part II canvasses the chief textualist approaches to interpreting facially ambiguous rulemaking grants. We conclude that the meaning of such grants will often remain inconclusive if interpreters examine them through a purely textualist lens, without any reference to history.
Part III turns to history and shows how Congress in the first half of the twentieth century relied on a convention for distinguishing between legislative and nonlegislative rulemaking grants. This convention emerged in the wake of two Supreme Court decisions, United States v. Eaton 22 and United States v. Grimaud, 23 that considered whether Congress could delegate to administrative agencies the authority to impose criminal penalties for violations of agency regulations. In answering this question, the Court focused on whether Congress had adopted legislation specifically authorizing criminal sanctions for rule violations, 24 and thus provided a point of reference for differentiating between legislative and housekeeping grants. After Grimaud, Congress repeatedly deployed the device of including or omitting sanctions for rule violations in a statute as a way of signaling whether an agency had received legislative rulemaking authority. We show that numerous important regulatory statutes enacted in the Progressive and New Deal eras followed this pattern, and that the convention can be discerned in the text and legislative history of the APA.
Despite the fact that legislative actors adhered to the convention through the adoption of the APA, Supreme Court decisions touching on rulemaking grants in the subsequent decades betrayed no awareness of the convention. As we recount in Part IV, the outcomes of the Court's decisions through the i960s were, with one exception, consistent with the convention. 2 5 Nevertheless, the Court's general failure to attend to the distinction between grants of legislative authority and grants of housekeeping authority increasingly caused lower courts and commentators to view agency intent, rather than congressional intent, as the key factor in identifying legislative rules.
As support for agency rulemaking grew in the i96os, advocates inside and outside certain agencies began to urge that facially ambiguous rulemaking grants that had been long regarded as authorizing only housekeeping rules provide a basis for legislative rulemaking. As chronicled in Part V, these efforts eventually led to the successful exer- cise of general legislative rulemaking powers by three federal agencies that had not previously wielded such powers: the FTC, FDA, and NLRB. 2 6 Opinions by Judges Wright and Friendly sanctioning the exercise of general legislative rulemaking authority by the FTC and the FDA in effect established a presumption that facially ambiguous rulemaking grants confer legislative rulemaking authority.
Part VI describes one vestige of administrative practice that deviates from the Wright-Friendly presumption: the tax world, which adheres to the view that the Treasury Department's general rulemaking grant authorizes only interpretive rules. This view, however, rests on the distinction between general and specific grants of rulemaking authority rather than on whether Congress attached legal sanctions to rule violations. Thus, even the tax world deviates from the original congressional understanding as reflected in the convention.
Finally, Part VII considers how courts should use this history in determining whether Congress has delegated power to agencies to make rules having the force of law, the inquiry that Christensen and Mead require. We consider how Congress, agencies, and courts might use the convention either as a key to understanding legislative intent on a statute-by-statute basis, or as a canon of presumptive congressional intent that overcomes Mead's greatest weaknessits failure to provide a clear rule for identifying the scope of the Chevron doctrine.

I. WHY THE MEANING OF RULEMAKING GRANTS MATTERS
Agency rules come in several types. 7 A basic distinction is between "legislative" rules and "nonlegislative" rules. 2 8 Legislative rules 26 The FDA's story is somewhat different from those of the FTC and the NLRB: although the agency's general rulemaking grant does not confer legislative rulemaking powers, according to the convention the FDA does have legislative rulemaking powers under several specific rulemaking grants. See infra section V.B.2, pp. 557-65. 27 For an overview of the different kinds of rules that agencies can make under the APA regime, see I KENNETH CULP DAVIS & RICHARD J. PIERCE, JR., ADMINISTRATIVE LAW TREATISE § 6.3 ( 3 d ed. 1094); Michael Asimow, Nonlegislative Rulemaking and Regulatory Reform, 1985 DUKE L.J. 381, 383-88 (distinguishing legislative from nonlegislative agency rules, describing the difficulty of distinguishing them in practice, and describing the functions of nonlegislative rules); William T. Mayton, A Concept of a Rule and the "Substantial Impact" Test in Rulemaking, 33 EMORY L. REV. 889, 895-910 (1984) (describing the rulemaking process and several proposed conditions for the creation of agency rules); Richard J. Pierce, Jr., Distinguishing Legislative Rules from Interpretative Rules, 52 ADMIN. L. REV. 547, 549-54 (20oo) (describing the differences between legislative rules and interpretive rules under the APA in terms of their effects and the procedures used to make them); Peter L. Strauss, The Rulemaking Continuum, 41 DUKE L.J. 1463, 1466-68 (1992) (outlining four kinds of rulemaking activity under the APA: "formal rulemaking," which has stringent procedural requirements, and "informal rulemaking," which has less stringent procedural requirements (both of which result in rules with the force of law); "publication rulemaking," which includes statements of policy and interpretive statements; and other "rules" of "lesser dignity" such as guidance documents and press releases). 28 See Asimow, supra note 27, at 383.
[Vol. I116:467 are those that have the force and effect of law. From the perspective of agency personnel, regulated parties, and courts, these rules have a status akin to that of a statute. 9 Nonlegislative rules do not have the force and effect of law; rather, they are simply statements about what an agency intends to do in the future. Another distinction is between "substantive" rules and "procedural" rules. Substantive rules regulate the primary behavior of parties outside the walls of the issuing agencyaddressing how much pollution they can emit, what they must disclose in proxy statements, and so forth. 30 The most important type of substantive rules are legislative substantive rules, which regulate primary behavior with the force of law. Such rules are typically referred to as either "legislative rules" or "substantive rules" for short; we will refer to them throughout this Article as "legislative rules." Substantive rules can also be nonlegislative in nature. "Interpretive rules" are nonbinding substantive rules that advise the public about how an agency interprets a particular statute or legislative rule that it administers. 3 1 "Policy statements" are nonbinding substantive rules that advise the public about how the agency intends to exercise some discretionary power that it has. 3 2 Procedural rules, in contrast to substantive rules, govern what happens inside an agencyhow it is organized, how it conducts hearings, and so forth. Like substantive rules, these too can be either legislative or nonlegislative in character. 33 This Article will not consider procedural rules in detail.
The distinction between legislative rules (that is, substantive legislative rules) and other types of rules is important in administrative law for several reasons. 34 First, the APA generally requires agencies to en- 29 See ATTORNEY GENERAL'S COMM. ON ADMIN. PROCEDURE, ADMINISTRATIVE PROCEDURE IN GOVERNMENT AGENCIES, S. DOC. NO. 77-8, at ioo (ist Sess. 194I) [hereinafter FINAL REPORT] (distinguishing "legally binding regulations" that receive statutory force from "interpretive regulations," under which the "statutes themselves and not the interpretations remain in theory the sole criterion of what the law authorizes"). We realize that to say that a legislative rule has a status akin to a statute begs the question of what qualities distinguish a statute. We have more to say about this in Part III, infra pp. 493-528, where we unpack the historical convention for differentiating between legislative and nonlegislative rulemaking grants. 34 In addition to the two reasons discussed in the text, three others deserve mention. First, because legislative rules are legally binding, in an enforcement action the agency need only show that a party violated the rule to prove it acted illegally; in contrast, if the agency invokes an interpretive rule, it is the statute itself, not the rule, that is controlling. Thus a legislative rule can ease an agency's burden of proof and narrow the issues to be litigated. See JERRY L. MASHAW ET gage in notice-and-comment rulemaking before developing legislative rules, but not before making procedural rules, interpretive rules, or policy statements. 35 In identifying legislative rules for the purposes of APA notice-and-comment requirements, recent appellate decisions have held that two conditions must be satisfied: "Congress has delegated legislative power to the agency and ... the agency intended to exercise that power in promulgating the rule. ''36 In practice, the decisions that apply this two-part test focus almost exclusively on the second partwhether the agency intended to make a rule that has the force and effect of law. 37 As far as we are aware, the first partwhether Congress has delegated power to the agency to make rules that have the force of lawhas never been the focus of any appellate opinion considering whether a particular rule is legislative and hence subject to the procedural requirements of § 553.
The judicial indifference to the first half of the test is puzzling. With rare exceptions, nearly all of the rulemaking grants adopted by Congress in the twentieth century do not specify whether they authorize legislative rules, procedural rules, interpretive rules, or policy statements. Instead, they simply speak of "rules and regulations" and leave it at that. 38 Examining the bare language of these statutes, therefore, courts cannot tell whether Congress "has delegated legislative power to the agency. '39 By focusing only on agency intent in determining whether a rule is legislative for § 553 purposes, then, courts implicitly assume that facially ambiguous rulemaking grants confer the AL., ADMINISTRATIVE LAW: THE AMERICAN PUBLIC LAW SYSTEM 448 ( 4 th ed. I998). Second, under some statutes a legislative rule, unlike an interpretive rule, is not open to judicial review in an enforcement action. See, e.g., 42 U.S.C. § 7 6o7(b)(I) (2000) (providing that petitions for judicial review of a rule made pursuant to certain sections of the Clean Air Act must be filed within sixty days of the rule's promulgation). Third, legislative rules are more likely than interpretive rules to be subject to judicial review before an agency brings an enforcement action.  Cir. 1983) ("A rule can be legislative only if Congress has delegated legislative power to the agency and if the agency intended to use that power in promulgating the rule at issue."); cf. Joseph, 554 F.2d at 1153 n.24 (noting that whether a rule is legislative depends on "the authority and intent with which [it is] issued"). 37 For example, in American Mining Congress, after stating the two-part test, the court devoted the balance of its opinion to discussing a variety of circumstances that indicate agency intent to make rules with the force of law. See American Mining Congress, 995 F.2d at 1109-13. 38 See infra section HIL.C, pp. 503-19 (describing grants of rulemaking authority in numerous twentieth-century statutes).
[Vol. I1I6:467 power to make legislative rules. Later in the Article, we explain the origins of this unstated presumption and question its validity. 40 A second reason that the distinction between legislative and other types of rulemaking is important concerns the standard that courts apply in reviewing agency interpretations of their statutory authority. The issue here is whether courts will review agency interpretations of statutes under the highly deferential Chevron test, 41 or the multifactor Skidmore standard, 42 or whether they will instead determine the meaning of the statute de novo. 43 The Supreme Court recently held in Christensen v. Harris County 4 4 and United States v. Mead Corp. 45 that Chevron applies only to agency interpretations that have the "force of law." '46 Agency interpretations that do not have the force of law receive only whatever deference is due under the Skidmore standard. 47 In elaborating the "force of law" criterion, the Court in Mead adopted a two-part test similar to that used to distinguish between legislative and nonlegislative rules in the APA rulemaking-requirements context: "We hold that administrative implementation of a particular statutory provision qualifies for Chevron deference when it appears that Congress delegated authority to the agency generally to make rules carrying the force of law, and that the agency interpretation claiming deference was promulgated in the exercise of that authority. '4 In applying the two-part Mead test, lower courts may do what they have done in the APA cases: ignore the first part and concentrate solely on the second. But this response is unlikely. The Court in Mead made clear that Chevron deference is grounded in a congressional intent to delegate primary interpretive authority to the agency. Its opinion referred throughout to congressional intent, expectations, contempla- 40  tions, thoughts, and objectives. 49 The Court also discussed a number of factors that led it to conclude that Congress had not delegated authority to the Customs Service to act with the force of law when it authorized the agency to issue tariff classification rulings. Specifically, the Court considered whether Congress required the agency to engage in relatively formal procedures before acting, whether Congress authorized the agency to adopt rules or precedents that generalize beyond a single case, and whether Congress authorized the agency to prescribe legal norms that apply uniformly throughout its jurisdiction. 50 The Court conspicuously framed each of these factors in terms of congressional rather than agency intent. If courts take seriously the relevance of Congress's intent to give an agency's rules the force of law, and heed the passages in Mead that appear to say that the inquiry into congressional intent should proceed by looking to all conceivably relevant factors, the result in most cases will be highly unpredictable. The central problem, again, is that Congress typically authorizes agencies to adopt "rules and regulations," but does not specifically say "legislative" rules and regulations. 51 And without any direction in the legislative text, the all-things-considered Mead framework provides little guidance to lower courts, agencies, and regulated parties about how to discern congressional intent in any given set of circumstances. Thus, it is likely to sow confusion among lower courts, inhibit agency planning, and generate additional legal costs. It is also likely to undermine congressional control over the allocation of interpretive authority, for while Congress can be explicit in new legislation, an unpredictable framework makes it difficult to determine how courts will construe existing legislation. 5 2 Given the dif- 49 See id. at 229-34. 50 See id. 51 Mead also creates problems in determining whether agencies should receive Chevron deference for interpretations embodied in procedural rules. Procedural rules, like substantive legislative rules, can have the force of law. See i DAVIS & PIERCE, supra note 27, § 6.6, at 252-54 (noting that procedural rules are sometimes legislative and sometimes merely guidelines). Moreover, there is little doubt that when Congress enacts a facially ambiguous grant of authority to an agency to adopt "rules and regulations," it intends at the very least to include in this grant the authority to make procedural rules. Thus, procedural rules will often satisfy Mead's basic requirement for Chevron deferencea delegation of authority from Congress to act with the force of law. Cf Merrill & Hickman, supra note 43, at 905-o6. On the other hand, Mead places considerable emphasis on whether Congress has required that an agency follow notice-and-comment procedures in issuing a rule, see Mead, 533 U.S. at 230-31, and the APA specifically exempts procedural rules from notice-and-comment procedures, see 5 U.S.C. § 553(b) (2000). Thus, one of the signals of congressional intent to make rules legally binding emphasized by Meadthe requirement of notice-and-comment proceduresis irrelevant. The Supreme Court recently reserved judgment on the degree of deference owed to a particular procedural rule. See Edelman v. Lynchburg Coll., 122 S. Ct. 1145, 1150 (2002). 48o [Vol. 116:467 ficulty in determining actual congressional intent, "some version of constructiveor perhaps more frankly said, fictionalintent must operate in judicial efforts to delineate the scope of Chevron. 5 3 The need to attribute some intent to Congress, in turn, raises the question whether there is any background principle that courts can apply in a manner that is reasonably faithful to Congress's actual intent and that simultaneously resolves disputes over the nature of delegated authority in a reasonably predictable manner.

RULEMAKING GRANTS
What does Congress mean when it gives agencies the power to make such "rules and regulations" as may be necessary to implement or administer an act? In Part III, we discuss how Congress historically followed a convention to signal its intent to grant either legislative rulemaking authority or merely housekeeping authority. 5 4 Once we understand this convention, we can discern Congress's probable intent in most instances. But to confirm the existence of this convention we must examine the legislative histories of numerous statutes. Using legislative history to resolve questions of statutory interpretation is controversial. Some interpreters, such as Justice Scalia, have insisted that questions of statutory interpretation should be resolved solely in terms of the objective meaning of a statute rather than by reference to Congress's subjective intentions. 55 Such "textualist" interpreters look to the language of a statute, its structure, and canons of interpretation in resolving statutory meaning, but ordinarily avoid legislative history, such as committee reports and floor debates. 5 6 However, a survey of the techniques of interpretation typically deployed by textualists sug-53 David J. Barron

2002]
gests that these tools cannot clearly resolve the meaning of ambiguous rulemaking grants in most instances.

A. Language
All interpreters regard a statute's language as the most important datum in statutory interpretation. 57 And indeed, sometimes the language of a rulemaking grant expressly specifies whether the grant includes the authority to adopt legally binding rules. For example, a few statutes state that the rules promulgated by an agency shall have "the force and effect of law. 58 Other rulemaking grants unambiguously limit their reach to procedural rules, providing, for example, that an agency has the power to issue "suitable procedural regulations to carry out the provisions" of an act. 59 More often, however, the language used in rulemaking grants is ambiguous, stating only that an agency has the power to make "rules and regulations as may be necessary to carry out the provisions of this title, ' 60 or "to make such rules and regulations ... as may be necessary in the administration of this Act. '6 1 It is possible that the varying language used in different grants could be given different meanings. For example, a grant that gives power to an agency to "enforce" an act or a provision of an act might be construed as authorizing legislative rules, whereas grants that speak only of "administering" an act might not. 62 57 Even those interpreters who formulate the inquiry in terms of ascertaining legislative purpose or intent routinely "start with the assumption that the legislative purpose is expressed by the ordinary meaning of the words used." Sec. Indus. Ass'n v. Bd. of Governors, 468 U.S. 137, 149 (1984) (quoting Richards v. United States, 369 U.S. i, 9 (1962)); see also Eskridge, supra note 56, at 621 (stating that "[t]he statute's text is the most important consideration in statutory interpretation"). Or, rulemaking grants might be differentiated on the ground that a grant that authorizes an agency to "carry out" the provisions of an act 63 is more legislative in tone than a grant that gives an agency the power to make rules for the "administration" of an act 64 or to promulgate rules "as may be necessary in the execution of its functions." 65 Briefs filed with the Supreme Court have occasionally made arguments of this nature. 66 But only rarely have courts suggested that different verbal formulations signify the conveyance of different types of powers. 67 In the end, whether any particular formulation is "legislative" or "interpretive" remains debatable. For example, it has been argued that a conferral of authority "to enforce" the provisions of an act is both a signal that legislative rulemaking authority has been given and a signal that the agency possesses only interpretive rulemaking authority. 68 Given the general language used in most rulemaking grants, it is not ordinarily possible simply by analyzing the text to determine whether such grants confer legislative as opposed to procedural and interpretive powers.

B. Structure
The overall structure of statutes provides another important consideration that meets with approval from textualist interpreters. As Justice Scalia has stated, a "provision that may seem ambiguous in isolation is often clarified by the remainder of the statutory scheme. '69 Structural arguments usually focus on the internal structure of a single ... merely concerns agency procedures and is solely administrative. It ignores the distinct reference in the section to enforcement." 63 See statutes cited supra note 6o. 64 See statutes cited supra note 6i. 65 Communications Act of 1934, ch. 652, § 4 (i), 48 Stat. io64, io68 (codified as amended at 47 U.S.C. § i 5 4 (i) (2000)) ("The Commission may perform any and all acts, make such rules and regulations, and issue such orders, not inconsistent with this Act, as may be necessary in the execution of its functions.").
66 For example, in Mourning v. Family Publications Service, Inc., 411 U.S. 356 (i973), the petitioner argued that a rulemaking grant to carry out "the purpose" of an act was broader than a grant to carry out "the provisions" of an act. See Brief for Petitioner at 24, Mourning, 411 U.S. 356 (i973) (No. 71-829). Similarly, in American Trucking, the appellants argued that the power the Motor Carrier Act gave to the ICC to "prescribe rules, regulations, and procedure for such administration" of the Act was specifically limited to rules and regulations for "administration." Appellants' Petition for Rehearing at io-i2, E. Motor Express, Inc., v. United States, 344 U.S. 298 (I953) (No. 35).
67 Nor are we aware of any evidence from legislative history that Congress has ever placed any significance on these divergences in verbal formulations.
68 Compare American Trucking, 344 U.S. at 311 (stating that the power of "enforcement" suggests a legislative grant), with Brief for Petitioner, supra note 66, at 23-24 (arguing that the power of "enforcement" suggests an interpretive grant).
act, but occasionally such arguments also consider the relationship between the act under consideration and related statutes. 70 Important clues about the meaning of a rulemaking grant can sometimes be gleaned from its placement in an act. One potential indicator of meaning is the grant's location in the administrative or organizational sections of an act. 71 For example, the general rulemaking grant in the Social Security Act of 1935 appears in Title XI, the "General Provisions" title, which contains statutory definitions, a severability clause, and the short title of the Act. 72 A general rulemaking grant in the Communications Act of 1934 also appears in the "General Provisions" title of the Act, which contains organizational provisions relating to the Federal Communications Commission (FCC) and its divisions, as well as statutory definitions. 7 3 Similarly, the general rulemaking grant in the Internal Revenue Code appears in Subtitle F, which pertains to "Procedure and Administration. '74 The placement of each of these grants may signal that the grant pertains to housekeeping matters only and hence does not authorize legislative rules.
In contrast, the placement of other rulemaking grants suggests that they authorize legislative rules. The most compelling example is a statute that includes both general and more specific rulemaking grants that overlap. For instance, a statute may contain both a general rulemaking grant that authorizes the promulgation of all rules and regulations necessary to carry out the administration of "this subchapter," and one or more narrower rulemaking grants that authorize rules necessary to carry out "this section. ' 48 Stat. 74, 85-86 (codified as amended at i5 U.S.C. § § 77s-77t (2000)) (giving the SEC general rulemaking powers in section i9(a) of the Act and then providing in section 20 that the SEC may institute proceedings against any person who violates "any rule or regulation" promulgated by the SEC under the authority of the Act); Natural Gas Act, ch. 556, § § i6, 20-21, 52 Stat.  other statutes, fit this pattern. 7 6 Although both the general grant and the specific grants are often ambiguous on their face, given the overlap it is reasonable to argue that the general grant confers only housekeeping powers, while the narrower grants confer legislative powers. After all, if both the general grant and the specific grants conferred the same type of power (either legislative or housekeeping), then the statute would include a significant redundancy." By construing the general grant to confer the more generally available housekeeping powersthe power to issue procedural or interpretive rulesand construing the specific grants to confer the more jealously guarded powerthe power to issue legislative rulesthe redundancy disappears. 78 The structural argument based on overlapping rulemaking grants is not necessarily decisive, however. One problem is that interpretive rulemaking (and perhaps procedural rulemaking as well) has long been viewed as an "inherent" power of all executive institutions. 7 9 Whether this power is inherent or not, in 1874 Congress granted to the heads of 76 With respect to the Securities Exchange Act, see i5 U.S.C. § 78w, which grants general rulemaking authority to the SEC; id. § 78m, which grants authority to make rules regulating periodical reports of registered securities; and id. § 78n, which grants authority to make rules regulating proxies. he power of an administrative officer or agency to prescribe interpretive regulations is not necessarily a delegated power. The authority to exercise such power need not be found in an Act of Congress."); John B. Olverson, Jr., Note, Legislation by Administrative Agencies, 29 GEO. L.J. 637, 640 (194) ("In the mere interpretation of a statute, the administrative agency need not be empowered by such statute to make regulations."). executive departments general authority to adopt rules governing their employees and affairs. s0 This "Housekeeping Act," which is codified at 5 U.S.C. § 301, provides that: "The head of an Executive department or military department may prescribe regulations for the government of his department, the conduct of its employees, the distribution and performance of its business, and the custody, use, and preservation of its records, papers, and property." ' Because executive agencies have long possessed these powers, one could argue that it would be redundant for Congress to give such agencies an additional general grant of power authorizing only interpretive and procedural rules. Therefore, to avoid redundancy between the Housekeeping Act and the agency's statute, the general rulemaking grant in the statute should be treated as legislative.
The matter is more complicated still. Section 30I's housekeeping grant applies only to the heads of executive departments and military departments, not independent administrative agencies. 8 2 As a result, it might not be redundant for Congress to give independent agencies such as the FTC, FCC, and NLRB general grants of power to adopt housekeeping rules. Such grants could simply be viewed as substitutes for the housekeeping grant already given to department heads. 8 3 80 See REV. STAT. § 16i (1873-1874). 81 S U.S.C. § 301 (2ooo). The Housekeeping Act consolidated into one place various housekeeping powers that had been conferred on department heads in prior acts. See, e.g., Act of July 27, 1789, ch. 4, § § 1-2, 1 Stat. 28, 29 (giving the Secretary for the Department of Foreign Affairs power over all the records, books, and papers of his office and giving the Secretary the power to "perform and execute such duties" as the President shall entrust to him); Act of Aug. 7, 1789, ch. 7, § 2, 1 Stat. 49, 5o (giving the Secretary of the Department of War power over its records, books, and papers); Act of April 30, 1798, ch. 35, § § 2-3, 2 Stat. 553, 554 (giving the Secretary of the Navy possession and charge of all books, records, and documents pertaining to the Department of the Navy). Notes presented to Congress by the commissioners who drafted the revised statutes explained the purpose of section 16i as follows: This section is suggested by section 8 of the act of 22 June, 1870, which reads as follows: "That the Attorney General is hereby empowered to make all necessary rules and regulations for the government of said Department of Justice, and for the management and distribution of its business." Substantially the same power seems to be conferred, though in vague and uncertain language, by the earlier acts organizing the other Departments. The commissioners have thought it advantageous that the power should be stated in terms, and be conferred alike on all the Secretaries. i REVISION OF THE UNITED STATES STATUTES AS DRAFTED BY THE COMMISSIONERS APPOINTED FOR THAT PURPOSE 80 (1872) (citation omitted). Thus, although the language used in REv. STAT. § 161 conferred what could be viewed as very broad procedural rulemaking powers on department heads, the commissioners' notes indicate that they were not attempting to enlarge the powers granted by earlier acts.
82 See 5 U.S.C. § 301; see also id. § ioi (defining the term "executive department"). 83 However, if Congress intended general rulemaking grants to independent agencies, such as the FTC and the NLRB, merely to be substitutes for section 305's housekeeping grant, one would expect that Congress would not give general rulemaking grants to executive departments (since section 30i's housekeeping grant already covers executive officers). But this is not the case, as Congress has frequently given executive department heads both general and specific rulemaking [Vol. 116:467 Thus, structural arguments about overlapping grants do not definitively resolve the meaning of facially ambiguous rulemaking grants, at least not in all cases.

