Michael K. Glenn, Gary R. Feulner, Chadbourne, Parke, Whiteside & Wolff, Washington, D.C., for American Paper Institute and The National Forest Products Ass'n.

Donald T. Bliss, David T. Beddow, O'Melveny & Myers, Washington, D.C., and Mark R. Steinberg, O'Melveny & Myers, Los Angeles, Cal., for Metal Finishing Ass'n of Southern California.

Barry S. Neuman (argued), Michael Steinberg (argued), George E. Henderson, Lee R. Tyner, Carol E. Dinkins, Asst. Atty. Gen., Jose R. Allen, Acting Chief, Environmental Defense Section, Lloyd S. Guerci, Joan Z. Bernstein, Anthony Z. Roisman, Ellen Maldonado, James W. Moorman, Donald W. Stever, Michael W. Stein, U.S. Dept. of Justice, Michael Dworkin (argued), Michael Murchison, Ellen Siegler, Daniel J. Berry, Environmental Protection Agency, Washington, D.C., for EPA; Of Counsel: Robert M. Perry, Associate Administrator and Gen. Counsel, Susan G. Lepow, Asst. Gen. Counsel Environmental Protection Agency, Washington, D.C., of counsel.

A. The Definitions of "Interference" and "Pass

Through"

B. Definition of "New Source"

AISI                 American Iron and Steel Institute
BAT                  Best Available Technology Economically
                      Achievable
BDT                  Best available Demonstrated Control
                      Technology
BPT                  Best Practicable Control Technology
                      Currently Available
CACI                 Chicago Association of Commerce and
                      Industry
CMA                  Chemical Manufacturing Association
EPA                  Environmental Protection Agency
FDF                  Fundamentally Different Factor
GM                   General Motors Corp.
IIPEC                Institute for Interconnecting and
                      Packaging Electronic Circuits
J. App.              Joint Appendix
Legis.  Hist.         Legislative History
Me degrees           Regulated Metal in Influent
MFASC                Metal Finishing Association of Southern
                      California
NAMF                 National Association of Metal Finishers
NPDES                National Pollutant Discharge
                      Elimination System
NRDC                 Natural Resources Defense Council
PM                   Precipitable Metals in Influent
POTW                 Publicly Operated Treatment Works
R. Add.              Addendum of Respondent
TSS                  Total Suspended Solids
TTO                  Total Toxic Organics
USBA                 United States Brewers Association
Xme                  Ratio of Me degrees to PM

Before: GIBBONS, HUNTER and BECKER, Circuit Judges.

Section 307 of the Clean Water Act1 directs the Administrator of the Environmental Protection Agency ("EPA") to promulgate regulations requiring industrial facilities to pretreat the pollutants that they discharge into public sewage treatment systems. The Administrator has promulgated both general pretreatment regulations2 and regulations establishing categorical pretreatment standards for existing electroplating sources.3 The petitioners in these consolidated cases seek review of the Administrator's actions in promulgating certain provisions of those regulations. Under section 509 of the Clean Water Act4 we have jurisdiction to exercise a limited review of the Administrator's actions. We may overturn those actions only if they are arbitrary, capricious or otherwise contrary to law.5 Under that standard of review, we find invalid certain provisions of the general pretreatment regulations. Because it is not for us to rewrite those provisions, we will remand them to the Administrator.

In 1972 Congress amended the Federal Water Pollution Control Act ("the Act" or "the Clean Water Act"),6 setting as a national goal the elimination, by 1985, of the discharge of pollutants into navigable waters, 33 U.S.C. Sec. 1251(a)(1) (1976). To reach that goal the Act directed the Administrator of EPA to promulgate regulations setting limits on the pollution that can be discharged by three general types of "point sources," see id. Sec. 1362(14) (1976 & Supp. I 1977).

First, the Administrator was to establish effluent limitations for point sources which discharge pollutants directly into navigable waters ("direct dischargers"). The Administrator had to define effluent limitations for categories or classes of point sources which would require existing direct dischargers to employ by 1977 the best practicable control technology currently available ("BPT"), id. Secs. 1311(b)(1)(A), 1314(b)(1) (1976), and to use by 1983-87 the best available technology economically achievable ("BAT"), id. Secs. 1311(b)(2) (1976 & Supp. I 1977), 1314(b)(2) (1976). For newly-constructed direct dischargers the Administrator had until 1974 to establish "new source" performance standards requiring the application of the best available demonstrated control technology ("BDT"). Id. Sec. 1316. The Administrator had to set the BPT, BAT, and BDT limitations by considering the factors specified in sections 304(b) and 306(b) of the Act, id. Secs. 1314(b), 1316(b). He was to apply those limitations to individual direct dischargers through the National Pollutant Discharge Elimination System ("NPDES") permit issued to the discharger under section 402 of the Act, id. Sec. 1342 (1976 & Supp. I 1977).

Second, the Act mandated that the Administrator set effluent limitations for publicly owned treatment works ("POTWs") engaged in the treatment of municipal sewage or industrial wastewater. See id. Sec. 1292(2) (1976 & Supp. I 1977). Under the Act the Administrator had to establish effluent limitations, based on "secondary treatment," which POTWs had to meet by 1977. Id. Secs. 1311(b)(1)(B), (C), 1314(d)(1) (1976). The limitations thus established were to be applied to each individual POTW through its NPDES permit. Id. Sec. 1342 (1976 & Supp. I 1977).

Third, section 307 of the Act addressed the "indirect dischargers," point sources which discharged their pollutants not directly into navigable waters but into POTWs. Congress recognized that the pollutants which some indirect dischargers release into POTWs could interfere with the operation of the POTWs, or could pass through the POTWs without adequate treatment. To prevent such discharges by existing sources, Congress directed in section 307(b)(1) of the Act:

(b)(1) The Administrator shall ... publish proposed regulations establishing pretreatment standards for introduction of pollutants into [POTWs] for those pollutants which are determined not to be susceptible for treatment by such treatment works or which would interfere with the operation of such treatment works.... Pretreatment standards under this subsection ... shall be established to prevent the discharge of any pollutant through [POTWs], which pollutant interferes with, passes through or otherwise is incompatible with such works.

33 U.S.C. Sec. 1317(b)(1) (1976); see also id. Sec. 1314(g) (Supp. I 1977). The Administrator had to designate the categories of existing sources to which each such standard would apply, promulgate the standards by 1973, and revise the standards as control technologies and industrial processes changed. Id. Sec. 1317(b). For newly-constructed indirect dischargers the Act directed that by 1974 the Administrator had to promulgate pretreatment standards for each category of new sources which "shall prevent the discharge of any pollutant into such treatment works, which pollutant may interfere with, pass through, or otherwise be incompatible with such works." Id. Sec. 1317(c). New and existing indirect dischargers did not need to obtain NPDES permits, but instead had pretreatment standards imposed directly upon them.

In 1977 Congress amended the Act by passing the Clean Water Act of 1977, Pub.L. No. 95-217, 91 Stat. 1566 ("the 1977 Amendments"). Section 54 of the 1977 Amendments added a sentence to section 307(b)(1) permitting a POTW to modify the pretreatment requirement of an existing indirect discharger if the POTW could successfully remove all or part of the toxic pollutants released by that discharger. Id. Sec. 54(a), 91 Stat. 1591 (amending 33 U.S.C. Sec. 1317(b)(1) (Supp. I 1977)).

The Administrator elaborated his regulatory approach to indirect dischargers in his National Pretreatment Strategy, 43 Fed.Reg. 27,759 (1978), and in the consent decree in NRDC v. Train, 8 Env't Rep.Cas. (BNA) 2110 (D.D.C.1976), modified sub nom. NRDC v. Costle, 12 Env't Rep.Cas. (BNA) 1833 (D.D.C.1979), aff'd in part sub nom. Environmental Defense Fund v. Costle, 636 F.2d 1229 (D.C.Cir.1980), modified on remand sub nom. NRDC v. Gorsuch, Nos. 2153-73 et al. (D.D.C. Oct. 26, 1982). The Administrator announced that he would promulgate two types of pretreatment standards.

The first type, "categorical" pretreatment standards, would establish numerical limits on the discharge, by twenty-one specific categories of industrial sources, of particular toxic pollutants which could cause interference with or pass through POTWs. 43 Fed.Reg. 27,760, 27,771-73 (1978); NRDC v. Train, 8 Env't Rep.Cas. (BNA) at 2130-36. Categorical pretreatment standards would be set to require the application of similar levels of control technology as the Act mandated for direct dischargers. 43 Fed.Reg. 27,760-63 (1978); 42 Fed.Reg. 6480 (1977). The Administrator agreed to promulgate categorical pretreatment standards "generally analogous to best practicable control technology currently available" (BPT) for eight industries by May 15, 1977. NRDC v. Train, 8 Env't Rep.Cas. (BNA) at 2128 p 13. For all twenty-one industrial categories the Administrator would then promulgate categorical pretreatment standards based on BAT for existing sources and BDT for new sources. 43 Fed.Reg. 27,760 (1978); see NRDC v. Gorsuch; NRDC v. Train, 8 Env't Rep.Cas. (BNA) at 2123-26.

The second type of pretreatment standard, the "prohibited discharge" standard, would not set numerical limits on the discharge of particular pollutants by specified sources. 43 Fed.Reg. 27,759-60 (1978). Rather, the prohibited discharge standard would establish a general prohibition on the release of any pollutants by any nondomestic source if those pollutants interfere with or pass through a POTW. Id.

The General Pretreatment Regulations for Existing and New Sources of Pollution, 40 C.F.R. Sec. 403.1.-16 (1982), serve to implement the two types of pretreatment standards. First, the general pretreatment regulations themselves contain the prohibited discharge standard generally forbidding interference and pass through, id. Sec. 403.5, and define the terms "interference" and "pass through," id. Sec. 403.3(i), (n). Second, the general pretreatment regulations establish the mechanisms and procedures governing the separately promulgated categorical pretreatment standards. The general regulations define whether a source is a "new source" under the standards. Id. Sec. 403.3(k). The general regulations contain a mechanism through which the existing industrial user of a POTW can obtain a variance from a categorical discharge limit if the user can show that during the development of the standard EPA had considered "fundamentally different" factors than those relating to the user's operation ("the FDF variance provision"). Id. Sec. 403.13. The regulations set up the procedure by which a POTW can revise an industrial user's categorical discharge limit to reflect the POTW's removal of the user's pollutants ("the removal credit provision"). Id. Sec. 403.7. Finally, the regulations provide a formula to calculate an adjusted categorical discharge limit where the industrial user mixes the effluent from the regulated process with other wastewaters prior to pretreatment ("the combined wastestream formula"). Id. Sec. 403.6(e).

The Administrator first proposed the general pretreatment regulations on February 2, 1977. 42 Fed.Reg. 6476 (1977). He promulgated the regulations on June 26, 1978. 43 Fed.Reg. 27,736 (1978). On October 29, 1979, the Administrator proposed amendments to the regulations, 44 Fed.Reg. 62,260 (1979), which he promulgated on January 28, 1981, 46 Fed.Reg. 9404 (1981). The Administrator then attempted to postpone indefinitely the effective date of first all and later part of the general pretreatment regulations. 47 Fed.Reg. 4518 (1982); 46 Fed.Reg. 19,936, 50,502, 50,503 (1981). After we declared that indefinite postponement invalid in NRDC v. EPA, 683 F.2d 752 (3d Cir.1982), the Administrator reinstated the regulations' effective date of March 30, 1981. 47 Fed.Reg. 42,688 (1982); see 46 Fed.Reg. 11,971 (1981). On October 4, 1982, we granted the petitioners' unopposed motion to extend the regulations' effective date until June 30, 1981. 48 Fed.Reg. 2774 (1983).

