Natural Resources Defense Council, Inc. v. U.S. Environmental Protection Agency

822 F.2d 104, 261 U.S. App. D.C. 372

Opinion

STARR, Circuit Judge:

The objective of the Clean Water Act is to “restore and maintain the chemical, physical, and biological integrity of the Nation’s waters.” 33 U.S.C. § 1251 (1982). Under the Act, the discharge of any pollutant into the navigable waters of the United States is unlawful. Id. § 1311(a). This basic rule admits of a critical exception— the discharge of pollutants is permitted if the source obtains and complies with a permit that limits the amounts and kinds of pollutants which can lawfully be discharged. Thus, the cornerstone of the Clean Water Act’s pollution control scheme is the National Pollution Discharge Elimination System (NPDES) permit program, established under the Federal Water Pollution Control Act Amendments of 1972. See 33 U.S.C. § 1342 (1982).

The original regulations implementing the NPDES program were promulgated by the Environmental Protection Agency in 1972 and 1973. Prompted by its experience during the “first round” of permitting, as well as statutory changes wrought by the Clean Water Act Amendments of 1977, EPA comprehensively revised the NPDES regulations in 1979. 44 Fed.Reg. 32,854 (June 7, 1979). Petitions for review of these regulations were filed in this and other circuits by numerous challengers, including trade associations, corporations, the Natural Resources Defense Council (NRDC), and Citizens for a Better Environment (CBE). Eventually, all petitions for review of both sets of regulations were consolidated in this court. NRDC v. EPA, 673 F.2d 392 (D.C.Cir.1980), cert. denied sub nom. Chemical Manufacturers Association v. EPA, 459 U.S. 879, 103 S.Ct. 175, 74 L.Ed.2d 143 (1982); Virginia Electric and Power Co. v. EPA, 655 F.2d 534 (4th Cir.1981).¹ At that time, the litigants identified about 55 issues under challenge. See Status Reports of Industry NPDES Petitioners, Respondents, and NRDC and CBE (filed Sept. 14, 1981).

After almost two years of settlement negotiations, EPA and the industry representatives (Industry) entered into an NPDES Settlement Agreement (Agreement) (filed June 9, 1982) covering 27 of 47 issues raised by Industry’s challenge.² In the wake of this development, our court remanded the record to the agency to permit implementation of the Agreement. Order (Aug. 6, 1982). After notice and comment, EPA promulgated final revisions to the NPDES regulations. 49 Fed.Reg. 37,-997 (Sept. 26, 1984). By virtue of the Agreement, Industry signatories were free to renew their challenges to the extent that the final regulations were not substantially the same as or altered the meaning of the terms of the Agreement. Because the final regulations reflected various changes from the Agreement, another flurry of petitions for review, both new and amended, were filed and consolidated in this court.

This opinion addresses the various challenges mounted by the Industry petitioners, on the one hand, and NRDC, on the other, to regulations which (1) define “new source”; (2) grant a ten-year grace period to new sources from more stringent technology-based standards of performance; (3) require permit applicants to identify all toxic pollutants used or manufactured in the industrial process; and (4) prohibit “bypasses,” that is, diversions of waste streams from effluent treatment facilities. Finally, we address petitioners’ challenges to an issue arising under both the Clean Water Act and the National Environmental Policy Act of 1969 (NEPA), 42 U.S.C. §§ 4321-4347 (1982) — whether EPA may ban the construction of new sources pending issuance of a permit.³ For the reasons that follow, we deny the petitions for review on each of these issues save for the construction ban.

I

The fundamental premise of the Clean Water Act is that “the discharge of any pollutant by any person shall be unlawful” except as otherwise permitted under the Act. 33 U.S.C. § 1311(a) (1982). A “discharge of a pollutant” is defined, in pertinent part, as “any addition of any pollutant to [the waters of the United States] from any point source.” Id. § 1362(12)(A). The term “pollutant” is broadly defined to include, among other things, solid waste; industrial, municipal, and agricultural waste; sewage sludge; biological or radioactive materials; wrecked or discarded equipment; heat; rock; sand; and cellar dirt. Id. § 1362(6). A “point source” is “any discernible, confined and discrete conveyance.” Id. § 1362(14).

Thus, the Act allows the discharge of pollutants from a point source only in compliance with limitations established in the Act. The Act imposes effluent limitations⁴ through two programs. The first applies water-quality based standards. Id. § 1313. It imposes on a point source effluent limitations that are based on the amounts and kinds of pollutants in the water in which the point source discharges. Id. § 1312(a). The second applies technology-based standards. It imposes on a point source effluent limitations based on how much of a reduction technology can achieve. Id. §§ 1311(b), (e), 1314(b).

To describe briefly the second approach, technology-based effluent limitations, as their name suggests, derive from standards formulated with reference to pollution control technology. See id. § 1314(b). The standard applicable in a particular case depends on the kind of pollutant — toxic, conventional, or non-conventional⁵ — and on whether the point source is a new or existing source.⁶ Under section 301, existing sources must achieve effluent limits on nonconventional pollutants that reflect the reduction in effluents that can be achieved through “the application of the best available technology economically achievable.” Id. § 1311(b)(2)(A). This is known as the “BAT” standard. New sources, on the other hand, are subject to stricter effluent limitations with respect to nonconventional pollutants. Section 30S prescribes new source performance standards (NSPS) that must reflect the “best available demonstrated control technology.” Id. § 1316(a)(1). This more stringent standard for new sources is known as the “BACT” standard.⁷ Toxic pollutants, whether from new or existing sources, are subject to effluent limitations based on application of the BAT standard. Id. § 1317(a)(2). Finally, by 1984 new and existing sources were to achieve effluent reductions in “conventional” pollutants that reflect the application of the “best conventional pollutant control technology,” known as a “BCT” standard. Id. §§ 1311(b)(2)(E), 1314(a)(4), (b)(4)(B).

In addition to technology-based standards, Section 302 of the Act provides for water-quality related effluent limitations. These limitations supplement technology-based standards and protect specific bodies of water. Id. § 1312. Whenever a technology-based effluent limitation is insufficient to make a particular body of water fit for the uses for which it is needed, the EPA is to devise a water-quality based limitation that will be sufficient to the task. Id. § 1312(a).

These national standards establish a nationwide floor for the achievement of pollution control. At the same time, the Act contemplates the delegation of pemit-writing and enforcement authority (as well as some standard-setting authority) to the several States. To date, thirty-seven States have assumed responsibility for issuing NPDES permits.

Both national and state effluent standards are enforced through the NPDES permit program. The NPDES permits thus “transform generally applicable effluent limitations and other standards ... into the obligations (including a timetable for compliance) of the individual discharger.” EPA v. California ex rel. State Water Resources Control Board, 426 U.S. 200, 205, 96 S.Ct. 2022, 2025, 48 L.Ed.2d 578 (1976). All dischargers are required to obtain a permit, which is issued after public notice and an opportunity for public hearing. 33 U.S.C. § 1342(a)(1), (b)(3).

Permits are issued only so long as the point source meets all applicable effluent limitations. Id. § 1342(a)(1). If no national standards have been promulgated for a particular category of point sources, the permit writer is authorized to use, on a case-by-case basis, “best professional judgment” to impose “such conditions as the permit writer determines are necessary to carry out the provisions of [the Clean Water Act.]” Id. Thus, compliance with a permit is generally deemed to constitute compliance with the Act's requirements. Id. § 1342(k).

The effluent limitations set forth in an NPDES permit are ordinarily drafted in terms of limitations on the amount, rate, or concentration of a specific pollutant. Permits may also include limitations on key generic parameters, such as pH or biochemical oxygen demand (BOD).⁸. Permits generally include other provisions, such as monitoring and reporting requirements, compliance schedules, and management practices. See id. §§ 1314,1318,1342(a)(2).

II

As should now be evident, the comprehensive NPDES regulations are pivotal to implementation of the Clean Water Act’s permit scheme. They include the full panoply of procedural and substantive requirements, including elaborations upon the statutory definitions, state program requirements, conditions to be incorporated into NPDES permits, and procedural requirements for revising or challenging a permit. It is to various portions of these final NPDES regulations that the present challenges are directed.

It is well established that the judiciary’s standard of review over such agency regulations is deferential. A “presumption of regularity” extends to agency rule-makings. Motor Vehicle Manufacturers Association v. State Farm Mutual Automobile Insurance Co., 463 U.S. 29, 43 n. 9, 103 S.Ct. 2856, 2866 n. 9, 77 L.Ed.2d 443 (1983). Under the standard of review laid down by Congress, such regulations may be set aside only if they are arbitrary or capricious, unsupported by the record, outside the scope of authority delegated to the agency by the operative statute, or not in accordance with law. 5 U.S.C. § 706(2)(A) (1982). As elaborated by the case law, an agency rule is arbitrary and capricious if the agency relies upon improper factors, ignores important arguments or evidence, fails to articulate a reasoned basis for the rule, or produces an explanation that is “so implausible that it could not be ascribed to a difference in view or the product of agency expertise.” State Farm, 463 U.S. at 43, 103 S.Ct. at 2867.

We are also mindful that EPA is in various respects construing the statute it is charged with enforcing. In such cases, we are to look to the statute and, if necessary, legislative history to divine the intent of Congress, which of course binds both agency and court. If Congress has not spoken to the “precise question at issue,” that is, if Congress’ intent on the issue at hand is unclear, then the agency’s interpretation must be sustained if it is reasonable in light of the language, legislative history, and policies of the statute. Chevron U.S.A. Inc. v. NRDC, 467 U.S. 837, 842-45, 104 S.Ct. 2778, 2781-83, 81 L.Ed.2d 694 (1984); see also United States v. Riverside Bayview Homes, Inc., 474 U.S. 121, 106 S.Ct. 455, 461, 88 L.Ed.2d 419 (1985); Chemical Manufacturers Association v. NRDC, 470 U.S. 116, 105 S.Ct. 1102, 1108, 84 L.Ed.2d 90 (1985); I.N.S. v. Cardoza-Fonseca, — U.S. -, 107 S.Ct. 1207, 1221-22, 94 L.Ed.2d 434 (1987).

We are, as to some aspects of this case, confronted with EPA’s modifications of positions previously advanced by the agency and incorporated in the Settlement Agreement. An agency’s change of course with respect to a particular policy does not in itself suggest a lack of reasoned decisionmaking. To the contrary, an agency volte face may evidence careful and thorough evaluation of the policy at issue. Analogously, the agency may promulgate a final rule that differs from the proposed rule, without triggering a new round of public commentary, if the changes represent a “logical outgrowth” of the prior notice and comments. American Paper Institute v. EPA, 660 F.2d 954, 959 n. 13 (4th Cir.1981); International Harvester Co. v. Ruckelshaus, 478 F.2d 615, 632 n. 51 (D.C.Cir.1973). We should also observe that a final rule which declines to embrace a proposed policy change and stays the course with the agency’s original approach can be “positive evidence of careful decisionmaking.” Small Refiner Lead Phase-Down Task Force v. EPA, 705 F.2d 506, 527 (D.C.Cir.1983). The ultimate issue is, of course, whether the agency engaged in reasoned decisionmaking. With these general principles in mind, we turn to the specific challenges addressed in this opinion.

