Full opinion text
OPINION OF THE COURT GARTH, Circuit Judge: These consolidated petitions challenge an amended final rule of the Environmental Protection Agency (EPA), known as the removal credit rule, 49 Fed.Reg. 31212 (1984) (codified at 40 C.F.R. § 403.7), promulgated pursuant to the Federal Water Pollution Control Act (“FWPCA” or “Clean Water Act”) of 1977, Pub.L. No. 95-217, 91 Stat. 1566 (codified at 33 U.S.C. 1251 et seq.). The Clean Water Act of 1972 set' as a national goal the elimination, by 1985, of the discharge of pollutants into the nation’s navigable waters. Pub.L. No. 92-500, 86 Stat. 816, § 101(a)(1) (codified at 33 U.S.C. § 1251(a)(1)). In 1977, recognizing that the 1972 Act’s regulatory mechanism for the control of toxics “ha[d] failed,” Congress amended the Clean Water Act to clarify and strengthen its provisions for dealing with toxic pollutants. Leg.Hist. 326-27 (comments of the House manager of the bill, Rep. Roberts); see also id. at 369 (statement of Rep. Clausen); id. at 427 (statement of Sen. Muskie). One aspect of the 1977 Act’s strengthened program to control toxic pollutants was a requirement that an indirect discharger i.e. an industrial discharger whose wastes flow into a public sewage system rather than directly into navigable waters, had to “pretreat” its waste waters so as to achieve, together with the Publicly Owned Treatment Works (POTW) that treated the waste before final discharge into navigable waters, the same level of toxics removal as was required of a direct discharger, which discharged directly into a river, lake, or ocean. At the same time, the Act allowed the indirect discharger to receive a “removal credit” from the POTW for the amount of waste removed from the stream of waste water by the POTW itself. 33 U.S.C. § 1317(b)(1). This provision, designed to avoid redundant treatment, permits an increased amount of pollutants to flow from the indirect discharger’s plant to the municipal treatment plant provided that the additional pollutants are removed by the municipal plant. By this process, the amount of pollutants ultimately discharged by the combination of the indirect discharger and the POTW will be no greater than the amount discharged by the direct dis-charger. The Act also required EPA to develop regulations for the disposal and utilization of POTW sewage sludge, so as to prevent contamination of the sludge with toxics removed from the effluent flowing through the POTW, and so as to encourage the productive recycling of sludge. In furtherance of this goal, the Act provided that POTWs could not grant removal credits to indirect dischargers for the POTW’s removal of toxics from the liquid waste stream if this removal and consequent transfer of toxics to the POTW sludge would render the sludge toxic and thus prevent sludge disposal in accordance with those regulations. 33 U.S.C. § 1317(b)(1). NRDC challenges four aspects of the removal credit rule. First, it argues that EPA’s method of calculating waste removal by POTWs violates the statutory requirement that direct and indirect dischargers be held to the same standard. Second, it argues that EPA’s decision to ignore sewer overflow events in calculating the amount of toxic waste removed by a POTW violates the same requirement. Third, NRDC claims that EPA’s action in modifying the test for determining when a credit must be withdrawn violates the Administrative Procedure Act, 5 U.S.C. § 551 et seq., and the Clean Water Act, 33 U.S.C. § 1251 et seq. Fourth, NRDC questions whether EPA may put into effect a relaxed removal credit rule when the sludge regulations that are a precondition for the issuance of such credits have not been promulgated. In each of these areas, we conclude that EPA’s 1984 removal credit regulations fail to meet statutory requirements. Furthermore, even extending the utmost deference to the Agency, we conclude that the promulgation of these regulations on the basis of the administrative record before us is arbitrary and capricious. Thus we grant NRDC’s petition for review. In contrast to NRDC, Petitioners Cerro Copper and Village of Sauget challenge the regulations as too strict and inflexible to take into account situations such as theirs. The petitions of Cerro Copper and the Village of Sauget will be denied. I. In order to understand the issues presented on this appeal, it is helpful to trace the history of the statute and regulations which were designed to eliminate the discharge of pollutants into our waters. A. The removal credit rule at issue here is a part of a complex regulatory framework mandated by the Federal Water Pollution Control Act Amendments of 1972 and 1977, 33 U.S.C. §§ 1251 et seq. The statute calls for a two-phase program to limit discharges of effluents. Direct dischargers of toxic wastes were to comply with the Best Practicable Control Technology (BPT) by July 1, 1977. 33 U.S.C. §§ 1311(b)(1)(A), 1314(b)(1). Between 1983 and 1987, direct dischargers of toxic wastes were to meet the more stringent standards consistent with the Best Available Technology economically achievable (BAT). 33 U.S.C. § 1311(b)(2). The statute also mandated that the EPA set effluent limitations for POTWs engaged in the treatment of municipal sewage or industrial wastewater. Id. §§ 1311(b)(1)(B)-(C), 1314(d)(1). Such limitations were to result in equal levels of treatment for all toxic discharges, whether issued directly into navigable waters or channelled by a sewage system through a POTW. Because secondary treatment by POTWs cannot deal adequately with toxic pollutants, the statute required that EPA establish national pretreatment standards, i.e. standards to which an indirect discharger must conform in treating its waste before such waste reaches the POTW. Those standards, applicable to indirect dischargers, provide for pretreatment which is equivalent to BAT standards. 33 U.S.C. 1317(b)(1). In order to avoid redundant treatment by the indirect discharger and the POTW (“treatment for treatment’s sake,” Leg. Hist. 343), section 307(b)(1) of the 1977 Act, 33 U.S.C. § 1317(b)(1), requires that the POTW be authorized to give credit to the indirect discharger for removal of pollutants achieved by the POTW. The removal credit provision thereby equates the amount of pollutants removed by the combined treatment of the POTW and the indirect discharger with the amount of pollutants removed by the direct discharger operating under the relevant BAT standard. 