C. Canons of Interpretation
Canons of interpretation are another tool accepted by most textualists. Three types of canons in particular are potentially relevant in determining the meaning of ambiguous rulemaking grants: the rule of lenity, nondelegation canons, and what this Article will call the Petroleum Refiners canon.

i.
The Rule of Lenity. -The rule of lenity requires courts to construe ambiguous statutes imposing criminal liability narrowly to provide fair notice to potential offenders and to constrain the discretion of prosecutors and courts. 8 4 Thus, where violation of an agency rule would expose persons to criminal sanctions, the rule of lenity provides a rationale for construing an ambiguous rulemaking grant as authorizing only interpretive rules. This application of the rule of lenity is particularly relevant where Congress has prescribed criminal sanctions for violating legislative agency rules and has given the agency multiple rulemaking grants, some of which are ambiguous. A potential offender would know that violating some agency rules could give rise to criminal sanctions but would not be able to tell whether rules promulgated pursuant to an ambiguous rulemaking grant are included in that category.
Of course, this canon applies only where rule violations are criminally punishable. Courts do not ordinarily apply the rule of lenity to statutes that impose only civil sanctions. 8 5 The Supreme Court has also curtailed the rule of lenity in recent years, and there is some question whether it applies at all in the context of judicial review of administrative regulations.

Nondelegation
Canons. -Canons based on the nondelegation doctrine 8 7 are also potentially relevant in interpreting the meaning of ambiguous rulemaking grants. 88 Courts have never vigorously enforced the original nondelegation principlethat Congress may not delegate powers that are "strictly and exclusively legislative." 8 9 Congressional delegations of significant discretionary powers, including rulemaking powers, began in the First Congress and have been a feature of our government ever since. 90 Nevertheless, courts have employed the nondelegation doctrine as a canon of statutory construction. 9 1 One familiar nondelegation principle provides that a congressional delegation of authority to an agency must contain an "intelligible principle" to guide and constrain the agency's behavior. 92 Courts could invoke the intelligible principle requirement as support for interpreting ambiguous rulemaking grants narrowly, on the ground that such grants fail to provide meaningful guidance to agencies about which sorts of rules are permitted. Thus, Stanley Surrey, a prominent tax  " of the Internal Revenue Code does not constitute a delegation of legislative rulemaking power, because such a sweeping grant lacks an intelligible  principle.   9 3 The intelligible principle argument, however, overlooks the fact that the content of an agency's legislative rulemaking will be constrained not only by the grant of rulemaking authority, but also by limitations in the relevant act's substantive provisions. The existence of this additional constraint on agency discretion, together with the Court's extreme reluctance to second-guess Congress regarding the degree of guidance that agencies should receive, probably means that any argument about ambiguous rulemaking grants based on the intelligible principle requirement would fail. 9 4 Another nondelegation principle that is perhaps more helpful in interpreting ambiguous rulemaking grants is that an agency may not bind the public with the force of law unless Congress has delegated it the power to do so. 9 5 In our system of separation of powers, it has always been assumed that the President, members of the executive branch, and federal administrative agencies have no inherent power to make law. 96 By the late nineteenth century, courts had recognized a corollary to this principle: administrative agencies cannot make legislative rules absent a delegation of this power from Congress.  ("[T]he general rule in this country is that the administrative authorities possess only the delegated ordinance powerthat is, they may issue ordinances only where the power to issue such ordinances has been expressly given to them by the legislature."); Morris M. Cohn, To What Extent Have Rules and Regulations of the Federal Departments the Force of Law, 41 AM. L. REV. 343, 345 (1o7) ("It has been held that heads of departments have no right, in the absence of statute, to make regulations upon a subject, so as to bind third persons.").
The understanding that a clear delegation is necessary before agencies can bind the public may be related to the rule, first set forth in the nineteenth century, that a municipality cannot act One historically significant manifestation of this understanding is Interstate Commerce Commission v. Cincinnati, New Orleans, & Texas Pacific Railway Co.,9 8 the now largely-forgotten Queen and Crescent Case, 99 in which the Supreme Court held that the Interstate Commerce Commission (ICC) had no power to prescribe future railroad rates. 0 0 The Interstate Commerce Act of I887101 unquestionably gave the ICC the power to determine whether current rates were reasonable. 0 2 But the Act contained only a narrow grant of procedural rulemaking authority. 1 0 3 The Court held that the power to adjudicate whether rates were reasonable could not be construed to imply the power to set future rates by rule, 1 0 4 because delegations of power to bind the public with the force of law are "never to be implied."' ' 0 5 The Court noted that whether Congress had given the ICC the power to set future rates had "been most strenuously and earnestly debated" by the parties, 1 0 6 and concluded that because the question was debatable, the Act had to be construed as not implying that power. 0 7 unless authorized to do so by the state of which it is a part. John F. Dillon explained the rule, which became known as "Dillon's Rule," as follows: It is a general and undisputed proposition of law that a municipal corporation possesses and can exercise the following powers, and no others: First, those granted in express words; second, those necessarily or fairly implied in or incident to the powers expressly granted; third, those essential to the declared objects and purposes of the corporation,not simply convenient, but indispensable. The Queen and Crescent Case suggests a nondelegation canon in the form of an "express statement" rule: all grants of rulemaking authority confer only housekeeping powers, unless Congress expressly confers the power to make legislative rules.'°0 Such a rule, if consistently followed, would eliminate any disputes about the extent of ambiguous rulemaking grants. Unless Congress states in the text of the grant that agency rules will have the "force of law," or includes other words to that effect, the grant authorizes only interpretive and procedural rules. But the Queen and Crescent Case is the only time the Court suggested such an express statement rule. 0 9 For whatever reason, the Queen and Crescent Case and its maxim that legislative rulemaking power is "never to be implied" were soon forgotten. 0 A similar but less draconian canon would insist not on an express statement, but only on some clear expression of congressional intent to confer power to act with the force of law. A number of decisions, including some of fairly recent vintage, contain language consistent with this proposition."' The difference between a "clear intent" rule and the ordinary rule that statutes should be given the meaning that Congress more likely than not intended is rather hazy. Presumably, under a clear intent rule, something more than a preponderance of the evidence is required to find a delegation of power." 12 Under the clear intent version of the canon, courts would construe ambiguous rulemaking grants to authorize only nonlegislative rules unless the ordinary tools of statutory interpretationincluding, of course, the language lished in 1904 read these decisions as resting on the proposition that only Congress can declare what action has the force of law: In and structure of the actclearly reveal that Congress intended to confer legislative rulemaking authority. 1 1 3 3. The Petroleum Refiners Canon. -A third potentially relevant canon lacks the pedigree of either the rule of lenity or the nondelegation doctrine. For want of a better term, this Article calls it the Petroleum Refiners canon, after the leading federal appellate opinion that adopted this approach. 114 The Petroleum Refiners canon provides that ambiguous grants of rulemaking authority should be construed to give agencies the broadest possible powers, so that they will have flexibility in determining how to effectuate their statutory mandates. In contrast to the nondelegation canons, this canon derives not from fundamental principles of separation of powers but from pragmatic considerations, including the convenience of allowing agencies to make legislative rules on all matters properly within their jurisdiction. The Petroleum Refiners canon thus bears an affinity with the Supreme Court's pronouncement in SEC v. Chenery Corp. 1 " that agencies should be free to choose between rulemaking and adjudication in determining how to formulate policy. 1 6 Chenery itself presented no question about the meaning of an ambiguous rulemaking grant." 7 Nevertheless, the Chenery doctrine is often invoked for the more general proposition that agencies should have discretion to choose the format in which they articulate policy. In this more general sense, the doctrine provides analogical support for the Petroleum Refiners canon.
While the Supreme Court has acknowledged the importance of construing rulemaking grants liberally so as not to "undermine the flexibility sought in vesting broad rulemaking authority in an administrative agency,"" 8 it has never explicitly endorsed the Petroleum Refiners canon. As we shall see, however, two prominent federal courts of appeals have construed ambiguous rulemaking grants as conferring legislative rulemaking authority, and these opinions can be read as embracing the Petroleum Refiners canon. 1 9 In effect, these courts have adopted the exact opposite of the Queen and Crescent Case 113 Section 558(b) of the APA, which provides that "[a] sanction may not be imposed or a substantive rule or order issued except within jurisdiction delegated to the agency and as authorized by law," 5 U.S.C. § 558(b) (2000), supports such a canon. See also infra pp. 525-26 (discussing the significance of § 5 5 8(b) in light of the convention discussed in Part III).  In sum, given their ambiguous language, the uncertain implications of structural arguments, and the unavailability of conflicting canons, the meaning of ambiguous rulemaking grants is often debatable if they are examined through a purely textual lens. This Article, therefore, turns to a consideration of what history teaches about the meaning of such grants.

III. THE CONVENTION CREATED
Although grants of authority allowing agencies to adopt "rules and regulations" appear to be facially ambiguous concerning whether they authorize rules having the force of law, the history of rulemaking during the Progressive and New Deal eras reveals that key participants in the legislative process did not regard such grants as ambiguous. Starting around World War I, Congress began following a convention for indicating whether an agency had the power to promulgate legislative rules. Under this convention, the requisite textual signal was provided by the inclusion of a separate provision in the statute attaching "sanctions" to the violation of rules and regulations promulgated under a particular rulemaking grant. If the statute prescribed a sanction, then the authority to make "rules and regulations" included the authority to adopt legislative rules having the force of law. If the statute did not include a sanction, the authority to make "rules and regulations" encompassed only interpretive or procedural rules.
The "sanctions" took various forms. 120 The clearest case, of course, was when Congress imposed criminal or monetary civil penalties on 120 During the relevant time period, the required legal consequences were generally referred to as "sanctions," but the concept of a "sanction" was given a meaning broader than criminal or civil monetary penalties. This understanding accords with the APA, adopted in 1946, which defined "sanction" as follows: "[S]anction" includes the whole or a part of an agency -(i) prohibition, requirement, limitation, or other condition affecting the freedom of any person; (2) withholding of relief; (3) imposition of any form of penalty or fine; (4) destruction, taking, seizure, or withholding of property; (5) assessment of damages, reimbursement, restitution, compensation, costs, charges, or fees; (6) requirement, revocation, or suspension of a license; or (7) taking other compulsory or restrictive action. Ch. 324, § 2, 6o Stat. 237, 238 (1946) (codified as amended at 5 U.S.C. § 551 (2000)).
persons who violated an agency's regulations. 1 21 On other occasions, however, the sanctions might take the form of the forfeiture or destruction of property, 122 the revocation of licenses,' 23 or the denial of bene- fits.124 In contrast, if the statute was silent regarding the legal consequences for failure to conform to regulations, it was understood as granting the agency the power to make only housekeeping rules. 1 2 S This convention can be seen as resting on two propositions that participants in the legal system widely shared in the first half of the twentieth century. The first is that "law" means roughly what John Austin defined it to mean: general commands backed by the threat of sanctions. 1 26 In other words, lawas distinguished from moral norms or principles of justiceis a directive enforced by the infliction of punishment or the imposition of other material sanctions by the state. 27 The second proposition is that only the legislature, as opposed to the executive and the judiciary, has the power to determine when directives will be given the force of law. This principle, of course, is   2000)) (declaring that the FTC has the power "to classify corporations and to make rules and regulations for the purpose of carrying out the provisions of this Act" but providing no sanctions for the violation of those rules and regulations). 127 As one authority put it: A rule of conduct which lacks a means of enforcement is not an expression of the will so much as of the mere wish of the state; and such a rule, if not enforceable by legal processes, should not even be graced with the title of a law of imperfect obligation. HART, supra note 20, at 28. On the continuity between Austin's command theory and the tenets of the Langdellian legal formalism of the late nineteenth and early twentieth centuries, see  Combined, the first and second propositions produce the convention: First, under the Constitution, only Congress has the authority to specify when an agency can act with the force of law (the nondelegation proposition). Second, agencies act with the force of law only when their rules are backed by sanctions (the Austinian proposition). Consequently, Congress can indicate whether agencies have the authority to act with the force of law by specifying whether the rules they promulgate will be backed by sanctions.
The convention did not emerge full-blown at any one moment. Rather, it gradually developed around the second decade of the twentieth century as Congress created new administrative entities and considered what kind of rulemaking authority to give them. Moreover, as we shall see, the convention was never explicitly memorialized in an authoritative text, such as a statute, a legislative drafting guide, or a prominent judicial decision. It remained part of the unwritten "common law" of legislative drafting in the first half of the twentieth century. Accordingly, the only way to establish the existence of the convention is to examine a significant number of regulatory statutes and their associated legislative histories, supplemented by contemporary writings by knowledgeable participants in the legislative and administrative processes.'

A. Early History: The Diversity of Rulemaking Grants
Although it was originally thought that Congress could not delegate powers that were "strictly and exclusively legislative,"' 130 congressional grants of rulemaking power actually began in the first session of Congress. The twenty-fourth statute enacted in 1789 provided that the 128 See supra pp. 490-91. 129 A note about our method of identifying facially ambiguous rulemaking grants is appropriate at this juncture. We began by identifying and analyzing numerous nineteenth-and early twentieth-century regulatory statutes, which we located through a variety of sources, including case law, legal articles and books about rulemaking, and by scanning the United States Code's Popular Name Table, which lists acts and the years in which they were enacted. We next identified those regulatory statutes that include facially ambiguous rulemaking grants and sought to determine whether there was any authority discussing whether these grants conferred legislative or merely housekeeping powers.
We did not systematically attempt to identify facially ambiguous rulemaking grants in statutes passed after the New Deal era; they undoubtedly number in the thousands. We did, however, encounter a number of these grants in the course of our research and we mention them in the Article when relevant. United States government would continue to pay previously granted pensions for one year "under such regulations as the President of the United States may direct. 1 1 3 1 Similarly, a statute passed in the second session of the first Congress prohibited unlicensed trade with American Indian tribes, instructed the executive department to issue licenses to individuals engaged in trade with Indians, and provided that the licenses were to be "governed in all things touching the said trade and intercourse, by such rules and regulations as the President shall prescribe. "1 3 2 From 1789 through most of the nineteenth century, delegations of rulemaking power generally ran to the President. 1 33 Occasionally, however, such grants ran to other officers. For example, in 1813 Congress passed a tax assessment and collection act that gave the Secretary of the Treasury the power to "establish regulations suitable and necessary for carrying this act into effect. 1 34 Notably, the statute expressly stated that these regulations "shall be binding on each assessor in the performance of the duties enjoined by or under this act. 135 About the same time, Congress passed an act involving taxation of foreign commerce in which it declared that "the Secretary of the Treasury shall give such directions to the collectors, and prescribe such rules and forms to be observed by them, as may appear to him proper for attaining the objects of this act." 1 36 Whereas early rulemaking grants were largely confined to military, foreign affairs, tax, and internal government matters, Congress in the late nineteenth century began to legislate over a wider range of activities, including the control and disposition of federal lands and the regulation of interstate commerce. Practical realities necessitated that the officers administering these statutes adopt implementing regula- [T]he legislature was inclined to share the burden of legislation with the Executive often during the first period; that this practice continued to some extent during the second; that Congress was liberal in delegating discretion during the third period; and that during the fourth period the practice was generally established. Furthermore, . . . although the President received a major portion of all delegated legislation during the first three periods, a division of labor was appearing in the administrative branch of the government and the basis for practically all legislative powers exercised at the present time by the major departments was being laid. Id. at 51. 134  Some of the statutes enacted by Congress expressly indicated that the regulations promulgated under their authority would have the "force of law. 1 39 In 1871, for example, Congress passed an act involving the safety of vessels that gave a supervising board the power to "establish all necessary rules and regulations required to carry out in the most effective manner the provisions of this act for the safety of life, which rules and regulations, when approved by the Secretary of the Treasury, shall have the force of law. '1 40 Similarly, in 187o, Congress passed an internal revenue act that gave the Secretary of the Treasury the power to make "all needful rules and regulations, not inconsistent with law, to be observed in the execution of this act, which shall have the force and effect of law. ' . 14 1 Other statutes did not explicitly state that agency rules and regulations would have the force of law, but their context implied this conclusion. In the Timber and Stone Act of 1878,142 for example, Congress authorized citizens in western states and territories to cut timber on public land. Section I of the Act authorized the Secretary of the Interior to prescribe rules and regulations "for the protection of the tim- 137 See Marshall Field & Co. v. Clark, 143 U.S. 649, 694 (1892) ("There are many things upon which wise and useful legislation must depend which cannot be known to the law making power, and, must, therefore, be a subject of inquiry and determination outside of the halls of legislation." (quoting Commonwealth ex rel. McClain v. Locke, 72 Pa. 491, 499 (1873)) (internal quotation marks omitted)). See generally i FRANK J. GOODNOW, COMPARATIVE ADMINISTRATIVE LAW 27-28 (1897) ("The needs of the government make it necessary that many details in the law be fixed less permanently than by statute. No legislature, however wise or far-seeing, can, with due regard for the interests of the people, which differ with the locality and change with the passage of time, regulate all the matters that need the regulation of administrative law."). 138 See COMER, supra note 133, at 51. 139 Congress's use of the "force of law" language during these years may have been influenced by court decisions holding that internal governmental regulations, navy regulations, and military regulations made pursuant to acts of Congress had the "force of law. ber and of the undergrowth growing upon such lands, and for other purposes. 143 Section 3 specified that any person violating the provisions of the Act or any rules and regulations made under it would face potential criminal charges, imprisonment, and a fine of up to $5oo. 1 44 Consequently, even though Congress never expressly declared that the rules and regulations would have the "force and effect of law," the unmistakable effect of the Act was to give the Secretary of the Interior the authority to bind persons with the force of law. The Tea Importation Act of 1897,45 which sought to regulate the quality of tea imported into the United States, provides another example of Congress implicitly endowing agency rules with the force of law. Section 3 of the Act gave the Secretary of the Treasury the power to establish uniform standards for tea purity, quality, and fitness. 1 4 6 Section 5 provided that any tea falling below the uniform standards set by the Secretary could not be released from the customs house unless it met those standards upon reexamination. 1 4 7 In other words, Congress conditioned the right to import tea on compliance with the Secretary of the Treasury's regulations, effectively giving these regulations the force of law. 148 By contrast, other statutes, exemplified by the landmark Interstate Commerce Act of 1887 (ICA), 14 9 included rulemaking grants that were expressly limited to matters of agency procedure and organization. Section 17 of the ICA gave the ICC the general power to "make or amend such general rules or orders as may be requisite for the order and regulation of proceedings before it, including forms of notices and the service thereof, which shall conform, as nearly as may be, to those in use in the courts of the United States.71 5 o 143 24 Stat. at 386 (emphasis added). In 1889, Congress added a provision to the Act that gave the ICC the power "to execute and enforce the provisions of this act." Amendments to the Interstate Commerce Act, ch. 382, § 3, 25 Stat. 855,858 (1889). This added provision made no mention of executing and enforcing the Act through rules and regulations; it simply gave the ICC the power "to execute and enforce" the Act. Id. In 1906 the ICA was amended by the Hepburn Act to give the ICC specific power to establish maximum rates by rule. See Hepburn Act, ch.

B. The Importance of Eaton and Grimaud
The increased use of rulemaking toward the end of the nineteenth century gave rise to two Supreme Court decisions that provided the impetus for a more uniform understanding about the proper form of rulemaking grants. No question was presented in either case about the meaning of a facially ambiguous rulemaking grant. Rather, the issue before the Court in both cases was one of authority: whether the federal government could impose criminal sanctions upon an individual for violating rules of conduct set forth in a regulation adopted by an agency pursuant to delegated authority. The Court's eventual answer focused on whether Congress had expressly provided by statute that those who violate a rule of conduct set forth in a regulation can be criminally punished. This response suggested a logical way of differentiating between delegations of legislative and nonlegislative authority. By specifically providing for the imposition of sanctions for the violation of a given regulation, Congress resolved any question of authority and also sent an unambiguous signal of its intent that the resulting rules have the force of law.
In United States v. Eaton, 5 1 the Court considered whether a wholesale oleomargarine dealer could be held criminally liable for failing to conform to a regulation, promulgated under the Oleomargarine Act, that required such dealers to keep books of receipts and disposals and to make returns to the Commissioner of Internal Revenue. Section 20 of the Act authorized the Commissioner to make "all needful regulations" for carrying out the Act's provisions. 1 52 The government argued that the prosecution was justified by section i8 of the Act, which made it a criminal offense for any manufacturer or distributor of oleomargarine to fail to do anything "required by law in the carrying on or conducting of his business. 15 3 The government urged that the regulation adopted by the Commissioner under section 20 created duties that were "required by law," and hence fell within the terms of the general prohibition of section 18.114 The Court rejected this contention. Relying on a blend of nondelegation and lenity precepts, the Court stated that the prosecution violated the principle that "a sufficient statutory authority should exist for declaring any act or omission a criminal offence."' 155 Although Congress had made it a criminal offense to neglect to do a thing "required by law," Congress had not expressly made it a criminal offense to neglect to do a thing required only by a regulation. 156 The Court did not hold that Congress could never make it a crime to violate a regulation adopted by an agency. It did suggest, however, that if such a delegation of legislative authority were ever to be permitted, Congress would have to speak "distinctly" in criminalizing failures to abide by agency regulations. '5 Eaton established one thing that Congress could not do: it could not delegate to an agency the authority to issue regulations backed by criminal penalties without explicitly identifying the regulations that, if violated, would give rise to such penalties. It was unclear, however, whether the decision stood for the broader principle that it was impermissible to impose any type of legal consequence, criminal or civil, for the violation of a regulation that lacked the proper delegation of authority from Congress. 158 The answer would depend on whether subsequent courts read Eaton as being grounded primarily in the doctrine of lenity, and hence limited to criminal sanctions, or whether they read it as being grounded in broader principles of nondelegation.
For nearly two decades, it remained uncertain whether Congress could delegate authority to an agency to adopt regulations that would give rise to criminal sanctions. 159 The Court finally resolved the ques- 154 Id. 155 Id. at 688. 156 Id. The defendants argued that Congress could never delegate to an executive agency the power to adopt rules and regulations punishable by criminal sanctions. The Supreme Court disagreed. In contrast to Eaton, which had rested on a blend of nondelegation and lenity reasoning, the Court in Grimaud framed its analysis exclusively in terms of whether the delegation was permissible. The Court read the history of the nondelegation doctrine to mean that: [W]hen Congress [has] legislated and indicated its will, it [can] give to those who were to act under such general provisions "power to fill up the details" by the establishment of administrative rules and regulations, the violation of which [can] be punished by fine or imprisonment fixed by Congress, or by penalties fixed by Congress or measured by the injury done. 164 Eaton was readily distinguished because that case had ruled that "while a violation of the regulations might have been punished as an offense if Congress had so enacted, it had, in fact, made no such provision so far as concerned the particular charge then under consideration.1 6 s The present case was entirely different: "[T]he very thing which was omitted in the Oleomargarine Act has been distinctly done in the Forest Reserve Act, which, in terms, provides that 'any violation of the provisions of this act or such rules and regulations of the Secretary shall be punished .... ,,, 166 Grimaud thus established what Congress could do: it could delegate power to an agency to adopt regulations subject to criminal penalties,  161 Id. at 5o9-1o. 162  provided that Congress itself legislated the penalties. 167 Moreover, because criminal sanctions are the most severe type of sanction for violating an agency regulation, there was little doubt after Grimaud that Congress could provide other types of sanctions for violating agency regulations as well. In other words, Grimaud established that Congress can delegate authority to agencies to promulgate regulations that have a variety of legal consequencesas long as Congress itself spells out by statute what those consequences are.
Neither Eaton nor Grimaud spoke directly to the question of how facially ambiguous rulemaking grants should be interpreted. 168 Nevertheless, the decisions established points of reference that Congress could use in signaling whether particular grants authorized rules and regulations having the force of law. If Congress specifically provided that the violation of a regulation would result in the imposition of sanctions, such as criminal penalties, then the rule would have the force of law (Grimaud). If Congress did not so provide, an agency could not enforce the rule with criminal penalties (Eaton), and it was doubtful whether it could be enforced with any type of civil sanction. 69 Both cases assumed that the place to look for determining whether a regulation has the force of law is the statute that delegated rulemaking power to the agency. This understanding, of course, was consistent with the version of the nondelegation principle underlying the Queen and Crescent Case, which the Court decided during the same era: 170 executive officers and agencies can act with the force of law only if Congress has delegated to them the power to do so.
After Grimaud, cases involving administrative rules and regulations appeared before the Court on occasion. Some of these decisions recognized the distinction between legislative and nonlegislative rules, and they reflected the understanding that legislative rules are those that impose legal consequences on persons who violate them.' 7 ' More 168 Significantly, however, in a case decided shortly after Grimaud, the Court dismissed out of hand the contention that either the Housekeeping Act, 5 U.S.C. § 301 (2000), or a general rulemaking grant given to the Commissioner of the General Land Office could serve as the source of a legislative rule. See United States v. George, 228 U.S. 14 (1913). Quoting these provisions in a footnote, the Court observed: "It will be seen that they confer administrative power only....
[A]nd certainly, under the guise of regulation legislation cannot be exercised." Id. at 20.
169 One scholar explained the significance of both cases as follows: "[Tlhe court will not presume the intended policy of allowing the commission or delegate to inflict a penalty or create a new crime in any case; therefore the authority to do that must at least be clearly expressed in the statute." John B. Cheadle. than thirty years elapsed, however, before the Court decided a case implicating the authority of an agency to promulgate legislative rules based on a facially ambiguous rulemaking grant. 1 7 2 Congress, in the meantime, entered a period of unprecedented activity in terms of the delegation of rulemaking authority to administrative agencies.