Ill

The Natural Resources Defense Council contends that the regulatory definition of “new source,” 40 C.F.R. § 122.2 (1985),⁹ is inconsistent with the language, legislative history, and policies of the Clean Water Act. As we noted earlier, “new sources” are subject to special treatment under the Act. Section 306 of the Act provides for “new source performance standards,” which are generally more stringent than standards for existing sources. 33 U.S.C. § 1316 (1982). See 49 Fed.Reg. at 38,043. Additionally, new sources are ineligible for variances from performance standards; variances are, on the other hand, potentially available to existing sources. See E.I. Dupont deNemours & Co. v. Train, 430 U.S. 112, 138, 97 S.Ct. 965, 980, 51 L.Ed.2d 204 (1977). Finally, in contrast to existing sources, NPDES permits for new sources are considered “major Federal actions” subject to NEPA’s requirements, 33 U.S.C. § 1371(c)(1).

New source means any building, structure, facility, or installation from which there is or may be a "discharge of pollutants,” the construction of which commenced:

(a) After promulgation of standards of performance under section 306 of CWA which are applicable to such source, or

(b) After proposal of standards of performance in accordance with section 306 of CWA which are applicable to such source, but only if the standards are promulgated in accordance with section 306 within 120 days of their proposal.

“New source” is defined by section 306 as “any source, the construction of which is commenced after the publication of proposed [new source performance standards] which will be applicable to that source, if such standard is thereafter promulgated in accordance with this section. ” 33 U.S.C. § 1316(a)(2) (emphasis added). Section 306 expressly requires that final regulations be promulgated within 120 days after publication of proposed regulations. Id. § 1316(b)(1)(B). Relying upon this latter provision, EPA in the regulation now challenged by NRDC defines “new source” as a facility on which construction commences after the proposal of applicable new source performance standards “but only if the standards are promulgated in accordance with section 306 within 120 days of their proposal.” 40 C.F.R. § 122.2. See supra note 6. Unconvinced by EPA’s approach in tracking the statutory language itself, NRDC argues that the Clean Water Act mandates application of new source performance standards to all facilities commenced after publication of proposed regulations, regardless of when the regulations are issued in final form.

A

The obvious difficulty of NRDC’s interpretation is that it rests upon one portion of the statutory definition and ignores the remaining part. To read out of a statutory provision a clause setting forth a specific condition or trigger to the provision’s applicability is, we should have thought, an entirely unacceptable method of construing statutes. 2A Sutherland, Statutes and Statutory Construction §§ 46.05, .06 (C. Sands rev. 4th ed. 1984). If there were any doubt that statutes must be read as a whole, and we cannot fathom how there would be, we will reaffirm that settled method of reading statutes. See, e.g., National Soft Drink Association v. Block, 721 F.2d 1348, 1352 (D.C.Cir.1983); National Insulation Transportation Committee v. ICC, 683 F.2d 533, 537 (D.C.Cir.1982); Greyhound Corp. v. ICC, 668 F.2d 1354, 1362 (D.C.Cir.1981). See generally Frankfurter, Some Reflections on the Reading of the Statutes, 47 Colum L.Rev. 527, 533, 538 (1947).

NRDC’s interpretation, while simplifying the statute, has the admitted and untoward effect of blue penciling out express conditional language in the statute. That language clearly seems intended to have “restrictive significance.” See Reiter v. Sonotone Corp., 442 U.S. 330, 339, 99 S.Ct. 2326, 2331, 60 L.Ed.2d 931 (1979). It would be illegitimate for the judiciary, in pursuit of some overriding Congressional goal (such as eliminating water pollution), to tear asunder a specific provision which Congress saw fit to enact. It scarcely needs repeating that statutes are rarely, if ever, unidimensionally directed towards achieving or vindicating a single public policy. See Board of Governors v. Dimension Financial Corp., 474 U.S. 361, 106 S.Ct. 681, 689, 88 L.Ed.2d 691 (1986); International Brotherhood of Teamsters, et al. v. Interstate Commerce Commission, 801 F.2d 1423, 1430 (D.C.Cir.1986), pet’n for reh’g granted, 818 F.2d 87 (D.C.Cir.1987). While a broad policy goal may well be the animating force driving the legislation, achievement of actual passage of the measure invariably requires compromise and accommodation. Board of Governors, 106 S.Ct. at 689. So here in this landmark environmental statute, the pertinent definition is quite clear, encompassing facilities commenced after the publication of proposed standards if that standard is thereafter promulgated in accordance with the subsequent provision of section 306. One such requirement expressly laid down by the statute, as we have seen, is promulgation of the final regulation within 120 days of publication of the proposed regulation.

When the statute is clear in evidencing Congress’ intent, there is ordinarily no warrant for resorting to legislative history. Burlington Northern Railroad Co. v. Oklahoma Tax Commission, — U.S. -, -, 107 S.Ct. 1855, 1858-60, 95 L.Ed.2d 404 (1987). Repair to debates and reports and other precursors to law is ever fraught with difficulty. See, e.g., International Brotherhood of Teamsters, 801 F.2d at 1428 n. 4; Abourezk v. Reagan, 785 F.2d 1043, 1054 n. 11 (D.C.Cir.1986), cert. granted, — U.S. -, 107 S.Ct. 666, 93 L.Ed.2d 718 (1986); Hirschey v. FERC, 111 F.2d 1, 7-8 & n. 1 (D.C.Cir.1985) (Scalia, J., concurring). Such recourse, we believe, is much to be avoided if no ambiguity infects the language which Congress enacted and the President signed into law. Eagle-Picher Industries v. EPA, 759 F.2d 922, 929 n. 11 (D.C.Cir.1985). But even accepting the invitation to leave that upon which Congress and the President acted,¹⁰ we find in the history of this enactment little that is instructive in this specific regard.

The original Senate bill defined “new sources” solely in terms of the date of publication of either proposed or final standards. See S. 2770, 92d Cong., 1st Sess. § 306(a)(2) (1971), reprinted in A Legislative History of the Federal Water Pollution Control Act Amendments of 1972, at 1623-24 (1973) (1972 Legislative History). That provision would have vindicated NRDC’s approach. But that approach was not destined for enactment. The House bill added the triggering “in accordance with” language. H.R. 11896, 92d Cong., 1st Sess. § 306(a)(2) (1972), reprinted in 1972 Legislative History at 990. The Conference Committee thereafter adopted the House’s approach. As is so frequently the case, the reason for the Conference Committee’s embracing the added language is nowhere revealed in the mists of legislative history. But what does seem clear is that the provision reflects Congress’ accommodation of two competing policies. On the one hand, as the Third Circuit has rightly said, “Congress intended to subject as many firms as possible to the new source regulations.” Pennsylvania Department of Environmental Resources v. EPA, 618 F.2d 991, 999 (3d Cir.1980). But that policy, standing alone, would argue in favor of universal coverage, treating that which is old as new. Congress did not go so far. Thus it was that the new source-existing source distinction is premised upon the policy determination that pollution controls implemented during the period of planning and construction of new plants was “the most effective and, in the long run, the least expensive approach to pollution control,” S.Rep. No. 414, 92d Cong., 1st Sess. 58 (1971), U.S. Code Cong. & Admin.News 1972, pp. 3668, 3724, reprinted in 1972 Legislative History at 1476. It was this approach that was to be preferred to the high cost of retrofitting, H.R.Rep. No. 911, 92d Cong., 2d Sess. 110 (1972), reprinted in 1972 Legislative History at 797, which would be the necessary consequence of treating that which was old (or more precisely, existing) as new. This policy choice is reflected in the 120-day window; as EPA well put it, “construction of a source to meet new source performance standards can only proceed in a meaningful way if final standards are available.” 44 Fed.Reg. at 32,858.

The Third Circuit has aptly described the 120-day limitation as:

[serving] to limit the period during which businesses contemplating construction, put on notice by a proposal for a standard, are left in a state of uncertainty with respect to final agency action. Congress said, in effect, that it is not unreasonable, once a business has been put on notice of a proposed standard affecting it, for that business to pattern its conduct for four months to the likely application of the standard.

Pennsylvania Department of Environmental Resources, 618 F.2d at 999. That is to say, Congress must have concluded that a period of uncertainty extending beyond 120 days was unreasonable, that economic life had to go on, and that American industry should not have to languish in doubt as to when non-final regulations would eventually enjoy the force of law. It is well-settled that in such cases it is not for the judiciary to “substitute its own construction of a statutory provision for a reasonable interpretation made by the administrator of an agency.” Chevron, 467 U.S. at 844, 104 S.Ct. at 2782. We reject NRDC’s attempt to write out the clear language that Congress saw fit to enact or, assuming the existence of ambiguity even where clarity nonetheless appears to us, see supra note 10, to overturn a reasonable agency interpretation of the statute.¹¹

IV

In enacting section 306 of the Clean Water Act, Congress intended to subject new sources to the most stringent regime of technology-based standards. In exchange for this exacting approach, Congress added a provision which provided new sources a measure of protection against early, legally mandated obsolescence of their investments in pollution control equipment. Specifically, section 306(d) of the Act grants a qualifying “new source” a ten-year period of exemption from any “standard of performance” more stringent than the standard to which it was subject at the time of construction. 33 U.S.C. § 1316(d).

NRDC challenges EPA’s interpretation of “more stringent standard of performance.” The reason for NRDC’s disagreement is that EPA’s interpretation is not limited to new source performance standards (NSPS), but rather provides the ten-year grace period from other more stringent technology-based standards as well, such as BAT, BPT (best practicable control technology), and BCT.¹²

Section 306(d) provides:

Notwithstanding any other provision of this chapter, any point source the construction of which is commenced after October 18, 1972, and which is so constructed as to meet all applicable standards of performance shall not be subject to any more stringent standard of performance during a ten-year period beginning on the date of completion of such construction or during the period of depreciation or amortization of such facility whichever period ends first.

33 U.S.C. § 1316(d). Since this provision applies only to “new sources,” the “applicable standards of performance” which must be satisfied to qualify for the grace period are clearly NSPS standards, as those are the standards applicable to new sources. Moreover, the term “standard of performance” in this provision is defined in a prior subsection, section 306(a)(1), to mean NSPS standards. Id. § 1316(a)(1). If this specific definition is indeed incorporated into the grace-period provision, then the literal reading of section 306(d), championed by NRDC, would grant new sources ten years of protection only against more stringent NSPS standards.