33 U.S.C. § 1317(b)(1). The removal credit provision was added to the statute by the Clean Water Act of 1977, Pub.L. No. 95-217, 91 Stat. 1589. That law was enacted in response to Congress’s recognition of the growing seriousness of the problems of toxic pollution, and of the woeful inadequacy of the 1972 Act in dealing with them. Leg.Hist. 326-27, 454-55, 862-65. The 1977 Act strengthened the controls over toxic pollutants in several ways. The 1977 Act explicitly codified the toxics consent decree issued by the District Court for the District of Columbia. That decree resulted from litigation brought by environmental groups to challenge EPA’s failure to promulgate the effluent standards mandated for toxic substances by the 1972 Act. See NRDC v. Train, 8 Env’t Rep.Cas. (BNA) 2120 (D.D.C.1976), modified sub nom. NRDC v. Costle, 12 Env’t Rep.Cas. (BNA) 1833 (D.D.C.1979), modified sub nom. NRDC v. Gorsuch, No. 72-2153 (D.D.C. Oct. 26, 1982), modified sub nom. NRDC v. Ruckelshaus, No. 73-2153 (D.D.C. Aug. 2, 1983 and Jan. 6, 1984). Following that decree, Congress required that BAT effluent guidelines, applicable to direct dischargers, be developed by July 1, 1980 for the 65 toxic pollutants listed in the decree. Pub.L. No. 95-217, § 53(a), 91 Stat. 1589 (codified at 33 U.S.C. 1317(a)(2)). Also following the toxics decree, the 1977 Act amended section 307(b)(1) of the Act, 33 U.S.C. 1317(b)(1), to require EPA to promulgate pretreatment regulations for indirect dischargers analogous to the BAT standards for direct dischargers. Congress stressed that pretreatment standards had to be based on BAT or more stringent limits. See 33 U.S.C. § 1317(a); Leg.Hist. at 271, 342, 403, 461, 690. While strengthening and clarifying the pretreatment requirements for toxics, Congress added the following provision to the statute, thus providing for removal credits to avoid redundant treatment: 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 1345 of this title, 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. 33 U.S.C. § 1317(b)(1). In addition, Congress sought to ensure that sewage sludge, instead of becoming a depository for toxic pollutants, should be usable as fertilizer or for other productive purposes. It did so by amending section 405 of the Clean Water Act to require EPA to develop regulations for sludge use or disposal within one year of the statute’s enactment. 33 U.S.C. § 1345(d). (See Exhibit B.) In conjunction with this amendment to section 405, Congress also amended section 307(b)(1) of the Act to ensure that EPA did not, in the course of administering the removal credit program, undermine the objective of rendering sewage sludge nontoxic and usable. By the same amendment, quoted above, that provided for removal credits, Congress also prohibited the granting of removal credits if such credits would lead to sludge contamination that would prevent the use or disposal of sludge in conformity with the section 405 regulations. FWPCA 307(b)(1), 33 U.S.C. 1317(b)(1). B. The removal credit regulation at issue here allows POTWs to grant removal credits to indirect dischargers so that the indirect discharger may increase its discharges of the pollutant by the amount that the POTW removes. The statute mandates that such credits be equal to the amount of toxics consistently removed by the POTW. The regulation requires that each POTW first determine its removal efficiency (or “consistent removal rate”) for each regulated pollutant. This is generally done through a process of measuring the concentration of a given pollutant found in the waste flowing into a POTW (i.e., the in-fluent) and then measuring its concentration in the waste flowing out of the POTW (i.e., the effluent). Removal is expressed as a percentage of the amount in the in-fluent. If, for example, a POTW consistently removes 60% of a particular pollutant that flows into it, it can grant a 60% credit to the indirect discharger. The indirect discharger would then be permitted to discharge more than twice as much of that pollutant to the POTW than it would otherwise have been allowed to discharge. The removal credit rule is one part of the General. Pretreatment Regulations for Existing and New Sources of Pollution, 40 C.F.R. § 403.1-16 (1984). These pretreatment regulations provide the framework for implementation of another set of regulations, the Categorical Pretreatment Standards, which establish specific limits for the discharge of particular toxic pollutants. The present removal credit rule is the fourth version of the rule promulgated by EPA on this subject. The first version, promulgated in 1973, was sketchy. It foreshadowed the statutory removal credit provision of section 1317(b)(1) enacted in 1977. 38 Fed.Reg. 30982-84 (1973). In 1977, EPA promulgated more complete regulations. Each of the two succeeding versions following the 1977 version has further relaxed the requirements that POTWs and indirect dischargers are required to meet. From the first regulations through all subsequent revisions, the regulations purport to require that indirect dischargers be held to the same standard as direct dischargers, subject to credit for removal of toxics by the POTW. In 1977, following the issuance of the toxics consent decree in NRDC v. Train, 8 Env’t Rep.Cas. (BNA) 2120, EPA proposed more developed pretreatment regulations. 42 Fed.Reg. 6176 (1977). While this proposal was pending, Congress enacted the Clean Water Act amendments of 1977, adding the removal credit provision to section 307(b)(1). After extensive comment and participation by interested parties, including four public hearings and sixteen public meetings, which generated testimony and comments from 400 individuals and groups, EPA, on June 26, 1978, promulgated its second removal credit regulation as a part of the General Pretreatment Regulations on June 26, 1978. 43 Fed.Reg. 27736 (1978). Following adoption of these regulations, industry and environmental groups brought several actions challenging the second removal credit provision. A settlement agreement between EPA and the industry parties led EPA to propose amendments to these regulations. 44 Fed.Reg. 62260 (1979). A third regulation was promulgated on January 28, 1981. 46 Fed.Reg. 9404 (1981). The revised rules made it substantially easier for POTWs to grant removal credits and to give larger crédito for the same pollutant removals. Several industry parties nonetheless again brought suit, contending that the new rules did not go far enough. These actions were consolidated with the actions of environmental groups challenging the 1978 and 1981 regulations, all of which were heard by this Court in National Association of Metal Finishers v. EPA, 719 F.2d 624 (3d Cir.1983) [hereinafter NAMF] reversed in part sub nom. Chemical Manufacturers Ass’n v. NRDC, — U.S. -, 105 S.Ct. 1102, 84 L.Ed.2d 90 (1985). In NAMF, this court upheld the 1981 removal credit rule, rejecting industry arguments similar to, or identical with, those made by EPA here. At the same time that they were attacking the 1981 regulations in court, the industry petitioners also urged EPA to suspend these same regulations so that they could be reconsidered by the new Administration. See, e.g., letters from Chemical Manufacturers Ass’n. to EPA, Mar. 9, 1981 & Mar. 19, 1981, App. 145-51. In fact, EPA did defer the effective date of the 1981 regulations indefinitely, and subsequently proposed to develop yet another removal credits provision. 47 Fed.Reg. 4520 (1982). In response to a suit brought by NRDC, this Court ruled that EPA’s suspension of the pretreatment regulations without notice or opportunity for public comment violated the Administrative Procedure Act, 5 U.S.C. § 553(b)-(e), and that therefore both this initial suspension and a subsequent rulemaking proceeding, extending that suspension in part, were illegal. The EPA was therefore ordered to reinstate all of the amendments retroactively to March 10, 1981. Natural Resources Defense Council v. EPA, 683 F.2d 752, 768-69 (3d Cir.1982). EPA complied with this order on September 28, 1982. 