Opinion for the Court filed by Circuit Judge HENDERSON.
HENDERSON, Circuit Judge:
The Environmental Protection Agency (EPA), acting pursuant to authority granted by the Comprehensive Environmental Response, Compensation and Liability Act of 1980 (CERCLA), 42 U.S.C. §§ 9601-9675, promulgated a rule regulating the reporting requirement for the release of radionuclides.1 54 Fed.Reg. 22,524 (May 24, 1989). In specific, the rule explains the meaning of “release” as used in CERCLA, provides administrative exemptions to CERCLA’s reporting requirement and sets the minimum level of release of radionuclides that must be reported to the EPA. We vacate the EPA’s interpretation of “release” as contrary to the express language of CERCLA, conclude that the administrative exemptions were not properly promulgated but leave them in place pending new rule-making and decline to reach the challenge to the level set for radon-222 because the petitioners failed to make their claim below.
I.
To address the growing dangers caused by the unregulated dumping and storage of hazardous wastes, Congress enacted CERCLA. As part of the overall program to “provide for a national inventory of inactive hazardous waste sites and to establish a program for appropriate environmental response action,” 1980 U.S.Code Cong. & Admin.News 6119, Congress vested the EPA with the authority to investigate and respond to the release, or threatened release, of hazardous wastes into the environment. 42 U.S.C. § 9604. To effectuate the EPA’s response authority, CERCLA requires parties to notify the EPA whenever a reportable quantity (RQ) of a hazardous substance is released into the environment. 42 U.S.C. § 9603. Additionally CERCLA vests the EPA with authority to determine what constitutes the RQ of any given hazardous substance and thereby enables the EPA to determine what releases must be reported. 42 U.S.C. § 9602(a).
In April 1985 the EPA promulgated a rule setting forth the RQs for many substances deemed “hazardous” under CERCLA. 50 Fed.Reg. 13,456 (April 4, 1985). The rule did not include a set of RQs for radionuclides. The EPA had intended to set RQs for these substances, see 48 Fed. Reg. 23,552 (May 25, 1983) (proposed rule-making), but failed to do so because it was unable to decide both how to measure the RQs for radionuclides and whether all radionuclides should have the same RQ (there are approximately 1,500 different radionuclides). 50 Fed.Reg. at 13,458. Consequently the EPA proposed that further studies be conducted and that new rule-making be initiated to determine the RQs for radionuclides.2
Nearly two years later, the EPA again set out to determine the appropriate RQs for radionuclides. See 52 Fed.Reg. 8,172 (March 16, 1987) (proposed rulemaking). Then, after another two years of notice and comment, the EPA issued a final rule. See 54 Fed.Reg. 22,524 (May 24, 1989). The rule includes three provisions that are challenged here. First, in the preamble to the rule, the EPA sets out in detail its interpretation of what constitutes a “release” of hazardous substances into the environment, thereby triggering the requirement to notify the EPA. According to this interpretation, “the placement of a hazardous substance into any unenclosed containment structure wherein the hazardous substance is exposed to the environment” constitutes a “release.” Id. at 22,526. Second, the rule sets forth several exemptions to [188]*188CERCLA's reporting requirement.3 These administrative exemptions enable certain industries to release radionuclides into the environment without reporting to the EPA. Finally, the rule establishes an RQ for each individual radionuclide, including radon-222, the RQ challenged by the petitioners. Id. at 22,538-42.
In response to the EPA’s rulemaking, numerous businesses and trade associations filed petitions for review with this court pursuant to 42 U.S.C. § 9613(a). The cases have been consolidated and two sets of petitioners and intervenors now challenge the rulemaking before this court. The first set of petitioners and intervenors, Consolidated Petitioners and Intervenors (Consolidated Petitioners), challenges the EPA’s interpretation of CERCLA’s reporting requirement as set forth in the preamble to the final rule. The second set of petitioners includes only two parties, the American Mining Congress and The Fertilizer Institute (AMC and TFI), and they challenge the administrative exemptions created by the EPA as well as the RQ established for radon-222. We address each of these issues in turn.
II.