There is an obvious drawback in embracing this fundamentalist approach to construing section 306(d). It would render utterly meaningless the grace period created by the statute. By definition, NSPS standards apply only to “new sources,” that is sources built after an NSPS standard is proposed. Id. § 1316(a)(2). (This assumes, of course, final promulgation within the requisite 120 days.) NSPS standards never apply to existing sources, as we discussed at some length in the preceding section of this opinion. Once a new source, subject to the applicable NSPS standards, becomes an existing source, subsequently promulgated NSPS standards cannot by law apply to that facility. While it stands to reason that NSPS standards will periodically be revised, the newest NSPS standards can only be applied to subsequently built sources (or, again, to those constructed within the 120-day period after the proposed regulations are published if they are finally promulgated within that period).

A

Our task, as always, is to divine Congress’ intent, mindful that to adopt NRDC’s approach would, in effect, write the grace period out of the law. It is to that inquiry that we now turn — what did Congress intend in crafting the ten-year grace period? First of all, it seems evident that Congress enacted an entire subsection of section 306 in order to create a ten-year period of protection against something. Congress was attempting to craft a provision providing new sources with predictability both as to the expected useful life of their pollution-control facilities and as to the standards those facilities would be required to meet. See H.R. Rep. 911, 92d Cong., 2d Sess. 103-04 (1972), reprinted in 1972 Legislative History at 790-91.

Describing this issue of statutory construction as a “vexing legal problem” which has concerned EPA since 1972, 44 Fed.Reg. at 32,872, the agency sought to give meaning to the grace-period provision by interpreting “standard of performance” to mean not only new source performance standards but all technology-based performance standards created by section 301, that is, BAT, BPT, and BCT. 40 C.F.R. § 122.29(d)(1). A modicum of textual basis for this interpretation is found in the opening clause of section 306(d), “Notwithstanding any other provision of this chapter,” 33 U.S.C. § 1316(d) (emphasis added). The “chapter,” it bears emphasis, refers to the entire Clean Water Act. This clause could thus be taken to suggest that the grace period is to be provided despite any inconsistencies or conflicts with other provisions of the Act, which would seemingly, if somewhat oddly, include the definition of “standard of performance.” This exclusionary language also indicates that the grace period could have worked an exception to the entire Act, once the new source meets the high statutory standard of “all applicable standards of performance.”

But EPA did not go so far as to exempt new sources from all such requirements. Specifically, the agency did not extend the protection period beyond technology-based performance standards to encompass water quality-based standards. See 40 C.F.R. § 122.29(d)(2)(i). EPA stopped short of that full measure of protection for two reasons. First, the term “standard of performance” is a term of art referring to technology-based standards.¹³ Second, an earlier, unenacted version of this provision granted a grace period from both technology-based and water quality-based standards. H.R. 11896, 92d Cong., 2d Sess. § 301(f) (1971), reprinted in 1972 Legislative History at 965-66 (exempting the facility from “any more stringent effluent limitations”).¹⁴ The narrower language of the statute as enacted suggests an intent to limit the protection provided by the grace period to technology-based standards.

B

In addition, EPA declined to construe “standard of performance” to include all technology-based standards, including those governing toxic pollutants.¹⁵ In EPA’s view, such an exception would be unfaithful to Congressional intent expressed in the 1977 Clean Water Act Amendments that the urgent problem of toxics be expeditiously addressed. See, e.g., S.Rep. No. 370, 95th Cong., 1st Sess. 52-53 (1977), U.S.Code Cong. & Admin. News 1977, p. 4326, reprinted in Legislative History of the Clean Water Act of 1977 at 685-86 (1978) (1977 Legislative History); 123 Cong. Rec. 38,959-61, 38,974, 38,991-92 (Dec. 15, 1977) (Reps. Roberts, Clausen, and Oberstar), reprinted in 1977 Legislative History at 326-29, 369, 406-07; 123 Cong.Rec. 39,171, 39,181-84, 39,219 (Dec. 15, 1977) (Sens. Muskie and Moynihan), reprinted in 1977 Legislative History at 426-27, 454-61, 549; 123 Cong. Rec. 26,695-96 (Aug. 4, 1977) (Sen. Muskie), reprinted in 1977 Legislative History at 862-65. This regulation furthers the clearly articulated legislative policy permitting imposition of a more stringent technology-based standard during the ten-year period if that standard controls a toxic pollutant, a hazardous substance, or other pollutants when control of such pollutants is a method of indirectly regulating toxic pollutants or hazardous substances. 40 C.F.R. § 122.29(d)(2)(H). EPA concluded that “[i]n light of the expressed intent of Congress to expedite the control of toxic pollutants, ... the intent of the Act would be ill-served by a policy of exempting classes of sources from compliance with toxic controls for ten years.” 44 Fed.Reg. at 32,873.

C

We are persuaded that EPA’s reading of the statute, while not the only plausible one, is reasonable. The regulation is drawn in order to give effect to all the language of the statutory provision. Nothing is blue-penciled out, which would be the unhappy consequence of NRDC’s reading. What is more, the agency’s interpretation cannot be said to depart from Congressional intent. Although the language of section 306(d) is confusing when read literally (and the legislative history offers no assistance as to the specific meaning of “standard of performance” as employed in this provision), EPA’s reading of section 306(d) reasonably balances and resolves the competing Congressional goals reflected in the provision. That is, under EPA’s approach, new source performance standards apply only to after-built sources, which is manifestly what Congress intended. In addition, effect is given to Congress’ language creating a ten-year grace period for new sources. Finally, EPA has circumscribed the scope of the grace period to avoid frustrating Congress’ subsequent directive that EPA expeditiously control toxic pollutants, an especially high priority item on Congress’ environmental agenda. The agency’s approach, we believe, reflects a reasonable accommodation of differing policy objectives. Guided by Chevron principles, we hold that the agency’s ten-year grace period regulation embodies a permissible construction of the statute.

V

The NPDES regulations require permit applications to include extensive factual information, such as effluent characteristics and treatment processes. These disclosure requirements have prompted the Industry petitioners to object to one such provision, namely that each applicant provide EPA with a list of “any toxic pollutant which the applicant currently uses or manufactures as an intermediate or final product or byproduct.” 40 C.F.R. § 122-21(g)(9). The regulation provides for a waiver or modification of this requirement if it would be “unduly burdensome” to an applicant and adequate information exists upon which to base issuance of the permit. Id.

Industry mounts several attacks on this regulation: that EPA lacks authority under the Clean Water Act to promulgate such a rule; that it is unsupported by the record; and that the promulgation of the rule violated the notice-and-comment provisions of the APA. EPA defends the information-gathering regulation as an appropriate adjunct to its toxic control scheme.

A

As we have indicated previously, toxic pollution stands high in the hierarchy of Congress’ environmental concerns. In 1972, Congress declared that elimination of toxic pollution constituted a national policy to be achieved through EPA's regulatory efforts. 33 U.S.C. § 1251(a)(3). Increased public concern about EPA’s lack of progress in controlling toxics, a problem compounded by the dearth of available data about such substances, led to litigation culminating in a consent decree which both structured and expanded the agency’s plan for regulating toxics. See Consent Decree, NRDC v. Train, 8 Env’t Rep.Cas. (BNA) 2120 (D.D.C.1976). Subsequently, the 1977 Amendments to the Clean Water Act, aimed at correcting the perceived inadequacies of the statutory scheme in eliminating toxic pollutants, adopted the regulatory approach reflected in the consent decree. H.R.Rep. No. 830, 95th Cong., 1st Sess. 85 (1977), reprinted in 1977 Legislative History at 269; 123 Cong. Rec. 39,197 (Dec. 15, 1977) (Sen. Randolph), reprinted in 1977 Legislative History at 497. The new toxic provisions were described as the “highlight” of the 1977 bill. 123 Cong. Rec. 38,959, 38,974 (Dec. 15, 1977) (Reps. Roberts and Clausen), reprinted in 1977 Legislative History at 326, 369. The indications are abundant that EPA was intended to possess broad latitude in identifying and regulating suspected toxics. S. Rep. No. 370, 95th Cong., 1st Sess. 52-53 (1977), reprinted in 1977 Legislative History at 685-86; 123 Cong. Rec. 38,959-61, 38,974, 38,991-92 (Dec. 15, 1977) (Reps. Roberts, Clausen, and Oberstar), reprinted in 1977 Legislative History at 326-29, 369, 406-07; 123 Cong. Rec. 39,171, 39,181-84, 39,219 (Dec. 15, 1977) (Sens. Muskie and Moynihan), reprinted in 1977 Legislative History at 426-27, 454-61, 549. But we need not, and do not, ground our analysis upon broad expressions of concern over toxic pollutants. It is, after all, a regulation promulgated under a statute that is before us for analysis. It is to the statute that we now turn.

B

EPA justifies this regulation as a reasonable exercise of the Administrator’s express statutory authority under section 308 to require point sources to make reports and to “provide such other information as he may reasonably require” in order to “carry out the objective of [the Act].” 33 U.S.C. § 1318(a)(A) (1982).¹⁶ Under section 402(a)(2), the EPA Administrator is authorized to impose permit conditions as to “data and information collection, reporting, and such other requirements as he deems appropriate.” Id. § 1342(a)(2). The breadth of this statutory grant of authority is obvious. In our view, the statute’s sweep is sufficient to justify broad information disclosure requirements relating to the Administrator’s duties, as long as the disclosure demands which he imposes are “reasonable.”

Industry nonetheless complains that the mandated information is not “reasonably required,” because EPA is empowered only to regulate discharges of pollutants into navigable waters. Therefore, the argument goes, EPA may not require information as to toxic substances present in a facility but is limited to requiring information as to toxic substances which are actually discharged.

We are unpersuaded that the statute can properly be read so restrictively. The Act charges the agency with the duty of policing toxic pollutants, whether deliberately discharged as an end result of the industrial process, or accidentally discharged at some point during the process. Industry’s crabbed interpretation of the Administrator’s powers would hamstring the agency by limiting its data-gathering function to information on toxic pollutants already identified by the agency. The agency could reasonably determine that it could not regulate as effectively as Congress intended without information as to what in the realm of possible toxic substances could be present in a discharge. It also seems reasonable for EPA to assume that toxic pollutants which go into or are produced by a facility might potentially be discharged, if only by accident.

But we are not left to speculate as to the agency’s reasoning on the reasonableness of this requirement. Cf. SEC v. Chenery Corp., 332 U.S. 194, 67 S.Ct. 1575, 91 L.Ed. 1995 (1947). In its preamble to the NPDES regulations, the agency set forth three examples of legitimate regulatory purposes to be served by this disclosure requirement. Each purpose, in our view, buttresses the determination that the disclosure requirement is reasonable. One is to ensure that permit conditions are adequate to control the potential discharge, for example, through imposing effluent limits for a particular toxic pollutant. 49 Fed.Reg. at 38,007. Second, information obtained from point sources would provide guidance as to the toxic substances in the effluent for which the agency should test, thus assisting EPA in performing compliance inspections. Id. Third, this information would assist EPA in its preventive function, namely in prescribing “best management practices” to prevent or control unintentional spillage, leaks, or drainage from storage of toxic substances which might otherwise result from poor management practices. See 33 U.S.C. § 1314(e).

