Full opinion text
TABLE OF CONTENTS Page I. FACTS AND PRIOR PROCEEDINGS.__________________- 98 A. Factual Background._______________________________ 98 B. Proceedings.---------1--------------------------- 99 II. APPLICABLE LAW. __________________________:------ 100 III. SUBSTANTIVE ISSUES CONCERNING TOXAPHENE. 103 A. Claims of the Parties.------------------------------- 103 B. Statutory Issues.__________________________________ 104 C. Scope of Review.__________________________________ 106 D. Adequacy of the Basis for EPA’s Regulations.---------- 107 1. EPA’s policy judgments.------------------------ 107 2. EPA’s factual determinations leading to the numerical standard______________________________________ 109 IV. SUBSTANTIVE ISSUES CONCERNING ENDRIN.-------- [¶] A. Statutory Issues.---------------------------------- 110 B. Review of the Support for EPA’s Policy Determinations. 114 1. Application factor.----------------------------- 114 ■ 2. Mixing zone factor. ----------------------- 115 Page V. PROCEDURAL ISSUES.______________________________117 A. Applicable Standards Under the Administrative Procedure Act_____________________________________________ 117 B. Reliance by the Administrator on Staff Assistance.______119 1. Background.__________________________________ 119 2. Administrator’s reliance on proposed findings and conclusions-------- 122 3. Contacts between judicial officer and rulemaking staff. 123 C. Omission of the Tentative Decision.------------ 128 VI. COMPLIANCE TIME. ________________________________ 130 VII. CONCLUSION.______________________________________ 131 Before BAZELON, TAMM and ROBINSON, Circuit Judges. Opinion for the court filed by TAMM, Circuit Judge. TAMM, Circuit Judge: We are called upon in these consolidated eases to review challenges to the Environmental Protection Agency’s (EPA) first regulations limiting discharges into the nation’s waterways of two toxic substances, toxaphene and endrin, under the Federal Water Pollution Control Act Amendments. For the reasons that follow, we uphold EPA’s regulations. 1. FACTS AND PRIOR PROCEEDINGS. A. Factual Background. Endrin is a chlorinated hydrocarbon first introduced about 1950. It has been used as a pesticide for several decades, and is currently used for pest control on crops including cotton and sugar cane. At present, there is only one domestic manufacturer of endrin, Velsicol Chemical Corp. (Velsicol), which produces three to six million pounds a year. 42 Fed.Reg. 2595, 2600 (1977). Its endrin manufacturing plant is located at Memphis, Tennessee, and its discharges eventually reach the Mississippi River and are carried through the lower Mississippi to the Gulf of Mexico. Evidence concerning the dangers of endrin to public health and the environment has been produced over the years since endrin was first introduced. In 1965, results from a study using rats suggested that it was carcinogenic. See 41 Fed.Reg. 31316 (1976). EPA has found that endrin has caused fifty-two reported fish and wildlife kills, largely the result of leakage or runoff during agricultural use. Id. at 31317. However, a massive fish kill in the lower Mississippi in 1963 was traced to discharges from Velsicol’s Memphis plant. 42 Fed. Reg. 2591. Endrin is suspected in particular of threatening the brown pelican, an endangered species sensitive to endrin. 42 Fed.Reg. 2591; 41 Fed.Reg. 31317. Regulatory authorities have taken a number of steps to control endrin. The Food and Drug Administration has established a ceiling on endrin levels in food. 42 Fed. Reg. 2591. Velsicol discharged, on the average, approximately seven pounds of endrin per day in the mid-1960’s, Joint Appendix (App.) I 182, but, under public and governmental pressure, it has taken steps to reduce this discharge. In June 1974, EPA issued Velsicol a National Pollutant Discharge Elimination System (NPDES) permit allowing an average daily discharge of no more than one pound of endrin. Velsicol failed to reduce its endrin discharge below an average of 2.5 pounds per day in 1975, App. I 184, and it was found in violation of its permit and was subject to civil penalties. United States v. Velsicol Chemical Corp., 9 ERC (BNA) 1723 (W.D.Tenn. 1976). Under the regulations now under review, Velsicol is required to limit its discharge to approximately .005 pounds of endrin per day. 42 Fed.Reg. 2594 & n.1. Toxaphene is also a chlorinated hydrocarbon pesticide. It has been used for several decades, and is currently used for pest control on cotton and livestock. In the recent past, toxaphene was produced by four manufacturers. See 41 Fed.Reg. 23590 (1976). Hercules, Inc. (Hercules) contends that it is now the only manufacturer that discharges any toxaphene into waterways, and thus, the only manufacturer affected by these regulations. In 1975, total domestic production of toxaphene was approximately one hundred million pounds. App. I 205-11. Hercules’ toxaphene manufacturing plant is located at Brunswick, Georgia, and its discharge flows into the Brunswick Estuary, an arm of the Atlantic Ocean. Evidence concerning the danger of toxaphene to public health and the environment, particularly to fish, has been produced in the years since the chemical was first introduced. Application of toxaphene to agricultural land surrounding several bird refuges resulted in massive bird kills in 1960-62. App. I 157. Toxaphene was identified as a cause of fish kills because of the characteristic “broken back” syndrome it produces in fish. In 1969, strobane, a pesticide closely related to toxaphene, was identified as a carcinogen in mice. App. I 148. Toxaphene has been frequently found in clarified and treated municipal drinking water. App. I 152. Prior to 1970, Hercules’ plant operated without a pollution treatment system; and, as Hercules concedes, its discharge “had an adverse effect upon the ecology” of local waters. Brief for Petitioner Hercules at 4. Subsequently, Hercules received an NPDES permit from the State of Georgia limiting its allowable toxaphene discharge. By 1975, Hercules reduced its average discharges of toxaphene to .27 pounds per day. App. I 206. Under the regulations now on review, Hercules is required to reduce its discharges to approximately .00375 pounds per day. 42 Fed.Reg. 2606 & n.2. B. Proceedings. The early attempts of EPA to limit discharges of toxic substances are described in the opinion of this court in EDF v. EPA [PCBs], 194 U.S.App.D.C.- at ---, 598 F.2d 62, at 68-70 (1978), decided today, and thus, there is no need to repeat that saga in detail. Endrin and toxaphene were on EPA’s July 1973 proposed list of toxic substances and its September 1973 final list. 38 Fed.Reg. 18044, 24342 (1973). In December 1973, EPA proposed standards for endrin and toxaphene, 38 Fed. Reg. 35388 (1973); but after the 1974 hearings, EPA failed to issue final standards. After the 1974 hearings, EPA subjected endrin and toxaphene to further investigation and analysis. EPA conducted its own experiments concerning the effects of endrin and toxaphene on a variety of species at its National Environmental Research Laboratory in Gulf Breeze, Florida. Based on these experiments and on other studies, the Chief of the Criterion Branch of EPA’s Office of Water Planning and Standards, Dr. Leonard J. Guarraia, prepared Criteria Documents summarizing the available information on endrin and toxaphene. On June 10, 1976, EPA proposed standards for endrin and toxaphene. 