C. The Convention Emerges
After Eaton and Grimaud, the structure of rulemaking grants in regulatory legislation assumed a much more uniform pattern. Express references to regulations having the "force of law," which were not infrequent in the nineteenth century, 1 7 3 became relatively rare, as did express qualifications limiting regulations to matters of procedure. Instead, "rules and regulations" became the standard refrain, but now Congress sometimes coupled the grant of authority to adopt rules and regulations with a specific provision imposing sanctions on rule violators, and sometimes it did not. The legislative histories of these statutes, to the extent they clarify the intended meanings of these grants, indicate that knowledgeable participants in the legislative process understood the presence or absence of a provision establishing a sanction for rule violations as the key variable differentiating legislative rulemaking grants from housekeeping ones. The convention that emerged, however, was most likely familiar only to staff attorneys working for Congress and for the agencies that Congress created. There is no evidence that knowledge of the convention extended to most members of Congress themselves, to the courts, orat least during the early years of this formative periodto the commentators who began making note of the growing importance of rulemaking. 7 4 the FCC had promulgated the rule within the scope of that authority); Ariz. Grocery Co. v. Atchison, Topeka & Santa Fe Ry. Co., 284 U.S. 370, 376-78 (1932) (holding that the ICC had no power to impose a sanction on a party for conduct that is "in direct conformity with the Commission's own prior valid legislative pronouncement"); Standard Computing Scale Co. v. Farrell, 249 U.S. 571, 574 (igig) (describing legislative regulations as those that "prescribe a course of action to be enforced by the power of the State"). Most notably, in Columbia Broadcasting System, Inc. v. United States, 316 U.S. 407 (0942), the Court discussed at some length the nature of rules that have the force of law. The specific holding was that such rules, when issued by the FCC, are subject to pre-enforcement judicial review under the Urgent Deficiencies Act. Id. at 417. The Court did not reach the issue of authority because the Communications Act of 1934 included a specific grant of authority to make the rules in question. Id i. Progressive-Era Rulemaking Grants. -Not long after Grimaud, Congress enacted a statute with a facially ambiguous rulemaking grant that will play a large role in our story: the Federal Trade Commission Act of 1914 (FTCA). 175 The Act created the FTC to serve as both an adjudicatory and an investigative body. Section 5 set forth the FTC's quasi-judicial adjudicatory functions, including its power to file complaints, hold hearings, determine whether violations of the FTCA had occurred or were occurring, and issue cease and desist orders. 1 76 Section 6 set forth the investigative powers of the FTC, including its power to demand reports from corporations and to publish information deemed to be in the public interest. 1 77 The FTCA contained only one reference to rules and regulations: section 6(g) provided that the FTC had the power "[f]rom time to time to classify corporations and to make rules and regulations for the purpose of carrying out the provisions of this Act." 78 The statute included no sanction for violations of such rules and regulations. In contrast, the Act empowered the FTC to bring suit to prevent violations of the Act 1 7 9 and set forth remedies for violations of the FTC's ceaseand-desist orders. 1 80 The failure to provide any sanction for the violation of rules adopted under section 6(g), along with the placement of should distinguish between grants of legislative authority and grants of merely procedural and interpretive authority. Particularly striking in this regard are Comer's rather hapless comments: [T]he terms of investment of rule-making power are not always so coupled with the law, or the subdivisions thereof, that the enforcing officer can readily see what his real power is. Thus, the general law of 1789, which gives to the heads of departments the power to prescribe rules and regulations not inconsistent with law, may or may not bestow upon the Secretary of the Treasury the authority to make rules and regulations for carrying out the provisions of the current income-tax law. A similar power bestowed specifically upon the Secretary of the Navy may or may not enable him to issue binding regulations for enforcing a particular policy entrusted to his care. A blank delegation of such power for enforcing the Packers Act may not give the Secretary of the Agriculture authority to promulgate rules for putting into effect specific provisions of that Act .... The officer must make the first guess and let the court determine whether such source is the correct one for effectuating the particular policy in question. COMER, supra note 133, at 133-34. Comer, a political scientist at Williams College, evidently drew his knowledge of rulemaking primarily from documentary sources, including, of course, appellate opinions. Since these sources contained no reflection of the convention as of the time he wrote (1927), it is not surprising that he regarded the meaning of facially ambiguous rulemaking grants as essentially a mystery. 180 Section 5 empowered the FTC to bring suit to enforce its cease and desist orders whenever any person failed to comply. Id. § 5, 38 Stat. at 719-20.
[Vol. I1I6:467 the rulemaking grant in section 6, which conferred the FTC's investigative powers, clearly suggests that Congress intended the rulemaking grant to serve as an adjunct to the FTC's investigative duties, regarding which Congress had not given the agency the authority to act with the force of law.
The legislative history of the Act supports the conclusion that the FTC's rulemaking grant did not confer legislative rulemaking authority. 1 8 1 Section 6(g)'s general rulemaking grant originated in the House Bill of 1914,82 which conferred only investigative powers on the FTC, such as the power to require reports from corporations and to classify corporations. 1 83 In contrast, the bill that passed the Senate granted adjudicative and investigative powers 18 4 but included no rulemaking provision at all. 185 As a consequence, when the Conference Committee met, the only rulemaking provision under consideration was the one included in the House bill. Under established practices for reconciling bills in conference, the Committee could not have granted the FTC legislative rulemaking powers, because neither bill granted the agency such authority.

86
Statements made during floor debates after the Conference Committee incorporated the rulemaking provision into the bill confirm that Congress did not intend section 6(g) to confer legislative authority. In response to questions about the scope of the FTC's powers, Representative Covington, a member of the Conference Committee, distinguished the powers given to the FTC from those delegated to the ICC under the Hepburn Act, which had given the ICC the power to prescribe future rates. 8 7 Covington stated that the "Federal Trade Com-181 See Burrus & Teter, supra note 14, at 1124 ("The legislative history of the Federal Trade Commission Act compels the conclusion that Congress had no intention to confer any 'legislative' rulemaking power upon the Commission."); Glen E. Weston, Deceptive Advertising and the Federal Trade Commission: Decline of Caveat Emptor, 24 FED. B.J. 548, 570 (1964) ("There is simply no indication in the legislative history of intent to confer any such power and plenty of evidence of an intent not to empower the FTC to make 'legislative' rules.").  2002] mission will have no power to prescribe the methods of competition to be used in the future." 188 In addition, he explained that the FTC, unlike the ICC, "will not be exercising power of a legislative nature" in issuing its orders. 1 8 9 In subsequent years, the courts, Congress, the agency, and knowledgeable commentators all shared the understanding that section 6(g) did not confer legislative rulemaking power on the FTC. In a leading separation of powers decision, Humphrey's Executor v. United States, 90 the Supreme Court surveyed the powers of the FTC and concluded that it exercised only "quasi-legislative" and "quasi-judicial" functions. The principal "quasi-legislative" power the Court identified under section 6 was "making investigations and reports [on unfair forms of competition] for the information of Congress."'' The Court did not mention the rulemaking grant in section 6(g), presumably because the litigants in the caseand the Court itselfassumed that this provision did not confer any power on the FTC to make legislative rules.
Similarly, the FTC described its own rulemaking power as excluding the authority to issue legally binding rules. In its 1922 Annual Report, for example, the FTC complained of the frequent confusion that existed over the extent of its authority: One of the most common mistakes is to suppose that the commission can issue orders, rulings, or regulations unconnected with any proceeding before it. It is frequently asked to do this, not only in a broad general way, but also to issue warnings to concerns alleged to be using unfair practices.... It is hoped, in time, to bring about a thorough understanding of the fact that the commission can not and will not function by any method not authorized in its organic act, whereby complete investigation, careful consideration, and, in short, due process of law and full respect for the moral and legal rights of both parties to controversies are assured. 92 power only to determine whether an existing rate was unreasonable, and, if it so found, to order the railroad to cease and desist from charging that rate. The Federal trade commission will have precisely similar power in regard to an existing method of competition. Lastly, the Final Report of the Attorney General's Committee on Administrative Procedure of 1941 reflects the same understanding. 193 The report, which was instrumental in leading to the APA, noted that "[r]elatively few of the administrative agencies studied by the Committee lack power to prescribe regulations for the control of activities which are subject to their authority.194 But one of the agencies that the Final Report pointed to as lacking legislative rulemaking powers was the FTC. 195 Similarly, a monograph on the FTC prepared by the Attorney General's Committee on Administrative Procedure concluded that rules issued by the FTC should be called "advisory interpretations" rather than "rules," because "[n]othing in the statutes administered by the Commission makes any provision for the promulgation of rules applicable to whole industries.' ' 196 The Packers and Stockyards Act of 1921,' 9 ' which regulated the meat packing industry, provides further evidence of the emergence of the drafting convention. The original House committee bill did not require either packers or stockyard operators to comply with any rules or regulations promulgated under the Act, nor did it include any general rulemaking grant. 19 8 By contrast, the Senate committee bill included a general rulemaking grant that unquestionably conferred legislative rulemaking authority. The bill declared that it was the "duty" of every packer and operator to comply with the Secretary's regulations and that conduct contrary to such regulations was "unlawful. ' " 199 The bill also set forth the procedures for ordering a packer or operator to cease and desist from violating a regulation. 20 0 During Senate floor debate, the provision of the Senate bill requiring that packers and stockyard operators comply with rules and regulations generated much discussion. Senator Stanfield criticized the provision, arguing that it gave the agency power to "practically enact a law under which the packer will become a criminal. '20 1 Similarly, Senator Wadsworth expressed concern that the agency might "issue[] a POL. Sci. 666, 671 (1926). For Rublee, a drafter of the FTCA and one of the agency's first Commissioners, it was "perfectly clear" that the FTC had no authority to give advance advice about the legality of business practices, and he did not even mention the rulemaking grant. Id. 193  rule or regulation which imposes an almost impossible condition of affairs upon a man engaged in one of these businesses. '20 2 Senator Norris defended the rulemaking provision, stating: I am one who has always looked with a jealous eye upon the granting of power to convict any man or establish a criminal offense for the violation of a rule or regulation that is not set out in the statute; but we can not set out hereno living man could set out hereevery rule and every regulation that will be necessary to properly care for this big business. 20 3 In the end, the bill that Congress passed into law reflected something of a compromise between the House and Senate views. The Act had four titles. Title I provided general definitions, Title II set forth the statutory provisions regulating packers, Title III regulated stockyards, and Title IV included the general provisions of the Act. 20 4 Title III subjected stockyard operators to penalties for violations of the rules and regulations promulgated by the Secretary under a specific rulemaking grant included in that title. 205 Title II, however, did not include a specific rulemaking grant with respect to the packers. In addition, Congress included a general rulemaking grant in Title IV, but no penalties were attached to this grant. 2 0 6 There is little doubt that the rulemaking grant in Title III was understood to confer power to adopt legislative rules. To be sure, the conferees deleted any reference to stockyard operators having a "duty" to comply with regulations or to violations of regulations being "unlawful." But the rulemaking grant in Title III was coupled with a provision imposing criminal sanctions for rule violations, 2 0 7 and was clearly understood to confer legislative rulemaking authority.
The critical question is what the members of the House and Senate understood the general rulemaking grant in Title IV to mean. This grant, unlike the one contained in Title III, contained no provision for sanctions for rule violations. Remarks by Senator Kenyon on the Senate floor explaining the compromise bill 20 8 indicate that Congress un- The Secretary's rules regarding the schedules were given legislative effect by Congress in section 3 o6(h), which provided that "[w]hoever willfully fails to comply with the provisions of this section or of any regulations or order of the Secretary made thereunder shall on conviction be fined not more than $i,ooo, or imprisoned not more than one year, or both." Id. § 3o6(a), (b), (h). 42  derstood the ambiguous grant in Title IV to authorize only procedural rules. He noted that this grant "gave the Secretary of Agriculture the right to establish rules and regulations of procedure," and characterized the provision as "not an especially important amendment.°2 0 9 The legislative history of the Packers and Stockyards Act thus offers an example of a grant coupled with sanctions understood to confer legislative power, combined with a grant lacking any provision for sanctions understood to authorize only housekeeping-type rules.
Congress also appeared to follow the convention when enacting subsequent statutes that created new agencies and delegated rulemaking power to them. 2 10 One prominent example is the Longshoremen's and Harbor Workers' Compensation Act of 1927.211 As enacted, the Act included a general rulemaking grant 2 2 but made no provision for sanctions for rules adopted under the grant. The Final Report of the Attorney General's Committee of 1941 confirms that the statute did not confer legislative rulemaking power: it noted that the United States Employees' Compensation Commission, the administrative agency initially charged with implementing the Act, was one of the "[r]elatively few" agencies that lacked the power to prescribe regulations having the force and effect of law.  2000)). Congress gave teeth to the rules and regulations promulgated by the SEC through two statutory sections. First, section 2o(b) gave the SEC the power to bring an action in any district court to enjoin acts or practices that constitute or will constitute a violation of the provisions of the Act, or any rule or regulations prescribed under the Act. Id. § 2o(b), 48 Stat. at 86 (codified as amended at 15 U.S.C. § 770. Second, section 24 provided that "[a]ny person who willfully violates any of the provisions of this title, or the rules and regulations promulgated by the Commission under authority thereof ... shall upon conviction be fined not more than $5,ooo or imprisoned not more than five years, or both." Id. § 24, 48 Stat. at 87 (codified as amended at 15 U.S.C. § 77x). According to the convention, these two statutory provisions indicate Congress's intent to give the SEC the authority to bind persons outside the agency with the force of law. 215 Ch. 404, 48 Stat. 881. Section 23(a) provided that "[t]he Commission and the Federal Reserve Board shall each have power to make such rules and regulations as may be necessary for the execution of the functions vested in them by this title." Id. § 23(a), 48 Stat. at 9oi (codified as amended at 15 U.S.C. § 78w(a)(i)). According to the convention, Congress indicated its intent to give legislative effect to these rules and regulations through two statutory provisions. First, section 2 i(e) gave the SEC the power to bring an action seeking to enjoin the violation of any rule or regulation promulgated under the Act. Id. § 21(e), 48 Stat. at 90 (codified as amended at i5 U.S.C. 78u). Second, section 32 set forth penalties for violations of any rules or regulations promulgated by the SEC. Id The National Labor Relations Act (NLRA) of 1935,218 by contrast, contained a facially ambiguous general rulemaking grant but lacked any textual signal indicating Congress's intent to attach legislative effect to such rules. Section 6(a) provided: "[t]he Board shall have authority from time to time to make, amend, and rescind such rules and regulations as may be necessary to carry out the provisions of this Act. '2 19 However, the Act did not give the NLRB the power to enjoin or prosecute violations of the rules promulgated under this grant, nor did it provide any statutory penalties for nonconformance with the NLRB's rules and regulations. According to the convention, these omissions indicate that Congress did not intend to grant legislative rulemaking powers to the NLRB.
The legislative history of the NLRA substantiates this conclusion. Upon consideration of different versions of the bill, one Senator made a proposal to limit the NLRB's rulemaking powers under what became section 6(a) to such "reasonable rules and regulations as may be necessary to carry out the provisions of this Act." 220 However, the proposal was rejected because, according to a Senate memorandum, "[in no case do the rules have the force of law in the sense that criminal penalties or fines accrue for their violation, and it seems sufficient that the rules prescribed must be 'necessary to carry out the provisions' of the act." 22 The Final Report of the Attorney General's Committee of 1941 confirms this legislative history: it notes that the NLRB's "power to 'make, amend, and rescind such rules and regulations as may be necessary to carry out the provisions of this chapter' has been assumed to extend only to matters of procedure. '222 Perhaps the strongest evidence of the continued operation of the convention during the New Deal years appears in statutes that contain multiple rulemaking grants.

trait.
As originally adopted in 1935, the Social Security Act contained both a general grant of rulemaking power and several more specific rulemaking grants. 22 6 Congress specified no legal consequences for violations of rules promulgated under the general grant. It was understood that this provision did not authorize legislative rules. 2 2 7 In contrast, the Act did attach sanctions to rules and regulations promulgated under specific rulemaking grants included in the statute, 2 2 8 and under the convention these would authorize legislative rules. 22 9 The Walsh-Healey Act, which Congress enacted to regulate federal government contracts, also contained both a general rulemaking grant and specific, narrower rulemaking grants. 230 The Act gave the Secretary of Labor the "authority from time to time to make, amend, and 224 Ch. 881, 49 Stat. 2036. 225 Ch. 675, 52 Stat. 1040. 226 See Social Security Act, § 1102, 49 Stat. at 647 (setting forth, in Title XI concerning the "general provisions" of the Act, a grant giving the Secretary of the 'IYeasury, the Secretary of Labor, and the Social Security Board the power to make rules and regulations "necessary to the efficient administration of the functions with which each is charged under this Act").
227 See FINAL REPORT, supra note 29, at 98 nn.i8-ig (listing the Social Security Board as an example of an agency that was given the power to act via case-by-case adjudication but was "not given power to elaborate the law they apply by adopting general regulations").
228 For example, section 8o8 of Title VIIE, which dealt with taxes with respect to employees, gave the Commissioner of Internal Revenue the power to "make and publish rules and regulations for the enforcement of this title." Social Security Act § 8o8, 49 Stat. at 638. Section 8io(a) gave teeth to the rules and regulations promulgated under section 8o8 by subjecting anyone who: [B]uys, sells, offers for sale, uses, transfers, takes or gives in exchange, or pledges or gives in pledge, except as authorized in this title or in regulations made pursuant thereto, any stamp, coupon, ticket, book, or other device, prescribed by the Commissioner of Internal Revenue under section 807 for the collection or payment of any tax imposed by this title, shall be fined not more than $i,ooo or imprisoned for not more than six months, or both. Id. § 8io(a), 49 Stat. at 638 (emphasis added). 229 In 1939, Congress adopted extensive amendments to the Social Security Act, including a new rulemaking grant. The new grant gave the Social Security Board "full power and authority to make rules and regulations." The rulemaking grant was facially ambiguous regarding whether it granted legislative, or merely interpretive, rulemaking authority. One provision of the amendments, however, mandated that when the Social Security Board denied a claimant benefits for failing "to submit proof in conformity with any regulation" issued under the general rulemaking grant, the court reviewing the denial of benefits "shall review only the question of conformity with such regulations and the validity of such regulations." Social Security Act Amendments of 1939, ch. 666, § 205(g), 53 Stat. 1360Stat. , 1370 In other words, the regulations, provided they were consistent with the Act, were to have legislative effect in determining eligibility for benefits. Because the statute specifically provided that a violation of a regulation could result in a sanctiondenial of benefitsthe new rulemaking grant authorized legislative rules under the convention.
230 Walsh-Healey Act, ch. 881, 49 Stat. 203649 Stat. (1936; see, e.g., id. § 4, 49 Stat. at 2038 (codified as amended at 41 U.S.C. § 38 (2ooo)) (granting the Secretary of Labor the authority to "administer the provisions of this Act ... and to prescribe rules and regulations with respect thereto"); id. rescind such rules and regulations as may be necessary to carry out the provisions of this Act." '231 The Act provided no sanctions for violations of the rules promulgated under this general grant. By contrast, rules and regulations issued pursuant to section 6, which authorized the Secretary to make rules setting reasonable "exemptions" from the minimum rate and maximum hour provisions of the Act, 2 32 are legislative according to the convention, because such exemptions necessarily would affect the conduct of contractors by exempting them from the Act's requirements. Support for this conclusion appears in a monograph on the Act prepared in 1939 for the Attorney General's Committee on Administrative Procedure, which summarized the significance of the differing treatment of rules and regulations promulgated under sections 4 and 6 as follows: The rules and regulations made by the Secretary under section 6 have vitality because they affect the nature of the performance required of a contractor. For the rules and regulations under section 4, however, no sanction is provided; they have no dispositive effect, except in so far as they operate as controls upon the Division itself; as administrative interpretations of the statute they may be given respectful judicial consideration if the interpretations be contested in court actions, but they are not binding as acts of subordinate legislation having the force and effect of law. 2 33 The Federal Food, Drug, and Cosmetic Act (FDCA) 2 34 similarly attached varying legal consequences to different rules and regulations based upon the source of rulemaking authority within the Act. The FDCA repealed the Pure Food Act of 1906,235 which did not confer legislative rulemaking authority over standards of identity for food or tolerances for poison residues in food. 236 Because a primary goal of the 1938 Act was to give greater powers to the Secretary of Agricul-  The key rulemaking provisions appear in section 701, which has been described as "schizophrenic" because of the mixed rulemaking provisions included within it.239 Section 701(a) granted general rulemaking power to the Secretary, stating that the Secretary could "promulgate regulations for the efficient enforcement of this Act. ' 240 Then sections 701(e), (f), and (g) set forth detailed procedures, including procedures for public hearings and judicial review of regulations, that the Secretary was required to follow when promulgating regulations under certain enumerated, specific rulemaking grants. 24 ' Notably, section 7oI(e) does not refer to section 701(a), and therefore rules promulgated under section 701(a) are not subject to section 701(e)'s procedural safeguards.
Under the convention, the specific rulemaking provisions that were subject to the procedural safeguards of sections 701(e), (f) and (g) conferred legislative rulemaking authority. 24 2 For example, section 401 gave the Secretary the power to promulgate regulations fixing standards of identity for food. 243 Those regulations are given legislative effect by various sections that expressly make violations of section 401 regulations unlawful and subject to criminal penalties. 244 237 See Lee, supra note 79, at is (noting that "attempts to change over from a system of interpretive regulations to legislative regulations embodying definitions and standards of identity for all foods" were a part of the push for new comprehensive food and drug legislation).
238 244 The following provisions gave section 401 regulations legislative effect: (i) section 403(g), 52 Stat. at 1047 (codified as amended at 21 U.S.C. § 343(g)), which declared that a food shall be deemed "misbranded" if it failed to conform to any standard of identity promulgated under section 4O1; (2) section 30I(a), 52 Stat. at 1042 (codified as amended at 21 U.S.C. § 331(a)), which made it unlawful to introduce into interstate commerce any "misbranded" food; (3) section 303(a), 52 Stat. at 1043 (codified as amended at 21 U.S.C. § 333(a)), which provided criminal penalties for introducing any "misbranded" food into interstate commerce; and (4) section 304, 52 Stat. at 1044-[Vol. 11I6:467 In contrast to the regulations subjected to the procedural safeguards of section 70I(e), nothing in the Act indicated that a regulation issued under the authority of section 70I(a) would subject the violator to any sanction, penalty, or other legal consequence. This silence suggests Congress's intent to withhold legislative rulemaking powers under that section. Legislative history supports this conclusion. During debate on the Act, Congress divided sharply over whether to give the Secretary the power to promulgate definitions, standards, and regulations having the force and effect of law. 246 Congress ultimately granted specific legislative rulemaking powers to the Secretary only after intense debate about what procedural safeguards the Act should employ to constrain those powers. 247 In contrast to the serious attention Congress gave to the procedural safeguards set forth in sections 701(e), (f), and (g), the general rulemaking grant of section 701(a) passed almost unnoticed. 24 As one scholar writing in 1 9 68 pointed out, the lack of attention paid to section 7oI(a), coupled with the absence of procedural safeguards attached to section 70I(a) rulemaking, carry great significance because 45 (codified as amended at 21 U.S.C. § 334), which subjected any food that was "misbranded" to seizure and condemnation.
A similar analysis applies to the other rulemaking grants covered by section 701(e). For example, the following provisions gave section 403(J) regulations legislative effect: (i) section 403(J), 52 Stat. at 1046 (codified as amended at 21 U.S.C. § 343(J)), which declared that food that "purports to be or is represented for special dietary uses" shall be deemed "misbranded" if its label fails to "bear[] such information concerning its vitamin, mineral and other dietary properties as the Secretary [prescribes] by regulations"; (2) section 30I(a), 52 Stat. at 1042 (codified as amended at 21 U.S.C. § 331(a)), which made it unlawful to introduce into interstate commerce any "misbranded food"; (3) section 303, 52 Stat, at 1043 (codified as amended at 21 U.S.C. § 333), which provided criminal penalties for introducing any "misbranded" food into interstate commerce; and (4) section 304, 52 Stat. at 1044-45 (codified as amended at 21 U.S.C. § 334), which subjected any misbranded food to seizure and condemnation. Similarly, the following statutory provisions gave section 4 o6(b) regulations legislative effect: (i) section 4 o6(b), 52 Stat, at 1049 (repealed I96O), which provided that "[t]he Secretary shall promulgate regulations providing for the listing of coaltar colors which are harmless and suitable for use in food and for the certification of batches of such colors"; (2) section 3oi(i), 52 Stat. at 1042 (codified as amended at 21 U.S.C. § 3 3 1(i)), which prohibited the unauthorized use of any mark, stamp, tag, or label authorized or required by the 4o6(b) regulations; and (,3) section 303, 52 Stat. at 1043 (codified as amended at § 333), which provided criminal penalties for violating any of the provisions of section 301, including its prohibition of the unauthorized use of any marks, stamps, tags, or labels authorized or required by section 406(b) regulations. 245 See Forte, supra note 236, at 693 n.29 (noting that there is no specific provision providing that a violation of section 701(a) is a violation of the Act and concluding that this omission is "probably due to the fact that Congress intended § 7oI(a) to authorize interpretive, not substantive, regulations"). 246 See id. at 692 (noting that serious questions were raised in Congress about both the advisability of authorizing legislative rulemaking and the procedural safeguards necessary to prevent arbitrary rules). 247 See id. at 692-94. 248 See id. at 693. "[i]t seems inconceivable that Congress, after five years of debate on the procedural limitations to be placed on the promulgation of some substantive regulations, would authorize the issuance of other regulations having the force and effect of law without debate and without any procedural safeguards. 2 49 Statements in Senate and House reports confirm that Congress did not intend section 701(a) to grant legislative rulemaking authority. A House Report, for example, described section 701 as follows: Section 701 relates generally to regulations. In the case of regulations, the violation of which constitutes an offense, it is required that appropriate notice of a public hearing be given and that adequate time shall be given after the promulgation of a regulation before it becomes effective.. This report indicates that key participants in the legislative process understood that by attaching sanctions to violations of certain rules and regulations, they elevated those regulations to legislative status. Given the understanding that these rules were legislative, Congress felt it necessary to attach important procedural safeguards to their issuance. No one urged the need to attach safeguards to rules promulgated under section 7oI(a), because those rules would carry only interpretive effect.
Overall, the text, structure, and legislative history of statutes Congress enacted through the end of the New Deal show a remarkably consistent adherence to the convention's framework for distinguishing between legislative and housekeeping grants.
The key was not whether the rulemaking grant was general or specific. Members of Congress referred to the presence or absence of sanctions as the basis for distinguishing between legislative and housekeeping grants on sev-249 Id. at 694; see also Becker, supra note 242, at 681-82 ("In light of this history of express and specific safeguards governing the authority to issue the substantive regulations authorized by the statute, it seems strange to suggest that the same statute was also intended to authorize the issuance of other unspecified regulations having the force and effect of law, whose violation could also result in the imposition of criminal penalties, without any provision for hearings and with judicial review essentially limited to whether the Agency action is arbitrary or capricious."); Merrill S. [Vol. 116:467 eral occasions when debating rulemaking grants 25 ' but did not refer to the generality of rulemaking grants as a basis for making such distinctions. In fact, Congress occasionally adopted very general rulemaking grants and attached sanctions to violations of rules issued under them in the same section of an act, thereby signaling that the rules promulgated under the grant would be treated as binding. 252 Despite the strength of the evidence supporting the existence of the convention, there is at least one rulemaking grant from the New Deal era in which the convention does not seem to capture congressional intent: the general rulemaking grant found in Title I of the Communications Act of 1934.253 The problem stems from the mixture of regulatory schemes that Congress borrowed from other acts in putting together the Communications Act. 254 Title I, which copied many of its provisions from the ICA, sets forth the Act's general organizational provision. Section 4 (i) of Title I contains a general rulemaking grant providing that the FCC "may perform any and all acts, make such rules and regulations, and issue such orders, not inconsistent with this Act as may be necessary in the execution of its functions. ' '255 Title II governs common carriers. Title III, which regulates broadcasting, borrowed many of its provisions from the Radio Act of 1927,256 including The Secretary of Interior shall make provision for the protection, administration, regulation, and improvement of such grazing districts as may be created under the authority of the foregoing section, and he shall make such rules and regulations and establish such service, enter into such cooperative agreements, and do any and all things necessary to accomplish the purposes of this Act and to insure the objects of such grazing districts, namely, to regulate their occupancy and use, to preserve the land and its resources from destruction or unnecessary injury, to provide for the orderly use, improvement, and development of the range; .. .and any willful violation of the provisions of this Act or of such rules and regulations thereunder after actual notice thereof shall be punishable by a fine of not more than $500. Id several specific rulemaking grants. 5 7 Title IV includes the Act's procedural and administrative provisions. Finally, Title V contains the Act's penal provisions. One provision of Title V, section 502, sets forth a general provision subjecting any person who violates any rule or regulation issued under the Act to criminal penalties. 2 5 8 If we apply the convention to the Communications Act, the omnibus provision of criminal sanctions for rule violations in section 502 has the effect of transforming all rulemaking grants in the Actincluding the grant in section 4 (i)into grants of legislative rulemaking authority. Given the structure and history of the Act, however, it is doubtful that this is what Congress intended. Section 4 (i) was based on section 17 of the ICA, 2 59 which granted only procedural rulemaking powers. In addition, section 4 (i) appears in Title I amid provisions relating to the general administrative organization of the FCCfar away from the penal provisions in section 502 of Title V.260 Finally, section 4 (i) was part of the original draft bill and passed through the 257 See, e.g., Communications Act, § 303(f), 48 Stat. at io82 (codified as amended at 47 U.S.C. § 303(f)) (giving the FCC power to "[m]ake such regulations not inconsistent with law as it may deem necessary to prevent interference between stations and to carry out the provisions of this Act"); id. § 3 0 3 (i), 48 Stat. at io82 ((codified at 47 U.S.C. § 3o3(i)) giving the FCC the "authority to make special regulations applicable to radio stations engaged in chain broadcasting"); id. § 3030), 48 Stat. at i082 (codified at 47 U.S.C. § 303(j)) (granting the FCC the power to "make general rules and regulations requiring stations to keep such records of programs, transmissions of energy, communications, or signals as it may deem desirable"). In 1937, Title III of the Communications Act was amended to add a provision authorizing the FCC to "[m]ake such rules and regulations and prescribe such restrictions and conditions not inconsistent with law, as may be necessary to carry out the provisions of this Act." Amendments to the Communications Act of 1934, ch. 229, § 6(b), 50 Stat. i89, 191 (1937) (adding section 303(r)). All of these various rulemaking grants included within Title III were given legislative effect by section 312(a), which provided for license revocation if any licensee violated any of the Commission's regulations. Communications Act, § 312, 48 Stat, at io86-87 (codified as amended at 47 U.S.C. § 312(a)).
258 Section 502, which was part of the 1934 Act, subjects any person "who willfully and knowingly violates any rule, regulation, restriction, or condition" imposed by the FCC under the authority of the Communications Act to a fine of up to $500 for each day during which such offense occurs. Communications Act, § 502, 48 Stat. at i ioo-oi (codified at 47 U.S.C. § 502). 259 A comparison of the language used in section 17 of the ICA with sections 4 (h) through () of the Communications Act shows that section 4 (i) was based upon section 17 of the ICA. Compare Communications Act, § 4 (i), 48 Stat. at io68 (codified as amended at 47 U.S.C. § i 5 4 (i)) ("The Commission may perform any and all acts, make such rules and regulations, and issue such orders, not inconsistent with this Act, as may be necessary in the execution of its functions."), with Interstate Commerce Act, ch. 104, § 17, 24 Stat. 379, 385-86 (1887) (repealed 1978) ("Said Commission may, from time to time, make or amend such general rules or orders as may be requisite for the order and regulation of proceedings before it, including forms of notices and the service thereof, which shall conform, as nearly as may be to those in use in the courts of the United States.").
260 Section 502, 48 Stat. at iioo, borrowed language from section 32 of the Radio Act of 1927. The Communications Act suggests that Congress was not infallibly attentive to the drafting convention in signaling which rulemaking grants are legislative and which are not. The most likely explanation for this oversight, in the case of section 4 (i) of the Communications Act, is that the general rulemaking grant drew so little attention, and was physically placed in the statute at such a great distance from section 502, that no one noticed section 502 was written in such a way that it literally applied to section 4 (i). Had someone noticed the interaction between section 4 (i) and section 502, we suspect that section 502 would have been revised to make clear that the penal provision reached only those rules issued under specific rulemaking grants included in the Communications Act, such as those rulemaking provisions borrowed from the Radio Act and placed in Title III. But apparently no one noticed. The Communications Act thus presents an example of a statute in which the inference of legislative intent drawn from the application of the convention should probably be disregarded, given other, contrary evidence of legislative intent.