The categorical pretreatment standards for the Electroplating Point Source Category, 40 C.F.R. Secs. 413.01.-84 (1982), are BPT-level pretreatment standards set pursuant to the NRDC v. Train consent decree. 44 Fed.Reg. 52,592, 52,608 (1978); see 8 Env't Rep.Cas. (BNA) at 2128 p 13(b). The categorical electroplating standards cover 7752 existing firms with electroplating operations, the firms falling in three broad groups: independent "job shops," firms performing electroplating as their primary line of business; independent manufacturers of printed circuit board; and "captive operations," electroplating sections of firms which perform electroplating as part of their manufacture of another product. See 44 Fed.Reg. 52,593 (1979); 43 Fed.Reg. 6561-62 (1978). The electroplating standards divide those firms into seven subcategories, based on the electroplating process employed.7 For each subcategory the standards, inter alia, set numerical limits on the dischargeable concentrations of cyanide and several metals (e.g., cadmium, chromium, copper, lead, nickel, and zinc). 40 C.F.R. Secs. 413.14.-84 (1982). Electroplating sources discharging less than 10,000 gallons per day of electroplating process wastewater have to meet limits for only lead, cadmium and amenable cyanide. Id. "Integrated" facilities, which combine the process wastestream from their captive electroplating operations with other wastewaters prior to pretreatment, are instructed to adjust their discharge limits using the combined wastestream formula. Id. Sec. 413.04; see id. Sec. 413.02(h).

The Administrator proposed the categorical electroplating standards on February 14, 1978, 43 Fed.Reg. 6560 (1978), and promulgated them on September 7, 1979, 44 Fed.Reg. 52,590 (1979), corrected, id. at 56,360. Following promulgation petitioners National Association of Metal Finishers and Institute for Interconnecting and Packaging Electronic Circuits filed petitions for review in this court. Nos. 79-2256, 79-2443. On March 7, 1980, those parties and EPA reached a settlement agreement ("the NAMF Settlement Agreement"). Addendum to Respondent's Brief at D-1 [hereinafter cited as "R.Add."]. Pursuant to that agreement the Administrator on July 3, 1980, proposed several amendments to the 1979 electroplating standards. 45 Fed.Reg. 45,322 (1980). In response to the petition for review of Ford Motor Co., No. 80-1008, EPA proposed other changes, 45 Fed.Reg. 19,245 (1980). Ford later filed a petition for reconsideration of the 1979 standards. J.App. at 2082. On January 28, 1981, the Administrator denied Ford's petition for reconsideration, 46 Fed.Reg. 9476 (1981), and promulgated the amendments to the electroplating standards, id. at 9462, corrected, id. at 30,625. The deadline for compliance with the electroplating standards for integrated facilities was set at three years from the effective date of the combined wastestream formula,8 while non-integrated facilities had a compliance date of May 12, 1982, 46 Fed.Reg. 9462 (1981), later modified to April 27, 1984, 48 Fed.Reg. 2775 (1983); 46 Fed.Reg. 43,973 (1982).

On August 31, 1982, the Administrator published the proposed Metal Finishing regulations, which established BAT pretreatment standards for most of the indirect dischargers presently covered by the electroplating standards. 47 Fed.Reg. 38,462-63 (1982). Only existing job shops and printed circuit board manufacturers would remain under the electroplating standards, which would be amended to restrict the discharge of toxic organic pollutants. Id. at 38,464, 38,468. On July 15, 1983, the Administrator promulgated the Metal Finishing regulations. 48 Fed.Reg. 32,462 (1983) (to be codified at 40 C.F.R. Sec. 433.10.-17).

As noted above, National Association for Metal Finishers ("NAMF"), Institute for Interconnecting and Packaging Electronic Circuits ("IIPEC"), and Ford Motor Co. ("Ford") filed petitions for review of the 1979 electroplating standards. Nos. 79-2256, 79-2443, 80-1008. Ford, NAMF, General Motors Corp. ("GM"), and Metal Finishing Association of Southern California ("MFASC") petition for review of the 1981 electroplating amendments. Nos. 81-1279, 81-1351, 81-1712, 81-2119. Ford also petitions for review of the Administrator's denial of its petitions for review of the Administrator's denial of its petition for reconsideration of the 1979 electroplating standards, No. 81-1214. We address that appeal in Ford Motor Co. v. EPA, 718 F.2d 55 (3d Cir.1983).

Petitioners Natural Resources Defense Council ("NRDC"), United States Brewers Association ("USBA"), and Chemical Manufacturing Association ("CMA") petition for review of the 1978 general pretreatment regulations. Nos. 81-1977, 81-1978, 81-1979. Petitioners Ford, NAMF, CMA, NRDC, Interlake, Chicago Association of Commerce and Industry ("CACI") and others seek review of the 1981 general pretreatment regulations. Nos. 81-1210, 81-1981, 81-1982, 81-1983, 81-1984, 81-1985, 81-2150, 81-2151.

Consideration of the cases was necessarily held pending our resolution in NRDC v. EPA of the challenge to the Administrator's indefinite postponement of the 1981 general pretreatment amendments. Judge Becker of this Court then presided over a series of conferences in which he consolidated the cases, set a briefing schedule, and, on October 29, 1982, limited the subjects of briefing.9

Under section 10(e) of the Administrative Procedure Act, we may not invalidate agency actions unless we find them to be "arbitrary, capricious, an abuse of discretion, or otherwise not in accordance with law." 5 U.S.C. Sec. 706(2)(A) (1976). This standard sets the level of deference with which we must review the agency's actions for their statutory authority, substantive validity and procedural regularity. See Weyerhaeuser Co. v. Costle, 590 F.2d 1011, 1024 (D.C.Cir.1978).

We must extend "great deference to the interpretation given the statute by the officers or agency charged with its administration." EPA v. National Crushed Stone Association, 449 U.S. 64, 83, 101 S. Ct. 295, 307, 66 L. Ed. 2d 268 (1980) (quoting Udall v. Tallman, 380 U.S. 1, 16, 85 S. Ct. 792, 801, 13 L. Ed. 2d 616 (1965)); American Iron & Steel Institute v. EPA ("AISI I"), 526 F.2d 1027, 1041-42 (3d Cir.1975), mandate recalled in part, 560 F.2d 589 (3d Cir.1977), cert. denied, 435 U.S. 914, 98 S. Ct. 1467, 55 L. Ed. 2d 505 (1978). If an act is susceptible to more than one reasonable interpretation, we must accept any reasonable interpretation chosen by the agency. Udall v. Tallman, 380 U.S. 1, 16, 85 S. Ct. 792, 801, 13 L. Ed. 2d 616 (1965); see NRDC v. Train, 421 U.S. 60, 75, 95 S. Ct. 1470, 1479, 43 L. Ed. 2d 731 (1975). If the agency rejects the reasonable interpretation of the statute, however, we must "honor the clear meaning of a statute, as revealed by its language, purpose and history." International Brotherhood of Teamsters v. Daniel, 439 U.S. 551, 556 n. 20, 99 S. Ct. 790, 800 n. 20, 58 L. Ed. 2d 808 (1979); see FEC v. Democratic Senatorial Campaign Committee, 454 U.S. 27, 32, 37, 102 S. Ct. 38, 42-45, 70 L. Ed. 2d 23 (1981).

Our inquiry into the substantive basis for the agency's actions must be searching and careful, but our review is a narrow one. As the Supreme Court has recently stated:

The scope of review under the arbitrary and capricious standard is narrow and a court is not to substitute its judgment for that of the agency. Nevertheless, the agency must examine the relevant data and articulate a satisfactory explanation for its action including a "rational connection between the facts found and the choice made." Burlington Truck Lines v. United States, 371 U.S. 156, 168 [83 S. Ct. 239, 246, 9 L. Ed. 2d 207] (1962). In reviewing that explanation, we must "consider whether the decision was based on a consideration of the relevant factors and whether there has been a clear error of judgment." Bowman Transp. Inc. v. Arkansas-Best Freight System, [419 U.S. 281, 285, 95 S. Ct. 438, 442, 42 L. Ed. 2d 447 (1974) ], Citizens to Preserve Overton Park v. Volpe, [401 U.S. 402, 416, 91 S. Ct. 814, 823, 28 L. Ed. 2d 136 (1971) ]. Normally, an agency rule would be arbitrary and capricious if the agency has relied on factors which Congress has not intended it to consider, entirely failed to consider an important aspect of the problem, offered an explanation for its decision that runs counter to the evidence before the agency, or is so implausible that it could not be ascribed to a difference in view or the product of agency expertise. The reviewing court should not attempt itself to make up for such deficiencies: "We may not supply a reasoned basis for the agency's action that the agency itself has not given." SEC v. Chenery Corp., 332 U.S. 194, 196 [67 S. Ct. 1575, 1577, 91 L. Ed. 1995] (1947). We will, however, "uphold a decision of less than ideal clarity if the agency's path may reasonably be discerned." Bowman Transp. Inc. v. Arkansas-Best Freight Systems, [419 U.S. at] 286 [95 S.Ct. at 442].

Motor Vehicle Manufacturers Association v. State Farm Mutual Automobile Insurance Company, --- U.S. ----, ---- - ----, 103 S. Ct. 2856, 2865-66, 77 L. Ed. 2d 443 (1983).

Our review of an agency's "observance of procedure required by law," 5 U.S.C. Sec. 706(2)(D) (1976), is more exacting. NRDC v. EPA, 683 F.2d 752, 760 (3d Cir.1982); see Weyerhaeuser, 590 F.2d at 1027-28. Under section 4 of the Administrative Procedure Act, an agency initiating informal rulemaking must first publish a general notice which includes "either the terms or substance of the proposed rule or a description of the subjects and issues involved." 5 U.S.C. Sec. 553(b)(3) (1976). Such notice must "fairly apprise interested persons" of the subjects and issues dealt with in the rule ultimately promulgated. American Iron & Steel Institute v. EPA ("AISI II"), 568 F.2d 284, 290-93 (3d Cir.1977); see Ethyl Corp. v. EPA, 541 F.2d 1, 48 (D.C.Cir.1976) (en banc), cert. denied, 426 U.S. 941, 96 S. Ct. 2663, 49 L. Ed. 2d 394 (1976). The agency must then give interested persons an opportunity to participate in the rulemaking through the submission of written comments. 5 U.S.C. Sec. 553(c) (1976). After considering the relevant comments submitted, the agency must incorporate in the promulgated rules "a concise general statement of their basis and purpose." Id. To ensure meaningful judicial review, the agency in that statement and in its supporting materials must articulate the rational basis for the choices it has made; however, as stated above, we "should not reverse an agency's decision that is not fully articulated where we can reasonably discern the basis for the agency's action." AISI I, 526 F.2d at 1047; see AISI II, 568 F.2d at 295-96.

Finally, we note that the Administrator's actions are entitled to a presumption of regularity. Citizens to Preserve Overton Park v. Volpe, 401 U.S. 402, 415, 91 S. Ct. 814, 823, 28 L. Ed. 2d 136 (1971). A party petitioning for review of an agency's regulations bears the burden of overcoming that presumption. Lewes Dairy v. Freeman, 401 F.2d 308, 316 (3d Cir.1968), cert. denied, 394 U.S. 929, 89 S. Ct. 1187, 22 L. Ed. 2d 455 (1969); accord Environmental Defense Fund v. Costle, 657 F.2d 275, 283 n. 28 (D.C.Cir.1981). If after adequate notice and opportunity to comment a petitioner claims on appeal that the agency overlooked technical, factual and policy issues not raised in comments before the agency, that petitioner will have less latitude in its complaints, Weyerhaeuser, 590 F.2d at 1028 n. 15, or in special circumstances will be barred altogether, AISI I, 526 F.2d at 1050; see American Frozen Food Institute v. Train, 539 F.2d 107, 134 (D.C.Cir.1976).

NRDC and all the other petitioners ("industrial petitioners") raise challenges to several provisions of the general pretreatment regulations, 40 C.F.R. Secs. 403.1.-16 (1982). We consider those challenges in the following order: (A) the definitions of "interference" and "pass through;" (B) the definition of "new sources;" (C) the FDF variance provision; (D) the removal credits provision; and (E) the combined wastestream formula.

A. The Definitions of "Interference" and "Pass Through"

Section 403.3 of the general pretreatment regulations defines "interference" and "pass through." 40 C.F.R. Sec. 403.3(i), (n) (1982). The industrial petitioners in their joint brief ("joint petitioners") and USBA contend that the breadth of the definitions of "interference" and "pass through" violates the Act because the definitions subject indirect dischargers to penalties without consideration of fault, causation or consequences. Joint petitioners argue that the definitions were improperly promulgated. We will grant the petitions for review in Nos. 81-1982, 81-1983, 81-1984, 81-2150, and 81-2151, and will remand the definition of both "interference" and of "pass through."