41 Fed.Reg. 23576 (1976). The standards were of two kinds: concentration limitations on the concentration of pollutant allowed in the discharges, and mass limitations on the amount of pollutant allowed in the discharges. For endrin, the proposed concentration limitation was 1.5 parts per billion, and the proposed mass limitation was 0.0003 kg/kkg (kilogram [kg.]/thousand [kilo] kilogram [kkg.]) of endrin produced. 41 Fed.Reg. 23595. For toxaphene, the proposed concentration limitation was 1.5 parts per billion, and the proposed mass limitation was 0.00001 kg/kkg of toxaphene produced. Id. In July 1976, EPA commenced an evidentiary hearing on the proposed endrin and toxaphene standards before an administrative law judge (ALJ), which was concluded on October 12 after sixteen days of testimony. EPA presented the testimony of twelve witnesses. Velsicol presented the testimony of four witnesses concerning endrin, and Hercules presented the testimony of six witnesses concerning toxaphene. The Environmental Defense Fund (EDF) presented the testimony of one witness concerning the carcinogenic effects of endrin and toxaphene. 42 Fed.Reg. 2592, 2602. Three other parties participated in the hearing and twelve persons filed written comments. At the close of the hearing the record was furnished to the EPA Administrator (Administrator) for his consideration. On January 3, 1977, the Administrator filed his final decision concerning endrin and toxaphene with the EPA hearing clerk, and on January 12, the decision was published. 42 Fed.Reg. 2588. The final decision adopted the proposed concentration limitations for endrin and toxaphene, but adopted relaxed mass limitations of 0.0006 kg/kkg of endrin produced and 0.00003 kg/kkg of toxaphene produced. On the day the final standards were filed, the parties engaged in a customary “race to the courthouse.” At 9:38 A.M., Hercules petitioned the United States Court of Appeals for the Fourth Circuit for review of the toxaphene regulations. However, the petition was premature as the standards were not filed until later in the day; Hercules filed a renewed petition for review two weeks later. At 11:13 A.M., the moment EPA filed its standards, EDF petitioned this court for review of the standards for both endrin and toxaphene. At 11:13 A.M., Velsicol also petitioned the United States Court of Appeals for the Sixth Circuit for review of the endrin regulations. Pursuant to motions by EPA, the Hercules and Velsicol petitions were transferred to this court. Subsequently, EDF’s petition for review was dismissed on voluntary motion. This court denied joint motions by Hercules and Velsicol for retransfer of their petitions and for expedition of the appeals. Compliance with the standards was required by January 12, 1978. II. APPLICABLE LAW. EPA adopted the regulations now under review pursuant to authority conferred by the Federal Water Pollution Control Act Amendments of 1972 (1972 Act or the Act), 33 U.S.C. §§ 1251-1376 (1976). Almost one year after EPA published its final regulations, while this case awaited oral argument in this court, Congress amended the toxics provision of the 1972 Act in section 53 of the Clean Water Act of 1977 (1977 Amendments or the Amendments), 33 U.S.C.A. § 1317 (1977). Velsicol argues that the subsequent congressional action requires us to remand the toxic effluent standards for endrin for new proceedings under the 1977 Amendments. We do not agree. Courts apply the statutory law as it exists at the time of decision unless the intent of Congress is otherwise. Bradley v. School Board, 416 U.S. 696, 715-16 & n.21, 94 S.Ct. 2006, 40 L.Ed.2d 476 (1974). We conclude that this is a case in which it was intended that the new legislation not affect regulations that were awaiting judicial review prior to enactment of the legislation. Compare In re District of Columbia Workmen’s Compensation Act, 180 U.S.App.D.C. 216, 219-222, 554 F.2d 1075, 1078-81, cert. denied, 429 U.S. 820, 97 S.Ct. 67, 50 L.Ed.2d 81 (1976) with De Rodulfa v. United States, 149 U.S.App.D.C. 154, 161, 164-167, 461 F.2d 1240,1247, 1250-53, cert. denied, 409 U.S. 949, 93 S.Ct. 270, 34 L.Ed.2d 220 (1972). Congress passed the 1972 Act in an effort to deal with the regulatory problems that had hindered efforts to control water pollution. EPA v. State Water Resources Control Board, 426 U.S. 200, 202-03, 96 S.Ct. 2022, 48 L.Ed.2d 578 (1976). The 1972 Act solved some regulatory problems, but not others. By 1977, Congress concluded that the toxics provision of the 1972 Act, which provided for regulation based on health criteria, had been a failure. As a reaction to inadequacy of the 1972 toxics provision, EPA commenced a new program to issue regulations under other provisions of the 1972 Act: it would regulate toxic discharges on an industry-by-industry basis and set standards based on the feasibility of control technology. EPA would also continue to issue health-based standards. EPA’s new program was sanctioned by the judicial decree in NRDC v. Train, 8 ERC (BNA) 2120, 2122 (D.D.C.1976) (“Flannery decree”), rev’d in part on other grounds sub nom. NRDC v. Costle, 183 U.S.App.D.C. 11, 561 F.2d 904 (1977). In the 1977 Amendments, Congress remodeled the toxics provision based on the judicial decree in NRDC v. Train. In section 53(a) of the 1977 Amendments, 33 U.S.C.A. § 1317(a)(2) (1977), Congress directed EPA to take into account, in issuing health-based regulations, “the extent to which effective control is being or may be achieved under other regulatory authority,” particularly EPA’s authority to promulgate feasibility-based regulations. See 33 U.S. C.A. § 1311(b)(2)(C) (1977). Velsicol contends that this provision shows that Congress intended EPA to withdraw the health-based toxics regulations promulgated under the 1972 Act, particularly the endrin regulations, and to regulate endrin discharges using feasibility-based criteria. EPA responds that the 1977- Amendments were intended to supplement its authority, and were not intended in any way to impede promulgation of health-based regulations. On a question of statutory interpretation such as this one, involving legislation enacted with EPA’s advice and cooperation, EPA’s view of the legislative intent concerning its ongoing efforts deserves some deference. See E. I. du Pont de Nemours & Co. v. Train, 430 U.S. 112, 134-35, 97 S.Ct. 965, 51 L.Ed.2d 204 (1977); Union Electric Co. v. EPA, 427 U.S. 246, 256, 96 S.Ct. 2518, 49 L.Ed.2d 474 (1976); Train v. NRDC, 421 U.S. 60, 75, 87, 95 S.Ct. 1470, 43 L.Ed.2d 731 (1975). Congress knew that EPA’s health-based regulations were in effect and were awaiting judicial review. See 123 Cong. Rec. H12929 (daily ed. Dec. 15, 1977) (staff report). Moreover, its attitude was not that EPA had done too much and should start all over, but rather that EPA had done too little. “Frankly, [Section 307(a)] has failed. . . . [0]nly six toxic chemicals — aldrin/dieldrin, endrin, DDT, toxaphene, benzidene and PCB’s — have been regulated. Six chemicals in 5 years.” Id. at H12927 (remarks of Rep. Roberts) (emphasis added). Congress’s strong support for continuing health-based regulation is shown not only by its impatience with EPA’s pace, but also by the great emphasis placed in the 1977 Amendments on making it easier for EPA to promulgate health-based regulations as a supplement to feasibility-based regulation. See 123 Cong.Rec. S19649 (daily ed. Dec. 15, 1977) (remarks of Sen. Muskie) (“these amendments reduce the evidentiary burden of establishing a toxic effluent standard”); S.Rep.No.370, 95th Cong., 1st Sess. 54 (1977); U.S.Code Cong. & Admin.News 1977, p. 4378 (lifting of formal rulemaking requirement for health-based regulation). The provision relied on by Velsicol merely directs EPA to “take into account” its “other regulatory authority”; it created no new threshold barrier in health-based regulation. Senator Muskie stated that the test for regulating a pollutant under health-based criteria instead of feasibility-based criteria is simply whether “there is sufficient information on toxicity to establish a separate nationwide effluent standard for that pollutant.” 