D. The Office of Legislative Counsel
One notable institutional development took place at approximately the same time Congress began systematically following the convention: the creation of the Office of Legislative Counsel in both the Senate and the House. 65 The offices began as a pilot project organized by Columbia Law School in 1916 but became permanent two years later.1 66 According to Frederic P. Lee, one of the lawyers who served in both the House and Senate offices in their early days, 267 the offices were interest to carry out the provisions of this Act." Amendments to the Communications Act, ch. 296, 52 Stat. 588, 588 (1938) (codified at 47 U.S.C. § 20i(b)). Although the language of this additional rulemaking grant appears to be very broad, the legislative history suggests that its intended scope was narrowly confined to matters concerning the furnishing of reports of positions of ships at sea. One senator explained during debate on the Senate floor that the amendment "relates only to information which comes from vessels at sea as to their location." 83 CONG. REC. 6291 (1938) (statement of Sen. White); see also S. REP. NO. 75-1652, at 3 (1938) (noting that the amendment allows free reporting services regarding ships at sea, but that it subjects this free service "to rules, regulations, and limitations which the Commission finds desirable in the public interest"). Nonetheless, in Iowa Utilities Board, the Supreme Court held that section 20i(b) constitutes a general grant of legislative rulemaking authority that extends to all of the FCC's jurisdiction over common carriers. Iowa Utilities Board, 525 U.S. at 377-78. Justice Scalia's opinion for the Court contained no discussion of the history of section 2oi(b), which suggests a far narrower purpose for that grant.
Whether Iowa Utilities Board was correct about section 2oi(b) if we view the matter through the lens of the convention is a closer call than the question whether section 4 (i) should be regarded as a grant of legislative rulemaking authority. Obviously, the omnibus penal provision in section 502 also applies to section 201(b), making section 201(b) presumptively legislative under the convention. And section 2oi(b), unlike section 4 (i), is not in a separate title of the Act dealing with definitions and administrative provisions, nor was it borrowed from another statute that conferred only procedural authority. The correct analysis here probably turns on whether one is willing to credit the sort of legislative history that appears in committee reports and floor statements that bear on legislative purpose (a large topic beyond the scope of this Article). If one is willing to credit such materials, then the presumptive conclusion about section 20i(b) drawn from the convention should probably be overcome by evidence that Congress intended this grant to apply only to locational signals from vessels at sea. If one is not willing to credit this kind of evidence, but only the inferences that can be drawn from the text of statutes (which of course is Justice Scalia's general position, see Scalia, supra note 55, at 29-37), then the presumptive conclusion based on the convention should probably stand and the Court was correct in Iowa Utilities Board to conclude that section 2oi(b) confers general legislative authority over all matters within the FCC's jurisdiction under Title II. strictly nonpartisan and professional. 2 6 Their functions included providing legal research to members of Congress, reviewing and commenting on proposed legislation, and most importantly, drafting bills. 2 69 In drafting legislation, the lawyers in the offices paid particular attention to problems of constitutionality, administrability, and judicial review. As Lee recounted: In all matters of drafting, knowledge of constitutional and administrative law is invaluable ....
[M]ost of the complex legislative problems today involve extensive executive machinery for enforcement and administration.
In the legislative provision for this machinery there must be met the many administrative and constitutional law problems involved in the form in which executive action may properly express itself and in the judicial review of such action. 270 We are not aware of any direct evidence about the kind of advice the Offices of Legislative Counsel provided to members of Congress concerning how to signal whether a statute authorized legislative rulemaking. But one very powerful piece of circumstantial evidence exists. Lee published another article in the Georgetown Law Journal in 194o detailing how to distinguish between legislative and interpretive regulations. 2 71 In this article, Lee used the term "substantive" to describe two types of rules: "legislative" regulations, which prescribe what the law shall be and have the force and effect of law; and "interpretive" regulations, which merely construe a statute and do not have the force and effect of law. 272 According to Lee, whether a "substantive" regulation is "legislative" or "interpretive" depends upon the grant of rulemaking authority in the statute, and in particular, whether Congress specified some sanction for violation of the rules. 27   In addition Congress may in the statute declare specifically that the regulations shall have the force and effect of law. More often a statute granting authority to make explained, if the "power to prescribe a substantive regulation is delegated by statute, but no sanctions are imposed by statute for failure to conform to the regulation, then it is interpretive." 2 7 4 Only if the statute provides legal sanctions for violations of the regulations do those regulations have the force and effect of law. 75 The implications of Lee's Georgetown article are considerable. His comments about how one identifies a grant of legislative, as opposed to interpretive, rulemaking authority exactly track the convention we have described. This understanding almost certainly reflects his extensive tenure as an attorney in the House and Senate Offices of Legislative Counsela tenure that coincided with the period during which Congress routinely observed the convention. Since the members of Congress frequently called upon the attorneys in the Offices of Legislative Counsel for assistance in drafting statutes to create administrative agencies and to confer powers on them, 2 7 6 and since the attorneys sought to implement faithfully the wishes of their superiors, 27 7 one can only conclude that the convention described by Lee was the device used by the attorneys in these offices to signal the intentions of Congress. 27 Lee's views about how to distinguish between legislative and interpretive grants of rulemaking authority are generally consistent with regulations attaches none of these sanctions to the regulations, is silent as to their force and effect, and is intended to cover only interpretive regulations or regulations affecting internal departmental functions. Particularly is this so of the usual broad grants that authorize "such regulations as may be necessary to carry out the provisions of this Act" or which are similarly phrased, sometimes with the additional requirement that the regulations be "not inconsistent with law." Id. at 19-2 (footnotes omitted other legal commentary in the New Deal period, up to and beyond the enactment of the APA. 2 7 9 For example, in a monograph written in the early 195os, Frank Cooper of Michigan Law School enumerated those circumstances in which congressional intent to authorize legislative rulemaking was clear: where "the statute specifically declares that the regulations shall have the force and effect of law"; where "the statute provides penalties that will result from noncompliance with the regulations"; where the statute makes "noncompliance with the regulations a ground for revocation of permits or licenses"; or where the statute authorizes "regulations which will relax a statutory rule otherwise applicable. '280 Cooper further noted that many statutes contain general rulemaking grants that are facially ambiguous regarding whether they authorize legislative or interpretive rulemaking. In such cases, he wrote, "the courts usually treat the regulation on the same basis as in cases where there can be no doubt but that the regulation is merely interpretative. "281

E. The Administrative Procedure Act
If the prevailing understanding was that agencies could make rules with the force of law only if Congress had provided by statute some sanction for violations of those rules, one would expect to find this understanding reflected in the language and legislative history of the APA.
Insofar as legislative history is concerned, such evidence abounds in the work of the Attorney General's Committee on Administrative Procedure, a blue-ribbon committee appointed by President Roosevelt in the late 1930s to undertake a comprehensive survey of administrative agencies and their procedures. 2 8 2 The committee pro-279 See sources cited infra note 281. A notable exception to this generalization is tax scholarship. As described more fully in section VI.C infra, pp. 574-75, tax specialists writing around the same time as Lee opined that the relevant distinction was between general and specific grants of rulemaking authority. See infra notes 604-612 and accompanying text. duced twenty-seven monographs that described the decisionmaking processes of various agencies and a summary report entitled Administrative Procedure in Government Agencies 2 8 3 better known as the committee's Final Report. 2 8 4 Both the monographs and the Final Report, which are considered "classics of administrative law scholarship" and which laid the intellectual groundwork for the drafting of the APA,285 contain statements that confirm the existence of the convention. The Final Report, for example, distinguishes between interpretive and legislative regulations by focusing on whether the statute imposes sanctions to compel observance of the regulations. 2 8 6 It explains that the "statutes themselves and not the regulations remain in theory the sole criterion of what the law authorizes or compels and what it forbids. 2 1 8 7 Specifically, the Final Report describes legislative regulations as follows: Many statutes contain provisions which become fully operative only after exercise of an agency's rule-making function. Sometimes the enjoyment of a privilege is made conditional upon regulations, as, for example, where Congress permits the importation of an article "upon such rules and regulations as the Secretary of the Treasury may prescribe," or allows utilization of public forests in accord with regulations to be laid down by administrative officers. Sometimes the extent of an affirmative duty is to be fixed by regulations, as, for example, where employers are commanded to pay wages not less than those prescribed in administrative regulations. Sometimes a prohibition is made precise by regulations, as, for example, where the sale of dangerous drugs is forbidden and the determination of what drugs are dangerous is left to administrative rules. In such instances the striking characteristic of the legislation is that it attaches sanctions to compel observance of the regulations, by imposing penalties upon or withholding benefits from those who disregard their terms.  The Final Report's description makes it clear that legislative rules can be the product of a broad, general rulemaking grant, such as a grant to make "such rules and regulations as the Secretary of the Treasury may prescribe. '2 89 The key, according to the Final Report, is not the general or specific nature of the rulemaking grant, but rather whether Congress attached sanctions in the statute to compel observance of the regulations.
In addition to the Final Report, at least two of the monographs prepared by the Attorney General's Committee recognize that facially ambiguous rulemaking grants sometimes confer authority to make only interpretive rules. 2 90 For example, the monograph on the Walsh-Healey Act explains that rules issued by the Secretary of Labor under the Act's general rulemaking grant to "make, amend, and rescind such rules and regulations as may be necessary" are not binding because "no sanction is provided." 29 1 Insofar as the text of the APA itself is concerned, the convention is less easy to discern. We would submit, however, that the influence of the convention appears in the seldom noticed section 558(b), which provides: "A sanction may not be imposed or a substantive rule or order issued except within jurisdiction delegated to the agency and as authorized by law. '292 Note the close association between "sanction" and "substantive rule," and the clear command that neither can be imposed by an agency unless Congress has authorized it to do so by law. Although section 558(b) does not expressly codify the conventionthat an agency may not issue legislative rules unless Congress has provided sanctions for rule violationsthe effect of the provision is the same. Legislative rules were and are widely understood to be those that, when violated, give rise to adverse legal consequencesin other words, "sanctions" as that term is broadly defined by the APA. 293 289 FINAL REPORT, supra note 29, at 27 (internal quotation marks omitted). 293 The APA's broad definition of "sanction" includes not only the imposition of a penalty or fine, the forfeiture of property, the assessment of damages, or the loss of a license, but also any other "compulsory or restrictive action" by an agency. 5 U.S.C. § 551(lO) (quoted in full supra note 120). 2002] Thus, what section 558(b) means, in the words of the Senate Report, is that "agencies may not impose sanctions which have not been specifically or generally provided for them to impose. 2 94 Once we understand the convention, we can read the APA in a new light and understand that it, too, presupposes that agencies can make rules with the force of law only if Congress has legislated statutory sanctions for rule violations.

F. Why a Convention?
Only rarely in the Progressive and New Deal years did Congress state expressly that agencies were authorized to issue rules with the force of law or were limited to issuing advisory interpretations and procedural rules. Instead, Congress repeatedly enacted legislation ambiguously authorizing agencies to issue "rules and regulations" and relied on a signaling device for indicating whether a grant was legislative: grants to make rules backed by sanctions authorized legislative rulemaking, whereas grants to make rules not backed by sanctions authorized only procedural or interpretive rulemaking.
As best we can tell, agencies consistently respected the convention during these years. Of course, adjudication dominated administrative law for the first six decades of the twentieth century, with rulemaking assuming central importance only in the 1970s and afterward. 95 Nevertheless, it is no accident that agencies operating under broadly worded rulemaking grants that, under the convention, conferred only nonlegislative rulemaking authorityincluding the FDA, NLRB, FTC, and United States Employees' Compensation Commissionmade no attempt during these years to adopt legislative rules. 2 96 In contrast, the FCC, SEC, Federal Power Commission, and ICC (with respect to specific rulemaking grants added to the ICA by amend-  [Vol. 11 6:467 ments) all had occasion to issue legislative rules. 29 7 Each agency in this latter group operated under broadly worded rulemaking grants that, under the convention, conferred legislative rulemaking authority.
Why did Congress use this convention to signal that it was conferring legislative rulemaking authority, rather than straightforwardly announcing that it was authorizing the agency to make "legislative" rules or rules with the "force of law"? We can only speculate about the answer. One possibility is economy in drafting. If Congress had explicitly authorized an agency to promulgate legislative rules, then the question immediately would have arisen: what sort of sanction attaches to persons who violate these rules? Eaton and Grimaud (and later the APA) could be read to say that Congress also must specify the sanctions. 2 98 So perhaps Congress thought it more economical to proceed immediately to the specification of the sanctions, which would by implication also convey the information that the rules would be legislative.
Another, more plausible possibility is that use of the convention would have been less likely to trigger political opposition than an explicit statement of authority to adopt rules with the force of law. An express delegation of legislative authority would in effect wave a red flag and alert opponents of the legislation. They could tap into deepseated unease over the idea of delegating legislative rulemaking powers to a body of unelected administrators and could use this unease to rally opposition to the measure. We have already seen examples of this phenomenon in the legislative histories of the Packers and Stockyards Act 2 99 and the FDCA. 300 Later, we will see another example in the history of the Internal Revenue statutes. 30 1 So we know the potential existed for opponents of particular bills to use the delegation of legislative rulemaking authority as a focal point of opposition.
Use of the convention obviously could not eliminate all such controversy about the delegation of legislative rulemaking authority. But at least it allowed regulatory statutes to confer legislative powers without using the inflammatory words "legislative" or "force of law," and hence it reduced somewhat the opportunities for opponents to raise this issue. This was especially true if the bill placed the sanctions at some distance from the grant of authority to adopt "rules and regulations." Such "acoustical separation" would further reduce the chances that a casual reader of the bill would notice that Congress was endowing the agency with legislative rulemaking power, even though knowledgeable insiders reading through the bill would grasp this implication immediately.
3 0 2 IV. THE CONVENTION IGNORED Given the secondary importance of rulemaking during the first six decades of the twentieth century, it is perhaps not surprising that more than thirty years passed after United States v. Grimaud 30 3 before the Supreme Court heard a case that potentially presented the question whether Congress had authorized an agency to make rules having the force of law under a facially ambiguous rulemaking grant. 30 4 Nor is it surprising that another ten years passed before the Court confronted a similar case again. 30 5 All in all, we have identified eight principal cases decided between 1943 and 1979 that potentially presented the Court with a question about an agency's authority to make legislative rules. 30 6 These cases involved disparate issues under disparate statutes and were rendered too infrequently to generate anything that could be described as a jurisprudence of rulemaking. In fact, the most striking aspect of the eight decisions, taken as a whole, is the absence of any recognizable theory regarding when grants of rulemaking authority confer lawmaking power and when they do not. 302 We postpone to Part VII the normative issues associated with whether Congress should be required to delegate legislative rulemaking powers through an express statement, as opposed to using a convention that signals its clear intent to do so. 303 220 U.S. 5o6 (igii), discussed supra pp. In retrospect, only one of the eight decisions generated an outcome inconsistent with the convention, 30 7 and in that case, the party challenging the rule did not raise the question whether the agency had been delegated the power to act with the force of law. 30 8 But any congruity between the convention and judicial outcomes during these years owes nothing to the Court's deliberative processes. Not one of the cases discussed or even acknowledged the convention introduced in Part III. Cumulatively, the only conclusion one can draw from reading the opinions is that the Justicesand the lawyers appearing before themhad no knowledge of the convention. This ignorance was to have fateful consequences. The Court's failure to offer any coherent basis for distinguishing between legislative and nonlegislative rulemaking grants created the opportunity for later courts and commentators to read the Court's opinions selectively to support the proposition that facially ambiguous rulemaking grants always confer legislative powers. Court, 1943Court, -1979 In an effort to compress our presentation, we divide the eight cases into four groups of two cases each. Within each group, the cases relate chronologically, present similar underlying issues, and reflect similar styles of reasoning in the way those issues are resolved.

A. Rulemaking Grants in the Supreme
i. National Broadcasting (1943) and American Trucking (1953). -The first two cases, National Broadcasting Co. v. United States 30 9 and American Trucking Ass'ns v. United States, 310 are separated by a full decade but present similar issues. National Broadcasting arose under the Communications Act of 1934311 and American Trucking under the Motor Carrier Act. 3 1 2 Broadly speaking, the issue in each case was whether the agency had authority to promulgate a legislative rule on a subject regarding which Congress had not delegated specific regulatory authority. National Broadcasting asked whether the FCC could adopt "chain broadcasting" rules regulating contractual relationships between networks and affiliated broadcasting stations. 31 3 The granting of a broadcasting license was conditioned upon a station's following these rules. 31 4 American Trucking asked whether the ICC could adopt regulations governing the practices of companies that leased trucks to licensed motor carriers. 3 15 The granting of operating licenses was conditioned upon a carrier's following these rules. 31 6 In both cases, the Court upheld the challenged regulations, citing multiple congressional rulemaking grants, including both specific and generic grants. In National Broadcasting, the Court relied on three rulemaking grants found in Title III of the Communications Act, one general and two relatively specific, each of which was facially ambiguous concerning whether it authorized rules with the force of law. 3 17 In American Trucking, the Court relied on two grants, one general and one specific, both of which were facially ambiguous in the same sense.