47 Fed. Reg. 42688 (1982). However, on that same day, EPA proposed to amend the removal credit rule once again, on the grounds that it “has been criticized as being so burdensome and unwieldy as to discourage POTWs from applying for and obtaining authorization to grant removal credits.” 47 Fed.Reg. 42698 (1982). The fourth and ostensibly final removal credit rule, which is the one at issue here, was promulgated on August 3, 1984. 49 Fed. Reg. 31212 (1984) (codified at 40 C.F.R. § 403.7) (1985). Among other changes, this rule adopted a new, more lenient, method of measuring the consistency of toxics removal by POTWs. The overall effect of the new rule is that it enables POTWs to grant more and larger credits than under the 1981 rule, which itself constituted a relaxation of the 1978 rule. As a consequence, the requirements for industrial pretreatment have been very significantly reduced. Petitioner NRDC filed a timely petition for review of the 1984 (fourth) rule, claiming that whereas the 1981 rule was in compliance with the statute, the 1984 rule has relaxed the conditions for granting removal credits beyond the bounds permitted by the Clean Water Act. Petitioners Cerro Copper and Village of Sauget, in contrast, seek to have the 1984 rule set aside on the ground that, as applied to them, the rule is more severe than the statute allows. In addition, all petitioners present procedural arguments as to why the rule is invalid. II. Our standard of review of the Agency’s informal rulemaking in this case is governed by section 10(e)(2) of the Administrative Procedure Act, 5 U.S.C. § 706(2), and in particular by subsections (A), (C), and (D), which provide that the reviewing court shall “hold unlawful and set aside agency action, findings, and conclusions found to be — (A) arbitrary, capricious, an abuse of discretion, or otherwise not in accordance with law; ... (C) in excess of statutory jurisdiction, authority, or limitations, or short of statutory right; [or] (D) without observance of procedure required by law.” 5 U.S.C. § 706(2)(A), (C), & (D). We must first determine whether the regulations are in conformity with the statute. In making this determination, the “view of the agency charged with administering the statute is entitled to considerable deference; and to sustain it, we need not find that it is the only permissible construction that EPA might have adopted but only that EPA’s understanding of this ‘very complex statute’ is a sufficiently rational one to preclude a court from substituting its judgment for that of EPA.” Chemical Manufacturers Association v. NRDC, — U.S. -, 105 S.Ct. 1102, 84 L.Ed.2d 90 (1985) (interpreting the Clean Water Act); see also Chevron v. NRDC, 467 U.S. 837, 104 S.Ct. 2778, 2781-83, 81 L.Ed.2d 694 (1984); NAMF, 719 F.2d at 637. On the other hand, this court may not abdicate its responsibility to interpret the statute in accordance with traditional principles of statutory construction. “[W]hile reviewing courts should uphold reasonable and defensible constructions of an agency’s enabling Act, ... they must not 'rubber-stamp ... administrative decisions that they deem inconsistent with a statutory mandate or that frustrate the congressional policy underlying a statute.’” Bureau of Alcohol, Tobacco, and Firearms v. Federal Labor Relations Authority, 464 U.S. 89, 97, 104 S.Ct. 439, 444, 78 L.Ed.2d 195 (1983) (quoting NLRB v. Brown, 380 U.S. 278, 85 S.Ct. 980, 13 L.Ed.2d 839 (1965)); see also Consumer Product Safety Commission v. GTE Sylvania, Inc., 447 U.S. 102, 120, 100 S.Ct. 2051, 2062, 64 L.Ed.2d 766 (1980); International Brotherhood of Teamsters, Chauffeurs, Warehousemen, and Helpers of America v. Daniel, 439 U.S. 551, 566 & n. 20, 99 S.Ct. 790, 800 & n. 20, 58 L.Ed.2d 808 (1979); Hi-Craft Clothing Co. v. NLRB, 660 F.2d 910, 915 (3d Cir.1981). Statutory interpretation is one of the traditional functions of courts. Of course, “[i]f the intent of Congress is clear, that is the end of the matter; for the court as well as the agency, must give effect to the unambiguously expressed intent of Congress.” Chevron v. NRDC, 467 U.S. 837, 104 S.Ct. 2781-82, 81 L.Ed.2d 694 (1984). Furthermore, agency action in fulfillment of a statutory mandate is 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). But this presumption is not “equivalent [to] the presumption of constitutionality afforded legislation drafted by Congress.” Motor Vehicle Manufacturers Ass’n., 463 U.S. 29, at 43 n. 9, 103 S.Ct. 2856, at 2866 n. 9, (1983). “The judiciary is the final authority on issues of statutory construction and must reject administrative constructions which are contrary to clear congressional intent.” Chevron v. NRDC, 104 S.Ct. 2778, 2782 n. 9. We must, therefore, first determine whether EPA’s position is at least “within the outer limits of its authority to interpret” the statute. Teamsters v. Daniel, 439 U.S. at 566, 99 S.Ct. at 800. Second, we must examine the challenged aspects of the rulemaking procedure carefully to determine independently that the Agency has not acted unfairly or in disregard of the statutorily prescribed procedures for notice and comment rulemaking. See, e.g., NAMF, 719 F.2d at 637-38; American Iron & Steel Institute v. EPA, 568 F.2d 284 (3d Cir.1977). Third, we must review the substantive aspects of agency action under the arbitrary and capricious standard of 5 U.S.C. § 706(2)(A). This standard has recently been described by the Supreme Court as follows: 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, Inc. 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 Transportation, Inc. v. Arkansas-Best Freight System, Inc., supra [419 U.S. 281 at 285, 95 S.Ct. 438, 442, 42 L.Ed.2d 447 (1974)]. 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 Transportation, Inc. v. Arkansas-Best Freight System, Inc. [419 U.S.] at 286 [95 S.Ct. at 442]. See also Camp v. Pitts, 411 U.S. 138, 142-43 (1973) (per curiam). Motor Vehicle Manufacturers Ass’n v. State Farm Mutual, 463 U.S. 29, 43, 103 S.Ct. 2856, 2866, 77 L.Ed.2d 443 (1983). See also Bowman Transportation, Inc. v. Arkansas-Best Freight System, Inc., 419 U.S. 281, 284, 95 S.Ct. 438, 441, 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). We must defer to an agency’s expert judgment when it is acting within the scope of the statute, but we cannot allow expertise to shield an irrational decision-making process. “[U]nless we make the requirements for administrative action strict and demanding, expertise, the strength of modern government, can become a monster which rules with no practical limite on its discretion.’ ... We have frequently reiterated that an agency must cogently explain why it has exercised its discretion in a given manner ....” Motor Vehicle Manufacturers Ass’n, 463 U.S. at 48-49, 103 S.Ct. at 2869-70 (citations omitted). Where, as here, an agency has reversed its established interpretation of a statute, the degree of deference accorded to the agency may be somewhat lessened. As this court has said, “sharp changes of agency course constitute ‘danger signals’ to which a reviewing court must be alert.” NRDC v. EPA, 683 F.2d at 760 (citation omitted). See also Motor Vehicle Manufacturers Ass’n v. State Farm Mutual, 463 U.S. at 42, 103 S.Ct. at 2866 (“A ‘settled course of behavior embodies the agency’s informed judgment that, by pursuing that course, it will carry out the policies committed to it by Congress. There is, then, at least a presumption that those policies will be carried out best if the settled rule is adhered to.’ ” (citation omitted)); Sierra Club v. United States Army Corps of Engineers, 772 F.2d 1043, 1046 (2d Cir.1985) (“A change in something from yesterday to today creates doubt. When the anticipated explanation is not given, doubt turns to disbelief.”) An agency is, of course, free to change its position, but it must supply adequate data and a reasoned analysis to support the change. We will apply these standards of review first to the challenges raised by NRDC (parts III-VI) and then to the issues raised by Cerro Copper and the Village of Sauget (part VII). III. A. EPA’s 1984 removal credit rule provides that the combined amount of toxics removed by an indirect discharger and a POTW must equal the amount of toxics removed by the direct discharger. However, as we discuss in greater detail below, EPA’s 1984 rule fails to require the same consistency in the removal of toxics by POTWs and indirect dischargers as is required of direct dischargers. It therefore violates the statutory requirement of section 307(b)(1) of the Clean Water Act that the indirect discharge of pollutants through a POTW must “not violate that effluent limitation or standard which would be applicable to such toxic pollutant if it were discharged by [a direct discharger].” 33 U.S.C. § 1317(b)(1) (1977). The BAT (Best Available Technology) limitations and standards issued under the statute require direct dischargers to remove a certain amount of each toxic pollutant with a certain degree of consistency. BAT limitations and standards have at all times contained these two requirements. Thus, the statutory mandate that the indirect discharger and the POTW together must achieve the same standard of treatment as is required of the direct dis-charger requires that the standard of treatment be the same both in total amount of toxics removed and in the consistency of such removal. The BAT limits applicable to direct dis-chargers are based on empirical studies of the amount and consistency of removal that can be achieved by a well-designed and operated plant. These limits require consistency of removal in two important respects. First, they are set so that a dis-charger can be in compliance virtually all the time. See, e.g., General Pretreatment Regulations, 43 Fed.Reg. 27743 (1978). “Guidelines are generally calculated with a 99% confidence level. Therefore, if a dis-charger exceeds the effluent limitations established by the guideline regulation, there is a 99% certainty that it was caused by discharger error rather than statistical variation.” National Pollutant Discharge Elimination System Permit Regulations, 49 Fed.Reg. 38019 (1984). Second, the limits fix precise daily maxima as well as monthly averages, neither of which may be exceeded by the direct discharger. The indirect discharger and the POTW must achieve an equivalent consistency in these two respects. When Congress enacted the 1977 amendments, many of the BAT effluent limitations applicable to direct dischargers had already been promulgated and others were near promulgation. All of them were set so that they could be met 99% of the time, and all of them contained monthly averages and daily maxima. See, e.g., 40 C.F.R. §§ 129.102-129.104 (1984) (effluent standards for endrin, toxaphene, and benzidine); 40 C.F.R. pts. 405-469 (1984) (effluent limitations guidelines for various industrial categories). Congress, in its 1977 amendment to section 307(b)(1) of the Clean Water Act, used the term “effluent limitation or standard,” thereby incorporating the consistency requirement of that term into the removal credit provision and making the requirement applicable to POTWs and indirect dischargers. As previously noted, the Clean Water Act of 1977 not only required indirect dischargers to meet standards equivalent to the BAT standards required of direct dischargers, but also provided for removal credits to avoid redundant treatment of wastes by the POTW and the indirect dis-charger. It is evident to us, from our reading of the statute, that such removal credits should be made available only when both the consistency and amount of treatment are indeed equivalent to BAT standards. Under the guise of preventing redundant treatment, EPA has defined the term “consistent removal rate” to encompass POTW pollutant removal that is not consistent at all. Yet in promulgating the first full removal credit rule, in 1978, EPA stated: Since direct dischargers are required to comply with their effluent limitations at all times, the EPA believes that the pollutant removal claimed by a POTW should be that removal which occurs virtually all of the time. As used in the regulation (§ 403.7), “consistent” removal is the removal capability that a POTW achieves in 95% of the representative samples' taken. 43 Fed.Reg. 27743 (1978). By permitting inconsistent removal, EPA also permits the granting of credits for treatment that is not redundant but mandatory under the statute. The 1978 rule required the POTW to conduct 12 influent/effluent samplings each year. The consistent removal rate was defined as the lowest of the 12 removal rates in the sample, under the assumption that the POTW would be removing at least that amount 95% of the time. Because POTW removal is more variable than that of direct dischargers, and because EPA believed that it would be too burdensome for POTWs to determine whether they are removing a toxic with 99% consistency, EPA adopted the 95% consistency figure for POTWs as substantially equivalent to the 99% figure applicable to direct dischargers. In response to complaints from indirect dischargers and POTWs that the 1978 removal credit provision was “unworkable” and that most POTWs would not apply for credits, this definition of consistent removal was amended in the 1981 rule to provide that the level of removal on which the credit would be based would be that achieved by a POTW 75% of the time. See 46 Fed.Reg. 9424 (1981). Seventy-five percent consistency was measured by averaging the lowest six of 12 POTW removal samplings. Id. As noted above, the regulated dischargers still claimed that the removal credit rule was “unworkable,” and pressed successfully for a further relaxation of the measurement of consistent removal. The 1984 rule, in fact, provides that “consistent removal is calculated ... as the difference between the average influent and effluent concentrations in all of the sample data.” 49 Fed.Reg. 31215 (1984). Instead of basing its measure of consistent removal on an average of the lowest six of the twelve samplings, as the 1981 rule required, the 1984 rule bases its measure on the average amount removed in all twelve