123 Cong.Rec. S19649 (daily ed. Dec. 15, 1977). EPA’s lengthy and detailed final decision on endrin, 42 Fed.Reg. 2590-601, both demonstrates and supports its view that there is sufficient information on endrin to establish such a standard. Velsicol emphasizes that Congress listed endrin among the sixty-five families of toxic substances in the 1977 Amendments. It maintains this must have been intended to require EPA to reconstruct its regulations under the feasibility-based provisions of the 1977 Amendments. However, the list of substances in the 1977 Amendments was modeled, in large measure, on the judicial decree in NRDC v. Train. See note 14 supra. That decree expressly directed EPA to proceed with health-based regulation of endrin at the same time as it proceeded with feasibility-based regulation of other substances. 8 ERC (BNA) at 2128-29. The 1977 Amendments required, “at a minimum,” section 307(a)(2), that EPA publish effluent limitations based upon best-available technology for all pollutants on the list; the list was in no way intended to preclude health-based standards if they were warranted. 123 Cong.Rec. S19663 (daily ed. Dec. 15, 1977) (remarks of Sen. Randolph). Thus, we conclude that the 1977 Amendments were intended to aid, not to impede, EPA’s health-based regulation. In accordance with EPA’s view that this intent would be frustrated if the regulations were nullified, we conclude that the applicable law in this proceeding is the 1972 Act and that no remand is necessary. III. SUBSTANTIVE ISSUES CONCERNING TOXAPHENE. A. Claims of the Parties. In setting discharge standards, EPA’s approach was to follow a procedure consisting in essence of determining, in the laboratory, which aquatic organisms were most vulnerable to toxaphene, measuring how much toxaphene they could take, and then, using these data, calculating a maximum tolerable discharge figure. Hercules challenges EPA’s laboratory-based approach, as well as particular details of EPA’s methodology. EPA’s approach may be viewed as consisting of six steps. In the first, EPA selected a half-dozen species of aquatic organisms for testing that were either important commercially in and of themselves (oysters), or important indirectly as food for other organisms (pinfish). Second, EPA conducted “bioassays” in the laboratory. In these bioassays, EPA exposed the organisms to toxaphene for ninety-six hours. EPA derived, from the bioassay results, the concentration of toxaphene that kills one-half of a group of aquatic organisms. The short-term lethal dose for the tested organisms ranged from 16 parts per billion for oysters to 0.5 parts per billion for pinfish. These figures reflect the vulnerability of fish and other aquatic life to toxaphene. Third, EPA estimated from the short-term lethal dose the maximum tolerable long-term concentration of toxaphene for the tested species. Fourth, EPA used the long-term tolerable concentration level to set an “ambient water criterion.” The ambient water criterion is intended to be the maximum concentration of the toxic substance allowed in the nation’s waterways — a concentration low enough that even the most vulnerable important life in the water will be safe. Fifth, EPA translated the ambient water criterion into the effluent discharge standard, the standard controlling how much toxaphene a plant may discharge legally. Finally, EPA set a “mass limitation” to supplement the effluent discharge limit. Whereas the effluent discharge limit is expressed in terms of a concentration of toxaphene, the mass limitation restates it in terms of the total amount of toxaphene produced. Such a supplemental figure compels a plant to meet the effluent discharge limit by removing toxaphene from its effluent. In the absence of such a mass limitation, a plant could meet the discharge limit simply by watering down its effluent until the effluent was diluted enough to meet the concentration limit. Hercules’ challenge to EPA’s laboratory-based approach relies on alternative data using field studies. Starting in 1970, Dr. Robert Reimold, a Georgia biologist, conducted a field study of aquatic life in Terry Creek, the arm of Brunswick Estuary into which Hercules’ plant discharges effluent. For six years, Dr. Reimold compared the variety of free-swimming species found in Terry Creek with the variety found in Duplin Estuary, a comparable nearby body of water not contaminated by toxaphene. Dr. Reimold believed that toxaphene was not harming the free-swimming aquatic life present in Terry Creek. The level of toxaphene in Terry Creek’s waters was between 1 and 2 parts per billion, which is between 200 and 400 times EPA’s “ambient water criterion” of .005 parts per billion. Accordingly, Hercules asserts that aquatic life can survive higher levels of toxaphene than EPA allows, and thus, that EPA’s regulations are overly strict. Hercules combines this frontal assault with several attacks on the specifics of EPA’s methodology. It contends that EPA’s use of the pinfish as a test species was misconceived, since the pinfish is not an important species in Brunswick Estuary. It also argues that EPA’s toxicity tests were conducted improperly, that the tests failed to support the standard with substantial evidence, and that EPA should not have adopted a mass limitation precluding dilution. EPA replies that its six-step approach was proper, since the toxics provision requires “categorical” standards for toxic pollutants, rather than standards based on local species and local receiving waters. EPA further responds that it was not required to give great weight to Hercules’ evidence, and that its own tests furnished an adequate basis for regulation. B. Statutory Issues. The court has discussed the standards of section 307(a), 33 U.S.C. § 1317(a) (1976), in EDF v. EPA [PCBs], 194 U.S.App.D.C. at -------, 598 F.2d at 79-81. As explained there, section 307(a)(2) of the 1972 Act lists six factors for EPA to “take into account” in proposing a toxic effluent standard: “the toxicity of the pollutant, its persistence, degradability, the usual or potential presence of the affected organisms in any waters, the importance of the affected organisms and the nature and extent of the effect of the toxic pollutant on such organisms.” EPA is to set discharge standards that provide an “ample margin of safety.” Section 307(a)(4). Under the “ample margin of safety” directive, EPA’s standards must protect against incompletely understood dangers to public health and the environment, in addition to well-known risks. See EDF v. EPA [PCBs], 194 U.S. App.D.C. at ---, 598 F.2d at 80-81. Hercules’ challenge to the regulations concentrates on the last three factors in section 307(a): “the usual or potential presence of the affected organisms in any waters, the importance of the affected