18
The Court upheld the regulations in both cases without specifying which rulemaking grants endowed the regulations with the force of law. It is not surprising that the Court proceeded in this manner. The challenger in each case claimed that the agency was acting beyond the scope of its regulatory jurisdiction. 3 1 9 In neither case did the challenger maintain that the agency lacked the power to adopt regulations having the force of law. Thus the Court probably felt no compulsion to discuss which statutory provisions supported legislative rulemaking.
Had the parties raised the question of the agencies' power to adopt legislative rules, and had the Court resolved the question in light of the drafting convention, the agencies' authority would have been sustained in both cases. With respect to the FCC's regulation of chain broadcasting, section 312(a) of the Communications Act provided for the revocation of licenses based on violations of the FCC's regulations, 320 315 See American Trucking, 344 U.S. at 300-02. 316 See id. at 302. 317 These were section 303(g), which provided that the FCC shall "generally encourage the larger and more effective use of radio in the public interest"; section 3 0 3 (i), which gave the FCC "authority to make special regulations applicable to radio stations engaged in chain broadcasting"; and section 303(r) (added by section 6(b) of the 1937 amendments to the Communications Act), which empowered the FCC to adopt "such rules and regulations and prescribe such restrictions and conditions, not inconsistent with law, as may be necessary to carry out the provisions of this and section 502 made the violation of "any rule, regulation, restriction, or condition" imposed by the agency under authority of Title III of the Act a criminal offense. 3 2 1 Thus, both the specific and the general rulemaking grants cited in National Broadcasting as authority for the FCC rule enabled legislative rulemaking, according to the convention. With respect to the ICC's leasing regulation, section 222(a) of the Motor Carrier Act provided for statutory fines for violations of the ICC's regulations, and section 222(b) gave the ICC the power to bring suit when any regulations were violated. 32 2 Under the convention, these provisions signaled Congress's intent to give legislative effect to rules and regulations that the ICC promulgated under the Act. The significance of National Broadcasting and American Trucking for the future lay not in what the Court said, but in what it did not say: the opinions demonstrated an apparent indifference to the question of the sources of the agencies' authority as legislative rulemakers. 323 Nowhere in his opinion for the Court in National Broadcasting did Justice Frankfurter attend to the distinction between grants of legislative and nonlegislative rulemaking powers. 324 Instead, the Court implicitly assumed that the Communications Act's rulemaking grants conferred the power to make legislative rules. Justice Reed's majority opinion in American Trucking quickly rejected the contention that the general rulemaking grant in the Motor Carrier Act "merely concerns agency procedures and is solely administrative. '32 5 He stated that the reference to "enforcement" in the general grant refuted this conten- Broadcasting to mean that a legislative rule "may rest upon an implied or an unclear grant of power" because the Court had sustained the FCC's rules even though the power to issue regulations governing the contractual relationships between networks and affiliates was not specifically conferred by the Communications Act. i KENNETH CULP DAVIS, ADMINISTRATIVE LAW TREATISE § 5.03, at 299 n.2 (ist ed. 1958). Davis thus saw National Broadcasting as running contrary to the older view, associated with Eaton and Grimaud, that the authority to issue binding rules must be specifically or explicitly delegated. Id. 324 See National Broadcasting, 319 U.S. at 2o9-26 (reasoning that the FCC's regulations were within its authority under the Communications Act). As will be discussed infra section V.B.2.(b), this silence proved to be significant when Judge Friendly later asserted that National Broadcasting supported the view that general grants of rulemaking authority confer the power to make legislative rules. See Nat'l Ass'n of Pharm. Mfrs. v. FDA, 637 F.2d 877, 88o (2d Cir. ig8I) ("The Supreme Court's decision in National Broadcasting Co. v. United States, which in retrospect seems to have inaugurated the modern approach, was not universally so recognized at the time, since the Court there relied in part on more specific grants of rulemaking power and the regulations at issue in that case, although legislative in effect, were clothed in the garb of procedural rules." (citations omitted)). In Storer Broadcasting, as in National Broadcasting, the FCC cited multiple grants of rulemaking authority to support its rules restricting multiple ownership, each of which was facially ambiguous. 333 In Texaco, only one rulemaking grant supported the Federal Power Commission: section 16 of the Natural Gas Act, which gave the Commission authority to prescribe such regulations "as it may find necessary or appropriate to carry out the provisions of this Act. '33 4 This provision was facially ambiguous in the same sense.
The thrust of the challenge in each case was not that the agency lacked power to make rules with the force of law, but that the structure of each act required that the issue in question be resolved through [Vol. 11I6:467 case-by-case adjudication. 33s The Supreme Court rejected the argument in both cases, holding that the statutory requirement of a hearing did not preclude the agencies from "particularizing statutory standards through the rulemaking process" and from rejecting without a hearing those who failed either to meet the regulation or to show why they were entitled to a waiver of the rule. 33 6 Given that the challengers in both cases objected to the rules as interfering with individual hearing rights, it is not surprising that the Court spent little time considering whether the agency had authority to promulgate rules having the force of law. The Court concerned itself only with the question whether the rules were within the scope of the agencies' authority, not with whether the rules were legislative. In Storer Broadcasting, Justice Reed's opinion for the Court concluded that the FCC's multiple ownership rules, although not specifically authorized by statute, were sufficiently related to the purpose of the Communications Act and therefore within the scope of sections 4 (i) and 303(r), which authorized rules "not inconsistent with the Act or law. '337 In Texaco, Justice Douglas's majority opinion similarly concluded that the Federal Power Commission's general rulemaking grant provided "ample" authority for rules that imposed conditions on individual applications because the rules were within the substantive bounds of the Natural Gas Act. 338 Had the Court addressed the question whether the agencies had authority to adopt legislative rules, and had it resolved the question in light of the drafting convention, then once again the Court would have upheld the agencies' authority in both cases. In the Communications Act, addressed in Storer Broadcasting, section 4 (i) probably does not support legislative rulemaking authority. 339 But section 303(r) appears to carry legislative effect under the convention for the same reasons discussed in connection with National Broadcastingrules promulgated under this grant are enforced by license revocations and criminal sanctions. 340 The Federal Power Commission's general rulemaking grant in Texaco would have been construed as conferring legislative rulemaking authority under the convention, because the Natural Gas Act authorized the Commission to bring suit against those who violate its regulations and to subject them to a fine. 34 1 The significance of Storer Broadcasting and Texaco does not lie in any holding by the Court about the meaning of facially ambiguous rulemaking grants. Rather, the cases are important largely because they encouraged the use of rulemaking by agencies in an effort to streamline the regulatory process and, more subtly, because they compounded the impression, inaugurated by National Broadcasting and American Trucking, that the question whether Congress has delegated to an agency the authority to make rules with the force of law is of little interest or significance to courts.
3. Thorpe (1969) and Mourning (1973). -The next two cases mark a departure from the Court's prior decisions in that, for the first time, the Court expressly considered whether particular rules adopted by agencies had the force of law. In Thorpe v. Housing Authority, 342 a tenant in a federally assisted housing project argued that a circular issued by the Department of Housing and Urban Development (HUD) entitled her to notice and an opportunity to respond prior to eviction by a local housing authority. 343 In Mourning v. Family Publication Services, Inc., 344 a company selling magazine subscriptions challenged a Federal Reserve Board regulation, promulgated under the Truth in Lending Act, that broadly defined consumer credit transactions to include any transaction payable in four or more installments. 34 S The disposition of the two cases abounds with irony from the perspective of the drafting convention. In Thorpe, the local housing authority did not argue that HUD lacked the authority to issue legally binding regulations on the subject of eviction procedures. 346  authority alternatively argued that if the Court ruled that the circular was intended to be mandatory, it could not be applied retroactively to pending eviction proceedings. 347 In an opinion by Chief Justice Warren, the Court held that the circular was mandatory 348 and thus was a legislative rule. The Court did not explain why the facially ambiguous grant of rulemaking authority found in section 8 of the Housing Act of 1937349 was a grant of legislative authority, other than to observe in a footnote that "[s]uch broad rule-making powers have been granted to numerous other federal administrative bodies in substantially the same language. '3s° The passage that immediately follows, and a passage later in the opinion, intimated that the only constraint on the agency was whether the rules were consistent with the general purposes of the Act. 3S ' Having established that the circular had the force of law, the Court then rejected the housing authority's argument that applying the circular to evictions already in process would constitute impermissible retroactive rulemaking.  § 717t (2000)) (which was also a legislative grant under the convention, see supra pp. 530-32). The author of the footnote was obviously unaware of the signaling device that we argue Congress had used in indicating which grants were legislative and which were nonlegislative. 351 In the passage immediately following, the Court addressed and rejected the local housing authority's contention that the circular, if understood to be a mandatory rule, was inconsistent with the statutory purpose of vesting in local housing authorities maximum responsibility for administering public housing programs. See Thorpe, 393 U.S. at 277-78. Later in the opinion, the Court observed that the circular was "reasonably related to the purposes of the enabling legislation under which it was promulgated," id. at 28o-81, which the Court said included providing "a decent home and suitable living environment to every American family," id. In contrast, the parties' briefs in Mourning did raise the question whether the facially ambiguous grant of rulemaking under the Truth in Lending Act 3 55 authorized the Federal Reserve Board to make rules having the force of law. 356 In doing so, however, the parties did not discuss the convention.
Following Thorpe, the Court ignored this threshold issue, focusing instead on the scope of the agency's rulemaking authority. As the Court framed the issue: Where the empowering provision of a statute states simply that the agency may "make . .. such rules and regulations as may be necessary to carry out the provisions of this Act," we have held that the validity of a regulation promulgated thereunder will be sustained so long as it is "reasonably related to the purposes of the enabling legislation.
'35 7 The Court held that the Federal Reserve Board's four-installment rule fell within the scope of the agency's authority and was reasonably related to the statute's purpose. 3 5 8 Had the Court sought to resolve the question whether the Federal Reserve Board's rules had the force of law under the convention, it would have concluded that they did. Section i i2 of the Truth in Lending Act provided for fines up to $50o0 and imprisonment for per- 353 See United States Housing Act of 1937, ch. 896, § 8, 50 Stat. 888, 891. Even if Congress had given HUD the authority to render legislative rules, there is no indication the agency intended that the circular would function as a legislative rule. Unlike most legislative rules, it was not promulgated in accordance with notice-and-comment procedures. See 5 U.S.C. § 55 3 (b) (2000). But see id. § 553(a) (exempting rules related to "grants"including perhaps HUD grants to local housing authoritiesfrom notice-and-comment requirements). Also, the agency did not publish the circular in the Federal Register. See Am. Mining Cong. v. Mine Safety & Health Admin., 995 F.2d 11o6, 0IO9 (D.C. Cir. 1993) (noting that legislative rules are usually published in the Federal Register). 354 See, e.g., Brecht v. Abrahamson, 507 U.S. 619, 631 (1993) (noting that if a prior precedent does not "squarely" address an issue, the Court is "free to address [it] on the merits" at a later date); United States v. L.A. Tucker 'Truck Lines, Inc., 344 U.S. 33, 38 (1952) (stating that an issue not "raised in briefs or argument or discussed in the opinion of the Court" cannot be taken as "a binding precedent on [the] point"); Webster v. Fall, 266 U.S. 507, 511 (1925) (stating that issues that merely "lurk in the record" and are not brought to the Court's attention should not "be considered as having been so decided as to constitute precedents"). sons who failed to disclose information required by any regulation issued under the Act. 35 9 In addition, the Act expressly gave the Board's regulations the force of law by providing that "[a]ny reference to any requirement imposed under this title or any provision thereof includes reference to the regulations of the Board under this title or the provision thereof in question. 3 60 Because of these statutory provisions, the regulation at issue in Mourningunlike the circular in Thorpewas legislative according to the convention. In Supreme Court litigation, Thorpe and Mourning represent the low-water mark in terms of attention to congressional delegations of power to agencies to act with the force of law. Neither the housing authority in Thorpe nor the consumer in Mourning referred to the convention, even though it would have been in their interests to do so. For its part, the Court focused solely on whether an agency's regulations fell within the general scope of its jurisdiction, as conferred by the act, ignoring entirely the question whether Congress has delegated authority to the agency to act with the force of law.
4. Gilbert (1976) and Chrysler (1979). -In the latter half of the 1970s, a reaction set in. In both General Electric Co. v. Gilbert 3 6 I and Chrysler Corp. v. Brown, 362 the Supreme Court held that particular agency pronouncements did not have the force of law, and for that reason were not valid legislative regulations. 363 In other words, the very issue ignored in Thorpe and Mourning suddenly returned to the fore. In neither case, however, did the Court point to the absence of congressionally mandated sanctions as the reason for denying these pronouncements the force of law. Thus, although the Court affirmed the importance of delegated authority to act with the force of law, it did not acknowledge the device Congress had employed for signaling an intention to delegate such authority.
Gilbert involved the claim that General Electric's employee disability plan violated Title VII of the Civil Rights Act of 1964 by excluding disability based on pregnancy. 364 One issue concerned the weight to be given to guidelines issued by the Equal Employment Opportunity Commission (EEOC) declaring that employers should apply disability benefits to pregnancy on the same terms and conditions as benefits applied to other temporary disabilities. 365 The Court noted that "Congress, in enacting Title VII, did not confer upon the EEOC authority 359 Truth in Lending Act, § 112, 82 Stat. at I5i (codified at i5 U.S.C. § i6ii). to promulgate rules or regulations pursuant to that Title. '366 The Court specifically contrasted EEOC guidelines with "regulations which Congress has declared shall have the force of law" or "regulations which under the enabling statute may themselves supply the basis for imposition of liability. '36 7 Because the EEOC guidelines did not have the force of law, the Court concluded that it would follow them only to the extent that their reasoning was persuasive. 36 8 The words the Court used to distinguish the guidelines from a legislative regulation -"Congress has declared," "force of law," "themselves supply the basis for imposition of liability"suggest that the Court was on the cusp of uncovering the logic of the convention. But the Court betrayed no awareness of the device Congress had used in signaling that agency regulations have the force of law. To be sure, such knowledge was unnecessary in Gilbert, because the rulemaking grant to the EEOC empowered the agency only to issue "suitable procedural regulations to carry out the provisions" of this subsection. 369 Thus, the Court easily reached the right resultthat Congress had not delegated to the EEOC the power to act with the force of lawwithout the need for recourse to the convention to unlock the meaning of a facially ambiguous rulemaking grant.
In Chrysler, a complicated "reverse-Freedom of Information Act" case, the Court considered whether regulations promulgated by the Department of Labor's Office of Federal Contract Compliance Programs (OFCCP), which required government contractors to furnish reports about their affirmative action programs, had the force of law. 370 The Court held that the OFCCP's regulations were not reasonably related to a grant of power given by Congress and therefore could not be treated as legislative. 371 366 Id. at 141. 367  A reverse-Freedom of Information Act (FOIA) suit is an action to enjoin the government from disclosing informationunder FOIAthat a party has submitted to the government. The Court in Chrysler held that FOIA did not authorize such suits, but discussed whether such an action might be permitted under i8 U.S.C. § 19o5, a criminal statute that prohibits agencies from disclosing information "not authorized by law." i8 U.S.C. § T905 (2000); see Chrysler, 441 U.S. at 294-95. This in turn presented the question whether the OFCCP regulations, which required disclosure of contractor reports about affirmative action compliance, were binding legislative regulations. Id. at 295.
One possible source of legal authority for the OFCCP regulations considered by the Court was 5 U.S.C. § 301, the "Housekeeping Act. '37 2 This act gives each head of an executive department general power to "prescribe regulations for the government of his department, the conduct of its employees, the distribution and performance of its business, and the custody, use, and preservation of its records, papers, and property. '37 3 The grant is facially ambiguous, because it does not specify whether the "regulations" to which it refers include legislative rules, or merely procedural and interpretive rules. The Court resolved the ambiguity by analyzing the language and history of § 301, concluding that it authorized only "rules of agency organization, procedure or practice," as opposed to legislative rules. 37 4 Application of the framework of the convention would have yielded the same conclusion with less effort, because § 301 does not contain any sanction for violating the rules promulgated under its authority. But the Court did not mention the convention.
Gilbert and Chrysler are especially revealing because then-Justice William Rehnquist wrote both opinions. In contrast to the authors of the other decisions we have considered, Justice Rehnquist was, at least at this time, highly sympathetic to the nondelegation doctrine. 3 75 This explains his endorsement of the proposition, as stated in Chrysler, that "[t]he legislative power of the United States is vested in the Congress, and the exercise of quasi-legislative authority by governmental departments and agencies must be rooted in a grant of such power by the Congress and subject to limitations which that body imposes. '376 Yet even Justice Rehnquist, with his sensitivity to the nondelegation doctrine, was unable to discern the convention. By 1979, as far as Supreme Court Justices and Supreme Court litigants were concerned, the convention had disappeared. 372 See id. at 3o8-12.

B. Why the Convention Was Ignored
After a half-century of legislation in which Congress employed the convention to signal whether rulemaking grants authorized legislative rules or housekeeping rules, why did the convention never surface in Supreme Court litigation? The most obvious explanationand in the end the only satisfactory oneis that no party in any of these cases ever described the convention in its briefs. 3 77 The Court is not omniscient and largely depends on the parties for information about the relevant law that applies to the many controversies it must resolve. If the parties fail to mention a source of law, the Court cannot be expected to discover it on its own.
It is possible, of course, that one or more Justices might have known about the convention from their own legal experiences. Many Justices of this period had congressional and executive experience. Justices Black, Vinson, Burton, and Minton had served in the Senate; Justices Reed, Jackson, and Clark had served in high level positions in the Department of Justice, and Justice Douglas had been Chairman of the SEC. But the convention functioned at a level of detail below that with which these individuals ordinarily would have been involved in their legislative and administrative capacities. Thus, it is not surprising that they would not have known about the convention, or if they once had known about it, that they would forget about it years later when presented with briefs in which it was not mentioned.
Justice Felix Frankfurter is the most difficult to account for, since he taught administrative law at Harvard before being appointed to the Court and was familiar with Eaton and Grimaud. 37 8 But Frankfurter was more an intellectual mentor to other administrative law scholars 377 If any party in these cases had intended to describe the convention, one would expect at a minimum to find some reference to United States v. Grimaud, 220 U.S. 5c6 (igii). In only one of the eight cases, however, did the parties make any reference to Grimaud in their briefs, and they did not do so to make the point that congressional sanctions for rule violations mark a grant of rulemaking as legislative. Compare Brief for Intervenor-Appellee Teamsters Union at 5o, Am. That the Justices were unfamiliar with the convention, however, only pushes the inquiry back one step: why did the parties fail to discuss it in their briefs? With respect to the first four cases in our survey, it is clear why the industry lawyers did not allude to the convention: the convention represented a losing argument for them. Their clients were anxious to secure a ruling holding particular regulations invalid, but in each case the convention indicated that the relevant agencies (the FCC, ICC, and Federal Power Commission) did have the authority to issue regulations having the force of law. Consequently, the industry lawyers concentrated on other arguments.
It is less clear why the government lawyers representing the agencies in these cases did not discuss the convention. Of course, their primary task was to defend the regulations against the attacks leveled by the industry lawyers, and achieving this goal did not require any reliance on the convention. But the government would have gained at least a psychological advantage by demonstrating that Congress clearly intended to delegate legislative rulemaking authority to the agencies in question. Perhaps the omission stems in part from the fact that the agencies are represented in the Supreme Court by lawyers in the Solicitor General's office. These lawyers specialize in appellate advocacy and derive most of their knowledge from studying Supreme Court opinions. Generally, they do not have a deep understanding of the law that surrounds the agencies they represent. Because the convention had never appeared in an appellate opinion and existed only in the "common law" of agency authority as understood by lawyers in Congress and the agencies, the Solicitor General's lawyers perhaps were not aware of it.
With respect to the lawyers who came of age after World War II, there is a more basic explanation for the lack of awareness of the convention. Administrative law scholarship in this period was dominated by scholars who were veterans of government service during the New Deal era. As a general rule, these scholars downplayed the importance of limits on agency powers, instead emphasizing the pragmatic case for

2002]
broad delegation of powers to administrative agencies and the importance of adapting administrative procedure to a judicial model of decisionmaking. 38 2 It is thus no coincidence that the early writings of these scholars contain little discussion of rulemaking (in contrast to adjudication) and no discussion of how to determine whether an agency has been granted legislative rulemaking authority. 38 3 Nor is it a coincidence that the instructional materials produced by members of this group after the New Deal devoted little coverage to rulemaking, and none at all to the interpretation of facially ambiguous delegations of rulemaking authority. 38 4 For this generation of scholars, the question of how agencies obtain the power to issue rules that have the force and effect of statutes was an awkward and potentially destabilizing one for the federal administrative state. Since no prominent appellate opinion dealt with the issue, these scholars evidently felt no obligation to explore the question in their instructional materials. 385 As best we can tell, no lawyer who attended law school after World War II would have learned about the convention from even the most assiduous study of the classroom materials in use during this period.
In any event, by the time we get to Thorpe and Mourning, considerations of legal relevance and litigational advantage cannot explain the failure of the parties to alert the Court to the convention. The only plausible explanation is collective ignorance. In Thorpe, the convention provided a winning argument for the housing authority, but the authority did not mention it. It is possible that the lawyers for the note 79, at 928-34. Although he cited some of the materials we discuss in this Article that recognize the convention, he did not explicitly describe the convention in any of his work; indeed, he did not seriously consider the possibility that the scope of an agency's power could be determined from examining the terms of its organic legislation.
385 In administrative law, as elsewhere in the law school curriculum, instructional materials tend to emphasize appellate opinions to the exclusion of virtually everything else. See CHASE, supra note 379, at 117-24. [Vol. 116:467 housing authority made a deliberate decision to concentrate solely on other arguments such as retroactivity. We learn from the Court's opinion that the authority had already begun complying with the circular, 386 and thus perhaps its only interest in continuing the litigation was to avoid an award of attorneys' fees. Still, it is unlikely that the authority's lawyers would have waived a winning argument had they been aware of its existence.
Mourning is even more revealing. The petitioner in Mourningwho supported the Federal Reserve Board's four-installment ruledevoted several pages of her brief to the issue of the Board's authority to promulgate legislative rules under the Truth in Lending Act's facially ambiguous rulemaking grant. 38 7 The petitioner set forth numerous arguments explaining why the grant authorized legislative rules. This presentation was thoroughly researched, but it included no mention of Congress's convention. 3 8 Given the petitioner's detailed analysis of the Board's general rulemaking grant, it is fair to assume that she would have relied upon the convention if she had been aware of it. Petitioner's failure to raise the convention thus suggests that the convention had disappeared from the consciousness of Supreme Court litigators by the time Mourning was argued and decided in I973.
The final two cases confirm that knowledge of the convention was lost to the circle of lawyers who argue cases before the Supreme Court. In Gilbert, petitioner General Electric and supporting amici had every incentive to show that the EEOC's guidelines did not have the force of law. Granted, the express language of the general rulemaking grant made it possible to do so without resort to the convention. Still, had the lawyers known of the convention, one would expect some mention 386  388 The petitioner argued that Supreme Court precedent established that general rulemaking grants arguably confer legislative powers even where several specific rulemaking grants are also included in the Act. Id. at 22-23. Relying on Thorpe, the petitioner claimed that "[t]he fact that the Act contains substantive provisions and authorizes the Board to issue interpretive regulations does not necessarily negate the grant of legislative rule making power to carry out the Act's more broadly phrased statement of purpose." Id. at 22. The petitioner acknowledged that the general rulemaking grant given to the Secretary of the Treasury under the Internal Revenue Code (IRC) was understood to authorize only interpretive rules. Id. at 23-24. However, the petitioner distinguished the Board's general rulemaking grant under the Truth in Lending Act from the grant included in the IRC on the ground that the Truth in Lending grant authorizes rules "to carry out the purposes" of the Act. Id. at 23 (emphasis added) (quoting I5 U.S.C. § 1604 (2000)) (internal quotation marks omitted). This language, the petitioner argued, was broader than the statutory language used in the IRC, which merely gives the power to issue rules "for the enforcement of this title." Id. (quoting 26 U.S.C. § 7805(a) (2ooo)) (internal quotation marks omitted). Finally, the petitioner supported her contention that the Board's general rulemaking grant authorizes legislative rules by pointing to section 16o2(g) of the Truth in Lending Act, which gives Board regulations the same legal force as statutory provisions. Id. at 24 (citing i5 U.S.C. § 602(g)).

2002]
of it. Yet the briefs' arguments only focus on the language of the rulemaking grant, never once referring to the convention. 38 9 Similarly, in Chrysler, the petitioner was anxious to demonstrate that 5 U.S.C. § 301 authorized only procedural and not legislative rules. Although the history of the rulemaking grant suggests it was, as its colloquial name indicates, only a housekeeping grant, one would expect some reference to the convention to confirm the point. The briefs, however, contained an exposition of the history of the particular grant but did not mention the convention. 3 9 0 It is tempting to attribute this collective silence about the convention to changes in legal beliefs and values, such as a decline in belief in the Austinian conception of law as commands backed by sanctions 39 ' or a diminished allegiance to the nondelegation doctrine. But it is not clear that any such explanation holds water. The decision in Thorpe is as "Austinian" in its understanding of law as anything found in the opinions of the late nineteenth century. Once the Court deemed the HUD circular to have the force of law, it regarded the circular as having a direct preemptive effect on the state law of eviction and on the leases between housing authorities and their tenants. And although the early decisions in our series authored by New Dealers like Justices Frankfurter, Reed, and Douglas may conceivably reflect a latent hostility toward the nondelegation doctrine, this cannot be said of the last two decisions in our series, authored by then-Justice Rehnquist. Gilbert and Chrysler, in fact, reflect the first steps in a short-lived effort by Justice Rehnquist to revive the nondelegation doctrine. 392 Even Judge J. Skelly Wright, author of the leading appellate opinion that conferred legislative rulemaking power in the absence of conventional sanctions, purported to be an enthusiast of the nondelegation doctrine, going so far as to publish an article urging its re- vival. 393 Thus, there would appear to be no correlation between support for the nondelegation doctrine and receptivity to, or even acknowledgement of, the convention.
In the end, the most parsimonious explanation for the fact that the Supreme Court ignored the convention is simple ignorance. The proximate cause of the Court's ignorance was the ignorance of the lawyers who appeared before it. The lawyers remained ignorant because, due to accidents of timing, no case presenting an opportunity for the Court to recognize the convention arose when the convention was most familiar to lawyers serving in Congress and the agencies. Suppose that the FTC or the NLRB had sought to engage in legislative rulemaking in the 1940s, and these actions had been challenged in litigation reaching the Supreme Court or even the courts of appeals. The upshot almost certainly would have been an appellate opinion explaining why the facially ambiguous rulemaking grants that Congress gave to these agencies did not include the power to make rules with the force of law. But this did not happen. And it did not happen, ironically enough, primarily because it was so clear to the staff attorneys at the FTC and the NLRB that their agencies lacked statutory authority to make such rules. Because lawyers tend to learn the law from reading appellate opinions, 394 and the convention never appeared in the Supreme Court opinions about rulemaking that accumulated in the years after the New Deal, the possibility of recovering knowledge of the convention slipped away.