C

In the face of EPA’s justification, Industry contends that the regulation is so burdensome that “it will be impossible for virtually all, if not all, dischargers, to comply fully.” Industry Brief at 61. Industry points to no evidence, however, in support of this broadside. A similar criticism of an earlier version of the disclosure regulation prompted EPA to make two changes in the current regulation aimed at reducing the burden on facility operators. The first modification was the elimination of a requirement that facilities predict what toxic pollutants might be used or manufactured in the next five years. See 45 Fed.Reg. 33,444 (May 19, 1980). In contrast to EPA’s earlier call for speculation about the future, the present regulation requires only a list of toxic pollutants currently used or manufactured. The second and more fundamental change was the addition of a provision permitting EPA to waive or modify the requirement if it would be “unduly burdensome” to the applicant.

Industry raises the specter of extraordinary disclosure demands so that, for example, compliance would require the detailed analysis of every trade-name cleaning compound which might occasionally be used in a facility. Industry Brief at 60-61. This sort of situation, however, would appear to be the sort of circumstance that the regulation’s waiver provision was intended to address. As EPA stated in its preamble to the regulations:

For example, applicants that use chemical solvents purchased under a brand name may be unaware of or unable to ascertain the specific toxic pollutant components that are in the solvent. To reduce these burdens, the regulation will allow the Director to modify or waive the requirement____ For example, the Director could modify the [requirement] to require only a listing of solvents by brand name.

49 Fed.Reg. at 38,007.

Unmoved by this display of administrative flexibility, Industry complains that it is too burdensome for a waiver applicant to demonstrate that the rule is “unduly burdensome.” We disagree. The “unduly burdensome” standard is facially reasonable. The regulation imposes no specific burden of proof, much less the burden of a “detailed submission” which Industry pessimistically anticipates. Indeed, we need not be detained by Industry’s conjectures in this respect, for it would be premature to entertain a challenge to the anticipated application of the waiver provision. That concern must wait for another day to bring before us a concrete setting.

Industry next suggests that the very existence of a waiver “exception” proves the unreasonableness of the rule. This position is strange indeed, for it is the exception from the rigors of the broad rule which helps tailor the rule to meet the requirements of the statutory standard of reasonableness. The waiver provision, fairly read, suggests EPA’s recognition of the breadth of its disclosure requirement and embodies a reasonable response to cabin, under appropriate circumstances, that potential sweep. This approach is entirely sensible. We decline Industry’s invitation to characterize that which in truth is a sign of reasonableness as an admission of bureaucratic excess.

Industry also argues that the rule is unreasonable because compliance might result, in some cases, in the divulgence of privileged information. This unhappy potential exists because EPA adheres to a longstanding policy of denying confidential treatment to NPDES applications. 40 C.F.R. § 122.7(b). Since 1978, EPA has interpreted the public disclosure requirement of section 402(j), 33 U.S.C. § 1342(j), as constituting an exception to section 308(b), which provides for confidentiality of trade secrets, 33 U.S.C. § 1318(b). See 49 Fed.Reg. 29,295 (July 19, 1984) (denial of petition for rulemaking to overturn regulation denying confidential treatment to trade secret information contained in NPDES applications). But this does not mean that confidential business information must automatically be divulged by virtue of these disclosure requirements. To the contrary, EPA expects the waiver provision to provide relief when such confidential information is not necessary for establishing permit limitations. 49 Fed.Reg. at 38,007-08.

But even accepting at face value Industry’s concerns, we are unwilling to brand the disclosure regulation as unreasonable. In a similar challenge to the information disclosure requirements of the Federal Insecticide, Fungicide, and Rodenticide Act (FIFRA), 7 U.S.C. § 136 et seq., another toxic substances control statute, the Supreme Court held that a manufacturer could constitutionally be required to provide confidential information where the Government had a legitimate regulatory interest in protecting the environment and public health. As the Court put it, “such restrictions are the burdens we must bear in exchange for ‘ “the advantages of living and doing business in a civilized community.” ’ ... This is particularly true in an area, such as pesticide sale and use, that has long been the source of public concern and the subject of government regulation.” Ruckelshaus v. Monsanto Co., 467 U.S. 986, 1007, 104 S.Ct. 2862, 2875, 81 L.Ed.2d 815 (1984) (citation omitted). While Industry is not, of course, mounting a constitutional challenge, the Court’s teaching in Monsanto suggests that we are not straying from the zone of reasonableness in vindicating the Administrator’s action in this respect. Indeed, upon analysis, the thrust of Industry’s objection in this regard does not go to the reporting regulation; rather, it represents an indirect quarrel with the agency’s confidentiality regulation, 40 C.F.R. § 122.7, which embodies the agency’s policy of refusing to grant trade secret protection to NPDES permit applications. That regulation, promulgated in 1979 after an extensive rulemaking, was not the subject of challenge and is, of course, not before us.

Finally, Industry fails to demonstrate why information as to toxic pollutants is deserving of an exception to this non-confidentiality policy, that is, why the disclosure of toxic pollutants in a facility presents a significantly greater threat to the confidentiality of trade secrets than disclosure of other, highly detailed information required of permit applicants. See 40 C.F.R. § 122.-21(f)-(g). In the absence of such a showing, we are. fortified in our conclusion that this requirement, accompanied as it is by the waiver provision, is reasonable.

D

Industry then turns from its substantive attack to a procedural challenge, arguing that the regulation was promulgated in final form (albeit in very different forms) twice, in 1980 and 1984, without notice and opportunity for comment. It is, of course, elementary in our law that rulemaking requires notice sufficient to “fairly apprise interested persons of the ‘subjects and issues’ before the Agency.” American Iron & Steel Institute v. EPA, 568 F.2d 284, 293 (3d Cir.1977); 5 U.S.C. § 553(b) (1982). The obvious function of this statutory requirement is to ensure public participation in administrative “legislating” and thereby, in theory, minimize the dangers of arbitrariness and inadequate information. See E.I. du Pont de Nemours & Co. v. Train, 541 F.2d 1018, 1025-26 (4th Cir.1976), aff'd in part and rev’d in part, 430 U.S. 112, 97 S.Ct. 965, 51 L.Ed.2d 204 (1977); Ethyl Corp. v. EPA, 541 F.2d 1, 48 (D.C.Cir.), cert. denied, 426 U.S. 941, 96 S.Ct. 2662, 49 L.Ed.2d 394 (1976); 1 K. Davis, Administrative Law Treatise §§ 6:25, :26 (2d ed. 1978).

In this instance, the notice provided by EPA was adequate to the task. In 1979, the proposed rules included an inquiry contained in the NPDES permit application which would have required applicants to list every “source” of toxic pollutants which were not deliberately discharged but might accidentally be released through various means such as runoff, leaks, and materials storage or handling. 44 Fed.Reg. at 34,358. The inclusion of this query in the proposed NPDES permit application was sufficient to alert interested parties to EPA’s intent to ask permit-tees to provide information as to toxics present in their facilities, regardless of whether the toxics would deliberately be discharged.

Nonetheless, assuming arguendo that the question set forth on the proposed NPDES application failed to provide sufficient notice, we are satisfied that any error was harmless, see 5 U.S.C. § 706 (“due account shall be taken of the rule of prejudicial error”), for the parties had abundant opportunity to comment on the proposed rule. In 1980, EPA explained that the final rule was prompted by comments received. 45 Fed.Reg. 33,290, 33,535 (May 19, 1980). The negative response to the 1980 final regulations induced the agency into settlement negotiations with Industry. Tellingly, during the protracted 1980-82 settlement negotiations, Industry representatives commented extensively on this specific rule and, indeed, were successful in persuading EPA to delete the disclosure requirement in the 1982 proposed rules. See 47 Fed.Reg. 52,072, 52,076 (Nov. 18, 1982). If anything, these negotiations provided Industry with a far greater opportunity to participate in the rulemaking than a plain vanilla notice-and-comment proceeding. To its credit, Industry concedes that in 1982 interested parties had notice and an opportunity to comment on the proposal to drop this rule. Industry Brief at 66. And, it is beyond dispute that EPA’s final decision in 1984 to retain a modified version of the disclosure rule did not trigger a renewed round of notice-and-comment procedures. International Harvester Co., 478 F.2d at 632 & n. 51.¹⁷

Industry’s challenge to the toxic pollutants reporting provision fails to persuade us that the regulation exceeds agency authority, is unreasonable, or was promulgated in violation of the APA’s procedural safeguards. That there exist alternative means of achieving the EPA’s purposes scarcely renders this regulation arbitrary or capricious. We are satisfied that the toxic pollutant reporting requirement is reasonable and represents a valid exercise of the broad power conferred by Congress upon the agency.

VI

In the 1984 final regulations, EPA chose to retain its regulation prohibiting bypasses, which are defined as “the intentional diversion of waste streams from any portion of a treatment facility.” 40 C.F.R. § 122.41(m)(l)(i) (1985). Bypasses which do not cause effluent limits to be exceeded are permissible in order to perform “essential maintenance to assure efficient operation.” Id. § 122.41(m)(2). The regulation also permits bypasses which may cause effluent limitations to be exceeded if such bypasses are “unavoidable to prevent loss of life, personal injury, or severe property damage.” Id. § 122.41(m)(4). The regulation thus ensures that treatment systems chosen by the permittee are operated as anticipated by the permit writer, that is, as they are designed to be operated and in accordance with the conditions set forth in the permit.

Industry attacks the bypass regulation as “arbitrary and unwarranted” on two principal grounds, namely, that the bypass prohibition is (1) unauthorized by the Clean Water Act, and (2) inconsistent with the policies underlying the Act. Industry Brief at 154-55. We conclude, however, that EPA enjoyed authority to adopt such a regulation; that the regulation directly promotes the goals of the Act; that it is fully consistent with the technology-forcing framework of the Act; and that the rule is reasonable.

To return for a moment to basic principles, the Clean Water Act is, as we have seen, quite broad in its sweep. EPA is not limited by statute to the task of establishing effluent standards and issuing permits, but is empowered by section 501(a) of the Act to prescribe regulations necessary to carry out its functions under the Act. 33 U.S.C. § 1361(a). It is also clear that permissible conditions set forth in NPDES permits are not limited to establishing limits on effluent discharge. To the contrary, Congress has seen fit to empower EPA to prescribe as wide a range of permit conditions as the agency deems appropriate in order to assure compliance with applicable effluent limits. 33 U.S.C. § 1342(a)(2); see also id. § 1314(e). Keeping in mind that, under Chevron, the court must defer to the agency’s construction as long as it is a reasonable interpretation of the statute, Chevron, 467 U.S. at 843-45, 104 S.Ct. at 2781-83, a permit condition prohibiting bypass seems to us to fall within the agency’s broad statutory authority.