organisms and the nature and extent of the effect of the toxic pollutant on such organisms.” Hercules argues that these factors preclude EPA from basing its calculations upon experiments performed on the pinfish, and require it to focus exclusively on organisms actually found in the particular receiving waters into which Hercules discharges. EPA’s testing scheme, Hercules contends, is “contrary to the statutory requirement that EPA design its standards to fit the characteristics of specific water bodies in terms of ‘the toxicity of the pollutant’ in those waters and ‘the importance of the affected organisms’ in those waters.” Brief of Petitioner Hercules at 59 (emphasis added). We agree with EPA’s interpretation of the statute. See, e. g., E. I. du Pont de Nemours & Co. v. Train, 430 U.S. at 134-35, 97 S.Ct. 965. As reported, the House version of the 1972 Act, H.R. 11896, 92d Cong., 2d Sess. § 307(a)(1) & (2) (1972) (emphasis added), contained a factor for toxic standards related to local conditions, as Hercules would prefer: “the usual or potential presence of the affected organisms in [the receiving] waters.” The Senate version did not include any factors relating to “affected organisms.” S. 2770, 92d Cong., 1st Sess. § 307 (1971). In conference, the House factors were accepted, but the phrase “in the receiving waters” was replaced by “in any waters.” The Conference Report notes: Sections 307(a)(1) and (2) are the same as the comparable provisions of the House amendment, except that the Administrator is required to take into account the usual or potential presence of the affected organisms in any water rather than in the receiving waters as is provided in the House amendment. 1 Senate Comm, on Public Works, 93d Cong., 1st Sess., A Legislative History of the Water Pollution Control Act Amendments of 1972 (hereafter, Legislative History) at 312 (Comm.Print 1973) (emphasis added). Senator Muskie, a conference manager, explained that the substitution was a compromise to preserve the “categorical determination” established by the Senate in lieu of the focus on particular “receiving waters” in the House version: With regard to toxic pollutant control, the Senate bill and the House amendment differed in provisions for determining whether a pollutant would, in fact, be toxic. The House amendment proposed that there be an examination of the effect of a pollutant on receiving waters to determine toxicity. The Senate bill established a general test in order to assure a categorical determination as to which pollutants were toxic and which were not. The Conference agreement provides specific tests for toxicity as proposed by the House but retains the categorical determination established by the Senate. 1 Legislative History at 173 (emphasis added). The Conference Committee’s focus on a “categorical determination” rather than on specific “receiving waters” reflected a drafters’ leitmotif in the 1972 Act. Pre1972 attempts at water pollution control had been based upon characteristics of local bodies of receiving water. For that reason, in part, they failed: The earliest version of the Federal Water Pollution Control Act was passed in 1948 and amended five times before 1972. Throughout that 24 year period, Congress attempted to use receiving water quality as a basis for setting pollution standards. W. Rodgers, Environmental Law 355-57 (1977). At the end of that period, Congress realized not only that its water pollution efforts until then had failed, but also that reliance on receiving water capacity as a crucial test for permissible pollution levels had contributed greatly to that failure. EPA v. State Water Resources Control Board, 426 U.S. 200, 202, 96 S.Ct. 2022, 48 L.Ed.2d 578 (1976). Weyerhaeuser Co. v. Costle, 191 U.S.App. D.C. 309 at 340, 590 F.2d 1011 at 1042 (1978). The 1972 Act remedied that shortcoming by shifting its focus from difficult-to-measure conditions and benefits in local receiving waters to readily determinable categorical consideration. Congress concluded that water pollution seriously harmed the environment, and that although the cost of control would be heavy, the nation would benefit from controlling that pollution. Yet scientific uncertainties made it difficult to assess the benefits to particular bodies of receiving water. . . . Under the new statutory scheme, Congress clearly intended us to avoid such problems of proof so that a set of regulations with enforceable impact is possible. Id. at 337-340, 590 F.2d at 1039-1042. See also E. I. du Pont de Nemours & Co. v. Train, 430 U.S. at 127, 97 S.Ct. 965. Applying a categorical approach, EPA did not limit its study to the aquatic life found in receiving waters for toxaphene discharges. Instead, EPA tested six important marine species, and from the results of these tests, it inferred the likely sensitivity of other species. Specifically, it chose the most' sensitive species actually tested (pin-fish), and used the test results from that species to set the ambient water criterion. The choice of the most sensitive species actually tested may be understood on at least two levels. Legally, section 307(a)(4)’s fundamental instruction is that EPA’s standards should provide an “ample margin of safety.” Basing the scheme upon the most sensitive species actually tested provides a margin of safety for less sensitive species. Scientifically, the scheme rests upon a statistical assumption: Reliable toxicity data exist for a relatively small percent of all organisms in the aquatic environment for any given substance. It seems a safe assumption to make, that as additional species are tested, some of them are going to be more resistant than those that have already been tested and some are going to be more sensitive. ... If we consider the thousands of species of aquatic organisms which exist in our waterways and assume that only one hundred of these species are important enough to protect for the purpose of man’s use; and if we further consider that perhaps only a half dozen species at most have been tested in most instances, it is highly probable that in any given water, species of greater sensitivity than those already tested, occur. Therefore, it would appear that the use of the most sensitive species tested, to establish acceptable levels, may not be overly conservative, since so few species have been tested and more sensitive important species surely exist. Furthermore, should one of the food chain species be sensitive and essential, an entire food pyramid would be jeopardized. App. I 255-56 (Affidavit of Dr. Donald I. Mount) (emphasis added). C. Scope of Review. The applicable scope of review is discussed in EDF v. EPA [PCBs], 194 U.S. App.D.C. at ---, 598 F.2d at 82-83. EPA’s determinations are subject to review under the “substantial evidence” test, which requires that EPA’s conclusions be supported by “such relevant evidence as a reasonable mind might accept as adequate to support [the] conclusion[s].” Consolidated Edison Co. v. NLRB, 305 U.S. 197, 229, 59 S.Ct. 206, 217, 83 L.Ed. 126 (1938). When agency rulemaking is based on judgments “on the frontiers of scientific knowledge,” this standard calls for deferential review of the agency’s conclusions that “depend to a greater extent upon policy judgments.” Industrial Union Department, AFL-CIO v. Hodgson, 162 U.S.App.D.C. 331, 338, 499 F.2d 467, 474-75 (1974). In this case, we are called upon to review a numerical standard for discharge. In reviewing a