The preamble to the EPA’s final rule states in part:
The Agency considers the stockpiling of an RQ of a hazardous substance to be a release because any activity that involves the placement of a hazardous substance into any unenclosed containment structure wherein the hazardous substance is exposed to the environment is considered a release. An unenclosed containment structure may allow the hazardous substance to emit, escape, or leach into the air, water, or soil. Thus, the placement of an RQ of a hazardous substance in an unenclosed structure would constitute a “release” regardless of whether an RQ of the substance actually volatizes into the air or migrates into surrounding water or soil. The same rule applies to the placement of material containing radionuclides in tanks or other containment structures outside a building. If the tank or containment structure is not totally sealed off from the environment, the placement into the containment structure of an amount of a hazardous substance that equals or exceeds an RQ constitutes a reportable release.
54 Fed.Reg. at 22,526 (footnote omitted). Consolidated Petitioners claim that this definition of “release” is invalid for two reasons. First they argue that the EPA’s failure to provide notice and comment before issuing this ruling violates the notice and comment requirements of the Administrative Procedure Act (APA), 5 U.S.C. § 553. Second, Consolidated Petitioners argue that the EPA’s interpretation contradicts the plain meaning of CERCLA and must be invalidated as contrary to the express intent of Congress.
A.
First we reject Consolidated Petir tioners’ contention that the EPA’s preamble runs afoul of the APA. Section 4(a) of the APA exempts from its notice and comment requirements those rules which are “interpretative” and not “legislative.” 5 U.S.C. § 553(b)(3)(A).4 We have consistently explained this distinction as follows:
An interpretative rule simply states what the administrative agency thinks the statute means, and only “ ‘reminds’ affected parties of existing duties.” Citizens to Save Spencer County v. EPA, 600 F.2d 844, 876 & n. 153 (D.C.Cir. 1979)____ On the other hand, if by its action the agency intends to create new [189]*189law, rights or duties, the rule is properly considered to be a legislative rule.
General Motors Corp. v. Ruckelshaus, 742 F.2d 1561, 1565 (D.C.Cir.1984) (citations omitted), cert. denied, 471 U.S. 1074, 105 S.Ct. 2153, 85 L.Ed.2d 509 (1985). Thus, as a general rule, an agency can declare its understanding of what a statute requires without providing notice and comment, but an agency cannot go beyond the text of a statute and exercise its delegated powers without first providing adequate notice and comment.
In the preamble passage, the EPA attempts to clarify the meaning of the term “release” as defined by CERCLA. By doing so, the EPA does not create any new right or duty but merely provides an interpretation of CERCLA’s reporting requirement.5 The EPA’s action, then, would seem to fit comfortably within the category of an interpretative rule as defined in General Motors. Consolidated Petitioners claim, however, that the EPA’s preamble passage cannot be considered interpretative because it creates a set of duties that did not previously exist. They also point out that the EPA did not attempt to determine Congress’s actual intent in defining “release” but instead based its definition on its own judgment and reasoning. Such a discretionary act, they claim, is not “interpretative” and, therefore, can be accomplished only with full notice and comment.
Consolidated Petitioners’ arguments reflect a fundamental misunderstanding of the distinction made by section 4(a) of the APA. This court’s holding in General Motors that an agency’s action is deemed to be legislative when the agency “intends to create new ... duties,” 742 F.2d at 1565 (emphasis added), does not mean that an agency’s action is legislative whenever it has the effect of creating new duties. To the contrary, as we reasoned in United Technologies Corp. v. EPA, 821 F.2d 714, 719-20 (D.C.Cir.1987), the proper focus in determining whether an agency’s act is legislative is the source of the agency’s action, not the implications of that action:
If the rule is based on specific statutory provisions, ... it is an interpretative rule. If, however, the rule is based on an agency’s power to exercise its judgment as to how best to implement a general statutory mandate, the rule is likely a legislative one.
Accordingly, the fact that the preamble may affect how parties act does not make the rule legislative — regardless of the consequences of a rulemaking, a rule will be considered interpretative if it represents an agency’s explanation of a statutory provision.
Similarly, the fact that the EPA’s interpretation of CERCLA’s reporting requirement does not include a full-blown inquiry into Congress’s intent does not prevent the preamble passage from being an interpretative rule. Consolidated Petitioners argue that an agency must “specifically discuss the language, purpose and legislative history” of a statute in order to consider the agency decision interpretative. Con-sol.Petit. Reply Br. at 19. The flaw in this argument is that it confuses the question whether the agency is interpreting a statute with the question whether the agency is thoroughly, or properly, interpreting the statute. Simply because an agency may fail to interpret a statute correctly does not mean that the agency has not in fact interpreted it. Here the EPA’s failure to discuss either the legislative history or the statutory language may have prevented the EPA from correctly construing the statute but that failure does not alter the fact that the EPA, at the very least, attempted to interpret the statute.