samplings. According to this formulation, the amount of removal achieved by a POTW on average, that is, 50% of the time, will determine the amount of credit an indirect discharger will receive. In short, EPA first changed the definition of “consistent” removal from its original meaning, i.e. removal that occurs 95% of the time, to a second meaning, i.e. removal that occurs 75% of the time. EPA, then, in its 1984 rule, revised its definition of consistency to refer to removal that occurs only 50% of the time. We find it difficult to fathom how a level of removal that is met one half of the time and exceeded one half of the time, and that contains no limit on the permissible amount of variability, can be termed “consistent.” EPA provides two justifications for the change in the definition of “consistent” removal from 75% consistency to 50% consistency. These justifications are the same ones advanced by EPA in 1981 for its shift from 95% to 75% consistency. The first justification is that the approach used in the 1978 rule, which based the measure of consistent removal on the lowest of 12 sample removals, is “statistically unsound.” Brief for Respondent at 32 (quoting 1981 Final Rule, 46 Fed.Reg. 9424 (1981)). In 1981, in justifying its shift from 95% consistency to 75% consistency, EPA outlined its thinking on this issue with commendable candor. EPA pointed out that using the lowest of the 12 samplings to define consistent removal might easily give rise to error, because “[t]he data at the extremes of this 12 point distribution have the greatest chance of being in error. Thus, if the lowest level of Removal identified were unrepresentatively low, the POTW would be held to an unreasonably small level of Removal.” 1981 Final Rule, 46 Fed.Reg. 9424. It would be better, EPA said, to estimate the amount removed 75% of the time rather than trying to estimate the amount removed 95% of the time, because the amount removed with 75% consistency could be measured reliably without increasing the number of samplings. EPA admitted, in 1981, that the choice between 95% consistency and 75% consistency was not a purely technical choice of the one statistically correct method. It stated frankly that it could estimate the amount of removal required to attain 95% consistency, and thus cure the “statistical unsoundness,” by increasing the number of annual samplings required of POTWs. The choice of the 75% level was “admittedly a compromise,” id., between two policy considerations. Because POTWs generally have little incentive to apply for authorization to grant removal credits, EPA was concerned that POTWs would not apply for such authorization if the sampling and reporting requirements imposed upon them were too onerous. On the other hand, EPA was concerned that the rule “ensure that a reasonably consistent level of removal is maintained.” 46 Fed.Reg. 9424 (1981). Even if EPA’s reasoning might have justified a change from a 95% to a 75% consistency requirement, that same reasoning cannot justify a change from a 75% to a 50% consistency requirement if such a change violates the command of the statute. That command, as we have repeatedly observed, requires that the amount and variability of toxic discharges through a POTW not exceed those that would be discharged by a direct discharger operating under BAT-level controls. The relevant question is not one of finding the one correct method of measuring consistent removal, nor of determining the easiest method of measuring consistent removal, but rather whether a removal credit based on the amount a POTW removes 50% of the time conforms with the requirement of the Act. We are convinced that it does not. In its brief, EPA makes much of the fact that using all twelve samplings gives “ ‘a more reliable estimate of the actual removal achieved than [did] the method employed in the 1981 amendments’.” Brief for EPA at 34 (quoting Proposed Rule, 47 Fed.Reg. 42700 (1982)) (emphasis added by EPA). EPA’s definition may indeed provide a more reliable estimate of the actual removal on average. But both the statute and EPA’s own previous regulations require that the indirect discharger, in combination with the POTW, meet the same standard as the direct discharger must meet. This means not only that the total amount removed by the POTW and the indirect discharger together must be at least equal to that removed by a direct discharger operating under BAT controls, but also that the POTW and the indirect discharger must consistently meet the monthly and daily limits that direct dischargers meet. EPA’s second justification for its new definition of consistency is that it gives full effect to Congress’ desire to avoid redundant treatment while still complying with the mandate of section 307(b)(1) that the ultimate discharge from the POTW must not be greater than that which would be allowed from a direct discharger. In its statement of basis and purpose accompanying the 1981 rule, EPA argued that although “[i]n most cases the proposed calculation will indeed result in higher removal allowances and consequently less stringent ... pretreatment limits,” this would not lead to the discharge of “unacceptable amounts of pollutants ... to navigable waters.” 46 Fed.Reg. 9424. EPA explained why, in its view, the change in the definition of consistent removal would not lead to POTW discharges in excess of BAT limits: Industrial users are subject to “daily maximum” and “long term average” pretreatment limits. If the Industrial User is to meet the long term average, the User can only infrequently approach the daily maximum number in its daily discharge. For the joint treatment provided by the Industrial User and POTW to be less effective than that required of a direct discharger, a discharge by the User at the daily maximum level would have to coincide with abnormally lower removal at the POTW. The statistical complexities of the situation do not permit a numerical estimate of the number of times this might occur, but EPA expects it to be small. One comment received by EPA on this subject stated that a computer simulation of the problem showed that violations of the daily maximum would occur less than 2% of the time. This simulation made certain assumptions concerning the statistical distributions which EPA is unable to verify, and EPA did not rely on this result in reaching a decision on the final method of calculation of Consistent Removal. 