concurring in part and dissenting in part:
In the Clean Air Act, Congress declared a national goal of restoring natural visibility in the country’s largest national parks and wilderness areas. In Part II of today’s opinion, the court adopts an interpretation of the Act that, in the view of the Environmental Protection Agency (EPA) and the National Academy of Sciences, will prevent the achievement of Congress’ goal. If that interpretation were required by the statutory language, we would of course be compelled to adopt it. But such an interpretation is not required. To the contrary, EPA’s construction of the Clean Air Act as permitting the group-BART provisions of the Haze Rule is a reasonable interpretation of the legislative language. It is therefore entitled to our deference under the standard announced in Chevron U.S.A. Inc. v. Natural Res. Def. Council, Inc., 467 U.S. 837, 842-43, 104 S.Ct. 2778, 2781-82, 81 L.Ed.2d 694 (1984). Accordingly, while concurring in most of the court’s opinion, I dissent from thé conclusions it reaches in Part II.
A
Chevron instructs courts to apply a two-step framework when reviewing an agency’s construction of a statute. First, we must ask “whether' Congress has directly spoken to the precise question at issue,” in which case we “must give effect to the unambiguously expressed intent of Congress.” Id. at 842-43, 104 S.Ct. at 2781-82. However, if the “statute is silent or ambiguous with respect to the specific issue,” we move to the second step and must defer to the agency’s interpretation as long as it is “based on a permissible construction of the statute.” Id. at 843, 104 S.Ct. at 2782; accord Barnhart v. Walton, — U.S. -, ----, 122 S.Ct. 1265, 1271-72, 152 L.Ed.2d 330 (2002).
My colleagues stop at Chevron’s first step, concluding that the language of the Clean Air Act (CAA) can be read in only one way. They adopt the view of the industry petitioners that under the Act, BART (“best available retrofit technology”) controls cannot be imposed on a source unless a state determines how much that particular source contributes to visual impairment in a downwind national park or wilderness area, as well as how *16much improvement in visibility would result from installing BART controls at that specific source. Op. at 7. EPA, by contrast, interprets the Clean Air Act as permitting a collective assessment of the impact that emissions from (and controls on) sources located in upwind regions have on visibility impairment in downwind areas.
Before considering the grounds for the court’s decision, it is important to understand why EPA decided to require a collective contribution approach, rather than a tracing of the effects of each individual source’s emissions. Congress added § 169A to the Clean Air Act “[i]n response to a growing awareness that visibility was rapidly deteriorating” in major national parks and wilderness areas (“Class I areas”). Chevron U.S.A., Inc. v. EPA, 658 F.2d 271, 272 (5th Cir.1981). The section establishes a national goal of restoring natural visibility in such areas,1 and expressly instructs EPA to issue regulations to “assure ... reasonable progress” toward meeting the national goal. 42 U.S.C. § 7491(a)(4). After examining the results of scientific studies, EPA concluded that such reasonable progress was not possible without a collective approach. The record compiled by EPA showed that visibility impairment in Class I areas is caused in large part .by long-range transport of combined emissions from multiple sources.2 Although it is practicable to trace emissions from an individual source into its surrounding region, and to model the transport of combined pollution from that region to a downwind Class I area,3 it is not possible to trace emissions from an individual source directly to such a downwind area without great time and expense4 —and even then the results would be of uncertain reliability.5 Citing the National *17Academy of Sciences’ conclusion that a program focused “on determining the contribution of individual emission sources to visibility impairment is doomed to failure,”6 EPA adopted the group-BART approach that is at issue here.
My colleagues do not dispute that we must defer to EPA’s expert opinion regarding the impracticability of tracing individual source emissions.7 Rather, they conclude that notwithstanding EPA’s view of the facts, the industry petitioners are correct that the Haze Rule’s group-BART provisions violate the plain meaning of the Clean Air Act by: (i) employing a group rather than souree-by-source standard in determining the appropriate BART controls for a particular source, and (ii) constraining the authority of the states to make their own BART-related decisions. These two contentions are considered in Parts B and C below. Because I conclude that there is nothing in the Clean Air Act that bars the approach taken by EPA, and that to the contrary the Haze Rule rests on a reasonable interpretation of the statutory language, I would follow the Supreme Court’s direction in Chevron and uphold the Rule.
B
As the court notes, the Haze Rule employs a group analysis in making two determinations required by the Clean Air Act: (i) whether a pollution-emitting source is subject to BART requirements at all, and (ii) what kind of BART controls should be placed on a subject source. The industry petitioners contend that the Clean Air Act prohibits the use of a group standard in making either of these determinations.