In sum, Consolidated Petitioners’ APA argument unduly emphasizes the consequences of the EPA’s rulemaking as well as the formulation of its interpretation. To determine whether the preamble is interpretative, the true emphasis must be on the “legal base upon which the rule rests.” United Technologies, 821 F.2d at 719. In [190]*190this case, the EPA’s preamble passage represents the agency’s attempt to interpret the meaning of a statutory provision. As such we have little difficulty in concluding that the rule is interpretative.
B.
Notwithstanding our conclusion that the EPA promulgated its interpretation of CERCLA’s reporting requirement properly, that interpretation runs contrary to the plain meaning of the statute and therefore must be reversed. As this court has oft repeated, we must give special deference to an agency’s interpretation of an act that the agency is charged with implementing. See, e.g., Consolidated Rail Corp. v. United States, 896 F.2d 574, 578 (D.C.Cir.1990) (citing Chevron USA, Inc. v. NRDC, Inc., 467 U.S. 837, 842, 104 S.Ct. 2778, 2781, 81 L.Ed.2d 694 (1984)). Nonetheless, when an agency attempts to give meaning to a statute that is plain on its face, we need not defer to the agency but instead must give effect to the intent of Congress. Chevron USA, Inc. v. NRDC, Inc., 467 U.S. at 842-43, 104 S.Ct. at 2781-82; Amalgamated Transit Union v. Skinner, 894 F.2d 1362, 1368 (D.C.Cir.1990). Because Congress has spoken clearly and directly to the issue addressed by the EPA in the preamble, it is our task to scrutinize the EPA’s interpretation to ensure that the EPA stays the course set by Congress.
CERCLA includes the following reporting requirement:
Any person in charge of a vessel or an offshore or onshore facility shall, as soon as he has knowledge of any release ... of a hazardous substance from such vessel or facility in quantities equal to or greater than those determined pursuant to section 9602 of this title ... [shall] immediately notify the National Response Center ... of such a release.
42 U.S.C. § 9603(a). CERCLA defines “facility” as:
[A]ny building, structure, installation, equipment, pipe or pipeline ..., well, pit, pond, lagoon, impoundment, ditch, landfill, storage container, motor vehicle, rolling stock, or aircraft, or ... any site or area where a hazardous substance has been deposited, stored, disposed of, or placed, or otherwise come to be located____
42 U.S.C. § 9601(9). A “release” is defined as “any spilling, leaking, pumping, pouring, emitting, emptying, discharging, injecting, escaping, leaching, dumping, or disposing into the environment____” 42 U.S.C. § 9601(22). Finally, “environment” is defined as “navigable waters ..., any other surface water, ground water, drinking water supply, land surface or subsurface strata, or ambient air____” 42 U.S.C. § 9601(8). When these in pari materia provisions of CERCLA are read together, it becomes clear that CERCLA requires reporting whenever a hazardous substance is released from a facility, such as a tank or a lagoon, into the environment, such as the ground or the air.
In the preamble to the final rule, the EPA interprets CERCLA to require notification whenever a hazardous substance is placed into an “unenclosed containment structure.” 54 Fed.Reg. at 22,-526. The EPA defines an unenclosed containment structure as “any surface impoundment, lagoon, tank, or other holding device that has an open side with the contained materials directly exposed to the ambient environment.” Id. at 22,526 n. 3. According to this interpretation, therefore, “the placement of an RQ of a hazardous substance in an unenclosed structure would constitute a ‘release’ regardless of whether an RQ of the substance actually volatizes into the air or migrates into surrounding water or soil.” Id. at 22,526.
The EPA’s interpretation of CERCLA’s reporting requirement cannot be reconciled with CERCLA’s express terms. Rather than defining a release as the movement of a substance from a facility into the environment (as does CERCLA), the EPA defines a release as the placement of a substance into a facility that is exposed to the environment. For the EPA’s interpretation to come within CERCLA’s meaning, the exposure of a substance to the environment must be equivalent to the release of that substance into the environment. But [191]*191this is not so. A simple example exposes the flaw in the EPA’s analysis: a company could place a non-volatile substance into an open-air storage container and the consequences of the open-air storage would be no different from those that would occur if the company had placed the substance to a closed container. Nonetheless, under the EPA’s interpretation, the company would have to report the transfer of the substance to the container notwithstanding the non-volatile substance’s inability to escape into the air.