1981 Final Rule, 46 Fed.Reg. 9424 (emphasis added). The same explanation is reiterated in EPA’s Rebuttal to NRDC’s Reply Brief. In essence, EPA’s argument rests on the claims that the POTW will rarely perform below its average, that the indirect discharger will virtually never exceed its limit, and that the frequency with which these two events will occur simultaneously is even lower. EPA is correct in asserting that the indirect discharger will rarely exceed its limit. That limit is set so that it can be complied with virtually all the time, and this remains true even when the limit is raised to credit the indirect discharger for the POTW’s additional removal of pollutants. But there is no evidence whatsoever to support the claim that the POTW will rarely perform below its average. In fact, this claim is blatantly contradicted by a wealth of evidence in the record, including repeated statements by EPA itself that POTW removal is extremely variable. In 1977, EPA cautioned that “extreme variability in pollutant removals experienced by POTWs” was among the factors to be taken into account in defining an approach to the granting of credits. Proposed Pretreatment Regulations, 42 Fed.Reg. 6485 (1977). In promulgating the 1981 rule, EPA noted that “industrial pretreatment provides much superior removal of pollutants than does treatment at the POTW.” 1981 Final Rule, 46 Fed.Reg. 9406. This is true, EPA said, first, because POTWs are not designed to remove toxic pollutants, and second, because “the dilution [by sewage, other industrial wastes, and, on some days, by rainwater] that occurs at the POTW causes less efficient removal ... than would be achieved by the Industrial User with its more concentrated waste stream____ Partially because removal of toxic pollutants by the POTW is incidental to its normal operations, it is also variable____ Removal of toxic pollutants by the POTW will ... be more variable than removal by treatment technologies designed to remove such toxics.” Id. at 9406, 9407. Most recently, in promulgating the 1984 rule, EPA once again emphasized the lack of consistency in POTW removal. 1984 Final Rule, 49 Fed.Reg. 31216. And, in its brief to this court, EPA declares that “the variability of daily removal estimates [for POTWs] may be quite large.” Brief for EPA at 35. The data from EPA’s 1978 study of 40-POTWs provide ample confirmation of the variability of POTW removal. These data clearly reveal that a POTW may remove virtually all of a particular pollutant on day one while removing little or none of that pollutant on day two. See, e.g., Determining National Removal Credits for Selected Pollutants for Publicly Owned Treatment Works, EPA 440/2 82-008, at B25-B28 (1982) (showing removal of silver on 6 different days for 16 POTWs). In contrast, most BAT limits provide that the discharge for any one day cannot be more than double the average limit for the month as a whole. See 40 C.F.R. pts. 405-469 (1984). The amount of variability (or inconsistency) in POTW removal is particularly important where the removal credit is great. In this situation, most of the removal required in reaching BAT-equivalent levels of treatment will be performed by the POTW, not by the indirect discharger. Thus, the consistency of performance that the indirect discharger achieves in removing the small share of the toxics that it must remove will be relatively unimportant in comparison to the large amount of variability in the POTW’s performance in removing the larger share for which it is responsible. See NRDC’s Reply to EPA’s Rebuttal. It is not at all unusual for removal credits to be in the 80 to 90 percent range. Especially using the method of calculating consistent removal permitted by the 1984 rule, such large credits are very common for many toxics. See, e.g., applications for removal credit authority from Chicago and Los Angeles County, App. 340-41, 344, 346-48. In such cases, the POTW by itself is responsible for removing between 80% and 90% of the pollutant. When this is so, the variability in the removal performance of the POTW will be almost fully reflected in variability in the final discharge into navigable waters. As noted above, EPA admitted in 1981 that it had no reliable evidence to support its contention that the joint treatment by the indirect discharger and the POTW operating under a 75% consistency requirement would only rarely exceed the BAT limits that direct dischargers must meet. Yet in 1984, again without presenting any evidence, and in the face of a substantial body of evidence to the contrary collected by EPA itself, including EPA’s own assertions, EPA claims that a measure of average removal (or 50% consistency) will not lead to violations of the variability limits imposed on BAT dischargers. In fact, as we have seen, the evidence on POTW variability in the record shows that the measure of consistent removal under the 1984 rule will frequently allow discharges very substantially above the daily and monthly maxima that BAT dischargers must not exceed. Under EPA’s rule, the parity in removals between direct and indirect dischargers that is mandated by the statute is guaranteed, if at all, only over the space of a year. If there could be any doubt regarding the fact that EPA’s 1984 rule is in derogation of the statute’s mandate, it is dispelled by a consideration of the legislative history of the 1977 Act. In enacting the removal credit provision in 1977, Congress had two purposes. Congress’ first, and overriding, concern was to ensure that the combined treatment by the indirect dis-charger and the POTW is equal to that by the direct discharger operating under BAT limits. Secondly, as a subordinate goal, Congress sought to avoid redundant treatment by the indirect discharger and the POTW insofar as this is possible without compromising the primary goal of parity. The first major water pollution control legislation, the Clean Water Act of 1972, stated in its first section, entitled “Declaration of Goals and Policy,” that “it is the national policy that the discharge of toxic pollutants in toxic amounts be prohibited.” 33 U.S.C. § 1251(a)(3). As we have noted above, Congress’s concern with the problem of toxic pollutants led it to strengthen the provisions for dealing with toxics in the 1977 Act. The Conference Report was explicit in stating that “[t]he combination of pretreatment and treatment by the municipal treatment works shall achieve at least that level of treatment which would be required if the industrial source were making a direct discharge.” Leg.Hist. 271 (emphasis added). This point was reiterated by Representative Roberts, the House manager, in presenting the conference bill to the House. Leg.Hist. 343. The Senate Report suggests that Congress wished removal credits to be granted sparingly. The Report notes that with the exception of a few POTWs specifically designed to treat industrial wastes, POTW treatment of wastes is too variable to allow the assumption of a specific level of removal. Comm, on Environment and Public Works, Clean Water Act of 1977, S.Rep. No. 370, 95th cong., 1st Sess. 58 (Leg.Hist. 691). The Report also points out that the toxics removed through secondary treatment by a POTW would simply be transferred to the sewage sludge, creating further disposal problems. “In the. long run,” the Report notes, “the only real solution to the problem of safe disposal of toxic or hazardous industrial pollutants is in their reuse and recycling by industry, not the transfer of such materials from [an] industrial waste stream into municipal waste streams____” Id. This can only be accomplished by pretreatment of the pollutants by the industrial discharger. Congress’s reasoning suggests that it did not intend the removal credit provision to be interpreted too generously, and it certainly did not intend that provision to be interpreted so as to nullify the protections it was then enacting against toxic pollution. Finally, the EPA itself has stated that the statute requires that POTW removal be highly consistent before a credit could be granted: In order to justify allowing a revision to a categorical pretreatment standard the pollutant removal claimed by a POTW must occur virtually all the time. The EPA believes that a stringent interpretation of “removal” is warranted in light of the policy of the Act to prohibit the discharge of toxic pollutants in toxic amounts (section 101(a)(3)), and the application of section 307(b)(1) to section 307(a)(1) toxic pollutants. 