Section 307(b) of the Act directs the Administrator to promulgate pretreatment standards to prevent the discharge of any pollutant through a POTW which "interferes with, passes through or is otherwise incompatible with such works." 33 U.S.C. Sec. 1317(b)(1) (1976 amended Supp. I 1977). Under that mandate the Administrator not only has promulgated the categorical pretreatment standards setting numerical limits upon discharges from certain regulated categories of industrial sources, but has also established a general prohibition applying to all non-domestic indirect dischargers whether or not they are subject to categorical pretreatment standards. See 40 C.F.R. Sec. 403.5(a) (1982). That "prohibited discharge" standard contains a general prohibition of the introduction into a POTW of pollutants that "Pass Through a POTW or Interfere with the operation or performance of the works." Id. Sec. 403.5(a). The prohibited discharge standard also specifically prohibits the introduction into a POTW of pollutants that in several specified ways cause interference.10 Violation of the prohibited discharge standard is unlawful and renders the violator liable to suit by the Administrator, by the State, by the POTW, or by any adversely affected party. 33 U.S.C. Secs. 1317(d), 1319(b), (c), (f), 1342(b)(7), 1365(a) (1976 & Supp. I 1977). Violations may carry civil penalties of up to $10,000 per day, and criminal penalties of up to $25,000 per day and two years in prison. Id. Sec. 1319(c)(1), (d). In addition, if the violation is likely to recur the POTW is required to develop and enforce such specific effluent limits for its users as are necessary to ensure the POTW's future compliance with its NPDES permit. 40 C.F.R. Sec. 403.5(c)(2) (1982).

Section 403.3 provides the definition of "interference" as that term is used in the prohibited discharge standard. As originally promulgated in the 1978 general pretreatment regulations, section 403.3 defined "interference" as "an inhibition or disruption of a POTW's sewer system, treatment processes or operations which contributes to a violation of any requirement of [the POTW's] NPDES Permit." 43 Fed.Reg. 27747 (1978) (emphasis added). In 1979 the Administrator proposed to narrow the ambit of the definition by requiring an inhibition or disruption which "causes or significantly contributes" to the violation of the POTW's permit, and by including a "safe harbor" provision exempting from the definition inhibitions and disruptions caused by an indirect discharger "in compliance with specific prohibitions or standards developed by Federal, State or local governments." 44 Fed.Reg. 62,260, 62,265 (1979). As promulgated, however, the 1981 general pretreatment amendments omitted the safe harbor provision and defined "significantly contributes" using three numbered categories. 46 Fed.Reg. 9413 (1981). The amended regulations thus redefine "interference" as:

an inhibition or disruption of the POTW ... which is a cause of or significantly contributes to either a violation of any requirement of the POTW's NPDES permit (including an increase in the magnitude or duration of a violation) or to the prevention of sludge use or disposal by the POTW .... An industrial user significantly contributes to such a permit violation or prevention of sludge use or disposal ... whenever such User:

(1) Discharges a daily pollutant loading in excess of that allowed by contract with the POTW or by Federal, State or local law;

(2) Discharges wastewater which substantially differs in nature or constituents from the User's average discharge; or

(3) Knows or has reason to know that its Discharge, alone or in conjunction with Discharges from other sources, would result in a POTW permit violation or prevent sewage use or disposal ....

40 C.F.R. Sec. 403.3(i) (1982).

Joint petitioners allege that the present definition is contrary to the Act because it renders an indirect discharger liable for interference even though its discharges did not cause the POTW's permit violation or sludge problem. They posit that an industrial user may be held liable if discharging more than average or beyond its contract limit, even though it is the discharge of another user of the POTW, or a malfunction or mistake at the POTW itself, that actually causes the inhibition or disruption. Joint petitioners contend that Congress did not intend to subject indirect dischargers to liability without proof of causation.

EPA argues that joint petitioners have misread the definition. EPA urges that the definition requires that causation be shown before liability is established. In its brief EPA emphasizes that an industrial user's discharge must "lead to" or "give rise to" the inhibition or disruption. Brief for Respondent (No. 79-2256) at 125-27. At oral argument EPA's counsel asserted that to prove liability the Administrator must show that the discharge both caused the inhibition or disruption and fell within the three categories defining "significantly contributes." Transcript of Oral Argument at 133, 136. We cannot agree. The words "leads to" and "gives rise to" do not appear in the definition. Instead, the promulgated definition requires only that the discharge "is a cause of or significantly contributes," and defines "significantly contributes" by substituting three categories of discharger misconduct, at least two of which exclude any necessity for proving that the discharge caused the inhibition or disruption. 40 C.F.R. Sec. 403.3(i)(1), (2) (1982).11 If the Administrator has not written the definition to require causation, we cannot rewrite the definition to match the representations of counsel.12

Given that section 403.3(i)'s definition of interference does not require causation to establish liability, we must now consider whether liability without causation is within the intent of Congress. We find that neither the language of the Act nor the intent of Congress appears to contemplate liability without causation. First, sections 307(b) and (c) requires that pretreatment standards "prevent the discharge of any pollutant ..., which pollutant interferes with such works." Id. Sec. 1317(b), (c) (1976 & Supp. I 1977) (emphasis added). Section 307(c) explains that such standards must be promulgated "to insure that any source introducing pollutants into a [POTW] ... will not cause a violation of the effluent limitations of such treatment works." Id. Sec. 1317(c) (1976) (emphasis added).

Second, Congress made plain its intent that "[i]n no event is it intended that pretreatment facilities be required for compatible wastes as a substitute for adequate municipal waste treatment works." S.Conf.Rep. No. 1236, 92d Cong., 2d Sess. 130, reprinted in 1972 U.S.Code Cong. & Ad.News 3668, 3776, 3807; accord H.R.Rep. No. 911, 92d Cong., 2d Sess. 113, reprinted in Senate Comm. on Public Works, 93d Cong., 1st Sess., Legislative History of the Water Pollution Control Act Amendments of 1972, at 753, 800 (1973) [hereinafter cited as "1972 Legis.Hist."].13 If the inhibition or disruption is caused not by the industrial user's discharge but by a mistake or malfunction at the POTW, the industrial user will be punished for failing to substitute its own pretreatment for the POTW's impaired treatment. We do not think that Congress intended such liability. See also AISI I, 526 F.2d at 1056 (rejecting penalties for circumstances beyond discharger's control).

We conclude that given the language and purpose of the Act, an indirect discharge cannot be liable under the prohibited discharge standard unless it is a cause of the POTW's permit violation or sludge problem. If the definition of "interference" required that an indirect discharger be both "the cause of" and "significantly contribute to" the POTW's permit violation, it would be consistent with that causation requirement. As written, however, the definition fails to require such causation, and thus violates the clear meaning of the Act.14 We will therefore remand the entire definition of interference15 to the Administrator.16

Joint petitioners allege that the definition of pass through in 40 C.F.R. Sec. 403.3(n) (1982) was promulgated without the notice and comment required under section 4 of the Administrative Procedures Act, 5 U.S.C. Sec. 553(c) (1976). They point out that the amendments proposed in 1979 contained no suggestion that the Administrator intended to use or to define the term "pass through" in the general pretreatment regulations. See 44 Fed.Reg. 62,260-71 (1979). The Administrator nonetheless promulgated the definition of "pass through" in the 1981 general pretreatment amendments, justifying his failure to first propose the definition by saying that it was "almost identical" to the promulgated definition of interference. 46 Fed.Reg. 9416 (1981).

EPA now admits that the definition of "pass through" was promulgated without the notice and comment required by the Administrative Procedure Act. Brief for Respondent (No. 79-2256) at 132-33. EPA suggests that for that reason we should remand the definition to the Administrator; nevertheless, it contends that we are not barred from passing on the definition's substantive validity. Id. at 133 & n. *. We believe that it would be fruitless for us to review the definition before it has been submitted for public comment. We will therefore remand the definition of "pass through" in section 403.3(n) to the Administrator.17

B. Definition of "New Source"

"New source" is defined in section 403.3(k) of the general pretreatment regulations, 40 C.F.R. Sec. 403.3(k) (1982). Under that definition, if the Administrator fails to promulgate a new source pretreatment standard within 120 days of its publication, those sources whose construction began after the publication but before the promulgation of the proposed standard are not considered to be new sources. Petitioner NRDC argues that by excluding those sources the definition is inconsistent with the Act and is contrary to our holding in Pennsylvania Department of Environmental Resources v. EPA, 618 F.2d 991 (3d Cir.1980). We agree, and will accordingly grant NRDC's petitions for review in Nos. 81-1977 and 81-1985.18

Under section 307(c) of the Act, the Administrator must promulgate new source pretreatment standards for any indirect discharger that would be a "new source" under section 306 of the Act if it were a direct discharger. 33 U.S.C. Sec. 1317(c) (1976). Section 306(a)(2) defines a "new source" as

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. Sec. 1316(a)(2). Section 306(b)(1)(B) directs the Administrator to promulgate proposed standards of performance within 120 days after the publication of the proposed regulations. Id. Sec. 1316(b)(1)(B).

Section 403.3(k) of the general pretreatment regulations defines "new source" as any source whose construction commenced

[a]fter proposal of Pretreatment Standards in accordance with section 307(c) of the Act which are applicable to such source, but only if the Standards are promulgated in accordance with section 307(c) within 120 days of their proposal.

40 C.F.R. Sec. 403.3(k)(2) (1982) (emphasis added). If the standards are not promulgated within 120 days of their proposal, only those sources whose construction began after promulgation are considered "new sources." Id. Sec. 403.3(k)(1); see id. Sec. 403.6(b).

In Department of Environmental Resources we considered a similar definition of "new source" promulgated for a category of direct dischargers. We rejected EPA's definition as inconsistent with the basic policies of the Act. Congress, we found, "intended to subject as many firms as possible to the new source regulations." 618 F.2d at 999. By its plain meaning the definition of "new source" in section 306(a)(2) achieved that goal by subjecting to new source standards all businesses which initiated new construction after being put on notice by the publication of the proposed standards. We stated that if dischargers wished to limit their period of uncertainty by forcing the Administrator to promulgate proposed standards within section 306(b)(1)(B)'s 120-day deadline, the proper remedy was not the exemption of new construction from new source standards, but was a citizen suit under 33 U.S.C. Sec. 1365 (1976) seeking EPA compliance with the deadline. We therefore held that section 306(a)(2) had to be given its plain meaning, and we struck down the EPA's definition. 618 F.2d at 1000.

In this case EPA has conceded that the "new source" definition in section 403.3(k) is invalid under our holding in Department of Environmental Resources. Brief for Respondent (No. 81-1977) at 14.19 Intervenor CMA argues nonetheless that the definition is valid under our holding and under the Act. CMA, however, fails either to distinguish Department of Environmental Resources,20 or to proffer any arguments on the proper construction of section 306 which were not considered and rejected in that decision.21 We hold, therefore, that the definition of "new source" in section 403.3(k) is invalid. We will remand the definition to the Administrator.

C. The Fundamentally Different Factor Variance

Section 403.13 of the general pretreatment regulations permits the Administrator to grant a variance from a categorical pretreatment standard to an existing indirect discharger within the category if the Administrator, in establishing the categorical standard, has considered factors "fundamentally different" from the factors relating to that source. 40 C.F.R. Sec. 403.13 (1982). Petitioner NRDC contends that the FDF variance is not authorized by the Act and is specifically prohibited insofar as it would permit the discharge of toxic pollutants. We need not determine whether the Administrator has authority to issue FDF variances, for we agree that such variances may not be issued for toxic pollutants. We will therefore grant NRDC's petitions for review in Nos. 81-1977 and 81-1985.

Section 307(b) of the Act directs the Administrator to promulgate pretreatment standards for existing indirect dischargers by category or categories of sources. 33 U.S.C. Sec. 1317(b)(1), (3) (1976 & Supp. I 1977). As he has chosen to regulate existing indirect dischargers in an analogous manner to direct dischargers, the Administrator bases the categorical pretreatment standards on the BPT and BAT levels of control technology set forth for direct dischargers in section 301(b) of the Act. Id. Sec. 1311(b). The Administrator determines those levels for existing indirect dischargers by considering the factors specified in section 304(b). Id. Sec. 1314(b).22

The fundamentally different factor variance in section 403.13 is also adopted from the regulatory scheme governing direct dischargers. Under the Consolidated Permit Regulations, 40 C.F.R. Secs. 125.30.-32 (1982), existing direct dischargers may obtain FDF variances from BPT and BAT effluent limitations. Terming the concept equally applicable to pretreatment standards, the Administrator modeled the FDF variance provision for existing indirect dischargers after the FDF variance provision for direct dischargers. See 46 Fed.Reg. 9435-36 (1981); 44 Fed.Reg. 62,264-65 (1979); 43 Fed.Reg. 27,73 8 (1978); 42 Fed.Reg. 6481 (1977).