Even in the final rule, the EPA cannot escape the inconsistency caused by equating the mere exposure of a substance to the environment to the release of that substance into the environment. At one point the EPA asserts:
An unenclosed containment structure may allow the hazardous substance to emit, escape, or leach into the air, water, or soil. Thus, the placement of an RQ of a hazardous substance in an unenclosed structure would constitute a “release” regardless of whether an RQ of the substance actually volatilizes into the air or migrates into surrounding water or soil.
54 Fed.Reg. at 22,526 (emphasis added). Implicit in this passage is the recognition that the mere exposure of a substance to the environment does not always result in the release of that substance into the environment. The EPA attempts to gloss over this incongruity by arguing that the placement of a substance into an unenclosed containment structure creates such a great threat of a release that the EPA is justified in treating it as an actual release. This argument, however, is precisely foreclosed by CERCLA. CERCLA provides that the EPA must be notified when a hazardous substance is actually released into the environment. 42 U.S.C. § 9603(a). Nowhere in the statute is there any language requiring that the EPA be notified when there is a threatened release. This omission is especially significant given the sections of CERCLA that expressly distinguish between threats of releases and actual releases. See, e.g., 42 U.S.C. §§ 9604, 9606. These sections give the EPA the authority to act with reference both to releases and to threatened releases. Under these circumstances, we must presume that Congress’s failure to subject threatened releases to the reporting requirement was intentional. See Russello v. United States, 464 U.S. 16, 23, 104 S.Ct. 296, 300-01, 78 L.Ed.2d 17 (1983). Accordingly we reject the EPA’s claim that the threat of a release from an unenclosed containment structure is sufficient to allow the EPA to require reporting when a reportable quantity of a hazardous material is put in the structure. Under CERCLA’s provisions, nothing less than the actual release of a hazardous material into the environment triggers its reporting requirement.
III.
The administrative exemptions created by the EPA in the final rule excuse parties from notifying the EPA when RQs of radionuclides are released by the following: (i) large land holdings that remain undisturbed {e.g., golf courses); (ii) large land holdings that are disturbed {e.g., construction sites or farms); (iii) coal and ash dumps; and (iv) utility and industrial facilities with coal burners. 54 Fed.Reg. at 22,529. AMC and TFI challenge these exemptions on two grounds. First, they claim that the EPA failed to provide notice and comment before the creation of the exemptions and therefore the exemptions lack a proper procedural foundation. Second, they claim that the EPA’s decision to exempt certain entities and industries and not others lacks a rational basis and the exemptions are therefore arbitrary and capricious. We agree with AMC and TFI’s claim that the EPA failed to provide adequate notice and comment and, accordingly, do not reach their second claim. Nonetheless, we allow those exemptions to remain in place until the EPA provides adequate notice and comment.
Section 4(a) of the Administrative Procedure Act requires an agency to publish notice of either the substance of a proposed rule or a “description of the subjects and issues” covered by a proposed rule. 5 U.S.C. § 553(b)(3). This court has [192]*192consistently interpreted that requirement to mean that an agency’s notice must “provide sufficient detail and rationale for the rule to permit interested parties to comment meaningfully.” Florida Power & Light Co. v. United States, 846 F.2d 765, 771 (D.C.Cir.1988), cert. denied, 490 U.S. 1045, 109 S.Ct. 1952, 104 L.Ed.2d 422 (1989). The proposed rule that was published by the EPA in this rulemaking makes no mention of the possible creation of administrative exemptions to CERCLA’s reporting requirement. The EPA explains this omission by pointing out that the proposed rule included a discussion of statutory exemptions to CERCLA. This discussion of the “subject of exemptions,” according to the EPA, was enough to put parties on notice that administrative exemptions might be created. EPA Br. at 34-35.
The EPA’s attempt to convert the topic of statutory exemptions into an invitation to send comments on administrative exemptions cannot be sustained. Under the EPA’s argument, the petitioners were responsible for inferring from the proposed rule’s reference to statutory exemptions that the creation of administrative exemptions might be considered. Implicit in this argument is the notion that a reference to a specific topic (here, statutory exemptions) can give rise to notice of the existence of a more general topic (here, exemptions in general) and that the general topic, in turn, can encompass notice of a second specific topic (here, administrative exemptions) only remotely related to the first specific one. Such reasoning, if accepted by the court, would turn notice and comment rulemaking into a guessing game in which the inclusion of one subject indicates that a distant cousin of that subject might be addressed. This reading of the APA cannot be sustained.