1978 Final Rule, 43 Fed.Reg. 27765 (1978). Nothing in the language of the statute or in the legislative history suggests that Congress intended to permit, let alone require, that removal credits be granted when these credits would result in discharges that would violate the daily maximum and monthly average limits under which direct dischargers must operate. We do not believe that Congress’ mandate for the removal of toxic wastes should be subject to exceptions dictated by the convenience of the dischargers. The reason for requiring polluters to meet daily and monthly limits as well as long-term limits is obvious: a single concentrated discharge of a toxic pollutant can do irreparable damage to the ecology of a body of water, killing fish and other life forms. Such excessive toxic discharges cannot be compensated for by a reduced discharge of water during subsequent months. Under EPA’s current definition of consistent removal, discharges could be above the limit for months at a stretch, so long as these above-average months were offset by below-average discharges in other months. We therefore hold that EPA’s definition of consistent POTW removal, i.e. removal that is achieved only 50% of the time, violates section 307(b)(1) of the Clean Water Act, 33 U.S.C. 1317(b)(1). B. Even had we not held that EPA’s 1984 definition of consistent removal violated section 307(b)(1) of the Clean Water Act, we would be obliged to hold that EPA’s enactment of the challenged regulation was arbitrary and capricious. According the utmost deference to the Agency, we still cannot find that EPA has given reasons for its new rule that “could lead a reasonable person to make the judgment that the Agency has made.” Weyerhaeuser v. Costie, 590 F.2d 1011, 1026-27 (D.C.Cir.1978). In 1981, when EPA reduced the consistency requirement from 95% to 75%, it acknowledged that this reduction might lead to violations of the required parity between indirect dischargers and POTWs on the one hand and direct dischargers on the other. Yet in 1984, when EPA further reduced the consistency requirement from 75% to 50%, it did not even address the key question of whether its new measure of “consistent” removal will assure such parity. Instead, in a one-sentence explanation of the change, EPA conclusorily stated that its new measure of consistency “provides a more accurate and equitable estimate of the actual removal achieved than the method employed in the 1981 regulation.” 1984 Final Rule, 49 Fed.Reg. 31215 (1984). This statement is misleading in that it focuses solely on the amount of removal while ignoring the consistency requirement of the statute. The available evidence, to which we have previously referred, indicates that the parity required by the statute will rarely be achieved under EPA’s 1984 rule. The fact that the Agency has “entirely failed to consider an important aspect of the problem [and has] offered an explanation for its decision that runs counter to the evidence before the agency” renders arbitrary and capricious its decision to change the measure of consistent removal to what is in reality a measure of average removal. Motor Vehicle Manufacturers Ass’n. v. State Farm Mutual, 463 U.S. 29, 43, 103 S.Ct. 2856, 2866, 77 L.Ed.2d 443 (1983). Without regard to Congress’s overriding intent that pollutants be eliminated from the nation’s waters, without evidentiary support in the record, and without adequate explanation, EPA has de facto deleted the consistency requirement from the removal credit rule. For this reason, without more, we are satisfied that EPA’s regulation is arbitrary and capricious under the test set out by the Supreme Court in Motor Vehicle Manufacturers Ass’n. v. State Farm Mutual, 463 U.S. 29, 43, 103 S.Ct. 2856, 2866, 77 L.Ed.2d 443 (1983). IV. A. Combined sewers are conduits that transport domestic wastewater, industrial waste-water, and, during periods of wet weather, storm water runoff. Combined sewer systems were built in major American cities before the turn of the century, at a time when the need for separate treatment of wastewater was not yet apparent. POTWs in these systems generally do not have the capacity to deal with the great increase in flow that occurs during rainfall or snowmelt. They therefore have overflow points and treatment plant bypasses to handle the excess flow during such periods. Events of overflow are referred to as “combined sewer overflows” (CSOs). The 1981 removal credit rule contained a provision that adjusted the amount of credit that an indirect discharger could receive so as to take CSOs into account. We conclude that EPA’s deletion of this provision from the 1984 rule, without any plausible explanation, violates section 307(b)(1), and in any event is arbitrary and capricious. The 1978 removal credit rule provided that removal credits could be granted by a POTW that experienced overflow at least once a year only if that POTW was implementing an approved plan to treat and control such overflows. 1978 Final Rule, 43 Fed.Reg. 27765. In response to complaints from indirect dischargers, this provision was modified in the 1981 rule to allow an alternative way for a POTW with overflow problems to grant credits. The POTW could calculate the number of hours per year during which overflow occurs. The removal credit would then be reduced by a percentage equal to the percentage of overflow time during the year. If, for example, overflow occurred during 15% of the year, the removal credit would be reduced by 15%. In promulgating the 1981 rule, EPA explained that if the regulations are to meet the statutory requirement of parity between treatment by the direct discharger and treatment by the indirect discharger plus POTW. it is obvious that the POTW should be credited only with that removal which it actually achieves. Thus, EPA has imposed, through the provisions of [40 C.F.R.] § 403.7, several requirements which ensure that industry standards are relaxed only to the extent that the POTW actually removes the pollutants in question. ****** [Therefore,] the provisions of § 403.7(b) provide that a removal allowance must reflect those periods where industrial pollutant-bearing wastes overflow the POTW and there is, consequently, no actual removal of these pollutants by the POTW. ****** Although some commenters have indicated that the foregoing requirements place unduly burdensome restrictions on the POTW wishing to request a removal allowance, the Agency believes that it has properly interpreted the statute to provide that the POTW will be credited only with that level of removal which is actually and consistently achieved. Support for this interpretation is found in the conference report accompanying the Clean Water Act and in the House debate on the Conference Report [as well as in the Senate debate on the Conference Report]. 