The purpose of the FDF variance provision for indirect dischargers is stated in section 403.13(b):

In establishing categorical Pretreatment Standards for existing sources, the EPA will take into account all the information it can collect, develop and solicit regarding the factors relevant to pretreatment standards under section 307(b). In some cases, information which may affect these Pretreatment Standards will not be available or, for other reasons, will not be considered during their development. As a result, it may be necessary on a case-by-case basis to adjust the limits in categorical Pretreatment Standards ... as they apply to a certain Industrial User within an industrial category or subcategory.

40 C.F.R. Sec. 403.13(b) (1982); see id. Sec. 125.30(b) (near-identical statement of purpose). Indirect dischargers, POTWs and other interested parties may request that an indirect discharger receive a variance. Id. Sec. 403.13(a), (b) (1982). Variances can be used to establish limits more or less stringent than that specified by the applicable categorical pretreatment standard. See id. Sec. 403.13(c)(2), (3). An industrial user seeking to obtain a discharge limit less stringent than required by the categorical standard must establish that the alternative limit is justified by factors relating to the discharge regulated by the categorical pretreatment standard which are fundamentally different from the factors considered by the Administrator in establishing the standard. Id. Sec. 403.13(b), (c)(1)(ii), (c)(2). In delineating the factors to be considered fundamentally different, section 403.13 includes most of the factors which section 304(b) directs the Administrator to consider in determining BPT and BAT standards. Id. Sec. 403.13(d); see id. Sec. 403.13(e).

NRDC argues that FDF variances from BPT and BAT pretreatment standards are contrary to the Act. NRDC correctly notes that while Congress expressly provided for modification of other discharge limits,23 the Act does not explicitly authorize FDF variances from the categorical pretreatment standards. EPA contends that the Act implicitly authorizes FDF variances for indirect dischargers, and relies on the approval given to the FDF variances for direct dischargers in E.I. du Pont de Nemours & Co. v. Train, 430 U.S. 112, 97 S. Ct. 965, 51 L. Ed. 2d 204 (1978).

In du Pont the Supreme Court held that the Administrator had to provide for variances for direct dischargers from BPT effluent limitations. Id. at 128, 97 S.Ct. at 975; see EPA v. National Crushed Stone Association, 449 U.S. 64, 72 & n. 12, 101 S. Ct. 295, 301 & n. 12, 66 L. Ed. 2d 268 (1980). The Court found that section 301(b)(1) required that "some allowance [be] made for variations in individual plants" under categorical BPT effluent limitations because that section spoke of "effluent limitations for point sources," 33 U.S.C. Sec. 1311(b)(1)(A) (1976), rather than "effluent limitations for categories and classes of point sources," id. Sec. 1311(b)(2)(A) (1976 & Supp. I 1977). 430 U.S. at 128, 97 S. Ct. at 975. As section 307(b) states that pretreatment standards apply to "categories of sources," id. Sec. 1317(b)(3) (1976), the Administrator is not required under du Pont to make any provision for variances from pretreatment standards.

Agreeing that a variance provision is not required, EPA asserts that the Administrator in his discretion may permit FDF variances from the pretreatment standards as an appropriate means to ensure that the categorical standards are not applied inequitably to a particular discharger. See NRDC v. EPA, 537 F.2d 642, 646-47 (2d Cir.1976).24 We need not consider whether the Administrator possesses the inherent authority to provide for variances from categorical pretreatment standards, however, because we find another of NRDC's contentions dispositive. NRDC, noting that the FDF variance provision is drawn to permit variances from pretreatment standards for toxic pollutants, argues that variances for toxic pollutants are forbidden by section 301(1 ) of the Act, 33 U.S.C. Sec. 1311(1 ) (Supp. I 1977). We agree.25

The elimination of the discharge of toxic pollutants has always received special emphasis under the Act. Id. Sec. 1251(a)(3) (1976); see id. Sec. 1362(13). In 1972 Congress directed the Administrator to list and develop effluent limitations for toxic pollutants under section 307(a) of the Act. Id. Sec. 1317(a). The discharge of toxic pollutants generated even greater congressional concern in 1977. E.g., Senate Committee on Environment and Public Works, 95th Cong., 2d Sess., Legislative History of the Clean Water Act of 1977, at 326 (1978) (statement of Cong. Roberts) [hereinafter cited as "1977 Legis.Hist."]; id. at 454 (statement of Sen. Muskie). In section 53 of the 1977 Amendments, Congress itself added toxic pollutants to the Administrator's list and required that he promulgate BAT effluent limitations for those pollutants by 1980.26 Section 53 also added subsection (1) to section 301:

(l) The Administrator may not modify any requirement of this section as it applies to any specific pollutant which is on the toxic pollutant list under section 1317(a)(1) of this title.

33 U.S.C. Sec. 1311(l) (supp. i 1977).

EPA does not dispute that the pretreatment standards mandated by section 307(b) are a "requirement" of section 301.27 Instead EPA argues that "modification" is a term of art in the Act, and that FDF variances are not modifications of a pretreatment standard but are simply the creation of a more appropriate standard based on factors previously overlooked by the Administrator. Under the Administrator's construction, section 301(l) deprives the Administrator only of his authority to "modify" BAT standards under section 301(c) and (g), 33 U.S.C. Sec. 1311(c), (g) (1976 & Supp. I 1977).28

The legislative history of section 301(1 ) does indicate that Congress was primarily concerned with prohibiting modifications under section 301(c) and (g).29 Nonetheless, it does not appear that Congress used "modification" as a term of art so as to exclude variance provisions from the proscription of section 301(1 ). Spokesmen for the 1977 Amendments used the terms "waiver" and "modification" interchangeably. 1977 Legis.Hist. 328-29 (statement of Rep. Roberts); id. at 458 (statement of Sen. Muskie). More important, Senator Muskie termed section 301(c) a "variance" provision. 1977 Legis.Hist. 461. As "modification" is thus not a term of art, section 301(1 ) includes variances in its broad prohibition.

EPA's attempt to distinguish the policy behind FDF variances from the policies behind the "modification" provisions is equally unsuccessful. The Supreme Court has stated that section 301(c)'s modifications of BAT limits serve the same function as FDF variances of BPT limits:

A Sec. 301(c) variance, thus, creates for a particular point source a BAT standard that represents for it the same sort of economic and technological commitment as the general BAT standard creates for the class.

National Crushed Stone, 449 U.S. at 74, 101 S.Ct. at 302. If Congress was willing to prohibit section 301(c) modifications where toxic pollutants are concerned, it is difficult to imagine why Congress would have permitted similar FDF variances for those same pollutants.

In Appalachian Power Co. v. Train, 620 F.2d 1040 (4th Cir.1980), NRDC argued that section 301(1 ) prohibited FDF variances from BPT effluent limitations for toxic pollutants. That court deferred to the Administrator's construction of the Act and upheld the FDF variance provision, remarking that "the best that can be said for Sec. 301(1 ) is that it is not clear." Id. at 1046-48. Because we find that section 301(1 ) is clear, we must disagree. Section 301(1 ) forbids modifications, and FDF variances are no less modifications than those provisions indisputably prohibited by that section. Given the clear congressional concern throughout the 1977 Amendments for discharges of toxic pollutants, we hold that FDF variances for toxic pollutant discharges are forbidden by the Act. We will therefore remand the FDF variance provision.

Section 403.7 of the general pretreatment regulations establishes the criteria and procedures by which a POTW may revise an indirect discharger's numerical discharge limit for a pollutant, as set in its categorical pretreatment standard, to reflect the POTW's removal of that pollutant. 40 C.F.R. Sec. 403.7 (1982). Joint petitioners, Interlake and CACI, argue that section 403.7 exceeds the Administrator's authority under section 307(b)(1) of the Act, is unworkable, and was improperly promulgated. We disagree, and will deny the petitions for review on this issue.

Section 307(b) of the Act authorized the Administrator to establish pretreatment standards for any pollutant that "interferes with, passes through, or otherwise is incompatible" with POTWs. 33 U.S.C. Sec. 307(b)(1) (1976). In enacting that section Congress indicated that pretreatment of compatible pollutants may not be necessary, and added that pretreatment should not be required as a substitute for adequate treatment by POTWs.30 In a further effort "to avoid treatment for treatment's sake," 1977 Legis.Hist. 343 (statement of Rep. Roberts), Congress in section 54(a) of the 1977 Amendments to the Act added a sentence to section 307(b)(1):

If, in the case of any toxic pollutant under subsection (a) of this section introduced by a source into a publicly owned treatment works, the treatment by such works removes all or any part of such toxic pollutant and the discharge from such works does not violate that effluent limitation or standard which would be applicable to such toxic pollutant if it were discharged by such source other than through a publicly owned treatment works, and does not prevent sludge use or disposal by such works in accordance with section 405 of this Act, then the pretreatment requirements for the sources actually discharging such toxic pollutant into such publicly owned treatment works may be revised by the owner or operator of such works to reflect the removal of such toxic pollutant by such works.

Pub.L. No. 95-217 Sec. 54(a), 91 Stat. 1591 (amending 33 U.S.C. Sec. 1317(b)(1) (Supp. I 1977)). The legislative history of the section made clear that "[i]n promulgating national pretreatment standards the Administrator shall include a provision recognizing the option of [a POTW] to modify the requirements to reflect the degree of reduction achieved by the treatment works." H.R.Conf.Rep. No. 830, 95th Cong., 1st Sess. 88, reprinted in 1977 U.S.Code Cong. & Ad.News 4424, 4463.

In the removal credits provision the Administrator has set conditions and procedures for such revision of categorical pretreatment standards. 40 C.F.R. Sec. 403.7 (1982). To be eligible to grant revisions to reflect the toxic pollutants it removes, a POTW must first have a pretreatment program approved by the responsible Approval Authority.31 Id. Sec. 403.7(b)(2).32 The POTW must then obtain authorization from the Approval Authority to revise the discharge limits for specific pollutants. Id. Sec. 403.7(b)(1). To obtain authorization the POTW must demonstrate "consistent removal" of each pollutant sufficient to justify the proposed revision. Id. Sec. 403.7(b); see id. at 403.7(a)(1), (2). If once a year or more untreated wastewaters overflow before they reach the POTW and thus bypass the POTW's treatment process, the POTW either must show that its indirect dischargers compensate for the overflows, or it must reduce the amount of consistent removal claimed. Id. Sec. 403.7(b)(3); see id. Sec. 403.7(a)(3). The POTW must also show that the revision will not prevent it from meeting applicable sludge management requirements. Id. Sec. 403.7(b)(4). Once authorization for the revision has been granted, the POTW must monitor and report semiannually on its capability to remove the specified pollutants. Id. Sec. 403.7(f)(1); see id. Secs. 403.7(d), 403.12(i), (j). If the Approval Authority determines that the discharge limit revision no longer meets the requirements of section 403.7, or is significantly contributing to a violation of the POTW's NPDES permit, the Approval Authority after an opportunity for corrective action may withdraw or modify the revision. Id. Sec. 403.7(f)(5).

Joint petitioners challenge the Administrator's authority under the Act to mandate that POTWs must have approved pretreatment programs before they may grant removal credits. See id. Sec. 403.7(b)(2). Joint petitioners first correctly observe that section 307(b)(1) does not expressly impose such a condition.33 They then claim that the Administrator has improperly transplanted that condition from section 402(b)(8) of the Act, 33 U.S.C. Sec. 1342(b)(8) (Supp. I 1977).

Section 402(b) sets the terms, conditions and requirements for permits issued under federal and state NPDES permit programs. Id. Sec. 1342(a)(3), (b) (1976 & Supp. I 1977). As amended in 1977, section 402(b)(8) authorizes the Administrator to insure that a POTW's permit includes conditions to require "a program to assure compliance with [section 1317(b) ] pretreatment standards by each [significant] source" introducing regulated pollutants into the POTW. Pub.L. No. 95-217 Sec. 54(c), 91 Stat. 1591 (amending 33 U.S.C. Sec. 1342(b)(8) (Supp. I. 1977)). The amended section 402(b)(8) and the removal credits provision were both added by section 54 of the 1977 amendments, and the legislative history makes clear that the two provisions are closely related. The conference report and spokesmen in the House stated that the conferees had added the provision to allow a POTW to revise pretreatment standards to reflect removal "in applying these pretreatment standards through its pretreatment program." H.R.Conf.Rep. No. 830, 95th Cong., 1st Sess. 87, reprinted in 1977 U.S.Code Cong. & Ad.News 4424, 4462; 1977 Legis.Hist. 342-43 (statement of Rep. Roberts); id. at 403 (statement of Rep. Anderson). Senator Muskie, the legislation's sponsor, informed the Senate that the new provision permitted POTWs to grant removal credits "[w]here a local compliance program is approved." 1977 Legis.Hist. 461. He explained:

Tying local [removal] credits to local compliance programs not only provides an incentive for local participation, but more importantly, it provides assurance that the removal levels which justified the local credits will be maintained by a publicly-owned treatment works committed to a sound pretreatment program.