numerical standard, we must ask whether the agency’s numbers are within a “zone of reasonableness,” not whether its numbers are precisely right. See, e. g., FPC v. Conway Corp., 426 U.S. 271, 278, 96 S.Ct. 1999, 48 L.Ed.2d 626 (1976) (electric power rates); Permian Basin Area Rate Cases, 390 U.S. 747, 796-98, 88 S.Ct. 1344, 20 L.Ed.2d 312 (1968) (natural gas rates); New York v. United States, 331 U.S. 284, 344, 67 S.Ct. 1207, 91 L.Ed. 1492 (1947) (rail rates); Bilingual Bicultural Coalition on Mass Media v. FCC, 193 U.S.App. D.C. 236 at 244, 595 F.2d 621 at 629 (1978) (en banc) (minority workforce statistics); Federation of Homemakers v. Schmidt, 176 U.S.App.D.C. 261, 264-65, 539 F.2d 740, 743-44 (1976) (food content standards). A principal rationale for the “zone of reasonableness” concept is that it frees the court from the minutiae of particular calculations, and correspondingly, it allows an agency discretion to adapt a general formula or methodology to the aspects of a particular case. Hercules argues that EPA failed to respond adequately to, or to rebut, its showing on particular issues. We shall discuss these issues individually. However, as a preliminary matter, we observe that the toxics provision places the burden of persuasion (in the agency hearing) on the person urging modification of the proposed standard. “[U]nless the Administrator finds, on the record, that a modification of such proposed standard (or prohibition) is justified based upon a preponderance of evidence adduced at such hearings, such standard (or prohibition) shall be promulgated.” Section 307(a)(2) (emphasis added). Since Hercules urged EPA to modify the proposed standard (that is, to promulgate a more relaxed, revised standard), the burden of proof in the agency hearing was on Hercules, not EPA. D. Adequacy of the Basis for EPA’s Regulations. 1. EPA’s policy judgments. Although EPA’s categorical focus led^it to favor a laboratory approach, that focus did not preclude, consideration of field data. Indeed, it considered past fish kills, bird kills, human poisonings, and other incidents, 42 Fed.Reg. 2602-03, as well as Dr. Reimold’s Brunswick Estuary studies, id. at 2604-05. However, EPA concluded that, in this particular case, the Brunswick Estuary study should not be accorded much weight both because of the inherent limitations of such a study and because of the particular limitations on Dr. Reimold’s technique. Id. Hercules vigorously disputes EPA’s conclusion about the Brunswick Estuary study. It argues that EPA failed to present evidence to rebut Dr. Reimold’s study; that neither the Administrator, nor the judicial officer assigned to the proceeding, Ms. Harriet Marple, had sufficient scientific training to reject Dr. Reimold’s approach; and, that the rejection of the field study was unsupported by substantial evidence. [¶] The scientific uncertainty attending the choice is demonstrated dramatically by the different alternatives advanced by Hercules and EPA, each propounded by different scientists. The choice of a suitable technique for estimating toxicity risks is a decision “on the frontiers of scientific knowledge.” The relative scientific and administrative advantages of different techniques are incapable of precise comparison by this court. Decision between the alternatives is a quintessential policy judgment within the discretion of EPA. We cannot accept Hercules’ notion that the administrator of the agency created to protect the environment lacked even the capability to exercise the discretion with which he was entrusted by Congress. Hercules contends that EPA failed to “rebut” the Reimold study with testimony from experts in Dr. Reimold’s field of expertise, ecological analyses. The short answer is that EPA had expert evidence on which it could, and did, base its refusal to credit Dr. Reimold’s study with more weight. Moreover, we would not impose on EPA a wooden requirement that when, as in this case, there is conflicting testimony from experts in different fields (here, toxicology and ecology), EPA may only accept the evidence from one field if it rebuts, point-by-point, the evidence from the other field. Certainly, EPA may compare evidence from different fields, and make a reasoned determination thereon. The issue is not whether EPA has substantial evidence from every scientific field, but whether it has substantial evidence on the record as a whole. Universal Camera Corp. v. NLRB, 340 U.S. 474, 488, 71 S.Ct. 456, 95 L.Ed. 456 (1951); see FPC v. Florida Power & Light Co., 404 U.S. 453, 462-67, 92 S.Ct. 637, 30 L.Ed.2d 600 (1972). Further, we uphold EPA’s decision to set a mass limitation restating its effluent discharge standard. See text at ---of 194 U.S.App.D.C., at 103-104 of 598 F.2d supra. EPA set a mass limitation in order that its effluent discharge standard would not be subverted; its standard would thus be met, not by dilution of effluent with uncontaminated water, but by treatment of the effluent. The 1972 Act favors toxic substance abatement rather than dilution. Section 101(a)(3), 33 U.S.C. § 1251(a)(3) (1976) (emphasis added) expresses a “national policy that the discharge of toxic pollutants in toxic amounts be prohibited.” Obviously, this national policy will be frustrated if discharge of a toxic substance continues" in the same amount, but merely in diluted form. During public hearings on the 1972 Act, then-Administrator Ruckelshaus testified that “we don’t believe that the solution to pollution is dilution.” 2 Legislative History at 1228. Congress accepted his position and “[t]he Conference substitute [for the Senate and House bills] specifically bans pollution dilution as an alternative to waste treatment.” 1 Legislative History at 284. In this respect, we draw support from cases involving clean air legislation that rejected dilution and dispersion of pollutants as a means of compliance. See Kennecott Copper Corp. v. Train, 526 F.2d 1149, 1153-60 (9th Cir. 1975), cert. denied, 425 U.S. 935, 96 S.Ct. 1665, 48 L.Ed.2d 176 (1976); Big Rivers Electric Corp. v. EPA, 523 F.2d 16, 20-22 (6th Cir. 1975), cert. denied, 425 U.S. 934, 96 S.Ct. 1663, 48 L.Ed.2d 175 (1976). See génerally Hines, A Decade of Nondegradation Policy in Congress and the Courts: The Erratic Pursuit of Clean Air and Clean Water, 62 Iowa L.Rev. 643 (1977). 2. EPA’s factual determinations leading to the numerical standard. The Criteria Document prepared for EPA and affidavits presented by EPA and EDF summarized an array of studies and evidence concerning the toxicity (including carcinogenicity), persistence, and degradability of toxaphene. Toxaphene was shown to be toxic to phytoplankton, App. I 133, aquatic invertebrates, App. I 133-35, 292, fish, App. I 135-42, 293-345, fish-eating birds, App. I 142-43, and mammals. App. I 143-49. Evidence was presented that toxaphene is carcinogenic. EPA summarized a study concluding that strobane, a substance related to toxaphene, causes cancer in mice. App. I 148. Preliminary data were presented indicating that toxaphene caused cancer in mice and rats. App. II 645-49. Toxaphene also was shown to cause enzyme induction. App. I 148. On this record, EPA cautiously refrained from declaring toxaphene to be a carcinogen, in part because the data were to recent to make a complete scientific analysis. However, EPA concluded that “the existence of these data underscores the Agency’s concern that discharges of toxaphene be minimized.” 