V. THE CONVENTION ERASED
The Supreme Court's failure to attend to the distinction between grants of legislative and housekeeping rulemakingand the American practice of distilling knowledge of public law from Supreme Court opinionsset the stage for the de facto erasure of the convention by the courts of appeals. As agencies and commentators began to perceive the advantages of rulemaking over adjudication, they began to rely on the language in several of the Court's opinions (including National Broadcasting, American Trucking, Storer Broadcasting, Texaco, Thorpe, and Mourning) as support for the proposition that agencies previously thought not to have legislative rulemaking authority in fact had enjoyed such power from their inception. The understanding that eventually emerged was exactly the opposite of the Queen and Cres- 393 See J. Skelly Wright, Beyond Discretionary Justice, 81 YALE L.J. 575, 582 (1972) (arguing that Congress should not be able "to vote itself out of business. There must be some limit on the extent to which Congress can transfer its own powers to other bodies without guidance as to how these powers should be exercised."). 394 See supra note 385.
cent Case express-statement canon: ambiguous grants of rulemaking authority are presumed to confer legislative rulemaking power unless Congress expressly indicates that the grant is limited to procedural or interpretive rules.

A. Commentators Advocate Expanded Use of Rulemaking
Beginning in the late 1950s and early i96os, administrative law commentators began to call for increased use of rulemaking. The case for rulemaking rested on two somewhat disparate sets of concernsfairness and efficacy. Some commentators emphasized the greater fairness of rules relative to adjudication. Making policy through adjudication can lead to inconsistent outcomes and frustrates expectations when policy changes retroactively. 395 Making policy through rulemaking is much more likely to result in standards that apply prospectively, providing clear notice of the law's requirements to all concerned. 396 Other commentators championed rulemaking as a means of making agencies more effective regulators. 3 97 For example, some argued that rulemaking allows agencies to promulgate blanket prohibitions against certain industrywide practices. 398 These somewhat conflicting aspirationsconstraining agency discretion and expanding agency powerconvinced numerous scholars and lawyers to join the call for more rulemaking. Warren E. Baker, the general counsel of the FCC, wrote a pioneering article in 1957 that embraced both themes. He argued that "rule-making is a sounder way of proceeding than the case-by-case method or general declarations of policy." 399 He pointed to Storer Broadcasting as "a perfect example" of how agencies could set policy through rulemaking. 40 0 He contended 395 See, e.g., KENNETH CULP DAVIS, DISCRETIONARY JUSTICE 66 (Illini Books ed. 1971) (1969) (arguing that the retroactive feature of adjudication may "be sufficiently unfair that good administrators ought to try to avoid it"). 396 See generally Shapiro, supra note 296, at 929-42 ("[A] rule declared in a regulation is more likely than a rule declared in adjudication to be limited in application to determining the legal status of future conduct . . . ."); see also DAVIS, supra note 395, at 66 ("[P]rospective rules often should be preferred to retroactive law-making through adjudication."). 398 See Shapiro, supra note 296, at 935 (noting the FTC's view that "when a practice is widespread in an industry, a rulemaking proceeding operates evenhandedly to bar that practice on the part of all, while an order directed only to one permits his competitors to gain an unfair advantage").
399 Two prominent public figures, James Landis and Judge Henry Friendly, soon added their voices to the call for increased agency rulemaking. In i960, Landis wrote his influential Report on Regulatory Agencies to the President-Elect, in which he contended, "A prime criticism of the regulatory agencies is their failure to develop broad policies in the areas subject to their jurisdiction. '4 0 3 Landis suggested that, rather than rely solely on adjudication, agencies should use rulemaking to set forward-looking policy. 40 4 Similarly, in a book published in 1962, Judge Friendly explained that a major source of dissatisfaction with the federal administrative agencies stemmed from their "failure to develop standards sufficiently definite to permit decisions to be fairly predictable and the reasons for them to be understood. '405 Although Judge Friendly devoted most of his book to the contention that agencies should develop more detailed adjudicative standards, he also urged that the case-by-case adjudication method "should be supplemented by much greater use of two other devicespolicy statements and rulemaking. 407 DAVIS, supra note 395, at 65. 408 Id. at 66.
tors toward earlier and more diligent use of the rule-making power, '40 9 and asserted that "[t]he typical failure in our system that is correctible is not legislative delegation of broad discretionary power with vague standards; it is the procrastination of administrators in resorting to the rule-making power to replace vagueness with clarity. '4 10 Davis acknowledged that some agencies may lack statutory power to issue legislative rules with the force of law. 411 He argued, however, that this was not a justification for failure to issue rules, because even agencies lacking legislative rulemaking powers could issue interpretive rules. 4 12 Davis approvingly pointed to the FTC as an example of an agency that lacked legislative rulemaking powers but that had nevertheless engaged in valuable interpretive rulemaking. 4  1965 that, given the momentum these decisions provided, battles over agency powers to issue legislative rules based solely on general rulemaking grants had "probably just begun.9 4 1 8

B. The FTC, FDA, and NLRB Exercise Legislative Rulemaking Powers
Professor Fuchs's prediction soon proved correct. As the advocacy of rulemaking mounted in the i960s, a number of agencies decided to test the limits of their rulemaking powers. The payoff was substantial. When the dust settled, three major agenciesthe FTC, FDA, and NLRBhad acquired legislative rulemaking powers that Congress had not originally granted to them.
i. The FTC.
-Consistent with the understanding that the FTCA did not confer any legislative rulemaking powers on the FTC, the agency made no attempt to promulgate rules with the force of law during the early decades of its existence. 419 Later amendments to the Act, if anything, only underscored the limited nature of the agency's rulemaking powers. 420 Congress expressly conferred legislative rulemaking authority on the FTC under several amendatory statutes, including the Wool Products Labeling Act 42 1 and the Fur Products Labeling Act. 42 2 If Congress had granted general legislative rulemaking author- 418 Id. at 8o6. 419 See Shapiro, supra note 296, at 925 (noting that in the first half of this century, the FTC relied on adjudicatory proceedings rather than regulations to control unfair practices). 420 A House Report issued in 1957 highlighted the limited nature of the FTC's rulemaking powers: The Federal Trade Commission, through its trade practice conference procedure, has promulgated rules with respect to the labeling and advertising of rayon, acetate, linen, and silk which move in interstate commerce. These rules, as such, do not have the force and effect of law but are rather advisory interpretations as to what may constitute unfair methods of competition and unfair and deceptive acts or practices in commerce, under the Federal Trade Commission Act. These rules, insofar as they go, have worked reasonably well, but their chief handicap is the limited jurisdiction of the Commission under its organic statute. H.R. REP. No. 85-986, at 2 n.I (957) (emphasis added). 421 Ch. 871, § 6(a), 54 Stat. 1128, ii31 (1940) (codified as amended at 15 U.S.C. § 68d(a) (2000)) (authorizing the FTC to make rules and regulations "as may be necessary and proper for administration and enforcement"). Congress gave the rules promulgated under this general rulemaking grant legislative effect through various statutory provisions. For example, section 3 provided that the introduction into commerce of any wool product that was misbranded within the meaning of the Act or of the rules or regulations promulgated by the FTC was unlawful, id. § 3, 54 Stat. at 1i29 (codified at i5 U.S.C. § 68h), and section io then subjected any person who willfully violated section 3 to criminal penalties, id. § 10, 54 Stat. at 1133 (codified at 15 U.S.C. § 6 9 f(b)). 422 Ch. 298, § 8(b), 65 Stat. 175, 18o (1950 (codified at i5 U.S.C. § 6 9 f(b)) (authorizing the FTC to promulgate rules and regulations "as may be necessary and proper for purposes of administration and enforcement of this Act"). Various statutory provisions gave legislative effect to the rules and regulations promulgated under this general grant. For example, section 3(a) made it unlawful to introduce into commerce any fur product that is misbranded within the meaning of the Act or of "the rules and regulations prescribed under Section 8(b)," id. § 3(a), 65 Stat. at 176 (codified at ity to the FTC in 1914 when it created the Commission, these subsequent grants of legislative rulemaking powers would have been superfluous.

2 3
The history of the Flammable Fabrics Act of 1953424 confirms most strikingly that Congress did not grant the FTC legislative rulemaking powers under the original FTCA. The Flammable Fabrics Act included a general rulemaking grant that authorized the FTC to "prescribe such rules and regulations as may be necessary and proper for purposes of administration and enforcement of this Act. '42 5 Congress wrote this grant in language similar to the general grant included in section 6(g) of the FTCA. Both stood alone, lacking any statutory sanctions to put teeth into the regulations. In 1967, however, Congress amended the Flammable Fabrics Act 426 by adding the following language to the rulemaking provision: "The violation of such rules and regulations shall be unlawful and shall be an unfair method of competition ... under the Federal Trade Commission Act. '4 27 Congress also gave the FTC the authority to enjoin any violations of the rules and regulations promulgated under the Act. 428 These amendments confirm that the rulemaking grant in the 1953 Act (which was identical to the grant in section 6(g)) did not confer legislative rulemaking powers. More important, they also show that, as late as 1967, when Congress wanted to signal that a particular rulemaking grant conferred legislative rulemaking authority, it added a provision imposing legal sanctions for rule violations.

(a) The FTC's Regulatory Practices from 1914 to 1962. -Because
Congress chose not to grant the FTC legislative rulemaking powers, for the first half-century of its existence the agency relied mainly on three methods of making policy. Case-by-case adjudication served as the primary enforcement tool. 42 9 Under the FTCA, the FTC possessed the power to issue a complaint, set a hearing, and decide whether to issue a cease and desist order requiring the respondent to stop any activities found to be unfair commercial practices or unfair methods of competition. 430 However, the shortcomings of adjudication as a method of setting policy soon became apparent. Adjudication failed to achieve widespread compliance, reaching only individual violators and specific acts rather than industrywide practices. 431 Soon after Congress established the FTC, the agency developed Trade Practice Conferences to help overcome the shortcomings of adjudication 432 and to identify practices that violated the laws administered by the FTC. 43 3 The procedure worked as follows: a specific industry, a consumer group, or the FTC would initiate a conference about a certain industry; proposed rules would be developed at the conference; the FTC would hold a public hearing on the proposed rules and then issue a Trade Practice Conference Rule (TPCR). 434 TPCRs were not regarded as legally binding. 435 Rather, the FTC relied on voluntary compliance by industry members. 436 If the FTC initiated suit based on a violation, the complaint would "charge [the 429 See Wegman, supra note 397, at 730 (noting that "[t]he cease and desist procedure is the usual route by which the FTC takes action against unfair commercial practices").
430 See Forte, supra note 416, at 887-88. 431 See Wegman, supra note 397, at 740 (concluding that adjudication "was not adequate to deal with certain industry-wide practices, since repeated adjudications involving identical facts not only were extremely cumbersome, but resulted in considerable unfairness to all parties concerned" party with] violation of the statutory provision on which the rules are premised, and ... not.., violation of the trade practice rule. '437 In 1955, the FTC added a third form of policymaking to its tool bag: Guides. 43 8 Like TPCRs, Guides did not have the force of law. 4 39 The Guides program consisted essentially of interpretive rules spelling out the FTC's understanding of the requirements of the FTCA. 440 For example, the FTC issued Cigarette Advertising Guides in I955 to summarize its view that it was unlawful to represent that cigarette smoking presents no health risks or to make references to that effect. 44 1 (b) The FTC Turns to Legislative Rulemaking. -After nearly a half-century of setting policy through adjudication, TPCRs, and Guides, in 1962 the FTC decided to institute a new procedure -Trade Regulation Rules (TRRs). Although the legal impact of TRRs was initially unclear, 4 42 it soon became apparent that the FTC did, in fact, intend to treat TRRs as legislative regulations.
The FTC cautiously tested political support for its new legislative rulemaking experiment by issuing several TRRs that dealt with fairly trivial matters. 443 None of these early rules stirred much contro-437 FTC Trade Practice Conference Rules, 20 Fed. Reg. 3o6I (May 6, I955). 438 See generally Auerbach, supra note 423, at  (explaining that the Guides program began in 1955 and that by October 31, 1962, the FTC had issued nine Guides). 439 See Note, supra note 436, at 394. 440 See Weston, supra note 181, at 567; see also Note, supra note 436, at 397 (noting that the "primary compliance tool in connection with the guides program has been education"). 441 See Note, supra note 436, at 396. 442 See Burrus and Teter, supra note 14, at 1119-20 (noting that the question still remained in 1966 concerning "whether the rules possess[ed] the force and effect of law or simply presentled] a prima facie case which may be rebutted"); see also Auerbach, supra note 423, at 455, 457 (arguing that "[a] trade regulation rule will not have the force and effect of law, in the sense that a violator of the rule will become subject to a penalty for its violation" but also noting that Commissioner MacIntyre contemplated the use of "substantive rule making power"). Auerbach's article was an outgrowth of his report to the Committee on Internal Organization and Procedure of the Administrative Conference. He argued in favor of a legislative amendment to give the FTC the power to issue legislative rules and regulations for the violation of which criminal penalties would attach. Id. at 458. He noted that the Commission staff was at that time divided on the desirability of legislation. The Director of the FTC's Bureau of Industry Guidance opposed legislation imposing penalties, but the Bureau of Deceptive Practices supported legislation authorizing rules enforceable by penalties. Id. 443 See Note, FTC Substantive Rulemaking: An Evaluation of Past Practice and Proposed Legislation, 48 N.Y.U. L. REV. 135, 143 (1973) (stating that "[t]he FTC's initial ventures into the area of rulemaking were marked by caution, reflecting not only the problems of adjustment to a new form of regulation, but also, perhaps, concern over its power to proceed and sensitivity to possible political ramifications. The Commission began with minor regulation of advertising and labeling in small businesses .... "). For example, in 1963 the agency promulgated a rule requiring that sleeping bags be marked with the size of the finished product rather than the dimensions of the material used in making the bags. 16 C.F.R. § 400 (1964). Similarly, a 1965 TRR declared that the practice of describing household electric sewing machines as "automatic," "fully automatic," or The actual effect of the rules was unclear because the FTC did not immediately attempt to bring any enforcement actions based on them. 445 However, in 1964, the FTC adopted its first major, controversial TRR, which dealt with unfair and deceptive practices in advertising and labeling cigarettes. 446 Perhaps anticipating opposition, the FTC included a lengthy discussion in the rule's statement of basis and purpose justifying the agency's use of legislative rulemaking. 447 The FTC cited Judge Friendly and other critics of the administrative process who had urged agencies to engage in more rulemaking. 448 It relied on National Broadcasting, Storer Broadcasting, and Texaco to support its claim that it possessed the power to promulgate the cigarette rule. 44 9 The agency also pointed to the general rulemaking grant in section 6(g) of the FTCA as a source of authority to issue binding rules. Finally, it argued that even if section 6(g) were not in the Act, the TRRs would not be ultra vires: "It is implicit in the basic purpose and design of the Trade Commission Act as a whole, to establish an administrative agency for the prevention of unfair trade practices, that the Commission should not be confined to quasijudicial proceedings.1 4 50 The FTC was correct in anticipating an adverse reaction to the cigarette rule and a challenge to its authority to promulgate it. The tobacco industry mounted a full-scale offensive against the TRR 45 1 and persuaded Congress to enact a weak labeling bill as a substitute for the strong restrictions contained in the FTC cigarette rule. 45  After its ill-fated attempt to regulate the cigarette industry, the FTC again resorted to promulgating fairly minor, uncontroversial TRRs. 456 Late in the I960s, however, criticism of the FTC for failing to act more aggressively in enforcing the FTCA rose to a new pitch. 45 7 Especially significant in this regard was The Nader Report on the Federal Trade Commission, 458 which pointed to the cigarette-advertising rule as an example of how the FTC could act to protect consumers, "if properly directed and motivated. 4 5 9 (c) National Petroleum Refiners Association. -Prompted by these criticisms, the FTC once again aggressively tested the limits of its TRR procedure in the late I96os and early 197oS. 4 60 The agency's pursuit of legislative rulemaking, in turn, led to closer scrutiny concerning whether the FTC in fact possessed legislative authority to is- PROCESSES 120 (1985) (noting that the Nader Report was instrumental in pushing the FTC to rely more heavily on its rulemaking powers).
[Vol. I1I6:467 sue such rules. 46 1 It also resulted in court challenges. The first case, decided in 1968, challenged a TRR based in part on the theory that the FTC lacked the statutory authority to promulgate legislative rules. 462 However, the U.S. District Court for the District of Columbia ducked the issue on ripeness grounds. 463 The National Petroleum Refiners Association brought the next challenge, asserting that the FTC lacked the authority to promulgate a TRR requiring the posting of octane ratings on gasoline pumps. 464 This time the U.S. District Court for the District of Columbia addressed the issue of the FTC's rulemaking authority head on. After thoroughly canvassing the history of the FTCA and its amendments, the court held that the FTC lacked any general authority to issue rules with the force of law. 465 The district court's decision created an uproar 466  466 See Note, supra note 443, at 142 (criticizing the district court's decision and arguing that there were "serious weaknesses" in the court's analysis). 467 See WEST, supra note 46o, at 121. 468 One lawyer, who served as General Counsel to the FTC during this time, recalled that he was busy "either arguing that the Commission had the power to promulgate rules or taking alternative steps, but, of course, maintaining the consistent argument that Congress should clarify the matter once and for all and set down the manner in which the Commission could promulgate Trade Regulation Rules." Ronald M. Dietrich, Business and Trade Practices, 28 FOOD DRUG COSM. L.J. 700, 70 (I973). joined by Chief Judge David Bazelon and Judge Spottswood Robinson, is a remarkable legal document. Judge Wright commenced with the usual disclaimers, observing that the FTC "is a creation of Congress, not a creation of judges' contemporary notions of what is wise policy. The extent of its powers can be decided only by considering the powers Congress specifically granted it in the light of the statutory language and background. ' 47 0 "As always," he continued, "we must begin with the words of the statute creating the Commission and delineating its powers. '471 There followed a discussion of section 5 of the FTCA, which authorizes the FTC to adjudicate complaints and issue cease and desist orders. Judge Wright rejected the industry's claim that this section was the exclusive source of the FTC's authority to act with the force of law. 472 There was no language in section 5 indicating that it was exclusive, Judge Wright observed. 47 3 And of course there was section 6(g), which expressly gave the FTC general authority to "make rules and regulations for the purpose of carrying out" various provisions of the Act, including section 5.474 Read together, Judge Wright explained, the two provisions suggest that the FTC could issue legislative rules that would then be applied and enforced in adjudications brought pursuant to section 5.
Judge Wright then shifted gears, noting that this reading "is reinforced by the construction courts have given similar provisions in the authorizing statutes of other administrative agencies. '476 There followed a skillful exposition and selective quotation from National Broadcasting, Storer Broadcasting, Texaco, American Trucking, and Mourning, which he read as supporting his claim that the FTC should be allowed to promulgate binding legislative rules. 477 Not surprisingly, Judge Wright's opinion reflected no recognition of a central difference between the rulemaking grants given to the agencies in these cases and the FTC's general rulemaking grant: namely, that the rulemaking grants in those cases, unlike Section 6(g), were coupled with statutory provisions imposing sanctions for rule violations.
The legislative history of the FTCA, as we have seen, provides significant evidence that Congress did not intend to grant legislative rulemaking authority to the FTC. 47s  nounced this history to be "ambiguous" regarding the meaning of Section 6(g), 47 9 and then relegated the details to an appendix for the especially diligent reader to consult. 480 Such ambiguity, he said, was not enough to overcome "the plain language of Section 6(g)," which, "read in light of the broad, clearly agreed-upon concerns that motivated passage of the Trade Commission Act, confirms the framers' intent to allow exercise of the power claimed here." '481 In the end, Judge Wright adopted what amounted to a new canon: unless the legislative history reveals a clear intent to the contrary, courts should resolve any uncertainty about the scope of an agency's rulemaking authority in favor of finding a delegation of the full measure of power to the agency.
The immediate significance of Petroleum Refiners was short-lived. Congress soon mooted the issue by adopting the Federal Trade Improvement Act of 1975,482 which expressly conferred legislative rulemaking powers on the FTC, subject to certain legislative and procedural limitations. 483 But the importance of Judge Wright's opinion went far beyond its impact on the FTC's rulemaking powers. His selfconfident tone and masterful blending of Supreme Court precedents provided the roadmap for a more general erasure of the convention and invited other agencies, including the FDA, to assert generalized legislative rulemaking powers that Congress had not expressly granted. 484 2. The FDA. -The FDA's rulemaking story is both similar to and different from that of the FTC. Much like the FTCA, the FDCA 4 8 5 included a facially ambiguous rulemaking grant that did not confer legislative rulemaking authority under the convention. 4 86 Unlike the FTCA, however, the FDCA also included several specific rulemaking grants that did authorize legislative rulemaking under the convention. 48 7 The drawback to these specific rulemaking grants, from the  § 57a (2ooo)). 484 Although Petroleum Refiners is the best-known decision of the D.C. Circuit broadly proclaiming that courts should presume congressional grants of general rulemaking powers to confer legislative rulemaking authority, it was followed by other, lesser-known decisions to the same effect. See, e.g., In re Permanent Surface Mining Regulation Litig., 653 F.2d 514, 523-25 (D.C. Cir. I98I) (en banc) (relying in part on Mourning and Petroleum Refiners to conclude that the general rulemaking grant in the Surface Mining Control and Reclamation Act confers legislative rulemaking powers); Citizens to Save Spencer County v. EPA, 60o F.2d 844, 873-74 (D.C. Cir. i979) (holding that section 3oI(a)(I) of the Clean Air Act, which gives the EPA general rulemaking authority, authorizes the agency to adopt legislative rules imposing preconstruction review requirements perspective of the agency, was that they required the agency to use formal rulemaking procedures when promulgating rules. Consistent with this understanding, for thirty years after the adoption of the FDCA the agency made no effort to use section 70I(a) to promulgate legislative rules. Rather, it set policy through case-by-case adjudication, interpretive rulemaking,;and legislative rulemaking under its specific rulemaking grants. Beginning in the ig6os and 1970s, however, the FDAprimarily in an effort to free itself from the formal procedural constraints that attached to the FDCA's specific rulemaking grants under section 70i(e)began to assert that section 70i(a) authorized it to issue legislative rules using only informal notice-andcomment rulemaking procedures. 488 (a) The Debate over the FDA's Rulemaking Powers Under Section 707(a). -The FDA's "belated discovery" of general rulemaking powers in section 70I(a) stemmed largely from the entrepreneurial efforts of Peter Barton Hutt during his tenure as the FDA's chief counsel. 4 9 In a paper presented to the Food and Drug Law Institute in 1972, Hutt expounded the theory that the FDCA should be viewed as a "constitution" that gave the FDA broad authority to implement "a set of fundamental objectives. '490 Specifically, he argued that the Act gave the FDA power to do anything not excepted or withheld by the Act, 491 and he cited the general rulemaking clause in section 70I(a) to support his conclusion that the Act "provide[d] ample legal authority" for the FDA to adopt procedures for the enforcement of FDCA requirements.