Faced with the breadth of the statutory language, Industry mounts its principal challenge — that the regulation is contrary to one element of legislative intent informing the Clean Water Act, namely to avoid dictating specific treatment technologies, thus encouraging experimentation with alternative means of pollution reduction. Statements in the legislative history indicate that this goal would be attained through imposition of end-of-the-pipe discharge limitations rather than a national prescription of a specific, uniform “best technology.” This is a policy, we hasten to note, articulated in the legislative history, although not in the Act itself. See S.Rep. No. 414, 92d Cong., 1st Sess. 59 (1971), reprinted in 1972 Legislative History at 1477; H.R.Rep. No. 911, 92d Cong., 2d Sess. 107 (1972), reprinted in 1972 Legislative History at 794.

The pedigree of the policy itself aside, Industry concedes that the bypass regulation does not, in fact, dictate that a specific treatment technology be employed; instead, the regulation requires that a system be operated as designed and according to the conditions of the NPDES permit. Industry Brief at 155 n. 1. Industry argues, however, that requiring the treatment system chosen by the discharger to be operated without bypass, even where the effluent limitations are not exceeded, is barred under a faithful discernment of Congress’ intent.

We do not agree. First, this argument requires the assumption that “on-off” regulation constitutes a choice of treatment technologies. Since that sort of option does nothing to further the goal of exploring diverse treatment technologies, we are unpersuaded that the “on-off” decision is the sort of technological choice Congress intended to leave entirely to the discharger.

Second, Industry ignores other, highly pertinent policies repeatedly expressed in the statute itself and which are furthered by this regulation. The first principle of the statute is, as we have seen, that it is unlawful to pollute at all. The Clean Water Act does not permit pollution whenever that activity might be deemed reasonable or necessary; rather, the statute provides that pollution is permitted only when discharged under the conditions or limitations of a permit. 33 U.S.C. § 1311(a); NRDC v. Costle, 568 F.2d 1369, 1374-77 (D.C.Cir.1977). The foremost national goal enunciated by Congress is the complete elimination of the discharge of pollutants. 33 U.S.C. § 1251(a)(1). This ambitious objective is underscored by the statutory instruction that “new source performance standards” not only reflect the best technology but also include, “where practicable, a standard permitting no discharge of pollutants.” Id. § 1316(a)(1) (emphasis added). In view of the far-reaching goals of the Act, compliance with an effluent standard cannot fairly be viewed as the ultimate object of the statute. In the context of a statute which seeks the elimination of pollution, it is difficult to believe that Congress intended that dischargers be entitled to shut off their treatment facilities and “coast” simply because they were momentarily not in danger of violating effluent limitations.

Third, Industry’s argument, if accepted, would undermine the statutory framework of technology-based standards. As we have seen, the regulatory scheme is structured around a series of increasingly stringent technology-based standards (beginning with the implementation of the best “practicable” technology (BPT), id. § 1311(b)(1), and progressing toward implementation of pollution controls to the full extent of the best technology which would become available (BAT), id. § 1311(b)(2). New sources would, again, be subject to the most stringent technology-based standards of all, namely “new source performance standards.” Id. § 1316(b)(1)(B).¹⁸ In setting new source standards, EPA is statutorily required to give serious consideration to a standard permitting no discharge of pollutants. Id. § 1316(a)(1). Thus, the most salient characteristic of this statutory scheme, articulated time and again by its architects and embedded in the statutory language, is that it is technology-forcing. See S.Rep. No. 414, 92d Cong., 1st Sess. 42 (1971), reprinted in 1972 Legislative History at 1460; 117 Cong.Rec. 38,808 (1971) (Sen. Montoya), reprinted in 1972 Legislative History at 1278; 118 Cong.Rec. 33,693, 33,696 (1972) (Sen. Muskie), reprinted in 1972 Legislative History at 163, 170; Tanners’ Council, Inc. v. Train, 540 F.2d 1188, 1195 (4th Cir.1976). The essential purpose of this series of progressively more demanding technology-based standards was not only to stimulate but to press development of new, more efficient and effective technologies. This policy is expressed as a statutory mandate, not simply as a goal. The bypass prohibition comports with this goal, because it requires that the applicable treatment technology, implemented for the purpose of achieving pollution reduction equivalent to the “best technology,” be operated as designed. Considering the nature of the statutory scheme, which pushes all dischargers to achieve ever-increasing efficiencies and improvements in pollution control, it is difficult to find a Congressional intent that dischargers who are achieving their current target levels of pollution control be at liberty to shut down temporarily for no reason other than the belief that they will not be in technical violation of their permit. Such bypasses appear, rather, at odds with the statutory scheme.

To be sure, the statutory scheme is one of effluent limitations enforced through permits. But to say that the agency may not prescribe technology in specific cases is not to say that the agency is only to set limits on the discharge levels of pollutants and to ignore the means employed to achieve pollution reduction. As we have previously observed, permits may include conditions other than effluent limitations. 33 U.S.C. § 1342(a)(2). Moreover, effluent limitations are premised upon the level of control attainable by proper use of the “best technology.” In the course of establishing effluent limitations, EPA is instructed to specify the factors to be considered in evaluating available technologies; utilize them to identify the best control technologies, measures, and practices available; and determine the degree of effluent reduction attainable using the best technology. Id. § 1314(b).¹⁹ As technology advances, EPA is instructed to revise its regulations at least annually, if necessary, and to revise effluent limitations every five years to reflect progress toward the goal of eliminating pollution. Id. §§ 1311(d), 1314(b).

We cannot help but observe, moreover, that nothing in the statutory language or legislative history reveals express Congressional intent that bypasses, which have no justification other than that they do not run afoul of effluent limitations, be permitted. At best, Industry divines such an intention in the policies and structure of the Act. Yet it is clear that the bypass regulation, far from being contrary to the policies and structure of the Act, in fact undergirds them. In view of the Act’s ambitious policies, we cannot say that the Act requires EPA to allow bypasses which are not provided for in the permit and which are unnecessary for maintenance purposes or to avoid harm to life or property. The statute’s goals are hardly fostered by allowing dischargers to shut off their systems at will whenever they are in compliance with the requirements represented by the effluent limitations.

In fact, individual permit provisions prohibiting bypass which are far more restrictive than this regulation have been upheld as reasonable. In Marathon Oil Co. v. EPA, 564 F.2d 1253 (9th Cir.1977), the Ninth Circuit approved a provision which permitted bypass only when unavoidable to prevent loss of life or severe property damage or when excess storm runoff threatened damage to the treatment equipment, thus rejecting the plea that bypass should be permitted in order to perform maintenance or repairs. Id. at 1274-75 (remanded for clarification of “severe property damage”). In contrast, the regulation before us is more generous than that sustained by our sister circuit in that it permits bypass for purposes of essential maintenance.

Industry also contends that the regulation is unreasonable because it prohibits bypass even when economic considerations militate in favor of that procedure. As we read it, however, the regulation either permits bypass or obviates the need for bypass in each of the three cases Industry identifies as requiring this procedure. First, Industry erroneously argues that bypass is forbidden even when needed for maintenance purposes. See Industry Brief at 155. “Routine maintenance” can be performed during normal downtime, and thus bypass is not needed. 49 Fed.Reg. at 38,-037. Necessary maintenance which cannot be performed during such periods is considered “essential maintenance,” and bypass that does not cause effluent limitations to be exceeded is expressly permitted by the regulation. 40 C.F.R. § 122.-41(m)(2); 49 Fed.Reg. at 38,037. In cases when continuous operation is the norm, Industry Brief at 162, EPA’s comments on the regulations suggest that it would probably consider maintenance to be “essential,” thus falling within the exception which permits bypass. 49 Fed.Reg. at 38,-037. Industry’s second example is seasonal or other fluctuations in a facility’s production schedule which result in little or no discharge, thus obviating the need for full operation (or any operation) of the treatment system. Industry Brief at 155. In such a case, EPA intends that such fluctuations are to be written into the permit. 49 Fed.Reg. at 38,037. If the permit conditions reflect such variations by allowing a facility to shut down treatment processes, then the resulting diversion is not considered a bypass. Id. Third, Industry contends that an event such as hydraulic flooding resulting from heavy rainfall may threaten to damage the system or render it inoperative, see Industry Brief at 155, 163. Then, however, even a bypass which exceeds effluent limitations would be permitted if it was “unavoidable to prevent ... severe property damage.” 40 C.F.R. § 122.41(m)(4).

Industry next argues that the regulation is a de facto effluent limitation for pollutants which the agency has chosen not to regulate specifically through a section 304 effluent limitation. This argument is prompted by EPA’s explanation of the bypass regulation as a means of minimizing the discharge of indirectly regulated pollutants. Industry’s argument suggests that the agency’s decision not to regulate a pollutant with an effluent guideline works a waiver of its power to regulate that pollutant. EPA replies, with persuasive force, that it frequently does not impose specific effluent guidelines for certain pollutants, especially in regulating toxics, but instead treats other “regulated” pollutants as “indicators” of the probable level of the unregulated pollutants because the model treatment technology (the basis for effluent guidelines) removes both. See 49 Fed.Reg. at 38,000 38,036-37. The agency admits that it cannot be absolutely certain that the “unregulated” pollutants are indeed being removed in all cases, for a dis-charger may utilize a dissimilar technology; the regulation, however, is defended as a workable and efficient (for both agency and permittee) means of regulating the numerous pollutants which do not have individual effluent limitations.