42 Fed.Reg. 2602. EPA also produced physical and chemical evidence that toxaphene is persistent, mobile, and adsorbed to sediments, and biological evidence that toxaphene is not degraded and tends to bioaccumulate in aquatic species. App. I 151-57, 249, 351. Against this background, EPA conducted bioassays in order to derive a numerical discharge standard. Systematic comparable bioassays were conducted on six species, using a uniform methodology. App. I 346-54. Hercules did not offer bioassays or other laboratory studies in opposition, but did criticize EPA’s studies. Hercules contended in the administrative proceedings, as it does now, that the pinfish toxicity data were used to set the ultimate numerical standard, and that the pinfish bioassay was unreliable because the fish suffered artificial “stress,” lowering resistance to toxaphene. EPA disputes the claim of artificial stress. Brief for Respondent in 77-1248 at 40-41. Hercules also objects to EPA’s reference on appeal to evidence of toxaphene’s carcinogenicity, since EPA’s own decision declined to make a finding of human carcinogenicity. We find EPA’s determination of the effluent discharge and mass limitation standards supported by substantial evidence and within the zone of reasonableness. As this court stated in Ethyl Corp. v. EPA, 176 U.S.App.D.C. 373, 410, 541 F.2d 1, 38 (en banc), cert. denied, 426 U.S. 941, 96 S.Ct. 2663, 49 L.Ed.2d 394 (1976), “[b]y its nature, scientific evidence is cumulative: the more supporting, albeit inconclusive, evidence available, the more likely the accuracy of the conclusion.” Although the pinfish bioassay played a significant role in setting the numerical standard, it should not be viewed in isolation. Its persuasive value was enhanced by prior studies of the similarity of many other aquatic species to toxaphene, and by bioassays using five other species performed contemporaneously. Although the support that other tests and studies lend to the standard is not as direct as that provided by the pinfish data, those tests and studies are mostly decidedly part of the agency’s cumulative evidence. With respect to the allegations that EPA’s pin-fish bioassay was itself flawed (e. g., that the fish were under stress) these allegations were refuted by testimony. App. II 726-27. Finally, we do not accept Hercules’ contention that this court may not consider EPA’s conclusion concerning carcinogenicity — that the data “underscore[ ] the Agency’s concern that discharges of toxaphene be minimized.” The record contains data concerning toxaphene’s carcinogenicity for animals. Thus, EPA’s limited conclusion, which articulates its “concern” about toxaphene’s carcinogenicity, is supported in the record. Although EPA did not make an affirmative finding of carcinogenicity, its concern was relevant to the setting of a standard providing an “ample margin of safety,” and therefore relevant in this appeal. See EDF v. EPA [PCBs], 194 U.S. App.D.C. at ---, 598 F.2d at 80-81. IV. SUBSTANTIVE ISSUES CONCERNING ENDRIN. A. Statutory Issues. Velsicol’s principal argument for invalidating the endrin standard is that EPA was required to give consideration to the feasibility of achieving the standard with existing pollution control technology, but failed to do so adequately. It contends that consideration of feasibility is required by “reasoned decision making” and principles of “rational” consideration. Velsicol maintains that the standard under review is, in fact, not achievable. Velsicol refers to Bunker Hill Co. v. EPA, 572 F.2d 1286, 1301 (9th Cir. 1977), a case arising under the Clean Air Act, for the proposition that “[t]he record must establish that the required technology is feasible, not merely possibly feasible.” EPA defends its standard on alternative grounds. It contends that section 307(a) simply does not require it to give any consideration to economic and technological feasibility. Reviewing the factors listed in that section, it points out that “[ajbsent from the list is any requirement that the Administrator consider technological feasibility or economic impact.” Brief for Respondent in No. 77-1349 at 41. See also id. at 53 & n. 30. EPA also contends that the endrin standard is, in fact, feasible. Id. at 53. We agree with EPA’s contention that section 307(a) does not require it to consider economic and technological factors. “[T]he starting point in every case involving construction of a statute is the language itself.’ ” Ernst & Ernst v. Hochfelder, 425 U.S. 185, 197, 96 S.Ct. 1375, 1383, 47 L.Ed.2d 668 (1976) (quoting Blue Chip Stamps v. Manor Drug Stores, 421 U.S. 723, 756, 95 S.Ct. 1917, 44 L.Ed.2d 539 (1975) (Powell, J., concurring)). Section 307(a)(2) requires EPA to “take into account” six factors in proposing toxics standards that comprehensively cover the fate of toxic substances in the environment, their effects on living organisms in general, and their effects on some organisms present and important in particular. EDF v. EPA [PCBs], 194 U.S.App.D.C. at ---, 598 F.2d at 79-80. None of the six factors speaks explicitly to the technical means used to control the discharge of substances, nor is there present any term commonly used to denote a feasibility consideration, e. g., feasibility, achievability, practicability, economic impact, or cost. Cf. section 304(b)(1) & (2), 33 U.S.C. § 1314(b)(1) & (2) (1976) (speaking of practicability and achievability). Section 307(a) itself provides for consideration of no other factors. Cf. section 304(b)(1)(B) & (2)(B) & (4)(B), 33 U.S.C.A. § 1314(b)(1)(B) & (2)(B) & (4)(B) (1977) (EPA may take into account “such other factors as the Administrator deems appropriate.”). An examination of other provisions of the 1972 Act reinforces this interpretation of section 307(a). Section 307(a)(4) directs EPA to set standards providing “an ample margin of safety,” without any mention of feasibility criteria. Further, we do not believe that section 307(a)(5) introduces feasibility criteria into the statutory scheme. Section 307(a)(5) states that “[w]hen proposing or promulgating any effluent standard (or prohibition) under this section, the Administrator shall designate the category or categories of sources to which the effluent standard (or prohibition) shall apply.” The legislative history makes clear that categories would be differentiated not by the feasibility vel non of pollution control, but by the extent to which a toxic substance is present in the industrial process. The legislative background explains why Congress focused on public and environmental protection, rather than discharge control technology, in the setting of toxic standards. The regulatory scheme is similar to that of the Clean Air Act Amendments of 1970, Pub.L.No. 91-604, 84 Stat. 1676, which distinguished between pollutants subject to technology-based regulation under section 111, and hazardous substances, subject to health-based regulation under section 112. Recognizing that “certain pollutants” required special treatment because of risk to health, Congress enacted section 112, dealing with hazardous pollutants, without provision for considerations of feasibility. 