Id. at 462. In light of this persuasive legislative history,34 we believe that the Administrator may require an approved pretreatment program as a condition upon a POTW's grant of removal credits.

Petitioner CACI, emphasizing that Congress in section 307(b)(1) authorized POTWs, not EPA, to grant removal credits, claims the Administrator may not require that POTWs obtain his authorization for each proposed removal credit. There is support, however, for such an authorization requirement in the legislative history. Senator Muskie stated that "EPA and the [states issuing NPDES permits] may approve case-by-case modifications of the national pretreatment standards," and listed several conditions the EPA might place on its authorization. 1977 Legis.Hist. 461. Moreover, the Administrator's authorization fits within the scheme of the Act as established by section 54 of the 1977 Amendments. As noted above, the Administrator may require that a POTW seeking to grant removal credits have an approved pretreatment program to assure compliance by its indirect dischargers with the section 1317(b) pretreatment standards. Section 54 also allowed the Administrator to bring an action to compel the POTW to enforce the pretreatment standards under its program. Pub.L. No. 95-217 Sec. 54(b), 91 Stat. 1591 (adding 33 U.S.C. Sec. 1319(f) (Supp. I 1977)). Together those provisions endow the Administrator with the power to deny authorization to a POTW's dispensation of removal credits. CACI acknowledges that power, but argues that the Administrator must set the conditions on his authorization by litigation rather than regulation. See Air Reduction Co. v. Hickel, 420 F.2d 592 (D.C.Cir.1968). We find it hard to believe that Congress required such a piecemeal approach. See Weinberger v. Hynson, Westcott & Dunning, Inc., 412 U.S. 609, 624-26, 93 S. Ct. 2469, 2480-81, 37 L. Ed. 2d 207 (1973). Section 501(a) of the Act empowers the Administrator "to prescribe such regulations as are necessary to carry out his functions under this chapter." 33 U.S.C. Sec. 1361(a) (1976); see du Pont, 430 U.S. at 132, 97 S.Ct. at 977; see also 33 U.S.C. Sec. 1251(d) (1976). Since such regulations would not deprive POTWs of the sole ability to grant, and the ultimate power to deny, removal credits, we conclude that under section 501(a) the Administrator may express the conditions on his authorization of removal credits in binding regulations.

Petitioners place more emphasis on their contention that the removal credits provision is simply unworkable. Joint petitioners raise two specific defects that they claim render the provision unworkable. First, they attack section 403.7(f)(5), which permits the Administrator to withdraw or modify removal credits if semiannual data reveals that the POTW issuing the credits is no longer attaining its predicted removal. Joint petitioners say that due to section 403.7(f)(5) they will be unable to rely on their removal-revised discharge limits and will be forced to install just as much control technology as if there were no removal at all. We agree with EPA, however, that such withdrawn or modified discharge limits, though unfortunate, are merely the recognition of the POTW's failure to remove the pollutant. By requiring such modifications, the Administrator prevents the granting of removal credits for toxic pollutants which the POTW simply discharges into navigable waters. Such a requirement is consistent with the mandate of section 307(b)(1) that any revision "reflect the removal of such toxic pollutants by such works." 33 U.S.C. Sec. 1317(b)(1) (Supp. I 1977). It is also consonant with the legislative history requiring "documented pollutant removals" and "a demonstration that the pollutant is degraded or treated," 1977 Legis.Hist. 461 (statement of Sen. Muskie).

Second, joint petitioners challenge the requirement in section 403.7(b)(3) that a POTW unable to prevent toxic overflows must reduce the amount of removal claimed in proportion to the number of hours of overflow. Joint petitioners claim that POTWs will be unable to make verifiable engineering estimates of the hours of overflow, and will thus be unable to grant removal credits. As the Administrator notes, however, section 403.7(b) simply implements the statutory requirement that removal credits be granted only for pollutants actually removed by the POTW. Moreover, a POTW unable to estimate the time, let alone the amount, of untreated wastewater overflow may not be able to accurately predict the proportion of pollutants which it will remove. Requiring such an estimate thus has a rational basis under the Act.

Joint petitioners and Interlake also make a generalized claim that the removal credits provision is unworkable. Such a generalized claim is necessarily less persuasive than a claim detailing the alleged errors made by the Administrator. We have nonetheless reviewed the bases cited by petitioners for the claim of unworkability. We find nothing in those sources that would cause us to invalidate the regulations as unworkable.35 Joint petitioners have thus failed to document their general assertion so as to overcome the presumption of regularity in the Administrator's conduct. We are accordingly not convinced that the regulations can be declared arbitrary and capricious as "unworkable." See AISI I, 526 F.2d at 1049, 1064.36 Accordingly, the petitions for review will be denied as to this issue.

In the general pretreatment regulations, section 403.6(e) establishes a formula to adjust the discharge limit set by a categorical pretreatment standard where the wastestream regulated by that pretreatment standard is combined with other wastewaters prior to pretreatment by the indirect discharger. 40 C.F.R. Sec. 403.6(e) (1982). Petitioner Ford, along with joint petitioners, argues that the very concept of such a formula is inconsistent with the structure of the Act. Joint petitioners, Interlake and GM argue that because the standards for a combined wastestream will change each time EPA regulates a process which contributes to the stream, the formula will lead to an arbitrary and capricious "moving target." Those petitioners also contend that the formula is invalid because EPA failed to consider the cost and feasibility of treating such combined wastestreams. Interlake makes that argument with reference to the iron and steel industry, and also contends that the formula is void for vagueness. Finally, GM asserts that the formula was improperly promulgated. Given our construction of the formula, however, we find nothing in those challenges that requires the invalidation of the formula.

Section 307(b) of the Act directs the Administrator to regulate discharges, not pollutant by pollutant, but by categories of sources. 33 U.S.C. Sec. 1317(b)(3) (1976). The Administrator establishes such categorical pretreatment regulations by "specific industrial subcategories." 40 C.F.R. Sec. 403.6 (1982); see 42 Fed.Reg. 6476 (1977). The Administrator has established such industrial categories by the process or operation used, rather than by the overall nature of an industrial facility. E.g., 40 C.F.R. Sec. 413.01(a) (1982) (applicable to "electroplating operations"); id. Sec. 420.01(a) (applicable to "production operations in the Iron and Steel Point Source Category"); see 46 Fed.Reg. 9419 (1982). Consequently, it is possible for a diversified industrial facility to have several different processes producing different wastestreams that are not regulated by the same categorical standard, or are not regulated at all. Id. Such a facility may segregate the wastestreams from each process and separately pretreat them, or it may combine some or all of its wastestreams prior to pretreatment (an "integrated" facility). See id. Similarly, an industrial facility may discharge diluting streams, such as cooling water, that it could segregate from or combine with its regulated wastestreams before pretreatment. The combination of streams obviously complicates the task of setting categorical pretreatment standards. As the Administrator has recognized, however, "[s]eparate treatment of wastes at an integrated plant can be costly, wasteful of energy, inefficient and environmentally counter-productive." Id. at 9420.

The difficulty of establishing national pretreatment standards for the universe of industrial sources is compounded by the way in which the level of pollutants in a discharge is measured. For most pretreatment standards the Administrator has decided to set numerical limits on the concentration of pollutants in the discharge.37 Of course, the concentration of pollutants in a wastestream can be reduced without actually reducing the amount of pollutants discharged simply by adding water.38 At the same time, however, the pretreatment of non-regulated, diluting wastestreams mixed with regulated streams may remove pollutants from the non-regulated streams that would otherwise go without pretreatment. 46 Fed.Reg. 9420 (1981).

The Administrator has attempted to strike a balance between those considerations by promulgating the combined wastestream formula. Id. The formula, applied when the wastestream from a regulated process is mixed prior to pretreatment with other wastewaters, derives from the numerical limit for the regulated wastestream an equivalent limit for the combined stream, weighted to reflect the relative flows of the contributing wastestreams ("flow-weighted"). As originally explained in the Administrator's National Pretreatment Strategy, and as later proposed for promulgation, the formula assumed that there was only one regulated process contributing to the combined stream, and that the non-regulated stream(s) contained no pollutants. Id.; see 44 Fed.Reg. 62,266 (1979); 43 Fed.Reg. 27,762 (1978). After receiving substantial public comment the Administrator recognized that those assumptions made combined pretreatment of wastestreams impracticable "by creating combined stream limits that were technically unattainable in most instances." 46 Fed.Reg. 9420 (1981). The Administrator promulgated a revised formula "to minimize the need for separation of wastestreams" while protecting against dilution. Id. Under the promulgated formula an alternative discharge limit is derived for the combined wastestream by considering the flow-weighted categorical concentration limit of each regulated stream, as well as the flow of any "dilute" streams. 40 C.F.R. Sec. 403.6(e)(1)(i) (1982).39 The formula has three basic effects. See 46 Fed.Reg. 9477 (1981). First, if a regulated stream is combined with a dilute stream, the concentration limit for the regulated pollutant becomes more stringent in proportion to the dilution. Second, if a regulated stream is combined with another regulated stream with different concentration limits for the same pollutant, the concentration limit for the regulated pollutant in the combined stream will be somewhere in between the two limits, in proportion to the flows and limits of two regulated streams. Third, if a regulated stream is combined with an non-regulated but non-dilute stream (an "unregulated" stream), the concentration limit for the regulated pollutant in the combined stream stays unchanged. Of course, if more than one of these combinations occurs, the effects are also combined.

We note at the outset that behind the promulgated formula's three effects lie three assumptions. First, the formula assumes that dilute streams as defined in 40 C.F.R. Sec. 403.6(e) (1982) are free of the regulated pollutant. 46 Fed.Reg. 9421 (1982). Second, the formula presumes that two regulated streams are just as pretreatable combined as they are segregated--that is, that they do not interfere with each other's pretreatment processes. Third, the formula assumes that unregulated streams are just as pretreatable as regulated streams.

Ford makes an assertion which, though raised in the context of the categorical electroplating regulations, would undermine the basic rationale advanced for the combined wastestream formula as applied to most indirect dischargers. Ford, with joint petitioners, contends that Congress intended that the Administrator regulate whole plants, not operations or processes, by industrial category. Thus, Ford asserts, the Administrator must regulate an integrated automobile manufacturing plant that combines its wastestreams before pretreatment, not by establishing separate categorical standards for its electroplating, rubber processing, iron and steel, etc., operations and then using the formula to create an alternative discharge limit for the combined stream, but by promulgating a single pretreatment standard for the facility, without any use of the formula. Ford analogizes to diverse though tangential sources to show that Congress intended that the Administrator was to regulate whole plants, but refers us to nothing which indicates that a whole plant can be subjected to only one categorical standard no matter how many processes are employed.40 Absent some indication that this was Congress' intent, we will defer to the Administrator's interpretation.

Joint petitioners, GM and Interlake put more emphasis on their assertion that the Administrator's power to promulgate pretreatment standards for each process does not allow him to impose such standards one by one on a single facility. Petitioners correctly point out that each time an unregulated contributing stream becomes regulated, application of the formula will change the combined wastestream's alternative discharge limit. Petitioners urge that as a result their facilities will be required to adjust to a "moving target," denying them finality and rendering the planning and construction of control technology impossible. EPA argues that the moving target is not the fault of the formula, but is inevitable where an agency of limited resources must promulgate standards for numerous categories which must also apply to integrated facilities.