Industry’s argument in this respect fails for the reason that the de facto limitations which Industry has identified are due to EPA’s practice of indirect regulation of pollutants, not to the bypass regulation. Industry does not challenge the practice of indirectly regulating pollutants without promulgation of specific effluent limits under section 304. This is unsurprising, for this practice has been upheld in several instances. See, e.g., Reynolds Metals Co. v. EPA, 760 F.2d 549 (4th Cir.1985) (total toxic organics (TTO) not regulated with BAT standard because TTO is controlled through regulation of oil and grease; copper, lead, zinc, and manganese not regulated with BAT standards because they are removed by model technology if operated efficiently enough to remove other pollution parameters).²⁰

Since the bypass regulation is not a de facto effluent limitation, Industry’s related argument — that the record shows the costs of regulation to outweigh its benefits — begins from a mistaken premise. The Clean Water Act requires EPA to undertake a cost-benefit calculus in establishing effluent limitations. See, e.g., 33 U.S.C. § 1314(b)(1)(B); 118 Cong.Rec. 33,696 (1972), reprinted in 1972 Legislative History at 169-70. Neither the statute nor the legislative history, however, imposes such a requirement when EPA issues regulations under its broad authority to implement the Act. See 33 U.S.C. § 1361(a). Indeed, the bypass regulation is not easily susceptible to cost-benefit analysis. As noted above, the agency admits that it is difficult to measure the reduction of pollutants that are indirectly regulated under this regulation. Thus, the costs and benefits of the regulation will vary, depending, among other things, on how it applies to a particular operation. Until it is so applied, the specific facts necessary to conduct a cost-benefit analysis are missing. In short, Industry’s argument about cost-benefit analysis has no basis in law or logic.²¹

We are mindful that Congress did not specifically address the question whether bypasses may be permitted. But, as Chevron teaches, in the face of Congressional silence, we cannot merely substitute our judgment for that of the agency. Our role is more limited, and that is to determine whether EPA’s decision to prohibit bypasses reflects an acceptable interpretation of the Clean Water Act. See, e.g., Young v. Community Nutrition Institute, 476 U.S. 974, 106 S.Ct. 2360, 90 L.Ed.2d 959 (1986). The agency’s adoption of a bypass regulation which incorporates two broad and sensible exceptions (bypasses which do not cause effluent limits to be exceeded for purposes of essential maintenance and bypasses which may cause effluent limits to be exceeded in order to avoid personal injury or severe property damage) is, in our view, reasonable and therefore lawful.

VII

We turn now from the Clean Water Act to another environmental statute which bears on this case. The National Environmental Policy Act (NEPA), 42 U.S.C. §§ 4321-4370 (1982), mandates that federal agencies consider “to the fullest extent possible” and seek to minimize the environmental impact of policies, regulations, or actions. Id. § 4332. This obligation is triggered for all “major Federal actions significantly affecting the quality of the human environment.” Id. § 4332(2)(C). Unless the agency makes an initial finding that the action will have “no significant impact” on the environment, NEPA requires preparation of an environmental impact statement (EIS) prior to a proposal for a “major Federal action.” The EIS must set forth a detailed statement as to various considerations such as “the environmental impact of the proposed action,” the “irreversible and irretrievable” commitments involved, and the alternatives available. Id. The agency is called upon to consider all “direct, indirect, and cumulative effects.” 40 C.F.R. §§ 1508.7-.8, 1508.18 (1985). Upon preparation, the EIS is to be circulated for comment to other federal agencies which have jurisdiction or special expertise with respect to any environmental impact involved, and both the EIS and the comments are to be made available to the public. 42 U.S.C. § 4332(2)(C).

The relevance of NEPA to the EPA regulations at issue is readily apparent from the Clean Water Act itself. The latter measure expressly provides that issuance of an NPDES permit to a new source is a “major Federal action” subject to NEPA review, 33 U.S.C. § 1371(c)(1) (1982); otherwise, EPA actions under the Clean Water Act, such as issuing permits to existing sources, are not subject to NEPA’s strictures. Id.

In its NPDES regulations, EPA has sought to effectuate this mandate by an extraordinary device, namely an outright ban on any on-site construction of a new source until after final issuance of an NPDES permit which incorporates appropriate EIS-related requirements, 40 C.F.R. § 122.29(c)(4)(i), or until 30 days after a finding of “no significant impact.” Id. § 122.29(c)(4)(ii). Violation of the ban constitutes grounds for denial of the permit. Id. § 122.29(c)(5).²²

The construction ban is designed to preserve the status quo for as long as necessary to complete NEPA review. Two exceptions, however, are provided to lend a measure of flexibility to the prohibition: (1) EPA’s Regional Administrators enjoy discretion to waive the ban upon a finding that pre-permit construction will not cause “significant or irreversible adverse environmental impact”; or (2) if the permit applicant executes a legally binding written agreement to comply with all EIS-related requirements. Id. § 122.29(c)(4)(i).

The construction ban is attacked by both Industry and NRDC. On the one hand, the Industry petitioners challenge EPA’s authority to prohibit construction of new sources pending completion of the EIS. On the other hand, NRDC assails the two exceptions to the pre-permit construction ban as violative of NEPA’s commands, which NRDC regards as absolute. Upon analysis, we agree with Industry that the construction ban cannot stand.

A

In determining whether EPA was empowered to impose the construction ban, we first consider whether Congress “has directly spoken to the precise question at issue.” Chevron, 467 U.S. at 842, 104 S.Ct. at 2781. If Congress’ intent is clear, “that is the end of the matter; for the Court, as well as the agency, must give effect to the unambiguously expressed intent of Congress.” Id. at 842-43, 104 S.Ct. at 2781.

EPA claims that the construction ban is authorized by section 511(c)(1) of the Clean Water Act, which requires EPA to comply with NEPA in issuing new source permits.²³ Although the clear language of section 511(c) expressly limits the “major Federal action” which is subject to environmental review to issuance of a permit to discharge pollutants, EPA argues that its implementing regulation, the construction ban, presages no dramatic expansion of agency authority; it is simply a recognition that EPA must consider all environmental effects of the proposed “major Federal action.” In addition, the agency takes note of its obligation to consider all reasonable alternatives to the proposed federal action, regardless of whether the agency has power to implement those alternatives. Natural Resources Defense Council, Inc. v. Morton, 458 F.2d 827, 834-35 (D.C.Cir.1972); 40 C.F.R. § 1502.14(c). EPA defends the construction-ban regulation as a reasonable procedure for preserving alternatives and ensuring that the agency’s decision is both objective and unprejudiced by the irreversible commitment of resources. See 42 U.S.C. § 4332(2)(C)(iii), (v); 40 C.F.R. §§ 1502.2(f), 1506.1; Scientists’ Institute for Public Information, Inc. v. AEC, 481 F.2d 1079, 1090 (D.C.Cir.1973).

In concluding that the construction ban regulation is not authorized by the Clean Water Act, we are guided by the Supreme Court’s analytical approach in a recent non-environmental case, Board of Governors v. Dimension Financial Corp., 474 U.S. 361, 106 S.Ct. 681, 88 L.Ed.2d 691 (1986). At issue in that case was the Federal Reserve Board’s attempt to assert jurisdiction over so-called “non-bank banks.” The Board accomplished this jurisdictional outreach by amending and broadening the statutory definition of “bank” in its implementing regulation. A unanimous Court held that Congress’ intended definition of “bank” was clear and unambiguous, and that the Board's expanded definition was contrary to Congressional intent. Id. at 686-88.

Upon analysis, we believe that we are confronted with an analogous situation in the case at hand. Here, Congress defined “major Federal action,” for purposes of the Clean Water Act, to be the issuance of a discharge permit. 33 U.S.C. § 1371(c). In section 511(c)(1), Congress unambiguously stated that:

Except for the provision of Federal financial assistance for the purpose of assisting the construction of publicly owned treatment works as authorized by section 1281 of this title, and the issuance of a permit under section 1342 of this title for the discharge of any pollutant by a new source as defined in section 1316 of this title, no action of the Administrator taken pursuant to this chapter shall be deemed a major Federal action significantly affecting the quality of the human environment within the meaning of the National Environmental Policy Act of 1969 (83 Stat. 852).

Although EPA concedes that private construction is not a Federal action, 49 Fed. Reg. at 38,018, much less a major one, the agency seeks through the construction ban to commence NEPA review at the pre-construction stage. NEPA review, however, is triggered only by a major Federal action. Therefore, unless the construction itself is pursuant to federal financial assistance, NEPA review may only be conducted with regard to the issuance of a discharge permit, which constitutes, of course, the major Federal action. Until the private owner applies for a discharge permit, then, EPA lacks authority to regulate the owner’s activities under NEPA and the Clean Water Act.²⁴ EPA’s construction ban thus seems directly contrary to Congress’ clear intent, as elucidated by the expressly limited statutory definition of “major Federal action.”

We find further support for this conclusion in the fact that although Congress full well knows how to confer the power to restrict construction pending issuance of a permit, it chose not to do so here. For example, the Act includes no requirement that permit issuance or permit review proceedings precede construction, as is the case with the Clean Air Act. Philadelphia Council of Neighborhood Organizations v. Coleman, 437 F.Supp. 1341, 1370-71 (E.D.Pa.1977), aff'd mem. 578 F.2d 1375 (3d Cir.1978); see 42 U.S.C. § 7475 (1982). Neither does the Clean Water Act confer upon EPA permitting authority over the construction of facilities, as do the Atomic Energy Act, 42 U.S.C. §§ 2133, 2235 (1982); see Calvert Cliffs Cordinating Committee, Inc. v. AEC, 449 F.2d 1109 (D.C.Cir. 1971), and the Federal Power Act, 16 U.S.C. § 797(e) (1982). Consequently, this is not a case in which a construction ban is clearly within the agency’s jurisdiction under its organic statute. See, e.g., Calvert Cliffs, 449 F.2d at 1128-29; Public Service Co. v. NRC, 582 F.2d 77, 81-84 (1st Cir.), cert. denied, 439 U.S. 1046, 99 S.Ct. 721, 58 L.Ed.2d 705 (1978); Detroit Edison Co. v. NRC, 630 F.2d 450, 452 (6th Cir.1980). The fact that Congress has vested some agencies with such power demonstrates that when Congress wanted to extend that power, “ ‘it knew how to do so and did so expressly.’ ” Transamerica Mortgage Advisors, Inc. v. Lewis, 444 U.S. 11, 21, 100 S.Ct. 242, 248, 62 L.Ed.2d 146 (1979) (quoting Touche Ross & Co. v. Redington, 442 U.S. 560, 572, 99 S.Ct. 2479, 2487, 61 L.Ed.2d 82 (1979)); see also CIA v. Sims, 471 U.S. 159, 169 n. 13, 105 S.Ct. 1881, 1888 n. 13, 85 L.Ed.2d 173 (1985).

EPA also looks to NEPA for its authority to impose the construction ban. NEPA’s mandate is, of course, procedural. It requires a specific decisionmaking process, broadening the range of factors that an agency must consider to encompass environmental effects, even those over which the agency has no control, Calvert Cliffs, 449 F.2d at 1123-26; Public Service Co., 582 F.2d at 81. NEPA’s mandate, however, stops at this requirement; NEPA does not require that certain outcomes be reached as a result of the evaluation. In the agency’s own words, “NEPA does not mandate that EPA take any particular action as a result of a NEPA review, but rather grants the Agency discretion to determine what action is appropriate.” 49 Fed.Reg. at 38,033.