1 Senate Comm, on Public Works, 93d Cong., 2d Sess., A Legislative History of the Clean Air Act Amendments of 1970 (hereafter, Clean Air Legislative History) at 133 (Comm.Print 1974) (remarks of Sen. Muskie) (The Senate subcommittee on Air and Water Pollution “was presented with strong evidence that any level of emissions of certain pollutants may produce adverse health effects that cannot be tolerated.” Id. at 227.). The 1972 Act continued this distinction, adopting technology-based standards for pollutants in general and health-based standards for toxic pollutants. Compare sections 301 & 304(b), 33 U.S.C. §§ 1311 & 1314(b) (1976) with section 307(a). “The Committee considers that the discharge of toxic pollutants are much too dangerous to be permitted on merely economic grounds.” 1 Legislative History at 800 (House Report). In light of this clear statutory wording and legislative history, it is not surprising that numerous commentators agree that considerations of technological and economic feasibility do not play a part in standard-setting for toxic substances. Velsicol argues that, if feasibility is not considered, the regulations will be promulgated at an irrationally strict level. However, the congressional selection of factors is a legislative determination that the need of the public and the environment for protection from toxic chemicals is more important than the problems of stringent regulation. This congressional determination is a rational response to the dangers presented by toxic substances. The meaning of the statute being clear, it is not this court’s prerogative to impose considerations of feasibility. See Tennessee Valley Authority v. Hill, 437 U.S. 153, 98 S.Ct. 2279, 57 L.Ed.2d 117 (1978); Union Electric Co. v. EPA, 427 U.S. 246, 96 S.Ct. 2518, 49 L.Ed.2d 474 (1976). It may also be argued that, as a practical matter, it will prove very difficult to administer the statute without considering feasibility, because of the resistance of industry representatives and environmentalists to what will be perceived by them, respectively, as overly or underly stringent regulation. However, our interpretation, which follows in large part EPA’s original interpretation of section 307(a), will not necessarily lead to great difficulty in administering the section, because it does not preclude EPA from gathering information about feasibility. Availability of information about feasibility may tend to induce industry and environmentalists to accept standards based on health criteria. Velsicol points out that EPA did consider feasibility in the proceedings leading to the regulations in issue. Indeed, Velsicol contends that EPA first set its endrin standard in accordance with feasibility criteria and then justified it with evidence related to the section 307(a)(2) factors. This argument raises the possibility that we must remand the agency’s standards on the grounds that EPA took an erroneous step in its deliberations. See SEC v. Chenery Corp., 332 U.S. 194, 196-97, 67 S.Ct. 1575, 91 L.Ed. 1995 (1947). However, upon careful examination of the manner in which EPA considered feasibility evidence, we conclude that a remand is unwarranted. EPA’s consideration of feasibility is more understandable in light of the history of toxics proceedings. In 1973 and 1974, EPA conducted proceedings for promulgating regulations under section 112 of the Clean Air Act and section 307(a). In both the air and the water proceedings, EPA expressly excluded from consideration evidence concerning the economics of pollution control, although it had gathered such evidence. The 1973-74 toxics proceedings under section 307(a) ended in failure; EPA did not adopt toxic pollutant regulations. See EDF v. EPA [PCBs], 194 U.S.App.D.C. at ---, 598 F.2d at 68-69 (discussing this period in detail). EPA’s failure was partially due to its fear of “almost certain challenge in court.” 41 Fed.Reg. 23577 (1976). Out of an abundance of caution, EPA altered its position after 1974 to allow limited consideration of feasibility. In part, it did so on the basis of judicial authority, since overruled, that directed it to consider feasibility in the context of another health-based statute. EPA’s position served to build an administrative record that could support its regulations on appeal if feasibility were ultimately held to be a factor under section 307(a). We have concluded that section 307(a) does not include feasibility as a factor for consideration. However, the fact that EPA considered evidence about feasibility does not require that the regulations be vacated. It is apparent that EPA calculated its endrin standards using health-based, not feasibility-related, evidence. It conducted bioassays of aquatic species, determined the maximum long-term dose of endrin that the most sensitive tested species could tolerate, set an ambient water criterion, and calculated effluent standards and mass limitations. Only then, upon calculating its standards, did it consider evidence about feasibility to provide an additional test. 42 Fed.Reg. 2597-600. It found that the standards were, in fact, feasible. Although consideration of feasibility is not contemplated by section 307(a), EPA’s consideration of it did not affect its conclusions based on the factors specified by section 307(a), but constituted an attempt to confer additional legitimacy on its conclusions. The supererogatory consideration did not prejudice Velsicol in any way. Indeed, Velsicol actually gained an additional opportunity to press for relaxed standards by presenting evidence that the standards were not achievable. We do not find support in the record or in the administrative decision for Velsicol’s contention that EPA proceeded “backwards” in setting the endrin standards. Velsicol invites the court to probe the Administrator’s mental processes in search of an alternative methodology for his decision other than the evidence and the reasons stated. We must decline this invitation. See Citizens to Preserve Overton Park, Inc. v. Volpe, 401 U.S. 402, 420, 91 S.Ct. 814, 28 L.Ed.2d 136 (1971); United States v. Morgan, 313 U.S. 409, 422, 61 S.Ct. 999, 85 L.Ed. 1429 (1941). B. Review of the Support for EPA’s Policy Determinations. Velsicol contests two of EPA’s determinations in calculating the endrin standards. We review the record, keeping in mind that, when issues are “on the frontiers of scientific knowledge,” the “substantial evidence” test calls for deferential review of EPA’s policy judgments. Industrial Union Department, AFL-CIO v. Hodgson, 499 F.2d at 474-75. Both of Velsicol’s contentions concern EPA’s alleged failure to deal adequately with local aquatic organisms and local receiving water conditions at the site of Velsicol’s plant in Memphis, Tennessee. We reject Velsicol’s contentions for the reasons underlying our rejection of Hercules’ argument that EPA was required to focus on conditions at the site of its Brunswick plant. 