We agree with petitioners that the "moving target" is not an inevitable dilemma but is the result of the Administrator's choice to regulate process-by-process rather than by a method which treated each industrial facility as an indivisible unit. See also 33 U.S.C. Sec. 1317(b)(1) (1976) (directing promulgation of pretreatment standards within 270 days). The Administrator's choice may well affect the costs and attainability of each categorical standard imposed on integrated indirect dischargers, and the Administrator must take such effects into account. We do not believe, however, that the existence of the moving target problem necessarily renders the Administrator's choice, and the combined wastestream formula, arbitrary, capricious or an abuse of discretion. The Administrator must regulate a vast array of indirect dischargers within the periods specified in the Act. The regulation of the variety of integrated facilities combining wastestreams of diverse character inevitably complicates the Administrator's task. Although petitioners offer several approaches the Administrator might have adopted,41 we cannot substitute our judgment for that of the Administrator. The process-by-process or "building block" approach may make up in relative simplicity and uniformity what it lacks in predictability.42 We cannot say that that approach, or the formula that implements it, lacks a rational basis.43

3. Attainability and Cost of Combined Pretreatment

Joint petitioners, Interlake and GM also contend that in promulgating the formula the Administrator must consider the pretreatability of combined wastestreams, and the cost of such pretreatment. They note that if a combined wastestream does not fit the formula's assumptions,44 the formulagenerated alternative discharge limit may be unattainable or more costly. If the Administrator has failed to consider the effluent reduction attainable or the cost, they argue, then the alternative discharge limits generated by the formula are invalid. They claim that as a result the formula itself is arbitrary and capricious.

Section 304(b) of the Act directs the Administrator in setting BPT and BAT limits to "identify ... the degree of effluent reduction attainable through the application" of the control technology, and to consider the cost of applying that technology. 33 U.S.C. Sec. 1314(b)(1), (2) (1976). Section 306(b) of the Act requires the Administrator to consider the same factors in setting BDT. Id. Sec. 1316(a), (b). When the Administrator sets pretreatment standards using the BPT, BAT, or BDT levels of technology, he must consider those statutorily-relevant factors for the wastestreams he regulates, whether they are segregated or combined. See generally AISI II, 568 F.2d at 304-05. Thus, the Administrator at some point must consider the effluent reduction attainable by pretreatment of combined wastestreams, and the cost of attaining that reduction.45

EPA admits that in promulgating the combined wastestream formula the Administrator "did not consider--in fact could not have taken into account--every relevant factor for every category." Brief for Respondent (No. 79-2256) at 78; see 46 Fed.Reg. 9422 (1982) (stating insufficiency of data). EPA urges instead that the questions of the attainable effluent reduction and the attainment cost of combined wastestreams "are best addressed in the individual categorical rulemaking" which sets the numerical discharge limits that are inserted into the formula. Brief for Respondent (No. 79-2256) at 78. Consequently, EPA argues that those questions are not ripe for judicial review after the promulgation of the combined wastestream formula, but must be considered only in reviewing the categorical standards applicable to combined wastestreams.

To determine whether a challenge to an administrative regulation is ripe for review,

a two-fold inquiry must be made: first to determine whether the issues tendered are appropriate for judicial resolution, and second to assess the hardship to the parties if judicial relief is denied at that stage.

Toilet Goods Association v. Gardner, 387 U.S. 158, 162, 87 S. Ct. 1520, 1523, 18 L. Ed. 2d 697 (1967); see Abbott Laboratories v. Gardner, 387 U.S. 136, 148-56, 87 S. Ct. 1507, 1515-19, 18 L. Ed. 2d 681 (1967). Applying that test, we believe that the issues raised by petitioners are not ripe for review.

First, the issues are not appropriate for judicial resolution in our review of the formula itself. We cannot determine whether the Administrator has properly considered the attainable effluent reduction and attainment cost of a combined wastestream's alternative discharge limit until that limit has been generated by the formula. The formula cannot generate an alternative limit for the combined wastestream until the Administrator promulgates a categorical standard setting numerical discharge limits for one or more of the process wastestreams contributing to the combined stream. Thus, the promulgation of a categorical standard provides "further factual amplification" necessary to decide the attainability and cost of an alternative discharge limit. Hooker Chemical Co. v. EPA, 642 F.2d 48, 52 (3d Cir.1981).46 It is only in our review of such categorical standards that we can resolve petitioners' claims.

Of course, once the Administrator promulgates a categorical standard applicable to a process stream contributing to a discharger's combined stream, the issues become appropriate for judicial resolution. Each such standard, by setting a new numerical discharge limit on the contributing stream, will result in a new alternative discharge limit for the combined stream. Because that alternative limit is then enforceable against the discharger, see 46 Fed.Reg. 9420 (1982), it must be based on a consideration of the relevant factors, including attainability and cost. Dischargers petitioning for review of the categorical standard may then argue that the Administrator failed to consider the cost and attainability of the alternative discharge limit, and the courts can resolve their claims.

Second, we do not believe that petitioners will suffer hardship if we now deny judicial review of their contentions. As noted above, petitioners will be able to seek review of the Administrator's consideration of attainability and cost each time the Administrator promulgates a categorical standard resulting in a new alternative discharge limit generated by the formula. They may then argue that that alternative limit is not based on a consideration of those relevant factors.47 Until such categorical standards are promulgated, no enforceable alternative limit exists and the impact on the petitioners is not "sufficiently direct and immediate." A.O. Smith Corp. v. FTC, 530 F.2d 515, 522 (3d Cir.1976) (quoting Abbott Laboratories, 387 U.S. at 152, 87 S.Ct. at 1517). As to future categorical standards, then, petitioners' arguments are merely postponed to another day.48

Petitioners claim, however, that without judicial review of the formula they will suffer hardship because they are subject to already-promulgated categorical standards. Interlake contends that in promulgating the new Iron and Steel Manufacturing categorical pretreatment regulations, 40 C.F.R. Secs. 420.01-.127 (1982), the Administrator failed to consider the attainability or cost of the alternative discharge limits generated by the combined wastestream formula. See 47 Fed.Reg. 23,267 (1982).49 Similarly, Ford challenges the Administrator's consideration of cost and attainability of combined pretreatment in promulgating the electroplating standards.

EPA argues that the challenges to the attainability and cost of the alternative discharge limits generated using the electroplating and iron and steel manufacturing pretreatment standards cannot be addressed in our review of the formula, but must be raised in the review of those categorical standards. First, EPA notes that Interlake is presently seeking review of the iron and steel standards in this court. National Steel Corp. v. EPA, Nos. 82-3225 et al. (3d Cir. filed June 10, 1982). EPA states that Interlake may raise its challenge in those cases. Brief for Respondent (No. 79-2256) at 93. Similarly, EPA asserts that Ford must press its arguments in its appeal from the denial of its petition to reconsider the categorical electroplating standards. See Ford Motor Co. v. EPA, 718 F.2d 55 (3d Cir.1983). EPA adds that if relief is warranted in those cases, the proper remedy would be the vacation of the categorical standards, not of the combined wastestream formula.

We agree with EPA that Interlake's challenges to the formula-generated alternative discharge limits for the iron and steel industry can be raised in our review of the categorical standards for iron and steel manufacturing. Similarly, we agree that Ford can challenge the alternative discharge limits for the electroplating industry in its appeal from the denial of its petition for reconsideration of the categorical electroplating standards. Those petitioners will then have the opportunity to question whether the Administrator properly considered the cost and attainability of those alternative limits. Because Interlake and Ford thus will not suffer hardship by our failure to review their challenges to the formula itself, we conclude that their challenges are not now ripe. We will therefore deny their petitions for review of the formula.

The categorical pretreatment regulations establish numerical limits, based on BPT-level technology, upon the discharge of certain pollutants by electroplating operations. 40 C.F.R. Secs. 413.01-.84 (1982). Petitioners make several challenges to these standards. First, Ford contends that the methodology behind the pretreatment standards is fatally flawed. Second, NAMF contends that the standards are not economically achievable for job shops and are thus arbitrary and capricious.50 Third, GM asserts that the June 30, 1984 compliance date for integrated electroplaters is arbitrary and capricious.

Ford contends that the Administrator has improperly calculated the discharge limits attainable using BPT. We disagree, and will deny Ford's petition for review.

In promulgating the electroplating standards the Administrator has adopted the BPT level of technology from section 301(b)(1)(A), 33 U.S.C. Sec. 1311(b)(1)(A) (1976). The requirements for determining BPT limits are set forth in section 304(b)(1), which directs the Administrator to "identify, in terms of amounts of constituents and chemical, physical, and biological characteristics of pollutants, the degree of effluent reduction attainable through the application of the best practicable control technology currently available for classes and categories of point sources." 33 U.S.C. Sec. 1314(b)(1) (1976). The stringency required by BPT is indicated in the legislative history:

The Administrator should establish the range of "best practicable" levels based upon the average of the best existing performance by plants of various sizes, ages, and unit processes within each industrial category.

1972 Legis.Hist. 170 (statement of Sen. Muskie); see S.Rep. No. 414, 92d Cong., 1st Sess. 50, reprinted in 1972 U.S.Code Cong. & Ad.News 3668, 3716.51 Unless the present practices of all sources in the category are "uniformly inadequate," 1972 Legis.Hist. 169-70 (statement of Sen. Muskie), "the average of the best" is a measure of BPT. Hooker Chemical & Plastics Corp. v. Train, 537 F.2d 620, 633 (2d Cir.1976); American Meat Institute v. EPA, 526 F.2d 442, 453 (7th Cir.1975); see National Crushed Stone, 449 U.S. at 76 & n. 15, 101 S. Ct. at 303 & n. 15; AISI I, 526 F.2d at 1057.

To set the BPT electroplating standards the Administrator first determined the pollutants of concern, e.g., cadmium, lead, cyanide, hexavalent and trivalent chromium, copper, nickel, and zinc. After determining the BPT pretreatment technology used by the average of the best plants, the Administrator used his sampling data52 to determine the effluent reductions achievable by such plants using that technology. The Administrator first derived a "long-term average effluent concentration" for each regulated pollutant, which represented the expected effluent concentration attainable over a year or more by using the best practicable technology. Because even plants using the best practicable technology experience routine fluctuations in their effluent concentration, the Administrator calculated "variability factors" representing the percentage increase normally occurring during one- and thirty-day periods. The Administrator then multiplied the long-term concentration average by the respective variability factors to obtain the one- and thirty-day pretreatment standards for each pollutant he regulated.

Ford challenges the data and methodology used by the Administrator in his calculations. Under the arbitrary and capricious standard our deference to the agency is greatest when reviewing technical matters within its expertise. In particular, the choice of scientific data and statistical methodology to be used is best left to the sound discretion of the Administrator. See BASF Wyandotte Corp. v. Costle, 598 F.2d 637, 655 (1st Cir.1979); American Petroleum Institute v. EPA, 540 F.2d 1023, 1036 (10th Cir.1976), cert. denied, 430 U.S. 922, 97 S. Ct. 1340, 51 L. Ed. 2d 601 (1977); FMC Corp. v. Train, 539 F.2d 973, 986 (4th Cir.1976); American Meat Institute, 526 F.2d at 457.

Ford first questions the Administrator's method of calculating the long-term average effluent concentration for pollutants other than cyanide and hexavalent chromium. For those two pollutants, the Administrator was able to base the long-term averages directly on empirical data from the average of the best plants. For copper, nickel, zinc, total (hexavalent plus trivalent) chromium, lead, and cadmium, however, the Administrator employed a multiple regression analysis, using three variables believed to be significantly related to the concentration of the regulated pollutant in the effluent: the concentration in the influent of the regulated Metal ("Me?"); the concentration in the influent of all Precipitable Metals ("PM"); and the concentration in the effluent of the Total Suspended Solids ("TSS"). J. App. at 1346. The Administrator simplified his equation by using the ratio of Me? over PM, called Xme. Id. at 1347. Finding that electroplaters using adequate pollution control could attain an average TSS of 25 mg/l and achieve an Xme equal to that of the 75th percentile of sampled firms, the Administrator then derived long-term averages for the regulated pollutants. See id. at 1357-65.

Ford first claims that a regression analysis using TSS and Xme explains very little. The Administrator thoroughly explained his use of the regression analysis and of TSS and Xme, and analyzed their predictive value, however. Id. at 1346-57, 1398.53 Ford also attacks as unattainable the values assigned to Xme and TSS. We believe, however, that both the assigned values are clearly attainable by the average of the best plants: the 75th percentile figure for Xme was already being met, by definition, by 75% of the sampled plants, and the 25 mg/l TSS figure was above that observed for the plants using BPT.54

Ford next challenges the Administrator's computation of the variability factors. Because neither Ford nor any other commentator criticized the Administrator's variability approach during the rulemaking leading to the 1979 standards, Ford has less leeway in demonstrating the invalidity of that approach. Ford questions both the Administrator's choice of data and his use of the median variability factor. We see nothing in Ford's criticisms that satisfies its burden.