NEPA, as a procedural device, does not work a broadening of the agency’s substantive powers. See, e.g., Winnebago Tribe v. Ray, 621 F.2d 269, 272-73 (8th Cir.), cert. denied, 449 U.S. 836, 101 S.Ct. 110, 66 L.Ed.2d 43 (1980); Gage v. United States Atomic Energy Commission, 479 F.2d 1214, 1220 n. 19 (D.C.Cir.1973); Kitchen v. FCC, 464 F.2d 801, 802-03 (D.C.Cir.1972); Movement Against Destruction v. Volpe, 361 F.Supp. 1360, 1383 (D.Md.1973), aff'd, 500 F.2d 29 (4th Cir.1974). Whatever action the agency chooses to take must, of course, be within its province in the first instance. In the circumstances at hand, the “major Federal action” which is subject to environmental review is the issuance of a permit to discharge pollutants; it is not the authorization of construction. As we read the Clean Water Act, EPA’s jurisdiction is limited to regulating the discharge of pollutants, such as formulating effluent standards, issuing or denying permits (including imposition of various conditions), and securing information pertinent to those tasks. NEPA review of new-source permitting, as required by section 511(c), is therefore limited to environmental review of the permit issuance and its alternatives.²⁵

B

Unable to support the construction ban with the language of the Clean Water Act, EPA advances an “enablement” theory, namely that “[c]onstruction of a discharge source generally proceeds in reliance on future Federal action: the issuance of an NPDES permit which is clearly within the Agency’s jurisdiction. Without the permit the source would be unable to operate as intended.” 49 Fed.Reg. at 38,018. Thus, the argument goes, the construction ban does not expand the agency’s organic jurisdiction, but merely protects it by preserving the balance of cost and benefit factors. Id. See, e.g., Greene County Planning Board v. FPC, 455 F.2d 412, 414-18 (2d Cir.) (Federal Power Commission’s authorization of power line constituted a major federal action), cert. denied, 409 U.S. 849, 93 S.Ct. 56, 34 L.Ed.2d 90 (1972); Silva v. Romney, 473 F.2d 287, 290-91 (1st Cir.1973) (private developer of project to which HUD funds had been committed could be enjoined pending preparation by HUD of EIS).

We are unpersuaded. While the discharge permit may, in practical effect, be needed for construction, it is most certainly not a legal condition precedent. That is to say, a rational decisionmaker would not likely build a facility prior to acquiring (or at least initiating the process of acquiring) an operating permit. We therefore acknowledge EPA’s observation that, as a matter of practice, facility planning and the permit process ordinarily proceed hand in hand, due to EPA’s close examination of the planned facility in establishing permit conditions. This practical reality, however, does not carry the day in law for the agency; rather, it only supports the point that, while the statute requires NEPA review to precede only the issuance of the permit, it would be most desirable for that review to precede construction as well. From this practical reality, however, it is a considerable leap, and one we decline to make in light of the statute’s silence, to conclude that the agency is vested with power to call a halt to construction activity.²⁶

EPA also seeks support for its enablement theory in section 306 of the Clean Water Act. In establishing NSPS standards for new sources under that section, EPA is required to examine all aspects of construction and operation of new sources, including available technology, operating policies, and alternatives, 33 U.S.C. § 1316(a)(1). The Act also specifically requires the agency to consider nonwaterquality related environmental impacts. Id. § 1316(b)(1)(B). Thus, the rationale for separate standards for new sources, which may be as stringent as a “no discharge” standard, id. § 1316(a)(1), was the conviction that the most effective and cost-efficient controls were those implemented during planning and construction. See S.Rep. No. 414, 92d Cong., 1st Sess. 57-58 (1971), reprinted in 1972 Legislative History at 1475-76. EPA argues that the close connection that generally exists between the planning of a new source project and the issuance of a NPDES permit justifies the agency’s ban on construction.

We disagree. First, section 306 of the Clean Water Act governs only the promulgation of national NSPS standards. It does not even address the issuance of individual permits, much less authorize a sweeping prohibition on building activity. Second, although issuance of a discharge permit is an absolute precondition to operation of a facility, we are persuaded that the NPDES process does not constitute sufficient federal involvement to “federalize” the private act of construction. See Winnebago Tribe v. Ray, 621 F.2d at 273 & n. 4 (Corps of Engineers permit does not “federalize” construction of power line); Save the Bay, Inc. v. United States Corps of Engineers, 610 F.2d 322, 326-27 (5th Cir.) (Corps of Engineers permit for outfall pipeline does not “federalize” the entire facility), cert. denied, 449 U.S. 900, 101 S.Ct. 269, 66 L.Ed.2d 130 (1980); Friends of the Earth, Inc. v. Coleman, 518 F.2d 323, 327-29 (9th Cir.1975). Through the public-private distinction created in NEPA, Congress chose not to impose NEPA oversight of private decisionmaking, instead electing to regulate private actions solely through other statutes (such as the Clean Water Act). It is, in short, the permitting, not the construction, which EPA has power to restrain pending NEPA review.²⁷

Finally, EPA contends that Congress intended NEPA review to be conducted prior to construction; otherwise, that review would be a hollow gesture, rendering meaningless Congress’ intention that siting alternatives be considered. Again, we are faced with an argument virtually identical to that advocated by the Board in Dimension Financial. There, the Board claimed that the expanded definition in the implementing regulation was necessary to fulfill the “plain purpose” of the legislation. Dimension Financial, 106 S.Ct. at 688. The Court firmly rejected the argument:

The “plain purpose” of legislation however, is determined in the first instance with reference to the plain language of the statute itself. Richards v. United States, 369 U.S. 1, 9, 82 S.Ct. 585, 590, 7 L.Ed.2d 492 (1962). Application of “broad purposes” of legislation at the expense of specific provisions ignores the complexity of the problems Congress is called upon to address and the dynamics of legislative action. Congress may be unanimous in its intent to stamp out some vague social or economic evil; however, because its Members may differ sharply on the means for effectuating that intent, the final language of the legislation may reflect hard fought compromises. Invocation of the “plain purpose” of legislation at the expense of the terms of the statute itself takes no account of the processes of compromise and, in the end, prevents the effectuation of congressional intent.

Id. at 688-89. Although we recognize the practical force of EPA’s argument, we are constrained to limit EPA’s power to that granted it by statute. In so doing, we remind the agency that it “has no power to correct flaws that it perceives in the statute it is empowered to administer. Its rulemaking power is limited to adopting regulations to carry into effect the will of Congress as expressed in the statute.” Id. at 689.²⁸

We conclude that the construction ban exceeds the agency’s authority under either the Clean Water Act or NEPA. As a result, we obviously have no occasion to address the issues raised by NRDC’s challenge to the exceptions to the ban.

VIII

For the reasons stated, we hold that the agency’s “new source” definition and ten-year grace period provision are harmonious with the Clean Water Act; that the promulgation of the regulation requiring permit applicants to report all toxic pollutants currently used or manufactured violated neither the Clean Water Act nor the APA; and that the bypass prohibition is fully consistent with the Clean Water Act and is not arbitrary. We further hold that the EPA lacks authority to ban construction of new sources pending permit issuance. Accordingly, the petitions for review, as to the issues addressed herein, are denied in part and granted in part.

It is so ordered.

1

. This court has jurisdiction to review these regulations under section 509(b)(1) of the Clean Water Act, 33 U.S.C. § 1369(b)(1) (1982). Natural Resources Defense Council, Inc. v. EPA, 673 F.2d 400, 403 (D.C.Cir.), cert. denied sub nom. Chemical Manufacturers Ass’n v. EPA, 459 U.S. 879, 103 S.Ct. 175, 74 L.Ed.2d 143 (1982).

2

. NRDC and CBE were not parties to that agreement.

3

. The issues to be addressed in other opinions by the court include three NEPA issues (effluent limits, non-water quality related conditions, admission of evidence); various procedural issues (public participation, APA continuance, non-adversary panel procedures, maximum state penalties, veto authority); storm water issues; toxicity limits; and three issues identified by the parties as "exceptions" (backsliding, net/gross, and upsets).

4

.Effluent limitations consist of restrictions on quantities, rates, or concentrations of pollutants and compliance schedules for the achievement of such restrictions. See id. § 1362(11).

5

.Under the Act, a pollutant falls into one of three categories. The Act defines "toxic pollutant” as follows:

The term “toxic pollutant" means those pollutants, or combinations of pollutants, including disease-causing agents, which after discharge and upon exposure, ingestion, inhalation or assimilation into any organism, either directly from the environment or indirectly by ingestion through food chains, will, on the basis of information available to the Administrator, cause death, disease, behavioral abnormalities, cancer, genetic mutations, physiological malfunctions (including malfunctions in reproduction) or physical deformations, in such organisms or their offspring.

Id. § 1362(13). "Conventional pollutants” include, but are not limited to, "pollutants classified [by EPA] as biological oxygen demanding, suspended solids, fecal coliform, and pH." Id § 1314(a)(4). "Nonconventional pollutants” are those which are neither toxic nor conventional.

6

. A “new source" is "any source, the construction of which is commenced after the publication of proposed regulations prescribing a standard of performance under this section which will be applicable to such source, if such standard is thereafter promulgated in accordance with this section.” Id § 1316(a)(2). All other sources are “existing sources.”

7

. Toxic pollutants are subject to health-based effluent standards in addition to the technology-based effluent limitations discussed in the text. Id § 1317(a). In other words, limits on toxic pollutants may be based not only on how much reduction advanced technology can achieve; they may be further limited based on health considerations.

8

. Generic parameters are measurements which describe a solution or aquatic environment rather than indicate the precise level of a specific chemical component or pollutant in it. The pH of a solution is a measure of its acidity or alkalinity. Biochemical oxygen demand (BOD) measures the biological or bacteriological pollution of the water. BOD is the amount of dissolved oxygen required by microorganisms in the water. It is used as an indicator of the concentration of organic matter in the water, e.g. sewage, upon which the microorganisms thrive. Unpolluted water normally is saturated with dissolved oxygen. As the concentration of organic matter increases, however, the growth of microorganisms increases the rate at which the organic matter is oxidized, thus increasing the demand for oxygen. When the oxygen is consumed more rapidly than it can be replenished from the air above the water surface, aerobic organisms and normal aquatic life die.

9

. 40 C.F.R. § 122.2 (1985) defines "new source" as follows:

10

. We have recently been reminded that what seems clear on its face may in fact be infected with ambiguity. Young v. Community Nutrition Inst., 476 U.S. 974, 106 S.Ct. 2360, 90 L.Ed.2d 959 (1986). Accordingly, we have not contented ourselves with exclusive reliance on the statute’s plain meaning, even though we must confess that the statute by its terms seems abundantly clear in this respect.