The EPA poses a second, more credible, argument. According to the EPA, its creation of administrative exemptions is proper because those exemptions represent the “logical outgrowth” of the proposed rule. In a nutshell, the EPA argues that it would have been logical for interested parties, in response to the notice that the EPA was considering the RQs for radionuclides, to have submitted comments that would support the creation of administrative exemptions. The conclusion that certain entities or industries are safe and need not be regulated, claims the EPA, should have followed naturally from a request for comments regarding the determination of hazardous levels of radionuclides. The EPA buttresses this argument by pointing out that several parties did in fact suggest that administrative exemptions be created and submitted supporting evidence. EPA Br. at 34-35.
This court has long recognized that an agency must be able to respond flexibly to comments and need not provide a new round of notice and comment every time it modifies a proposed rule. See, e.g., Small Refiner Lead Phase-Down Task Force v. EPA, 705 F.2d 506, 547 (D.C.Cir.1983); International Harvester Co. v. Ruckelshaus, 478 F.2d 615, 632 n. 51 (D.C.Cir.1973). It follows that an agency may issue rules that do not exactly coincide with the proposed rule so long as the final rule is the “logical outgrowth” of the proposed rule. In determining whether a final rule is the logical outgrowth of the proposed rule, the key focus is on whether the purposes of notice and comment have been adequately served. See Small Refiner Lead, 705 F.2d at 547. This means that a final rule will be deemed to be the logical outgrowth of a proposed rule if a new round of notice and comment would not provide commenters with “their first occasion to offer new and different criticisms which the agency might find convincing.” United Steelworkers of America v. Marshall, 647 F.2d 1189, 1225 (D.C.Cir.1980) (quoting BASF Wyandotte Corp. v. Costle, 598 F.2d 637, 642 (1st Cir.1979)), cert. denied, 453 U.S. 913, 101 S.Ct. 3148, 69 L.Ed.2d 997 (1981).
Contrary to the EPA’s assertions, the administrative exemptions contained in the final rule cannot be considered a logical outgrowth of the proposed rulemaking. First the EPA granted exemptions for certain types of industries and entities, not for certain types of radionuclides. To arrive at those exemptions, the EPA needed evidence and arguments pertaining to the operations of, and hazards surrounding, entities that use or produce radionuclides. Yet [193]*193the proposed rulemaking stated only that the EPA would be determining the hazards of the radionuclides themselves. Because comments appropriate to a determination of the hazards of radionuclides would not necessarily be appropriate to a determination of the hazards of entities using or producing radionuclides, we conclude that the notice published by the EPA did not provide interested parties with an adequate opportunity to comment.
The fact that some commenters actually submitted comments suggesting the creation of administrative exemptions is of little significance. Commenting parties cannot be expected to monitor all other comments submitted to an agency. As we stated in Small Refiner Lead Phase-Down Task Force, “the EPA must itself provide notice of a regulatory proposal. Having failed to do so, it cannot bootstrap notice from a comment.” 705 F.2d at 549 (emphasis in the original). See also AFL-CIO v. Donovan, 757 F.2d 330, 339-40 (D.C.Cir.1985). We conclude that the EPA’s notice was not sufficient to advise interested parties that comments directed to the creation of administrative exemptions should be made. Indeed, we note that one of the reasons the EPA failed to grant exemptions that would have included AMC and TFI is that, according to the EPA, there was not enough information available about those industries. See EPA Br. at 40.