2
Hutt's theory sparked a debate over the FDA's rulemaking authority 493 and drew strong criticism from industry attorneys. 4 94 For example, one attorney argued that "[i]t is for some a weird and dissonant note to hear that in the sensitive area of congressionally delegated authority, a well-motivated administrative agency can legally do what it alone deems desirable unless Congress has in advance specifically prohibited it. ''4 9 s Similarly, in a paper delivered at an FDA-sponsored 488 See I JAMES T. O'REILLY, FOOD AND DRUG ADMINISTRATION § 4:02 (i979) ("The Food and Drug Administration grew during the late ig6os and the 1970S from a law enforcement agency which brought deterrent actions against violators, into a more paper-bound generator of rules and regulations."). 489  The debate surrounding the FDA's decision to use section 701(a) to expand its rulemaking powers reached the courts in the early I97OS. 4 9 7 The first major cases to touch on the issue were four related 1973 Supreme Court decisions usually referred to as the Hynson Quartet. 498 These cases assumed without discussion that the FDA had authority to issue legislative regulations under section 70I(a). For example, in Commission. Since they have jurisdiction over labeling as well as advertising, we will simply ask the FTC to repromulgate them." Id. 509 Although all three judges agreed that the FDA had legislative rulemaking authority under section 701(a), see id. at 695-98, Judge Lumbard wrote a concurring opinion arguing that in cases in which an agency engages in legislative rulemaking under a general rulemaking authorization, the courts should interpret the phrase "arbitrary, capricious, [or] an abuse of discretion" in the APA in such a way as to ensure that especially rigorous judicial review of agency action is provided, id. at 705 (Lumbard, J., concurring) (quoting 5 U.S.C. § 7 o6(2)(A) (2000)) (alteration in original). promulgate binding regulations without complying with the procedural safeguards set forth in section 701(e).s'. 2 In reaching this conclusion, judge Mansfield relied on three factors. First, he noted that while courts had once demanded proof of a specific delegation of legislative authority to an agency purporting to issue legislative rules,. modern courts "have learned from. experience to accept a general delegation'as sufficient in certain areas of expertise." ' 13 Second, he observed that "over the last decade rule-making has been increasingly substituted for adjudication as a regulatory technique, with the support and encouragement of courts" in cases such as Petroleum Refiners. 5 1 4 Third, he concluded that any "doubts ... regarding the FDA's power under § 701(a) to promulgate binding regulations were dispelled" by the Supreme Court's decisions in the Hynson Quartet. 515 One factor that did not significantly influence the Second Circuit's holding in Nutritional Foods was the legislative history of the FDCA, which, as we have seen, provides strong evidence that Congress intended to grant legislative rulemaking authority to the FDA only pursuant to specific rulemaking grants. 5 1 6 judge Mansfield made just one brief reference to the legislative history in his opinion, observing that the court's attention had not been directed to anything in the legislative history of sections 70I(a) or (e) that militated against the court's decision.

17
After Nutritional Foods, many courts treated the issue of the FDA's legislative rulemaking authority under section 70I(a) as settled. 518 But, perhaps because of the court's failure to address the crucial legislative historyand because of the possibility of Supreme Court review in some future caseother interested observers, including both scholars and FDA officials, did not view the issue as completely re-512 See id. at 695-98. 513   In a lengthy opinion by Judge Friendly, the Second Circuit again upheld the power of the FDA to issue legislative rules pursuant to section 70I(a). Judge Friendly admitted at the outset that "[i]n the interest of historical accuracy, it should be noted that at one time it was widely understood that generalized grants of rulemaking authority conferred power only to make rules of a procedural or an interpretative nature, and not binding substantive regulations, for which a specific delegation was thought necessary. 5 2 9 However, reading the language of section 70I(a) "with the eyes of i98o," Judge Friendly concluded that this section did in fact give the FDA the power to issue both "substantive as well as procedural" regulations. 5 30 Judge Friendly based this conclusion primarily on arguments drawn from precedent. He began by reviewing numerous non-FDA cases involving facially ambiguous rulemaking grants, including National Broadcasting, American Trucking, and Petroleum Refiners. 5 31 He read these decisions to stand for the proposition that general rulemaking provisions are ordinarily understood "to endow agencies with power to issue binding rules and regulations. 5  Friendly discussed these cases at length even though the parties had cited only one, American Trucking, in their briefs. 5 33 Judge Friendly's invocation of non-FDA cases not briefed by the parties suggests that he aspired to build on Judge Wright's Petroleum Refiners opinion and develop a canon of interpretation regarding general rulemaking grants with applications beyond FDA cases. Judge Friendly also relied on the Supreme Court's decisions in Abbott Laboratories and the Hynson Quartet as well as the Second Circuit's earlier decision in Nutritional Foods, 5 34 describing these FDA cases as "formidable authority to the effect that § 70I(a) itself is a grant of power to issue binding regulations. " 535 Only after considering these precedents did Judge Friendly turn to the legislative history of the FDCA. He acknowledged that "if the 1962 Congress had made the detailed examination of the legislative history and contemporary understanding of the 1938 Act, which plaintiffs' counsel have now made at long last, it might well have concluded that § 70I(a) in fact very likely was not intended to confer power to issue binding substantive rules. 5 36 But it was not clear, he continued, whether the Congress that passed the CGMP amendments in 1962 was familiar with the 1938 history 537 In any event, the belated excavation of the legislative history, Judge Friendly asserted, was insufficient to overcome the force of stare decisis created by the Supreme Court's decisions construing ambiguous rulemaking grants to confer legislative powers; nor should that history invalidate more specific rulings such as Nutritional Foods. 5

38
Given Judge Friendly's acknowledgement that the ruling conflicted with the original understanding of section 701(a), his decision constituted an even more dramatic reversal of the Queen and Crescent Case canon than did Judge Wright's opinion in Petroleum Refiners. Judge Wright had suggested that facially ambiguous rulemaking grants should be presumed to confer legislative rulemaking power unless clear evidence to the contrary existed in the legislative history. 5 39 Judge Friendly appeared to say that facially ambiguous rulemaking grants should be construed to confer legislative rulemaking power, even in the face of clear evidence to the contrary in the legislative history. In effect, Pharmaceutical Manufacturers adopted what we have called the Petroleum Refiners canon in full-blown form as an express statement rule: all rulemaking grants are conclusively presumed to confer legislative rulemaking authority unless Congress expressly indicates in the text of the statute that the grant is limited to procedural and interpretive rules.
3. The NLRB. -Much like the FDA and the FTC, the NLRB claimed no general rulemaking powers in its early years, in keeping with the understanding that the NLRA did not grant the NLRB legislative rulemaking powers. 5 40 From its creation in 1935 through the early 1970S, the NLRB formulated policy exclusively through adjudication.
Beginning in the ig6os and continuing through the ig8os, numerous scholars urged the NLRB to turn to rulemaking to formulate its policy. 5 4 1 Most of the arguments advanced by these proponents focused on the utility of rulemaking, which assertedly yielded greater precision, certainty, uniformity, and longevity than did adjudication. 5 4 2 Only a few scholars attended to the threshold issue of whether the NLRB in fact had legislative rulemaking power. 543 In the face of the legislative history indicating that Congress had not given the NLRB such authority when creating it in 1935, these scholars pointed to an amendment to section 6(a) of the NLRA added by the Taft-Hartley Act in I947. s 44 As amended, the rulemaking grant read: "The Board shall have authority from time to time to make, amend, and rescind, in the manner prescribed by the Administrative Procedure Act, such rules and regulations as may be necessary to carry out the provisions of this 540 For a discussion of why the NLRB's general rulemaking grant is nonlegislative, see supra section III.C.2, p. 511. 541  Act. '545 These scholars argued that the addition of the reference to the APA, which prescribed the manner of adopting only legislative rules, would be meaningless unless it also referred to the adoption of legislative rules. 5 46 Thus, the 1947 amendment was said to embody Congress's understanding that section 6(a) conferred legislative rulemaking authority on the NLRB. 547 This argument fails, however, for two reasons. First, it ignores the fact that the APA does impose certain procedural requirements on the manner of adopting procedural and interpretive rules. Section 3(a)(2) of the APA requires that "the nature and requirements of all formal or informal procedures" be published in the Federal Register; section 3(a)(3) requires the same for "statements of general policy or interpretations formulated and adopted by the agency for the guidance of the public. 5 -48 Thus, the added reference to the APA would not be meaningless when applied to procedural and interpretive rules adopted under the authority of section 6(a). 5 4 9 Second, the argument ignores that an anti-labor, Republican-led Congress passed the Taft-Hartley Act in order to address "what many viewed as a tendency by the NLRB toward overzealous regulation of employer conduct through its unfair labor practice jurisdiction. '550 It is implausible that Congress would have chosen to grant the NLRB expanded rulemaking powers in 1947 given that the primary objective of the Taft-Hartley Act was to rein in the NLRB. [Vol. 11I6:467 Later, Congress joined scholars in pressuring the NLRB to resort to rulemaking. In 1977, the House adopted a bill called the Labor Reform Act, 55 1 which would have required the NLRB to promulgate rules with specified purposes, such as defining appropriate bargaining units and expediting elections. The Act did not clarify or seek to change the language of the NLRB's general grant of rulemaking powers found in section 6(a); rather, it sought to encourage the agency to use its preexisting section 6(a) rulemaking powers, which the drafters implicitly assumed to include legislative rulemaking. For example, the bill provided that the NLRB "shall, to the fullest extent practicable, exercise its authority under [its general rulemaking grant in section 6(a)] to promulgate rules declaring certain units to be appropriate for the purposes of collective bargaining. ' '5 52 Although the Senate defeated the bill after six cloture attempts, 55 3 the effort served as an important barometer of the pressure exerted at the time to force the NLRB to resort to rulemaking.

4
The judiciary represented a final source of pressure on the NLRB to resort to rulemaking. Most prominently, Judge Friendly authored numerous opinions urging the NLRB to use its rulemaking powers to issue legislative rules. 555 Perhaps the most influential judicial stimulus to use legislative rulemaking, however, came in 1969 when the Supreme Court decided NLRB v. Wyman-Gordon Co.1 56 Wyman-Gordon began as an NLRB adjudication in which the NLRB enforced its socalled Excelsior Underwear rule, which required employers to furnish unions with the names and addresses of employees eligible to vote in bargaining elections. 55 7 The rule originated in Excelsior Underwear, 558 a prior adjudication in which the NLRB had announced the rule but limited its operation to future cases. 559 The employer in Wyman-Gordon challenged the Excelsior Underwear rule on the ground that the NLRB had in effect engaged in rulemaking without complying with the procedures required by the However, six Justices strongly criticized the NLRB for not adopting its Excelsior Underwear rule by notice-and-comment rulemaking. 5 6 3 Not surprisingly, some commentators heralded Wyman-Gordon as recognizing that the NLRB possessed legislative rulemaking powers. 5 64 Immediately after Wyman-Gordon, the NLRB gave in to the pleas for rulemaking and promulgated a minor jurisdictional rule declaring that it would assert jurisdiction in proceedings involving private, nonprofit colleges and universities with gross annual revenues of one million dollars or more. 565 Two additional narrow jurisdictional rules soon followed. 5 6 6 Although each of these rules was couched in procedural terms, they carried legislative effect in that they announced how the NLRB would exercise its enforcement authority in the future.
Then the Supreme Court decided NLRB v. Bell Aerospace Co. 567 Reversing a Second Circuit decision authored by Judge Friendly that had directed the NLRB to engage in rulemaking, S68 the Court reaffirmed in strong terms the NLRB's discretion to announce new principles through adjudication. 569 The agency happily reverted to its old 563 The six votes derived from Justice Fortas's plurality opinion plus Justice Douglas's and Justice Harlan's separate dissenting opinions. See id. at 764-65; id. 775-76; id. 78o-8i. The NLRB argued that even if its Excelsior requirement did represent a rule rather than an order, this was at most a procedural rulenot a legislative oneand thus was exempt from the notice-andcomment requirements of the APA. See Brief for the National Labor Relations Board, supra note 56o, at 30-31. 564 See, e.g., Bernstein, supra note 414, at 604 ("Wyman-Gordon suggests that rule makinginstead of adjudicationmay be required in some circumstances."  § 103.1 (200)). 566 One rule provided that the NLRB would assert jurisdiction in proceedings involving symphony orchestras with gross annual revenues of not less than one million dollars, Jurisdictional Standards, Symphony Orchestras, 38 Fed. Reg. 6177 (Mar. 7, 1973)  In 1987, however, the NLRB decided once again to test the scope of its rulemaking powers. The result was the NLRB's first broad-scale legislative rule, which specified the employee bargaining units it would find appropriate in various kinds of health care facilities and hospitals. 571 In the Notice of Proposed Rulemaking, the NLRB explained that its authority to promulgate the rule derived from the general rulemaking grant in section 6 of the NLRA. 57 2 The agency also noted that for years members of Congress, the courts, and scholars had urged it to engage in rulemaking. 573 The American Hospital Association challenged the validity of the NLRB's rule on three grounds: that section 9(b) of the NLRA requires the NLRB to make bargaining unit determinations "in each case" and therefore precludes the board from using rules to define bargaining units; that the rule violated a congressional admonition to the NLRB to avoid the undue proliferation of bargaining units in the health care industry; and that the rule was arbitrary and capricious. 5 7 4 The Association did not raise the more fundamental question whether the NLRB possessed legislative rulemaking authority in the first place. Rather, the parties, the Seventh Circuit, and the Supreme Court all assumed that Congress had given the NLRB the power to promulgate legislative rules. 5 75 Writing for the Seventh Circuit, Judge Posner remarked that "there is broad although not unanimous agreement in the legal community, which we and other courts have remarked approvingly, that the exercise of the Board's dormant substantive rulemaking power is long overdue." 57 6 Similarly, Justice Stevens, writing for the Supreme Court, noted in one perfunctory sentence that the general rulemaking grant given to the NLRB "was unquestionably sufficient to authorize the rule at issue in this case unless limited by some other provision in the Act," which limitation the Court did not find. The willingness of the parties in American Hospital to accept that the NLRB had been delegated legislative rulemaking powers most likely stemmed from two sources. First, the pathbreaking opinions of Judges Wright and Friendly that had treated ambiguous rulemaking grants as presumptively authorizing legislative rules had by then been on the books for a decade or more. 5 78 Second, although Wyman-Gordon did not expressly consider the NLRB's rulemaking powers under section 6(a), the opinions in that case, as well as the Court's treatment of the rulemaking versus adjudication issue in Bell Aerospace, implicitly suggested that the NLRB possessed legislative rulemaking powers. Thus, by the time the Court decided American Hospital in i991, counsel for the Association no doubt concluded it was not worth the effort to challenge the NLRB's exercise of legislative rulemaking powers. As a consequence, the courts, guided as usual by the submissions of the parties, apparently did not perceive any issue of authority either.

VI. TAX EXCEPTIONALISM
Although the successful assumption of legislative rulemaking powers by the FTC, FDA, and NLRB would appear to reflect a complete triumph of the view that facially ambiguous rulemaking grants confer legislative power, at least one important vestige of the earlier understanding remains. The tax world continues to adhere to the notion that the facially ambiguous general rulemaking grant in section 7805(a) of the Internal Revenue Code (IRC) confers only interpretive, not legislative, rulemaking authority. [Vol. 11I6:467

A. Section 78o5(a) of the Internal Revenue Code
Section 7805(a) of the IRC grants the Secretary of the Treasury the power to "prescribe all needful rules and regulations for the enforcement" of the tax laws. 5s 0 This general grant of rulemaking authority appears to have originated in section oo5 of the Revenue Act of 1917,81 which gave the Commissioner of Revenue the power to promulgate, with the approval of the Secretary of the Treasury, "all needful rules and regulations for the enforcement of the provisions of this Act. 58 2 Congress did not attach any statutory sanctions to violations of the regulations promulgated under this general grant. In contrast, sections iooi and 1002 of the 1917 Act included specific grants of authority for the Commissioner to regulate returns, 58 3 and section 1004 attached penalties to the failure to make any returns required by regulation S 4 Under the convention described in Part III of this Article, therefore, the general grant conferred only interpretive power upon the Commissioner, while the specific grants conferred legislative power.
The Revenue Act of 1918585 and the Revenue Act of 1921586 included the same combination of general and specific rulemaking grants, using similar language. 58 7 When Congress sought to include the usual mixture of rulemaking grants in the Revenue Act of 1924, debate and confusion arose over whether a general grant of rulemaking authority would give the Commissioner the power to promulgate binding regulations. In particular, Representative Deal expressed concern that the general rulemaking grant would give the Commissioner too much power s During debates on the House floor, Representative Deal argued: Stat. at 309. Like the Revenue Act of 1917, the 1918 Act did not attach any statutory sanctions to violations of the rules and regulations promulgated under the general rulemaking grant. Yet penalties were provided in the 1918 Act for violations of rules and regulations promulgated under specific rulemaking grants. Like the 1917 and 1918 Revenue Acts, the 1921 Act attached statutory sanctions only to rules and regulations promulgated under certain specific rulemaking grants. and every morning before breakfast if they feel like doing it. The tax-answer may be that many of the key players understood that the general grant conferred only interpretive authority. 5 96 When the IRC was enacted in 1939 and subsequently amended in 1954, Congress continued to attach statutory sanctions solely to rules and regulations promulgated under specific rulemaking grants. Under the convention's framework, this absence of sanctions leads to the conclusion that the rules and regulations promulgated under section 7805(a)'s general rulemaking grant are interpretive, which is in fact the common understanding today among tax lawyers.

B. The Supreme Court Confirms That Section 78o5(a) Is Interpretive
Notwithstanding the gradual repudiation of the convention outside the tax world, the revisionism of Judges Wright and Friendly is unlikely to spread to the IRC. One reason for this resistance is that, in contrast to what happened in the regulatory context, the Supreme Court has endorsed the outcome dictated by the convention in decisions explicating the meaning of section 7805(a) of the IRC. For example, in Rowan Cos. v. United States, 597 the Court noted that "the Commissioner interpreted Congress' definition [of the word 'wages'] only under his general authority to 'prescribe all needful rules.' 26 U.S.C. § 7805(a). '598 Because the regulation was merely interpretive, the Court held that it deserved "less deference than a regulation issued under a specific grant of authority. 5 99 Similarly, in United States v. Vogel Fertilizer Co.,600 the Court considered a regulation, issued by the Commissioner under section 7805(a), which interpreted the statutory term "brother-sister controlled group. '6 0 1 The Court again observed that because the Commissioner had issued the regulation under his general rulemaking grant, the interpretation was entitled to "less deference than a regulation issued under a specific grant of authority to define a statutory term or prescribe a method of executing a statutory provision.

2
Vogel Fertilizer and Rowan confirm that in the tax worldin contrast to other administrative realmsa facially ambiguous general rulemaking grant authorizes only interpretive, not legislative, rules. 60 3 Because the Supreme Court has endorsed that view, it will probably remain secure, unless and until Congress amends the Code.

C. Why the Tax World Thinks Section 7805(a) Is Interpretive
When we trace the history of the understanding that section 7805(a) authorizes only interpretive rules, we uncover an anomaly that seems, at first, to cut against the convention. This anomaly made its first appearance in a series of articles written in the 1940s by several eminent tax scholars. 60 4 They include Erwin Griswold, who later became Dean of Harvard Law School and Solicitor General of the United States, and Stanley Surrey, who has been called "the most influential tax theorist of his generation. '60 5 These authors did not focus on the absence of sanctions for violations of section 7805(a) as the reason for construing the grant to authorize only interpretive rules. 60 6 Instead, they argued that the relevant distinction was between general and specific grants of rulemaking authority. General grants were said to authorize only interpretive rules, whereas specific grants authorized legislative rules. 60 7 Surrey buttressed this contention with two arguments: first, the specific rulemaking grants in the IRC would be redundant if the general grant were construed to give the agency general legislative rulemaking authority; 60 8 and second, the delegation of a general grant . 604 See Surrey, supra note 78, at 558 (arguing that revenue acts do not "support a delegation of legislative power"); see also Alvord, supra note 78, at 256-61 (stating that for the great majority of Treasury regulations, the Commissioner has no legislative authority); Erwin N. Griswold, A Summary of the Regulations Problem, 54 HARV. L. REV. 398, 400-01 (194) (explaining that interpretive regulations differ from legislative regulations, but only as a matter of degree); cf. Coverdale, supra note 579, at 69 (reporting that Surrey noted as early as 1940 that the general rulemaking grant given to the Treasury was not sufficient to delegate legislative powers). 605 Coverdale, supra note 579, at 69. 606 The distinction between interpretive and legislative rules caught Surrey's and other tax scholars' attention due to debate over the "reenactment doctrine," which was well entrenched by the 1940s. This doctrine provided that when an agency interprets a statutory provision, the administrative construction is automatically given the force of law when Congress later reenacts the legislative language that the administrative rule construes. In arguing against the reenactment doctrine, Surrey pointed to the fact that most Treasury regulations were merely interpretive because they were promulgated under the IRC's general rulemaking grant. See Surrey, supra note 78, at 557-58.
607 Id. For a discussion of the distinction between general and specific grants, as compared with the conventional "sanctions" test, see supra section ILA, pp. 482-83. 608 See Surrey, supra note 78, at 558 (contending that rules issued under specific rulemaking grants should be taken to possess different attributes than rules issued under the general rulemak-[Vol. I116:467 of legislative rulemaking authority would raise serious constitutional questions under the nondelegation doctrine, because the general grant lacks an intelligible principle to guide agencies in making rules. 60 9 As we have seen, the distinction between general and specific grants of authority was not the basis for distinguishing between legislative and nonlegislative rulemaking grants under the drafting convention. Indeed, many New Deal-era statutes, including the Natural Gas Act, the Securities Acts, the Communications Act, and the Taylor Grazing Act, 6 10 contain very broad rulemaking grants that nevertheless have always been understood to authorize legislative rules. We have also seen that the reasons that tax scholars cite in support of the general/specific distinctionthe structural argument about redundancy and the invocation of the intelligible principle canonremain open to question.

11
We can only conjecture about how Surrey and other tax scholars came to adopt an explanation for the interpretive nature of section 7805(a) that is inconsistent with the logic of the convention. Part of the explanation may be that no appellate opinion or other written source has described the convention. Surrey had worked in the Treasury Department and was undoubtedly familiar with the received understanding that section 7805(a) authorizes only interpretive rules. He sought an explanation for this assumption that fit the facts of the tax world, and came up with the general/specific distinction. Since there was no judicial opinion or other written source that contradicted this explanation, and Surrey's arguments were at least superficially plausible, his explanation became the conventional wisdom of the tax world. 6 12 ing grants because "otherwise the careful particularization of Congress in these other sections would be without meaning"); see also Alvord, supra note 78, at 257 (arguing that "some difference was intended by Congress," or it would not have granted the Treasury both specific and general rulemaking power).

VII. WHAT THE CONVENTION MEANS TODAY
Our purpose in looking backward to uncover the historical understanding of facially ambiguous rulemaking grants is, ultimately, to advance understanding of administrative authority in ways that are relevant to ongoing controversies. As recounted in Part I, there are two contexts today in which it is critical to determine whether Congress has delegated power to an agency to make rules with the force of law. One of these contexts is when courts consider whether agencies must comply with the procedural requirements for promulgating legislative rules under the APA. The other is when courts decide whether to apply Chevron's strong deference doctrine. This latter issue, in particular, is a matter of continuing debate within the Supreme Court. Eight Justices agreed in United States v. Mead Corp. 613 that Chevron rests on congressional intent, and that the requisite intent is to delegate authority to an agency to make rules that have the force of law. But the Court was much more divided the previous Term in  . 1265 (2002), Justice Breyer's opinion for the Court recited factors similar to those mentioned in Skidmore in support of the proposition "that Chevron provides the appropriate legal lens through which to view the legality of the Agency interpretation here at issue." Id. at 1272; see also id. at I269-72. This echoing of Skidmore suggests that Justice Breyer may continue to adhere to his view, set forth in Christensen, 529 U.S. at 596-97 (Breyer, J., dissenting), that Chevron is simply a special case of Skidmore deference. But see Barnhart, 122 S. Ct. at 1274 (Scalia, J., concurring in part and concurring in the judgment) (noting that one of the factor relied upon by Justice Breyerwhether the agency interpretation is "longstanding"is an "anachronism" after Chevron, and criticizing the majority for failing to explain why certain agency interpretations were sufficiently authoritative under Mead to qualify for Chevron deference). In Edelman v. Lynchburg College, 122 S. Ct. 1145, 1150 (2002), the Court declined to resolve whether an interpretation set forth in an EEOC procedural rule warranted Chevron deference. The reason that Justice Souter's majority opinion gave for avoiding the question was that the agency interpretation was "the position we would adopt even if there were no formal rule and we were interpreting the statute from scratch." Id. This language seems disturbingly inconsistent with the delegated lawmaking rationale of Chevron, which directs courts to uphold reasonable agency interpretations even if they conflict with the interpretation the court would adopt on its own. Chevron U.S.A. Inc. v. Nat'l Res. Def. Council, 467 U.S. 837, 843 n.iI (1984). As other Justices suggested in concurring opinions, Edelman also calls into question the Court's apparent reaffirmation of the delegated lawmaking rationale in Mead. See Edelman, 122 S. Ct. at 1153 (Thomas, J., concurring) (urging acceptance of the agency interpretation because "the EEOC possessed the authority to promulgate this procedural regulation, and . .. the regulation is reasonable, not proscribed by the statute, and issued in conformity with the APA"); id. at 1154-55 (O'Connor, J., joined by Scalia, J., concurring in the judgment) (arguing that the EEOC regulation was entitled to Chevron deference because the agency had been delegated authority to adopt procedural regulations, and the regulation had been re-promulgated using formal notice-and-comment procedures).

A. The Role of the Convention in Statute-Specific Determinations of
Legislative Intent Mead appears to contemplate that courts will seek to determine on a statute-by-statute basis whether Congress has delegated power to agencies to make rules with the force of law. 616 The convention we have described would seem at a minimum to be an important datum in undertaking this kind of ad hoc inquiry into congressional intent. The fact that during the formative years of the administrative state Congress followed a convention in signaling whether it was giving agencies the authority to make rules having the force of law is important contextual information in understanding the meaning of facially ambiguous rulemaking grants adopted both during the New Deal and afterwards.
With respect to rulemaking grants that predate the i96os, we would go further. We are aware of no Supreme Court decision, prior to Thorpe in 1968, that is inconsistent with the convention; nor are we aware of any evidence that Congress had stopped following the convention. To be sure, there is little evidence that the typical member of Congress was aware of the convention, and some memberssuch as Representative Dealclearly were not. Thus, we are not suggesting that Congress in any sense collectively intended that courts follow the convention. But there is substantial circumstantial evidence that staff attorneys, in drafting individual statutes, used the convention as a signaling mechanism to agencies and other informed observers. Consequently, even if the convention was unknown to most legislators themselves, it may nonetheless provide a reliable guide to the type of rulemaking authority that Congress had in fact agreed was appropriate under each particular statute. Even if courts consider other evidence of congressional intent, such as the structure of a statute or canons of interpretation, the convention is a sufficiently reliable indicator of congressional intent that it should ordinarily outweigh these other factors. Still, we are aware of no evidence suggesting that Congress ever repudiated the convention or acquiesced in the Wright-Friendly view. Even today, when Congress includes a rulemaking grant in a statute, the legislation frequently specifies whether or not sanctions attach to violations of an agency's rules and regulations. 6 19 In a statute-bystatute assessment of the legislative intent of statutes enacted after 198o, however, the convention undoubtedly becomes a more problematic guide.