11

. In so holding, we respectfully disagree with the Third Circuit’s approach both in this statutory arena and its decision which remanded similar "new source” definitions in the coal mine and general pretreatment regulations. See National Ass’n of Metal Finishers v. EPA, 719 F.2d 624 (3d Cir.1983), rev'd on other grounds sub nom. Chemical Manufacturers Ass'n v. NRDC, 470 U.S. 116, 105 S.Ct. 1102, 84 L.Ed.2d 90 (1985); see also Pennsylvania Dep’t of Environmental Resources v. EPA, 618 F.2d 991 (3d Cir.1980). In Pennsylvania Dep’t of Environmental Resources, our sister circuit concluded that the only interpretation of section 306 consistent with the Clean Water Act’s policy of maximizing the number of businesses subject to the new source standards was that advanced here by NRDC. 618 F.2d at 999-1000. We are persuaded, however, that that policy is but one of several underlying section 306, see id. at 999. While EPA's incorporation of the 120-day limit into the "new source” definition is arguably less than the optimal resolution of these policies, it is nevertheless at minimum a reasonable construction of the statute. We can ask no more. Chevron, 467 U.S. at 842-45, 104 S.Ct. at 2781-83.

12

. The Act imposes a two-phase pollution control scheme for existing industrial sources which requires the attainment of progressively more stringent standards by established deadlines. By 1977, existing sources were required to achieve effluent limitations reflecting the application of the "best practicable control technology" (BPT). This standard generally re-fleeted the "average performance of the best existing plants.” Tanners’ Council, Inc. v. Train, 540 F.2d 1188, 1191 (4th Cir.1976); American Meat Inst. v. EPA, 526 F.2d 442, 453 (7th Cir.1975). By 1984, existing sources were to satisfy a more demanding standard of the "best available technology economically achievable” (BAT), which reflected the greatest attainable level of effluent reduction which could be achieved. In determining this standard, EPA could consider even plant processes or control technologies "that have not been applied as long as the record demonstrates that there is a reasonable basis to believe that the technology will be available by 1983.” Tanners’ Council of America, 540 F.2d at 1195. Achievement was required by 1984 of effluent limitations for "conventional” pollutants which reflect the "best conventional pollution control technology."

13

. The concept of technology-based standards or "standards of performance,” as distinct from pollution standards based upon the quality of the receiving medium (air or water), was borrowed from the Clean Air Act. In 1970, Congress defined "standard of performance" as a technology-based standard, 42 U.S.C. § 7411(a)(1), and required EPA to establish such standards for new sources. Id. § 7411(b)(1)(B). In 1977, the Clean Air Act was amended to provide for standards of performance for existing sources as well. Id § 7411(d).

14

. We note that this statutory language was described by the Conference Report as “comparable to” an earlier version of section 301(f) of the House Amendment, S.Rep. No. 1236, 92d Cong., 2d Sess. 128 (1972), reprinted in 1972 Legislative History at 311. But that observation, found again in the uncertain terrain of legislative history, does not suggest that no difference was perceived between the two versions.

15

. The analysis which leads to the conclusion that "standard of performance" must mean all technology-based standards would seem to compel the conclusion that the term encompasses technology-based toxic pollutant standards as well. EPA has rejected this interpretation, however, instead excluding toxic or hazardous substance standards from the ten-year grace period. See 40 C.F.R. § 122.29(d)(2). We emphasize that this exclusion has not been challenged in this case.

16

. Section 308(a) provides:

(a) Whenever required to carry out the objective of this chapter, including but not limited to (1) developing or assisting in the development of any effluent limitation, or other limitation, prohibition, or effluent standard, pretreatment standard, or standard of performance under this chapter; (2) determining whether any person is in violation of any such effluent limitation, or other limitation, prohibition or effluent standard, pretreatment standard, or standard of performance; (3) any requirement established under this section; or (4) carrying out sections 1315, 1321, 1342, 1344 (relating to State permit programs), and 1364 of this title—

(A) the Administrator shall require the owner or operator of any point source to (i) establish and maintain such records, (ii) make such reports, (iii) install, use, and maintain such monitoring equipment or methods (including where appropriate, biological monitoring methods), (iv) sample such effluents (in accordance with such methods, at such locations, at such intervals, and in such manner as the Administrator shall prescribe), and (v) provide such other information as he may reasonably require; and

(B) the Administrator or his authorized representatives, upon presentation of his credentials—

(i) shall have a right of entry to, upon or through any premises in which an effluent source is located or in which any records required to be maintained under clause (A) of this subsection are located, and (ii) may at reasonable times have access to copy any records, inspect any monitoring equipment or method required under clause (A), and sample any effluents which the owner or operator of such source is required to sample under such clause.

33 U.S.C. § 1318(a).

17

. Industry suggests that the final rule was unsupported by the record, see 5 U.S.C. § 706(2)(E), apparently because the volume of opposing comments of companies and trade associations on behalf of "hundreds of companies and thousands of dischargers,” Industry Brief at 66, should be deemed to outweigh the supporting comments of several States, an environmental group, and the EPA’s own enforcement arm. If we correctly understand Industry’s point, it scarcely needs saying that its approach misses the mark. The substantial-evidence standard has never been taken to mean that an agency rulemaking is a democratic process by which the majority of commenters prevail by sheer weight of numbers. Regardless of majority sentiment within the community of commenters, the issue is whether the rules are supported by substantial evidence in the record. The number and length of comments, without more, is not germane to a court's substantial-evidence inquiry. Cf. Association of Data Processing Service Organization v. Board of Governors, 745 F.2d 677 (D.C.Cir.1984) (discussing substantial evidence review).

18

. Such sources must implement the "best available demonstrated control technology," which presumably would include more effective alternatives than those available to existing sources.

19

. As a supplement to effluent limitations, EPA is also empowered to prescribe actual management practices to control the accidental discharge of toxic substances. 33 U.S.C. § 1314(e).

20

. See also Natural Resources Defense Council, Inc. v. Costle, 568 F.2d 1369, 1380 (D.C.Cir.1977) ("[W]hen numerical effluent limitations are infeasible, EPA may issue permits with conditions designed to reduce the level of effluent discharges to acceptable levels. This may well mean opting for a gross reduction in pollutant discharge rather than the fine-tuning suggested by numerical limitations. But this ambitious statute is not hospitable to the concept that the appropriate response to a difficult pollution problem is not to try at all.").

21

. The Third Circuit addressed a similar argument in its review of the agency’s formula for calculating general pretreatment regulations under section 307 of the Act, 33 U.S.C. § 1317. See National Ass'n of Metal Finishers v. EPA, 719 F.2d 624, 653-56 (3d Cir.1983), rev'd in part on other grounds sub nom. Chemical Mfrs. Ass’n v. NRDC, 470 U.S. 116, 105 S.Ct. 1102, 84 L.Ed.2d 90 (1985). In that context, the court rejected as unripe a challenge to the costs and benefits of the formula. It found that EPA was obligated to consider costs and benefits only when it applied the formula to obtain numerical discharge limits. Id. at 654-56.

22

. This rule applies when EPA is the permitting authority. Since NEPA imposes duties only on the federal government, not the States, this regulation does not apply when a State is issuing a NPDES permit under its delegated NPDES permitting authority.

23

. While EPA’s claim of authority to ban construction pending issuance of a NPDES permit can be traced to an EPA General Counsel opinion written in 1976, EPA General Counsel's Opinion No. 76-18 (Sept. 23, 1976), the construction ban provision first appeared in the 1979 NPDES regulations. 44 Fed.Reg. at 32,-871-72, 32,915. In 1982, the agency reversed course and proposed to rescind this portion of the regulation, for the stated reason that the Clean Water Act "does not regulate construction of facilities, only discharges from them.” 47 Fed.Reg. at 52,077. Comments received thereafter prompted EPA to reembrace its former policy. Concluding that the construction ban was authorized by both the Clean Water Act and NEPA, the agency reinstated the construction ban in the 1984 final NPDES regulations. 49 Fed.Reg. at 38,017. We are thus not confronted with a consistent agency interpretation of its powers. See I.N.S. v. Cardoza-Fonseca, — U.S. -, 107 S.Ct. 1207, 1221 n. 30, 94 L.Ed.2d 434 (1987) (”[a]n agency interpretation of a relevant provision which conflicts with the agency's earlier interpretation is ‘entitled to considerably less deference’ than a consistently held agency view."); cf. Haig v. Agee, 453 U.S. 280, 291, 101 S.Ct. 2766, 2774, 69 L.Ed.2d 640 (1981) (noting that a consistent administrative construction of a statute granting broad rulemaking authority must be followed by the courts absent compelling indication that it is wrong).

24

. As EPA concluded in 1982, the Clean Water Act regulates the discharge of pollutants, not the construction of sources. The Act does not prohibit construction of a new source without a permit; facilities of any size, shape, or description may be built in any location without running afoul of the Act. The Act only prohibits new sources from discharging pollutants without a permit, 33 U.S.C. § 1311(a), or in violation of existing NSPS standards, id. § 1316(e). It does not prevent such sources from being built.

It is this jurisdictional limitation that distinguishes this regulation from regulations such as the bypass prohibition. The bypass regulation, namely a prohibition of certain discharges not provided for in the permit, is well within EPA’s proper ambit under the Clean Water Act of regulating actual discharges. In contrast, as discussed in the text, the construction ban is outside the agency’s Clean Water Act jurisdiction; if it is authorized at all, its source must lie in NEPA.

25

. To support its claim of power under NEPA, EPA cites section 1506.1(a) of the Council on Environmental Quality’s (CEQ) regulations. Section 1506.1(a) provides in part that "Until an agency issues a record of decision ... no action concerning the proposal shall be taken which would: (1) Have an adverse environmental impact; or (2) Limit the choice of reasonable alternatives.” 40 C.F.R. § 1506.1.

EPA’s argument is deficient in two respects. First, section 1506.1(a) regulates action concerning the proposal, which under the Clean Water Act, is the application for a discharge permit. It is not until a permit is sought that section 1506.-1(a) is pressed into operation. Thus, the regulation confers no authority to impose a pre-application ban on construction.

Second, the CEQ mandate provides that the regulations are binding "except where compliance would be inconsistent with other statutory requirements." 40 C.F.R. § 1500.3. Since NEPA does not broaden an agency’s existing power, use of the CEQ regulation to expand EPA’s power would be inconsistent with both NEPA and the Clean Water Act.

26

. The issue presented is not the proper timing of either NEPA review or NPDES permit issuance, but whether the agency has authority to ban construction. We do not speak to the question whether EPA has authority to impose timing requirements on the permit application process so as, for example, to require that NPDES applications precede construction.

27

. Similarly, the environmental effects of the construction siting decision cannot be deemed to be either direct or indirect effects of EPA’s issuance of a discharge permit. The environmental impacts of siting are not "effects,” in any practical sense, of the permit issuance.

28

. Such a ban on construction, we note, is unenforceable. Section 309, 33 U.S.C. § 1319, which provides for state and federal enforcement, and section 505, id. § 1365, which provides for enforcement through citizen suits, authorize civil and criminal prosecution only for polluting in violation of a permit or effluent standard. See also id. § 1342(k).