1981 Final Rule, 46 Fed.Reg. 9423. Notwithstanding its own interpretation of the statutory requirement, EPA resorted to what it termed a “justifiable compromise” that went beyond giving credit for toxic waste which is actually and consistently removed. 46 Fed.Reg. 9426. In fact, adjusting the credit granted to the indirect discharger by the percentage of time that overflows occur does not take into account the fact that CSOs cause sudden increases in pollutant discharges into navigable waters. “[An] important characteristic of CSO ... is the intermittent nature of the discharge____ The impact of a large combined sewer overflow event on any viable aquatic biota element in the receiving water can be extremely detrimental.” Report to Congress on Control of Combined Sew er Overflow, supra, at ES-4. A strict interpretation of the statute would, therefore, need to take into account the POTW’s performance during an overflow. However, here, as with the definition of consistent removal, the Agency was concerned not to adopt an approach that would be “unduly burdensome for POTWs to implement,” or that would reduce too greatly the number of POTWs able to grant removal credits to their industrial dischargers. 46 Fed.Reg. 9426. Although the 1981 CSO provision was a relaxation of the 1978 rule, and an admitted compromise, it was challenged by industry petitioners in this court. National Association of Metal Finishers v. EPA, 719 F.2d 624 (3d Cir.1983) (NAMF). The petitioners argued that “POTWs will be unable to make verifiable engineering estimates of the hours of overflow, and will thus be unable to grant removal credits.” NAMF, 719 F.2d at 649. This court, upholding the CSO provision, responded that “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.” Id. In 1982, EPA proposed a further relaxation: it would entirely eliminate the overflow compensation requirement from the calculation of removal credits, on the ground that “the overflow adjustment makes a negligible difference in the final removal credit.” 1982 Proposed Rule, 47 Fed.Reg. 42701. It based its conclusion on a 1978 study of 15 POTWs. See Report to Congress on Control of Combined Sewer Overflow, supra, ch. 6 passim. “Relying on frequency modeling of the rainfall characteristics of [10 of the 15] sites, the report found that combined sewers will overflow an average of 7.3% of the time.” 1982 Proposed Rule, 47 Fed.Reg. 42701. A 7.3% adjustment in the removal credit, EPA stated, will led to a negligible adjustment in the discharge limit the indirect discharger is required to meet. The 1984 rule did in fact eliminate the CSO provision. B. The deletion of the CSO provision violates section 307(b)(1) of the Clean Water in two respects. First, EPA may not base removal credits on a purported average figure for the frequency of CSOs. To do so ignores the great variability among POTWs in the number and duration of CSO events and in the amount of bypass that occurs during such events. For many POTWs, overflow would require an adjustment of substantially more than 7.3%. Second the statute does not permit EPA to raise pollutant discharge limits for indirect dischargers by 7.3%, or by any amount that is not de minimis, over those that are required of direct dischargers simply because it feels that an increase in pollutant discharge is not very large. This violates the requirement that treatment by indirect dischargers plus POTWs must equal that by the direct discharger, because it permits the indirect discharger and the POTW to discharge a greater total amount of pollutants, and because it permits POTWs and indirect dischargers to discharge waste that is largely untreated, in complete disregard of any requirement of consistency. The type of average measure relied upon by EPA in deciding to delete the CSO provision is similar to the national removal rate measure proposed (in 1982) and subsequently rejected (in 1984) by the Agency as a method of determining how much credit to grant to indirect dischargers. According to EPA’s 1982 proposal, a national removal rate for each pollutant was to have been set at the level which was met by roughly 75% of 40 POTWs studied by EPA. 1982 Proposed Rules, 47 Fed.Reg. 42699-700. In rejecting the national removal credit proposal as proposed in 1982, EPA stated: The Agency has concluded, upon reconsideration, that Congress intended that a removal credit be granted for a particular pollutant only to the extent that a particular POTW can demonstrate that it removes the pollutant. The language of the statute, buttressed by the legislative history, indicates that removal credits are to be based upon case-by-case removal determinations, rather than upon a nationally determined rate. 1984 Final Rule, 49 Fed.Reg. 31212-13. EPA correctly relied on the language of section 307(b)(1) of the Clean Water Act, on the legislative history of the 1977 Act, and on this court’s holding in NAMF, 719 F.2d 624, to conclude that the enormous variability of removal performance among POTWs made the use of an average removal figure unlawful. For the very same reasons, EPA’s reliance on an average rate of “nonremoval,” i.e. of overflow, cannot support its deletion of the CSO adjustment in the 1984 rule. If EPA concedes, as it does, that removal rates for individual POTWs cannot be determined on the basis of a national sample, it is difficult to see how the duration and importance of CSO events at individual POTWs can be determined on the basis of two studies of a small number of POTWs, (one of ten POTWs, the other of eight POTWs) neither of which can be deemed representative of the entire universe of 1300 municipalities. By using an average 7.3% figure for overflows, EPA fails to take into account that in some cities the percentage of time in which overflows occur is substantially greater. In Philadelphia, for example, overflows occur at nearly double the average rate. Chicago’s 1984 application for authority to grant removal credits, included in the record in this case, states that in two typical pumping stations, overflows occurred 10.9% of the time. See Request of Authority to Revise Categorical Pretreatment Standards 25 (Metropolitan Sanitary District of Greater Chicago, 1984) (App. 343). In addition to ignoring variability in the frequency and duration of CSOs, EPA also bases its decision on the asserted “insignificance” of a 7.3% adjustment in the discharge limit applicable to indirect dischargers. It seems obvious that a 7.3% difference in the total amount of pollutant discharged into navigable waters over the long term is not de minimis. As NRDC points out, EPA has in the past deemed relatively small changes sufficient to warrant promulgating new rules. For example, the Agency promulgated new pretreatment standards for electroplaters for nickel, even though the new limits were only 3% to 8% more stringent than the old ones. Compare 46 Fed.Reg. 9472 (1981) with 48 Fed.Reg. 32487 (1983).