Finally, Ford claims that, in contrast to the other pollutants for which the Administrator used the regression analysis, the lead and cadmium discharge limits are totally unsupported.55 For copper, nickel, zinc, and total chromium, the Administrator had had adequate data on the influent and effluent concentrations of each pollutant to determine individualized coefficients in the regression equation. Those coefficients, different for each pollutant, served to ensure the "best fit" for each pollutant's long-term average. J.App. at 1347-49, 1359. For lead and cadmium, however, the Administrator recognized that he had inadequate data to derive individualized coefficients. Having computed a "group average" of the coefficients determined for copper, nickel, zinc, and total chromium (expressed in his Equation 7), J.App. at 1347, 1349-50, the Administrator stated:

Because of the small number of plants plating [cadmium] or discharging [lead], it is not feasible to develop best fit equations for these metals. However, Equation predicts, quite well, the discharge concentrations of the metals for which adequate data were available. Therefore, this equation is used to derive average [cadmium] and [lead] limits as well.

J.App. at 1359, 1361 (reference and footnote omitted). The Administrator then adopted the average of the coefficients of the four metals as the coefficients for lead and for cadmium. In other words, the Administrator chose to predict the treatability of lead and cadmium using data from other metals, explaining only that the data predicted well the treatability of those other metals.

Ford correctly notes that nothing in the Administrator's statement explains why the data for the four metals will predict well the treatability of lead and cadmium. We can reasonably discern, however, that the Administrator found lead and cadmium to be equally as treatable as the other metals. Ford has failed to rebut that implicit assumption, for it has never demonstrated, either in the administrative record or before us, that lead and cadmium are not equally treatable.

We note, moreover, that the Administrator buttressed his conclusion using what data he possessed on lead and cadmium. See Weyerhaeuser, 590 F.2d at 1054 n. 70. In a footnote to his explanation for using the average coefficients, the Administrator noted that the long-term average effluent concentrations for lead and cadmium derived using the average coefficients were higher than the observed average concentration for both lead and cadmium at both the three lead-discharging plants and the three cadmium-plating plants which the Administrator determined had usable data. J.App. at 1361 n. 9.56 Although, as Ford points out, the Administrator had found the cited data insufficient to develop lead and cadmium coefficients directly, we believe that the data nonetheless provides some support for his conclusion reached through use of the group average.

We conclude that Ford has failed to overcome the presumption that the Administrator's decision was rational. The Administrator's implicit finding, supported by the available data, that lead and cadmium are equally treatable as the four other metals has not been rebutted by contrary evidence. We therefore reject Ford's challenge to the Administrator's use of the average coefficients to derive the long-term effluent concentrations for lead and cadmium.57

NAMF charges that the electroplating pretreatment standards are not economically achievable. EPA both disputes the validity of NAMF's contention and argues that NAMF is barred from raising its contention. We hold that NAMF is not barred from challenging the electroplating standards. We also hold that the Administrator did not abuse his discretion in concluding that the costs imposed by the standards were justified by the reduction in the discharge of pollution. We will therefore deny the petitions for review.

Section 304(b)(1) of the Act states that in promulgating BPT discharge limits the Administrator must identify "the degree of effluent reduction attainable through the application of the best practicable control technology currently available" and must consider "the total cost of application of technology in relation to the effluent reduction benefits to be achieved from such application." 33 U.S.C. Sec. 1314(b)(1)(A), (B) (1976). The legislative history indicates that Congress intended to require "a limited cost-benefit analysis" to determine whether the required technology was indeed practicable. 1972 Legis.Hist. 170 (statement of Sen. Muskie). The Administrator's consideration of cost was to include both the dollar outlays of dischargers to comply with the standards and the potential unemployment and economic dislocation caused by the closing of dischargers unable to comply. AISI I, 526 F.2d at 1053 n. 57.

In promulgating the 1979 electroplating standards the Administrator noted that the standards would have a significant economic impact, particularly on the economically-vulnerable job shops and printed circuit board manufacturers. The Administrator stated that to reduce costs he had imposed less stringent standards on plants with an electroplating process wasteflow of less than 10,000 gallons per day, and had eliminated the limits on hexavalent chromium. 44 Fed.Reg. 52,590-91 (1979); 43 Fed.Reg. 6562 (1978). Nonetheless, he estimated that the cost of compliance for electroplaters would be $1.34 billion in capital costs and $425 million in annual cost. Id. at 52,593-94. He also estimated that, rather than attain compliance, 21.5% of indirectly-discharging job shops (587 firms and 9,653 workers) and 3.1% of indirectly-discharging printed circuit board manufacturers (10 firms and 321 workers) might close, and that 3.0% of the employees of indirectly-discharging captive electroplating operations would lose their jobs when those operations shut down (140 firms and 2610 workers). 44 Fed.Reg. 52,594 (1979).58 At the same time the Administrator estimated that compliance with the standards would remove 140 million pounds per year of toxic pollutants. Id. at 52,591. In response to comments that the standards were not achievable,59 the Administrator stated:

Congress realized that some businesses would close as a result of the promulgation of technology-based standards. Congress determined that long term environmental benefits were more important than short term dislocations. The Administrator has considered the costs and benefits of this regulation, as evidenced by his exemption of small platers from some requirements.

Id. at 52,602. The Administrator added that he did not conduct a strict cost-benefit analysis because such an analysis "was discouraged by Congress during the development of the Clean Water Act." Id. at 52,606.1. The NAMF Settlement Agreement

NAMF and IIPEC sought review of the 1979 electroplating standards in this court. Those parties and the Administrator then entered into the NAMF Settlement Agreement. R.Add. at D-1. Under that agreement the Administrator agreed to propose several specific amendments to the electroplating standards. In addition, the Administrator agreed to include as part of the preamble to those amendments a statement "concerning the relationship between the proposed regulations ... and possible further best available control technology economically available (BAT) regulations." R.Add. at D-2. The statement affirmed that the BAT standards would be based on technology compatible with that underlying the BPT standards, and that the cumulative impact of the BPT standards would be considered in determining what was economically achievable under the BAT standards. R.Add. at D-8. The statement also said that the Administrator "does not plan to develop more stringent new pretreatment standards" for job shops and printed circuit board manufacturers "in the next several years." Id. at D-9. In return, NAMF and IIPEC stated that if the final regulations and preamble did not "differ significantly" from the proposed regulations and preamble, then they would not challenge the 1979 electroplating standards. Id. at D-3.

We need not consider whether the Administrator had the power to enter into the NAMF Settlement Agreement because we find the Administrator has failed to live up to its terms. The parties to the Settlement Agreement agree that NAMF and IIPEC offered to exchange their right to contest the 1979 standards if the Administrator, inter alia, would both publish the suggested language in the preamble and abide by that language. Instead of publishing the language of the Settlement Agreement when he promulgated the 1981 amendments, however, the Administrator wrote a preamble indicating that he could not give the requested assurances. 46 Fed.Reg. 9464 (1981). The Administrator's belated correction does not alter the fact that he failed to comply with the Settlement Agreement. See 46 Fed.Reg. 30,625-26 (1981). More important, the Administrator's subsequent actions in the Metal Finishing regulations were not fully in accord with the language of the preamble. By proposing and promulgating an additional requirement on job shops and printed circuit board manufacturers in the form of the limit on total toxic organics ("TTO"), the Administrator has developed a "more stringent pretreatment standard" for those sources. See 48 Fed.Reg. 32,462 (1983); 47 Fed.Reg. 38,464 (1982).60 Imposition of such a new requirement differs significantly from the absence of further requirements contemplated by the Settlement Agreement. We hold, therefore, that NAMF and IIPEC are free to challenge the standards.61

NAMF contends that the 1979 electroplating standards are not "economically achievable." Brief for Petitioners NAMF et al. at 21. It points in particular to the Administrator's estimate that approximately 20% of indirectly discharging job shops, employing almost 10,000 workers, may close as a result of the standards. Because the Act requires that pretreatment standards be "economically achievable," NAMF argues, the electroplating standards are thus arbitrary and capricious. Id. at 27.

As its language suggests, NAMF asserts that the electroplating standards are not BPT standards. Instead NAMF contends that the standards are based on the best available technology economically achievable--BAT.62 The consent decree in NRDC v. Train, however, required the Administrator to promulgate BPT pretreatment standards for the electroplating industry by May 15, 1977, before promulgating BAT standards for that industry. 8 Env't Rep.Cas. (BNA) at 2128. Although in proposing the electroplating standards the Administrator was imprecise regarding the level of technology on which the standards were based, see 43 Fed.Reg. 6560-62, 6564-65, 6568 (1978), it appears that as promulgated the electroplating standards are intended to represent BPT-level technology, see 44 Fed.Reg. 52,592, 52,608 (1979).63

NAMF argues that even if the standards are based on BPT, they must nonetheless be economically achievable to be "practicable." Reply Brief for Petitioners NAMF, et al., at 4-5. Section 304(b)(1) of the Act does not include economic achievability among the requirements listed for BPT, however. 33 U.S.C. Sec. 1314(b)(1) (1976); see 1972 Legis.Hist. 231 (statement of Rep. Jones). Instead that section directs the Administrator to consider "the total cost of application of technology in relation to the effluent reduction benefits." 33 U.S.C. Sec. 1314(b)(1)(B) (1976); see also CPC International, Inc. v. Train, 540 F.2d 1329, 1341 (8th Cir.1976), cert. denied, 430 U.S. 966, 97 S. Ct. 1646, 52 L. Ed. 2d 357 (1977). We believe that that limited cost-benefit analysis, and BPT itself, have a role in the statutory scheme quite different from the requirement that the best available technology be "economically achievable."

BPT does not limit the amount of pollution control required of a discharger or an industry to its economic capability. Rather, this first phase requires the elimination of all pollutant discharges where "the costs imposed on the industry are worth the benefits in pollution reduction." National Crushed Stone, 449 U.S. at 76, 101 S.Ct. at 303. A discharger not making such "inefficient" discharges need make no further effort toward curtailing pollution, even if he can afford it. Id. at 75, 101 S.Ct. at 303. A discharger making inefficient discharges must raise his performance to BPT standards; if he cannot afford it he must go out of business. Id. at 76, 101 S. Ct. at 303. The second phase, BAT, "assumes that the 1977 BPT standard has been met" and that all such inefficient pollutant discharges have been eliminated. See id. at 74, 101 S. Ct. at 302: The Act then demands "reasonable further progress toward the national goal of eliminating the discharge of all pollutants," 33 U.S.C. Sec. 1311(b)(2)(A) (1976), progress that can only be made by requiring the remaining dischargers to eliminate "efficient" discharges, where the costs outweigh the benefits of pollution reduction.64 It is only at this stage that the Act requires that the standards be "economically achievable." Id. The remaining dischargers need only commit "the maximum resources economically possible," National Crushed Stone, 449 U.S. at 74, 101 S.Ct. at 302; if the BAT standard is not within the "economic capability" of a discharger making sufficient progress, he need make only such efforts as are economically achievable for him. 33 U.S.C. Sec. 1311(c) (1976); see National Crushed Stone, 449 U.S. at 74, 101 S.Ct. at 302.

In National Crushed Stone, the Supreme Court held that, so long as the Administrator had properly found the effluent reduction benefits were worth the costs imposed on an industrial category, it "would be inconsistent with this legislative scheme" to excuse an individual discharger from BPT requirements because those requirements were not economically achievable for him. 449 U.S. at 75, 76-77, 101 S. Ct. at 302, 303-304. We hold that it would be equally inconsistent to require that BPT regulations be economically achievable for even a major proportion of an industrial category. Congress anticipated that the BPT regulations "would cause economic hardship and plant closings" because they would impose on "a substantial number of point sources" within each industrial category additional costs which "must be borne or the point source eliminated." Id. at 76, 78-83, 101 S. Ct. at 303, 304-307; see Association of Pacific Fisheries v. EPA, 615 F.2d 794, 808-09 (9th Cir.1980); AISI I, 526 F.2d at 1052; see also 1977 Legis.Hist. 404 (statement of Rep. Anderson). The closing of 20% of the job shops and the loss of 10,000 jobs, while a severe hardship, will not invalidate the electroplating standards unless the Administrator has failed to consider those costs in relation to the effluent reduction benefits, or has improperly concluded that the benefits are worth the costs.