B. The Case for a Canon
An alternative to Mead's statute-by-statute approach to ascertaining legislative intent would be the adoption of a canon that would serve as an aid to identifying congressional intent. 620 We assume that 617 See Nat'l Ass'n of Pharm. Mfrs. v. FDA, 637 F.2d 877. 889 (2d Cir. ig8i). 618 As confirmation of this, consider the views published by the Administrative Conference of the United States on the subject in the early 198os: Though only a minority of statutes contain ... forthright instructions to make rules, most regulatory agencies have no difficulty in pointing to statutory language authorizing them to make "such rules and regulations as may be necessary to carry out the provisions of the Act." Such an authorization clearly enables an agency to promulgate procedural, organizational or other "housekeeping" rules, and it also clearly enables an agency to issue non-binding guidelines or interpretations of its statutory authority. These powers are now quite accepted and may even be deemed within an agency's "inherent" authority. . 620 Barron and Kagan, supra note 53, also recognize the need for a canon or "background rule" that would serve as a presumptive guide to congressional intent. The canon they suggest is based on whether "the official Congress named in the relevant delegation" has "personally assumed responsibility for the decision prior to issuance." Id. at 235. One of the problems with this proposal is that most agencies have legal authority to subdelegate functions to subordinate officials. Most prominently, in the Reorganization Act of 1949, Congress delegated standing authority to the [Vol. I1I6:467 such a canon would serve only as a presumption and would be rebuttable based on other evidence from the text, structure, and history of the statute in question. But structuring the inquiry under such a canon would offer several advantages over an ad hoc consideration of legislative intent. 62 1 The first and most important reason for adopting a general canon would be to facilitate communication between Congress and the courts. 622 Christensen and Mead make clear that Congress has the ultimate authority to turn Chevron deference on and off. Congress can do this either by delegating power to an agency to act with the force of law with respect to a particular issue (Chevron on) or by not delegating such power to the agency (Chevron off). But Congress can exercise this authority only if it knows what to say in a statute to delegate such power. 623 If a canon provides a general presumption, then Congress will generally know what it must do to confer primary interpretive authority on an agency. Conversely, if acting with the force of law is identified under an ad hoc inquiry, as Mead suggests, then Congress will not know if it has succeeded in designating an agency (or the courts) as the primary interpreter until after the issue is litigated. Adopting a canon thus could help to preserve and protect the role of Congress, which the Court has identified as the very foundation of the Chevron doctrine.
A possible objection to this argument is that Congress could, in the future, specify whether it has given agencies the authority to make interpretations that have the force of law. Congress could simply include or not includelanguage in a given statute that agency rules will have "the force and effect of law." This suggestion, however, overlooks the problem of what happens under thousands of existing rulemaking grants. Congress has neither the time nor the institutional capacity to President to issue reorganization plans that provide for "the authorization of any officer to delegate any of his functions." Reorganization Act of 1949, ch. 226, § 3, 63 Stat. 203, 203. Thus, Congress's naming of an official as the decision maker in an agency's organic act is only a default rule, making it a problematic guide to congressional intent if the default has been changed. Also, the Barron and Kagan proposal only provides an answer to the question whether Chevron deference applies to any particular agency interpretation; it does not address the broader issue of how to interpret facially ambiguous rulemaking grants.  review and amend existing legislation to ensure that it has allocated interpretive authority properly. If the Court were to adopt a general canon about which kinds of delegations should sustain Chevron deference, Congress and affected agencies and interest groups would be better able to predict how courts will interpret existing delegations. In this way, the relevant parties would be in a better position to know which statutes should be targeted for revision.
A second reason to adopt a general canon is to facilitate agency planning. Agencies regard Chevron deference as a good thing, and they understandably want courts to accept their legal interpretations so that they can fulfill their statutory obligations as they see fit. Yet agencies must make a significant investment in administrative procedures to obtain the Chevron payoff. In the vocabulary of Christensen and Mead, agencies must take whatever procedural steps are necessary to demonstrate that they intend to exercise authority to make rules with the force of law. In the context of rulemaking, this ordinarily means committing considerable time and resources to notice-andcomment procedures. 624 The ability to know, in advance, whether such an investment will pay off would be tremendously helpful to agencies.
A third reason to adopt such a canon is to facilitate control of the lower federal courts by the Supreme Court. A canon would function like a presumptive rule, and rules are generally better than broad standards for exercising control over subordinate actors in a hierarchy. 62 5 In the federal judicial system, the problem of control is exacerbated by the fact that the Supreme Court can review only a limited number of cases each year. 626 If the Court had the capacity to review every court of appeals decision, it could define the scope of the Chevron doctrine by engaging in a statute-by-statue analysis. But given its limited resources, the Court might better control the behavior of lower courts by adopting a presumptive rule.
Finally, a general canon would allow lawyers to provide more accurate advice to clients about the probable outcome of litigation than  627 Similarly, when such litigation occurs, arguments about the proper application of a rule will generally require less effort to develop than arguments about the proper application of a standard. Adopting a general canon would therefore reduce legal costs.

C. Choosing a Canon
The discussion in Parts II and III suggests three candidates for a canon that could be used to interpret facially ambiguous rulemaking grants. First, the canon implicit in the Queen and Crescent Case would provide that rulemaking grants are presumed to authorize only procedural and interpretive rules unless Congress expressly states otherwise. Second, the convention we describe in Part III would provide that rulemaking grants are presumed to authorize the imposition of rules with the force of law either when Congress states so expressly or when it provides for sanctions for rule violations. Third, the Petroleum Refiners canonthe approach endorsed by Judges Wright and Friendlywould provide that rulemaking grants are always presumed to authorize legislative rules unless Congress clearly intends (or, in the Friendly version, expressly states) otherwise. We assess these three candidates against three criteria: how well they respect the constitutional principle that only Congress can delegate authority to agencies to act with the force of law; how well they track probable congressional intent concerning the powers that agencies can exercise; and the disruption to reliance interests entailed by the transition from the current regime to the proposed canon.
i. The Queen and Crescent Case Canon. -The Queen and Crescent Case express-statement canon has one great virtue: it would robustly enforce the principle that agencies have no inherent authority to act with the force of law unless Congress delegates that authority to them. Under this canon, agencies could engage in legislative rulemaking if and only if Congress has expressly authorized them to make rules with "the force and effect of law" or the equivalent. With the benefit of hindsight, the Court might have been well advised to generalize such an express-statement rule from the Queen and Crescent Case. The nondelegation principle would have been secured; Congress, agencies, and the courts would have had a very clear signaling device for determining when authority to make legislative rules had been conferred; and no major reliance interests would have been frus- 627 See FREDERICK SCHAUER, PLAYING BY THE RULES 145-49 (iggi); cf Louis Kaplow, Rules Versus Standards: An Economic Analysis, 42 DUKE L.J. 557, 579-84 (1992) (arguing that rules are more efficient than standards if fewer resources can be expended in determining the content of the law ex ante rather than ex post). 2002] trated, since nearly all major delegations of such authority occurred after the Queen and Crescent Case.
But the Court did not generalize such an express-statement rule from the Queen and Crescent Case. Instead, it held in Grimaud that Congress could make it a criminal offense to violate an administrative regulation, as long as Congress itself legislated the sanction. 62 8 This decision presupposed that Congress was not required to confer legislative rulemaking authority through an express statement, but it could do so by clear implication. Relying on this ruling, Congress proceeded over the next half-century to enact thousands of rulemaking grants in which it conferred authority to make legislative rules by clear implication; that is, by prescribing sanctions for violations of "rules and regulations." To adopt an express-statement rule now would massively frustrate congressional intent, as manifested in these many enactments. It would be paradoxical to vindicate the principle of legislative supremacy with a rule that would undermine congressional intent on such a vast scale. Effectively, an express-statement rule would be the undoing of the administrative state. These costs are surely too great for the Queen and Crescent Case express-statement canon to tempt the Supreme Court today.
2. The Convention as Canon.
-The convention we describe in Part III has a much stronger claim to being a viable canon for ascertaining presumptive delegatory intent. The convention would not vindicate the nondelegation principle with the same vigor as the Queen and Crescent Case canon, but Congress would still be required to confer the necessary authority by clear implication, most commonly by legislating sanctions for rule violations. In other words, some affirmative legislative act would be requiredeither the express conferral of legislative power or the express adoption of sanctions for rule violationsbefore the transfer of authority would be deemed to have occurred.
The convention would also harmonize much better with actual congressional intent. As we have shown, Congress followed the convention quite faithfully until about i96o, while it was adopting most of the regulatory statutes that serve as the foundation for the modern administrative state. We are less sure about what has happened since then, and in particular what has happened since ig8i, when Judge Friendly's opinion in Pharmaceutical Manufacturers endorsed the Petroleum Refiners canon. 62 9 However, we are unaware of any evidence showing that Congress has relied on the Petroleum Refiners canon in drafting modern regulatory statutes.
The big question is how disruptive it would be to existing reliance interests to adopt the convention as a canon at this late date. For the past twenty years, any administrative agency that obtained a legal opinion on the subject would have been advised that a facially ambiguous rulemaking grant would likely be held to confer legislative rulemaking authority. It is unclear how many legislative regulations have been promulgated on this understanding, under statutes that the convention would deem not to authorize such regulations. We cannot even begin to provide a complete accounting, given the thousands of federal statutes containing tens of thousands of rulemaking grants, large and small. A systematic review of these rulemaking grants would pose a Herculean task. All we can offer are some observations drawn from the more historically significant examples of facially ambiguous rulemaking grants and from the controversies we have surveyed in this Article.
First, it appears that the vast majority of administrative agencies that need legislative rulemaking authority do in fact have such authority under the convention. Among the agencies we have reviewed, this includes the FCC, the SEC, the Federal Power Commission (now the Federal Energy Regulatory Commission), the Social Security Administration, the EPA (as to most functions), the IRS (through multiple individual grants), and the Federal Reserve Board (under the Truth in Lending Act). Other agencies that did not originally have legislative rulemaking authority have acquired it through statutory amendments; these include the FTC, the ICC (now the Surface Transportation Board), and the Department of Labor (under the Longshoremen's and Harbor Workers' Compensation Act).
Second, those agencies that lack legislative rulemaking authority under the convention have often been reluctant to engage in legislative rulemaking even when told that they could do so. This describes the NLRB, which, notwithstanding repeated prodding to engage in legislative rulemaking, has continually reverted to a system that relies exclusively on adjudication. It also describes the IRS, insofar as the general rulemaking grant in section 7805(a) is concerned.
Beyond these two categories of agencies are those agencies that lack general legislative rulemaking power under the convention but that have relied on decisions by the courts of appeals authorizing them to use nonlegislative rulemaking grants as sources of legislative rulemaking. The FDA is the most prominent agency in this category and is likely the agency that would be most severely affected by the adoption of the convention as a canon. Yet even here the convention would not necessarily result in the invalidation of all existing FDA regulations issued under the authority of section 70i(a). We assume that some of these rules could have been adopted under specific rulemaking grants covered by section 7oi(e), which requires the use of formal rulemaking procedures. Any challenge to these rules for failure to comply with the procedural requirements of section 70r(e) would likely now be barred as untimely. Thus, with respect to these rules, the primary consequence of adopting the convention would be to deny the FDA Chevron deference for interpretations embodied in these rules. 630 Courts, however, would likely regard other FDA rules, such as the Current Good Manufacturing Practice rules, as interpretive rather than legislative. 631 These rules would not only lose Chevron deference, but they would also be open to challenge in individual enforcement proceedings.
Other agencies undoubtedly also lack general legislative rulemaking authority, although they may benefit from a greater number of specific grants of authority than does the FDA. The Clean Water Act (CWA), for example, grants general rulemaking authority to the EPA in section 50i(a). 6 3 2 The CWA nowhere prescribes any sanctions for violations of rules issued under the authority of this grant. As a result, under the convention this provision is not a source of legislative rulemaking authority. However, the CWA also contains numerous specific rulemaking grants permitting the EPA to establish a variety of particular practices and standards. 633 Violations of these regulations are punishable by a variety of civil and criminal penalties and are enforced by other sanctions such as permit denials; thus the regulations authorize legislative rules under the convention. Still, these specific rulemaking grants are not exhaustive, and occasionally circumstances arise in which the EPA's authority to promulgate legislative rules under one of these specific provisions is unclear. 634 Lower courts have occasionally invoked section 50i(a) as a source of legislative rulemaking authority in such cases, 635 but this avenue would be foreclosed under the convention.
630 This is because the rules would violate the second step in the Mead test: whether the rules were adopted pursuant to a grant of authority to act with the force of law. See United States v. Mead Corp., 533 U.S. 218, 227 (2ooi). The rules could have been adopted pursuant to such a grant, but in fact were adopted pursuant to section 7oI(a), which authorizes only interpretive rules. Whatever the merits of adopting the convention as a presumptive guide to congressional intent, no significant precedential barrier prevents the Supreme Court from adopting such a canon. Nearly all of the Court's decisions that recognized specific rules to be legislative are consistent with the convention. The principal exceptions -Thorpe, the Hynson Quartet, and American Hospitalare cases in which the parties did not brief the question whether the agency had authority to act with the force of law. These decisions cannot, therefore, be characterized as holdings binding on the Court as a matter of stare decisis. 642 Perhaps more importantly, none of the reasoning in the Court's cases is inconsistent with the convention. The statements in National Broadcasting, American Trucking, Storer Broadcasting, Texaco, and Mourning regarding the importance of liberally construing grants of rulemaking authority are all unexceptional, because the rulemaking grants in those cases did in fact authorize legislative rules under the convention. And the holdings in Petroleum Refiners and Pharmaceutical Manufacturersno matter how important to the formation of general legal attitudes since I980 -are decisions of inferior courts not binding on the Supreme Court.
3. The Petroleum Refiners Canon. -The third possible canon presents the mirror image of the virtues and vices of the Queen and Crescent Case as a canon. The Petroleum Refiners canon does the least to further the nondelegation principle that agencies may act with the force of law only pursuant to a specific delegation from Congress. This canon requires no unambiguous affirmative step to effectuate a delegation of legislative rulemaking authority. Courts would construe ambiguous references to rulemaking as transferring legislative authority, placing the burden on Congress to achieve a different outcome. Congress would still be able to restrict the agencies' ability to make rules with the force of law, but only by denying the agency all rulemaking authority or by clearly limiting rulemaking grants to procedural and interpretive rules. Such a strong presumption in favor of delegation seems like an inadequate level of protection for a principle that serves as a cornerstone of our system of separation of powers.
Use of the Petroleum Refiners canon would also do considerable violence to actual congressional intent. Not only would the FTC, the FDA, and the NLRB receive legislative rulemaking authority under grants not intended to confer such powers, but so would the IRS and the Labor Department under statutes like the Longshoremen's and Harbor Workers' Protection Act.
The affirmative case for the Petroleum Refiners canon is that it would not disturb any reliance interests grounded in existing legisla- 642 See cases cited supra note 354. [Vol. 116:467 tive rules, since the canon would legitimate all known legislative rules. In particular, we would not have to worry about the fate of the FDA's Current Good Manufacturing Practice rules and other rules adopted pursuant to section 701(a) and intended by the agency to be legislative rules.
Whether the plight of the FDA and other possible reliance issues are sufficiently troubling to warrant embracing the Petroleum Refiners canon in light of its other shortcomings is a matter of opinion. Our judgment is that the advantages of the convention in terms of respecting nondelegation values and achieving fidelity to congressional intent are sufficiently great, and the systemic costs of adopting the convention as a canon sufficiently confined, that the convention represents the best candidate for a general canon of presumptive delegatory intent.

D. The Chevron Paradox
There is one final wrinkle that may prove to be an insuperable barrier to resurrecting the convention as a guide to the meaning of facially ambiguous rulemaking grants.
The Court has, not surprisingly, treated Chevron as the lodestar situation in which Chevron deference is warranted. 643 However, it turns out that the rule at issue in Chevron did not have the force of law under the convention, and hence under Mead should not have been afforded Chevron deference. This outcome is likely to give the Court significant pause before endorsing the convention as a presumptive guide to congressional intent.
Chevron involved an EPA interpretation of the term "stationary source" as used in the 1977 Clean Air Act Amendments, applicable to states out of compliance with national ambient air quality standards. 644 The EPA adopted this interpretation in a rulemaking proceeding conducted under the authority of the general rulemaking grant in the Clean Air Act, which was facially ambiguous regarding whether it authorized rules with the force of law. 645 The agency used noticeand-comment procedures in adopting the rule. 646  Approval and Promulgation of Implementation Plans, 46 Fed. Reg. 50,766 (Oct. 14, I98I). The fact that the agency employed notice-and-comment procedures is not, in itself, a sufficient basis to conclude that the rules were legislative. Mead indicates that "[i]t is fair to assume generally that Congress contemplates administrative action with the effect of law when it provides for a relatively formal administrative procedure" such as notice and comment. Mead, 533 U.S. at 23o. But Congress did not direct the EPA to use notice-and-comment procedures under its general rulemaking grant. The decision to do so was the EPA's alone, and unilateral action by an agency ment of basis and purposes did not clearly indicate whether the agency regarded the rule as legislative or interpretive. 647 The general rulemaking grant upon which the EPA relied in issuing the stationary-source rule was enacted as part of the 1963 amendments to the Clean Air Act. 648 The original Clean Air Act, adopted in 1955, was little more than a grant-in-aid program designed to stimulate cooperative federalism. 649 In 1963, as part of a general revision of the Act, Congress added a rulemaking grant authorizing the Secretary of HEW (later the Administrator of the EPA) to "prescribe such regulations as are necessary to carry out his functions under this Act. '650 Congress did not prescribe statutory sanctions for violations of the regulations issued under this grant. Under the convention, the absence of sanctions meant that Congress did not intend to allow the regulations issued under the Act to bind persons outside of the agency with the force of law.
The legislative history of the 1963 amendments also lacks any clear expression of congressional understanding that regulations issued under the general rulemaking grant would be legally binding. During debates on the 1963 amendments, a proposal was advanced in the House to add the word "procedural" to the general rulemaking provision, so that the amendments would only authorize the power to "prescribe such procedural regulations. '65 1 Representative Taft, who offered the amendment, explained that he did not believe it was the intention of the House to authorize regulations that would have the force of law, and he argued that it should be made clear that the general rulemaking provision was intended to refer only to procedural regulations. 652 The amendment passed the House; however, the Conference Committee did not agree to insert the word "procedural" into the grant. Although the Conference Report stated that the word "procedural" was eliminated "as being too restrictive upon the authority cannot establish the required delegation of power from Congress. For further discussion of this aspect of Mead, see Merrill, supra note 52, at 814-15.
647 Requirements for Preparation, Adoption and Submittal of Implementation Plans and Approval and Promulgation of Implementation Plans, 46 Fed. Reg. 50,766, 50,770 (Oct. 14, 1981). Some aspects of the EPA's discussion suggest that the rule was interpretive. For example, the agency said that it would not require states to follow the definition in their implementation plans, but only that it would not disapprove plans that followed the definition. Id. at 50,769. Other aspects suggest that the agency intended the rule to be legislative. For instance, the EPA construed the rule as being subject to pre-enforcement review in the D.C. Circuit, id. at 50,770, which is generally a characteristic of legislative rules, see supra note 34. which the Secretary needs to carry out the act, '6S3 it did not explain whether the Conference Committee understood the general grant to authorize legislative rulemaking. Representative Taft raised concerns on the House floor about the deletion of the word "procedural" from the Conference Committee bill. 6 -4 However, other House members assured him that the general rulemaking provision included in the Act was a limited one. Representative Roberts explained that despite the elimination of the word "procedural," he felt the Conference bill sufficiently addressed Representative Taft's concerns because "the power of the Secretary is very adequately tied down in this bill as to what he can do and as to what kind of information he must act upon, and the extent of the data upon which he may proceed." 6 5 Representative Rogers similarly assured Representative Taft that the rulemaking grant was narrow, stating that there was "no rulemaking power for abatement" included in the bill. 65 6 In the end, when Congress passed the bill without Representative Taft's proposed qualification on the general rulemaking grant, the House understood the rulemaking grant to be fairly limited. But it was not clear just how limited the rulemaking grant really was. 657 Given that the legislative history is at best ambiguous, the general rulemaking grant would not be viewed as legislative under a canon adopting the presumption that only rulemaking grants providing sanctions for violations reflect a congressional delegation of authority to make rules with the force of law.
In 1990, Congress thoroughly overhauled the Clean Air Act and amended the statute to include sanctions for violations of rules issued under the general rulemaking grant. 65 8 Post-I990, therefore, the Act's general rulemaking grant clearly qualifies as a source of legislative rulemaking authority under the convention. 6 59 But the EPA adopted the rule at issue in Chevron under the 1977 version of the Actwell before Congress amended the general rulemaking grant to give the EPA general authority to issue legislative rules implementing the Clean Air Act. This situation presents the ultimate paradox: Christensen and Mead hold that Chevron's strong doctrine of deference applies only where agencies have been delegated authority to make rules with the force of law. Yet the agency that Chevron held to be entitled to this strong deference did not, according to the convention, have authority to make rules having the force of law until many years after Chevron was decidedat least not with respect to the particular rule considered in Chevron.
In principle, this paradox should not really matter. The important propositions for which Chevron standsthat Congress often impliedly intends that an agency rather than the courts be the primary interpreter of gaps and ambiguities in a statute, and that in such cases, courts are to accept reasonable agency interpretations of the statuteremain unaffected even if Chevron incorrectly determined that Congress had implied such a delegation in the particular provision at issue. Only much later, in Mead, did the Court specify that the key question is whether Congress has delegated to the agency the authority to make rules with the force of law. Chevron did not fully anticipate this development, and it contains no holding concerning whether Congress had given the EPA authority to make rules with the force of law. Indeed, the Court in Chevron did not even cite the rulemaking grant under which the EPA issued its interpretation. Still, whether the Court would ever be willing to endorse the convention as the key to ascertaining the meaning of facially ambiguous rulemaking grants, when the convention generates this kind of internal dissonance in the Court's case law, remains very much to be seen. CONCLUSION Our story is about legislative supremacy and the signals Congress uses in its exercise. In the abstract, all agree that agencies have no inherent authority to act with the force of law. 660 Agencies can issue edicts that have the effect of statutes only if Congress delegates to them the authority to do so. But what happens if Congress speaks ambiguously in delegating authority to an agency, referring to "rules and regulations" without specifying whether this means rules and nificant because the general rulemaking grant authorizes "regulations," not "rules." If Congress had intended regulations issued under the general rulemaking grant to be enforced via the added sanctions in § 7413, one would think that it would have applied the sanctions to "regulations" as well as "rules." 660 See authorities cited supra notes 95 and 97. [Vol. 116:467 regulations with the force of law, or merely procedural and interpretive rules and regulations? For a long time, Congress used one set of signals to indicate the meaning of such facially ambiguous grants. Most members of Congress were probably blissfully unaware of these signals. But the signals functioned well enough that the attorneys drafting the legislation and advising agency officials about the scope of their authority could do their jobs.
The signals were not, however, made sufficiently explicit that appellate lawyers or courts had easy access to them. Because of the vagaries of litigation, no Supreme Court decision enshrined the signals in an authoritative opinion. When times changed, other actorsagency critics, a new generation of agency lawyers, and judges and their law clerksinterpreted the signals differently. Because there was no controlling legal authority to stop them, these actors in effect engineered a major transfer of delegated power beyond anything that Congress contemplated. While courts continue to incant the principle that agencies have no inherent authority to act with the force of law, in practice courts have enabled every agency with a general grant of rulemaking authority to decide in its discretion whether to act with the force of law.
Another debate has recently emerged concerning which institution in the administrative state is to exercise primary authority in the interpretation of law. The Supreme Court has decided that this question is also a matter of legislative supremacy and that Congress is entitled to signal whether an agency can exercise primary interpretive authority in a given instance. 661 In an effort to describe in general terms which signal will be examined to answer this question, the Court has said that the threshold question is whether Congress has delegated authority to the agency to make rules with the force of law. 66 2 Thus, by a quirk of fate, the Supreme Court has come around to the question it never answered during the formative years of the administrative state: what language must Congress use to indicate that an agency has been given power to make rules with the force of law?
The Court will soon have to decide whether ambiguous rulemaking grants should be interpreted in accordance with the original set of signals or in accordance with the revisionist understanding that such grants always confer authority to act with the force of law unless Congress specifically limits the grant. We have argued that it is important that the question of delegatory intent be resolved by a presumptive rule, rather than ad hoc. Regarding the choice of a presumptive rule, 661 See United States v. Mead Corp., 533 U.S. 218, 229 (2001). 662 See id. at 231-32.