1. Application factor. EPA used the same six-step technique in setting its endrin standard that it used in setting its toxaphene standard. See App. I 355-59. In considering toxicity, it summarized studies showing that endrin was toxic to a broad range of invertebrates, fish, birds, and mammals. App. I 94-107. Evidence concerning the carcinogenicity of endrin in mice and rats had been produced, 41 Fed.Reg. 31316-17 (1976) (establishing “rebuttable presumption” against continued registration of endrin as a pesticide) and new evidence concerning this was presented in these proceedings, 42 Fed.Reg. 2591-93. Endrin was demonstrated to be persistent and mobile in some circumstances, and while it was sometimes degraded or eliminated, it was bioaccumulated in many organisms. App. I 108-17. With this general data on toxicity, persistence and degradability as background, EPA compared the short-term lethal endrin doses for a number of important freshwater and marine organisms. Pink shrimp was the most sensitive species actually tested, with a short-term lethal dose of .037 parts per billion. To estimate a long-term safe dose for shrimp, EPA multiplied the short-term lethal dose by an “application factor” of .1, a figure derived from endrin tests on an invertebrate, one type of stonefly that is found in mountain streams. EPA then calculated the long-term safe dose for pink shrimp, and it set the ambient water criterion for endrin at .004 parts per billion. See 42 Fed.Reg. 2593-94. Velsicol challenges EPA’s use of shrimp toxicity data and stonefly application factor data. It notes that pink shrimp are not exposed to Velsicol’s endrin discharges until the endrin travels over 700 miles along the Mississippi River to the Gulf of Mexico, and that stoneflies of this species are not exposed to Velsicol’s discharges at all. See Brief of Petitioner Velsicol at 12-16. It argues that data were available for species that do live in the Mississippi, and that EPA should not have preferred the shrimp and stonefly data over data for species in the local receiving waters. As we discussed in connection with EPA’s tests on pinfish in setting toxaphene standards, Congress intended that EPA would use categorical determinations in setting standards for toxic pollutants, rather than limit its consideration to species present in receiving waters. EPA’s decision sets forth at length the reasons for selecting the shrimp and stonefly data as the basis for the standards. 42 Fed.Reg. 2593-94. Choice among scientific test data is precisely the type of judgment that must be made by EPA, not this court. “It is in just such matters that the findings of [EPA], because of its experience and the assistance of its technical staff, should be accorded the greatest weight and the courts should be most hesitant to substitute their judgment for that of [EPA].” FPC v. Florida Power & Light Co., 404 U.S. at 466, 92 S.Ct. at 645 (quoting United States ex rel. Chapman v. FPC, 191 F.2d 796, 808 (4th Cir. 1951), aff’d, 345 U.S. 153, 73 S.Ct. 609, 97 L.Ed. 918 (1953)). 2. Mixing zone factor. Once effluent is discharged, it mixes with the waters at the site of discharge. Having set an ambient water criteria for endrin of .004 parts per billion, EPA set an effluent discharge standard using a “mixing zone” factor. It employed a mixing factor of 375, anticipating that one part of effluent would be mixed with about 375 parts of water before aquatic organisms had significant contact with endrin. As a backstop, EPA also provided for a “tightening variance” — a smaller mixing zone factor for any discharges in waters where mixing is likely to be less thorough. 42' Fed.Reg. 2609, 2614-15. Velsicol contends that the evidence supports a larger mixing zone factor, up to 4,000 in view of the volume and flow of the Mississippi River, and that EPA must also provide for “loosening” variances (large mixing zone factors for thoroughly-mixed waters). We sustain EPA’s decision with respect to the mixing zone factor. The issue of mixing zones is a controversial one. One leading commentator on section 307(a) argued that the 1972 Act “probably does not allow mixing zones at all,” since Congress banned pollution dilution as an alternative to waste treatment. K. Hall, The Control of. Toxic Pollutants Under the Federal Water Pollution Control Act Amendments of 1972, 63 Iowa L.Rev. 609, 628 & n. 109 (1978). On the other hand, Velsicol urges that EPA should take into account the full diluting capacity of the Mississippi, based on the volume, flow, geographical configuration, and other factors that go into calculations about particular receiving waters. EPA’s decision to use a moderate mixing zone factor and to make provision only for “tightening,” not loosening, variances must be deemed a rational, although simple, compromise. By prescribing that EPA focus on categorical factors rather than on characteristics of local receiving waters, Congress sought to keep toxics regulation from becoming ensnarled in disputes over the characteristics of particular local waters, a complex matter Congress left to the setting of water quality standards under section 303, 33 U.S.C. § 1313 (1976). Section 307 regulation was to focus on toxicity and related characteristics of pollutants. Accordingly, Congress left EPA to choose among the available approaches for translating toxicity data into end-of-pipe standards. “An agency confronted with a complex task may rationally turn to simplicity in ground rules, and administrative convenience, at least where no fundamental injustice is wrought.” American Public Gas Association v. FPC, 186 U.S.App.D.C. 23, 63, 567 F.2d 1016, 1056 (1977) (quoting Gulf Oil Corp. v. Hickel, 140 U.S.App.D.C. 368, 374, 435 F.2d 440, 446 (1970)). Having chosen a methodology, the compromise figure set by EPA was rational. A much higher figure for the mixing zone might well have frustrated two congressional policies: that the “margin of safety” provided by the standards be “ample” and that pollution concentration be controlled by treatment, not dilution. We find that the choice of 375 as the mixing zone factor was within the “zone of reasonableness” and, within that zone, the choice of a precise figure is left to EPA. V. PROCEDURAL ISSUES. Hercules and Velsicol challenge EPA’s procedures in two basic respects. First, they contend that the Administrator and the judicial officer impermissibly relied on submissions from, and engaged in contacts with, EPA staff. Second, Hercules argues that EPA committed reversible error in not issuing a tentative decision after the close of the evidence and before the final decision. A. Applicable Standards Under the Administrative Procedure Act. Petitioners maintain that EPA was required to follow the procedure set forth in section 5 of the Administrative Procedure Act (APA), 5 U.S.C. § 554 (1976). They support this contention basically by two arguments. First, section 307(a) states that the Administrator, after proposing standards, shall hold a “hearing” and that he shall make certain findings “on the record” of the hearing. Petitioners maintain that section 554 applies to all proceedings “on the record.” See Reply Brief for Petitioner Hercules at 12-13. Second, petitioners argue that a “functional” analysis of these proceedings shows that EPA must follow section 554. We cannot agree. Section 554 is entitled “Adjudications” and, by its express terms, applies in cases of “adjudication required by statute to be determined on the record after opportunity for an agency hearing.” 5 U.S.C. § 554 (emphasis added). It is well settled that section 554 applies only in eases of adjudication, and not to rulemaking proceedings. Hoffmann-La Roche, Inc. v. Kleindienst, 478 F.2d 1, 13 (3d Cir. 1973); Willapoint Oysters, Inc. v. Ewing, 174 F.2d 676, 692-94 (9th Cir.), cert. denied, 338 U.S. 860, 70 S.Ct. 101, 94 L.Ed. 527 (1949). Thus, the issue before us is whether section 307(a) requires adjudication or rulemaking. We commence our analysis with the APA definitions. “Adjudication” is defined in the APA as “agency process for the formulation of an order.” 5 U.S.C. § 551(7) (1976). An “order” means “the whole or a part of a final disposition . . . in a matter other than rule making.” 5 U.S.C. § 551(6) (1976). “Rule making” is defined as “agency process for formulating, amending, or repealing a ru