Despite our conclusion that the EPA failed to provide adequate notice and comment on the issue of administrative exemptions, we allow the exemptions to remain in place until the EPA conducts a new round of notice and comment. Ordinarily, when a regulation is not promulgated in compliance with the APA, the regulation cannot be “afforded the ‘force and effect of law.’ ” Chrysler Corp. v. Brown, 441 U.S. 281, 313, 99 S.Ct. 1705, 1723, 60 L.Ed.2d 208 (1979) (citing Morton v. Ruiz, 415 U.S. 199, 94 S.Ct. 1055, 39 L.Ed.2d 270 (1974) and Allegheny-Ludlam Steel Corp., 406 U.S. 742, 758, 92 S.Ct. 1941,. 1951, 32 L.Ed.2d 453 (1972)). Yet, when equity demands, an unlawfully promulgated regulation can be left in place while the agency provides the proper procedural remedy. See, e.g., Western Oil & Gas Assoc. v. EPA, 633 F.2d 803, 813 (9th Cir.1980); Chemical Mfrs. Ass’n v. EPA, 870 F.2d 177, 236 (5th Cir.1989), cert. denied, — U.S.-, 110 S.Ct. 1936, 109 L.Ed.2d 299 (1990); but see Alaniz v. OPM, 728 F.2d 1460, 1469-70 & n. 7 ,(Fed.Cir.l984) (declining to leave in place procedurally inadequate regulations that are not “harmless”). Because the removal of the EPA’s exemptions may affect the EPA’s ability to respond adequately to serious safety hazards, we are reluctant to remove the exemptions here.6 As the Ninth Circuit reasoned in Western Oil and Gas, “[o]ur intervention into the process of environmental regulation, a process of great complexity, should be accomplished with as little intrusiveness as feasible.” 633 F.2d at 813. Also we are influenced by the fact that no party challenged the specific exemptions set by the EPA. Indeed AMC and TFI’s only complaint about the exemptions is that they are not broad enough. We emphasize, however, that, by leaving the EPA’s exemptions in place, we are not relieving the EPA of its burden to conduct notice and comment rulemaking ab initio, i.e., without giving preference to the exemptions left in place in the interim.
IV.
Finally we decline to reach AMC and TFI’s challenge to the RQ level set for radon-222. As a general rule, this court will not address arguments that were not initially raised before the agency. See, e.g., Etelson v. OPM, 684 F.2d 918, 923 (D.C.Cir.1982). Nonetheless this exhaustion requirement is prudential and must be applied “ ‘flexibly, with an eye toward its underlying purposes.’ ” Cutler v. Hayes, 818 F.2d 879, 890 (D.C.Cir.1987) (citing Etelson, 684 F.2d at 923). As we stated in Cutler, these purposes include:
(1) discouraging the frequent and deliberate flouting of administrative pro[194]*194cesses; (2) protecting agency autonomy by allowing an agency the first opportunity to apply its expertise, exercise its discretion, and correct its errors; (3) aiding judicial review by promoting the development of facts during the administrative proceeding; and (4) promoting judicial economy by reducing duplication, and perhaps even obviating judicial involvement.
818 F.2d at 890-91 (quotation marks and citations omitted.)
Regarding the EPA’s determination of the RQ for radon-222, AMC and TFI argue that the EPA based its calculations on a physically impossible assumption and that the EPA used a scientific methodology inconsistent with the EPA’s stated policy. Neither of these arguments was raised before the agency during the rule-making. Indeed, AMC and TFI concede that they did not discover the supposed flaws in the EPA’s analysis until after the final rule was issued and then analyzed by an outside consultant. . AMC & TFI Reply Br. at 19 n. 8. AMC and TFI attempt to avoid the consequences of their failure to raise their claim before the EPA by relying on several comments in which parties made generalized complaints about the EPA’s methodology. See, e.g., JA 405 (RQ should not be calculated according to most hazardous chemical form); 409 (EPA calculations are “oversimplified”); 413 (EPA fails to consider effects of a release from large-acreage facilities). Yet none of these comments makes, even tangentially, the arguments raised by AMC and TFI for the first time in this court.
We decline to reach their claim regarding the RQ level of radon-222 because the agency should have full opportunity to apply its expertise and discretion to scientific and technical matters such as this. As we have emphasized in the past, this court is not made up of “engineers, computer modelers, economists or statisticians.” Sierra Club v. Costle, 657 F.2d 298, 410 (D.C.Cir.1981), rev’d on other grounds, 463 U.S. 680, 103 S.Ct. 3274, 77 L.Ed.2d 938 (1983). To decide the issue raised by AMC and TFI, however, we would have to assume such roles. The evaluation of the scientific data relied on by AMC and TFI is best left to the agency with expertise in the area and we will not excuse their failure to tap that expertise.
Y.
In conclusion, we vacate that portion of the final rule that interprets CERCLA to require parties to report the placement of an RQ of a hazardous substance into an unenclosed containment structure. We also hold that the EPA provided inadequate notice and comment regarding the creation of administrative exemptions to CERCLA’s reporting requirement. We exercise our equitable discretion, however, and allow those exemptions to remain in place until adequate notice and comment is completed. Finally, we decline to address the issue of whether the RQ set for radon-222 is arbitrary and capricious because of the petitioners’ failure to raise the issue before the agency.
It is so ordered.