NAMF challenges the Administrator's consideration of the costs and benefits of the electroplating standards. The Administrator is accorded considerable discretion in weighing costs and benefits. AISI I, 526 F.2d at 1052 & n. 54. As Senator Muskie stated:

The balancing test between total costs and effluent reduction benefits is intended to limit the application of technology only where the additional degree of effluent reduction is wholly out of proportion to the costs of achieving such marginal level of reduction for any class or category of sources.

1972 Legis.Hist. at 170 (statement of Sen. Muskie); see Association of Pacific Fisheries, 615 F.2d at 805, 809; BASF Wyandotte, 598 F.2d at 656.

Contrary to NAMF's assertion, it appears that the Administrator did perform the required cost-benefit analysis for the BPT electroplating standards. He calculated that the benefits would be an effluent reduction of 140 million pounds of toxic pollutants per year, and that the total costs would be $1.34 billion plus $425 million annually, resulting in the closing of 737 electroplating operations and the loss of 12,584 jobs. He then stated that he had considered the costs and benefits in promulgating the electroplating regulations. From this we can reasonably discern that the Administrator concluded that the benefits were worth the costs.

NAMF argues that the Administrator's cost-benefit analysis was fatally flawed, however, because he failed to consider less burdensome alternatives. It claims that if the electroplating standards were five percent less stringent the costs to electroplaters could be cut in half. EPA denies that the Administrator must make any such marginal analysis for BPT.

Section 304(b)(1) directs the Administrator to consider "the total cost" in relation to the effluent reduction benefits resulting from the application of control technology. 33 U.S.C. Sec. 1314(b)(1)(A) (1976).65 The legislative history of this requirement leads us to conclude that Congress intended that the Administrator consider "the additional degree of effluent reduction" in relation to "the costs of achieving such marginal level of reduction." 1972 Legis.Hist. 170 (statement of Sen. Muskie) (emphasis added); see AISI I, 526 F.2d at 1076 n. 19 (Adams, J., concurring). Indeed, given the place of BPT standards in the two-phase statutory scheme, a balancing solely of net costs and net effluent reduction benefits would make no sense under the Act. By setting as a national goal the elimination of pollutant discharges, Congress at least preliminarily has weighed the costs and benefits of achieving such a goal and has determined that society would thus be better off--that the net benefits exceed the net costs. See Weyerhaeuser, 590 F.2d at 1037. If the BPT cost-benefit analysis were to be conducted on a net basis, the national goal could be attained by BPT standards alone. Congress envisioned BPT standards as only a first stage, however. It provided for the second-stage BAT standards to make further progress towards the national goal, and at the same time indicated that the BPT cost-benefit analysis served "to limit the application of technology" required of dischargers under BPT standards. 1972 Legis.Hist. 170 (statement of Sen. Muskie). To perform its limiting function, and to preserve any role for BAT standards in the statutory scheme, BPT cost-benefit analysis must be conducted on a marginal basis.

In Weyerhaeuser Corp. v. Costle, 590 F.2d 1011 (D.C.Cir.1978), the petitioners argued that the Administrator had to make an incremental balancing of costs and benefits in promulgating certain BPT effluent limitations. Id. at 1047. The court replied:

A requirement that EPA perform the elaborate task of calculating incremental balances would bog the Agency down in burdensome proceedings on a relatively subsidiary task. Hence, the Agency need not on its own undertake more than a net cost-benefit balancing to fulfill its obligation under section 304.

However, when an incremental analysis has been performed by industry and submitted to EPA, it is worthy of scrutiny by the Agency, for it may "avoid the risk of hidden imbalances between cost and benefit."

Id. at 1048 (quoting AISI I, 526 F.2d at 1076 n. 19 (Adams, J., concurring)); accord BASF Wyandotte, 598 F.2d at 656 & n. 37. The Weyerhaeuser court examined the marginal analysis submitted by the petitioners and found no hidden imbalance between the marginal costs and benefits. Id.

While we agree that for BPT "the cost of compliance was not a factor to be given primary importance," AISI I, 526 F.2d at 1051 (emphasis added), both cost and benefit remain factors that the Administrator must consider and compare. See Weyerhaeuser, 590 F.2d at 1045-46. Such comparison is meaningless unless conducted on a marginal basis. Marginal analysis may indeed be an elaborate task, see AISI I, 526 F.2d at 1076 n. 19 (Adams, J., concurring), but Congress anticipated that the Administrator would have to engage in "complex balancing." 1972 Legis.Hist. 181 (statement of Sen. Muskie); see H.R.Rep. No. 911, 92d Cong., 2d Sess. 107, reprinted in 1972 Legis.Hist. 753, 794. Moreover, while we agree that only marginal analysis will reveal hidden imbalances between cost and benefit, we cannot understand why the Act would require such analysis only on request. We therefore conclude that the Administrator on his own must undertake a sufficient marginal analysis to indicate that the marginal cost is not wholly out of proportion to the marginal effluent reduction benefit. See also American Paper Institute v. EPA, 660 F.2d 954, 961 (4th Cir.1981).

We note that despite his legal position in this case the Administrator apparently employed marginal cost-benefit analysis in setting the electroplating standards. See AISI II, 568 F.2d at 297. He stated:

Although the Clean Water Act does not require consideration of alternative timing, or alternative methods of ensuring compliance, EPA has considered alternative stringency levels, and alternative types of regulations.

44 Fed.Reg. 52,593 (1978); see J.App. at 1693. The Administrator lifted many requirements from electroplaters with smaller flows, finding that his action would "greatly [reduce] the projected economic impact of the standards while relaxing controls on less than one percent of the flow." 43 Fed.Reg. 6561 (1978). He set the required flow rate at 10,000 gallons per day by balancing the marginal economic impact against the effluent reduction benefits. 44 Fed.Reg. 52,603-04 (1979). Similarly, the Administrator eliminated the hexavalent chromium limits because it reduced the cost of the electroplating standards without significant environmental effect. Id. at 52,591.

NAMF claims, however, that it demonstrated a hidden imbalance between marginal costs and benefits by submitting a less burdensome alternative in its comments on the 1978 proposed pretreatment standards. In those comments NAMF suggested that the standards be made less stringent so that electroplaters could release their rinse waters without pretreatment. J.App. at 853. As rinse waters comprise 90% of the volume of the electroplating process wastestream, 43 Fed.Reg. 6565 (1978), NAMF argued that the remaining wastewaters, containing higher concentrations of pollutants, could be pretreated in smaller pretreatment facilities at less than half the cost. J.App. at 854-56. Unlike NAMF's brief in this case, however, its comments failed to calculate the effluent reduction benefit lost by permitting rinse water to go without treatment.66 Without knowing the incremental benefit, it is impossible to determine whether the economic costs imposed are indeed wholly out of proportion. We therefore reject NAMF's assertion that it has demonstrated an imbalance between the incremental cost and benefit.

NAMF makes several challenges to the validity of the Administrator's determination of the cost and economic impact of the electroplating standards. We have examined each contention and have found that the Administrator's challenged decisions were not arbitrary and capricious.67

We are thus left with NAMF's assertion that the net costs of the 1979 electroplating standards are wholly out of proportion to the net effluent reduction benefits. We cannot say that the Administrator was arbitrary and capricious when he determined that the removal of 140 million pounds per year of toxic pollutants was worth $1.34 billion plus $425 million annually, with the loss of 737 firms and 12,584 jobs.

C. The Compliance Deadline for Integrated Facilities

Section 413.01(a) of the electroplating standards directs that integrated electroplaters must comply with the standards by three years after the effective date of the combined wastestream formula, 40 C.F.R. Sec. 403.6(e) (1982). 40 C.F.R. Sec. 413.01(a) (1982). GM claims that the Administrator's unsuccessful effort to indefinitely postpone the effective date of the formula has left integrated electroplaters only 21 months in which to achieve compliance. GM contends that that reduced time for compliance renders the three-year deadline arbitrary and capricious. We disagree, and will deny GM's petition for review on this issue.

Section 307(b) of the Act directs the Administrator to promulgate categorical pretreatment standards for existing sources, and requires that such standards "shall specify a time for compliance not to exceed three years from the date of promulgation." 33 U.S.C. Sec. 1317(b)(1) (1976 & Supp. I 1977). When the Administrator promulgated the electroplating standards in 1979, he set the compliance date for all facilities at the maximum three years, October 12, 1982, "because of the high projected economic impact of these pretreatment standards." 44 Fed.Reg. 52,595 (1979); see 43 Fed.Reg. 6562 (1978). He subsequently exempted integrated facilities from the electroplating standards until the proposed combined wastestream formula became effective. 45 Fed.Reg. 19,246 (1980). When the Administrator promulgated the formula, denied Ford's petition for review, and promulgated the electroplating amendments on January 28, 1981, he announced that integrated electroplaters would not have to comply with the 1979 electroplating standards or the 1981 electroplating amendments until three years from the formula's effective date of March 13, 1981. 46 Fed.Reg. 9464 (1981); see id. at 9404. The Administrator justified the extension by stating that the formula "would have to be promulgated in final form before integrated facilities would understand their compliance obligations under the electroplating standards." Id. at 9464.

After an initial postponement of the formula's effective date to March 30, 1981, the Administrator then indefinitely postponed the formula's effective date. See 47 Fed.Reg. 4518 (1982); 46 Fed.Reg. 50,502-03 (1981); see also id. at 11,971 (1981). He explained that because he had received numerous comments criticizing the "highly controversial" formula's effect on integrated facilities, he believed the formula should be deferred "while the Agency studies the implications of the present formula further." 47 Fed.Reg. 4519, 4520 (1982). The Administrator recognized that the indefinite postponement of the formula also postponed the date by which integrated facilities had to comply with the electroplating standards. 46 Fed.Reg. 43,973 (1981).

In NRDC v. EPA, 683 F.2d 752 (3d Cir.1982), we found that the Administrator's attempt at indefinite postponement violated the notice and comment requirements of the Administrative Procedure Act. Although we noted that the attempted postponement of the formula had effectively postponed the compliance date for integrated electroplaters, id. at 756-57, we ordered the Administrator to reinstate the combined wastestream formula, effective March 30, 1981. Id. at 768-69. We stated:

Our decision does not, of course, forestall future agency action with regard to the [combined wastestream formula], provided such action is taken in compliance with the Administrative Procedure Act.

Id. at 768-69. Soon after our decision, on August 10, 1982, GM filed a petition for reconsideration asking the Administrator to suspend the effective date of the formula. The parties informed us at oral argument that GM's petition had been denied. Transcript of Oral Argument at 235, 241.

We recognize the dilemma for integrated electroplaters caused by the Administrator's attempted indefinite postponement. We took that dilemma into account in deciding NRDC v. EPA, however. We were cognizant of the effect our decision would have upon the compliance date for integrated electroplaters, and we nonetheless reinstated the combined wastestream formula. We left to the Administrator any postponement of the effective date of the formula. GM's petition initiated that administrative process. GM's recourse is to petition for review of the Administrator's denial of that petition, not to raise the issues in this proceeding. We will therefore deny GM's petition for review on this issue. We hold that GM has failed to show that the Administrator has abused his discretion.

We will grant the petitions for review in Nos. 81-1977, 81-1982, 81-1983, 81-1984, 81-1985, 81-2150, and 81-2151. We will deny all other petitions. We will also deny EPA's motion regarding MFASC.

(a) 40 C.F.R. Sec. 403.3(i) (1982), establishing the definition of "interference; "

(b) 40 C.F.R. Sec. 403.3(k) (1982), establishing the definition of "new source; "

(c) 40 C.F.R. Sec. 403.3(n) (1982), establishing the definition of "pass through; " and

(d) 40 C.F.R. Sec. 403.13 (1982), containing the fundamentally different factor variance provision.

GIBBONS, Circuit Judge.

I join in the opinion of the court. I write separately only to note that if the interference rule, 40 C.F.R. Sec. 403.3(i) (1982), is clarified to reflect the interpretation which the government urged at the oral argument on this appeal, it will be consistent with 33 U.S.C. Sec. 1317(b) and (c). The interference must be caused by a pollutant. If it is established that the interference is caused by a pollutant, and a user of the POTW is a source of such pollutant, the three methods set forth in 40 C.F.R. Sec. 403.3(i) (1982) for determining responsibility for the interference satisfy both the Clean Water Act and due process.