Under the Act, a source is subject to BART requirements, and hence a state implementation plan must require such a source to install BART controls, if it “emits any air pollutant which may reasonably be anticipated to cause or contribute to any impairment of visibility in any [Class I] area.” CAA § 169Á(b)(2)(A), 42 U.S.C. § 7491(b)(2)(A). Under the Haze Rule, a state must “find that a BART-eligible source is ‘reasonably anticipated to cause or contribute’ to regional haze if it can be shown that the source emits pollutants within a geographic area from which pollutants can be emitted and transported downwind to a Class I area.” Regional Haze Regulations, 64 Fed.Reg. at 35,740. That is, a source is subject to BART requirements, without proof of that source’s individual contribution to visibility impairment in a Class I area, as long as the source emits pollutants into an upwind area from which pollutants may be transported to a downwind Class I area. Id.
The industry petitioners contend that CAA § 169A(b)(2) unambiguously provides that a source is subject to BART require*18ments only if a state can show the extent to which that particular source contributes to impairment in a Class I area. That section, however, requires states to impose BART controls on any source that “emits any air pollutant which may reasonably be anticipated to cause or contribute to any impairment of visibility in any [Class I] area.” 42 U.S.C. § 7491(b)(2)(A) (emphasis added). Far from plainly compelling the petitioners’ reading, the italicized words pile ambiguity upon ambiguity and virtually invite the reader to adopt the construction favored by EPA. See Merriam-Webster’s Collegiate Dictionary 252 (10th ed.1996) (defining “contribute” as “to give or supply in common with others,” or “to give a part to a common ... store”) (emphasis added); Central Ariz. Water Conservation Dist. v. E.P.A., 990 F.2d 1581, 1541 (9th Cir.1993) (“The phrase ‘may reasonably be anticipated’ suggests that Congress did not intend to require EPA to show a precise relationship between a source’s emissions and all or a specific fraction of the visibility impairment within a Class I area.” (quoting with approval National Researoh Council, Haze in the Grand Canyon: An Evaluation of the Winter Haze Intensive Tracer Experiment 5 (1990))). If a source is one of several that emit pollutants into an upwind area, and if pollution from that area is' transported downwind to a national park,8 then it can hardly be unreasonable to conclude that the pollutants issued by the source “may reasonably be anticipated” to “contribute” to “any” impairment in the park.
My colleagues wisely do not accept the industry petitioners’ contention that § 169A(b)(2) bars a collective determination of whether a source is subject to BART. (As discussed in Part C infra, they do conclude that EPA may not require the states to employ such a mode of analysis.) They do, however, accept the petitioners’ contention that to determine the kind of BART controls that should be imposed on a subject source, a state must determine how much that particular source contributes to visual impairment in the downwind Class I area, Op. at 7, as well as the degree of improvement in visibility that would occur in the downwind area if that particular source installed such controls, id. at 7. The Haze Rule, by contrast, provides that once a state has concluded that a particular source is subject to BART requirements, in determining the kind of BART controls to place on the source the state must consider the degree of improvement that would be achieved in the downwind area by imposing BART controls on all subject sources in the contributing upwind area. See 40 C.F.R. § 51.308(e)(l)(ii)(B); Regional Haze Regulations, 64 Fed.Reg. at 35,741.
The industry petitioners rest their contention that the statute unambiguously bars this collective assessment approach on § 169A(g)(2), which states:
[1]n determining best available retrofit technology the State ... shall take into consideration [1] the costs of compliance, [2] the energy and nonair quality environmental impacts of compliance, [3] any existing pollution control technology in use at the source, [4] the remaining useful life of the source, and [5] the degree of improvement in visibility which may reasonably be anticipated to result from the use of such technology.
42 U.S.C. § 7491(g)(2). According to both the industry petitioners and the court, this section requires the state to take into con*19sideration each of the five listed factors on a source-by-source basis. Since the Haze Rule does require source-by-source consideration of the first four factors, see Regional Haze Regulations, 64 Fed.Reg. at 35,740-^41; Op. at 6, the only question is whether such consideration is also required of the fifth factor: “the degree of improvement in visibility which may reasonably be anticipated to result from the use of such technology.”
There is nothing in the statutory language that requires a source-by-souree application of the fifth factor. Section 169A(g)(2) requires an assessment of the degree of improvement that may reasonably be anticipated “from the use of such technology,” but it does not say whether that improvement must be from the use of such technology by a single source or by all sources in the upwind area.9 Although the court says that the statute does not permit any of the five factors to be treated differently from any of the others, the statute itself does not say so. Moreover, the first four factors are different in kind from the fifth: the first four all go to the cost of imposing controls on a particular source and permit a determination of the most cost-effective control technology for each such source. Regional Haze Regulations, 64 Fed.Reg. at 35,740-41. The fifth factor, by contrast, goes to the benefit to be derived from using the most cost-effective controls. In EPA’s expert view, that benefit can best be determined by considering the total benefit that would accrue if each source in the upwind area used the kind of controls most cost-effective for that source.
The industry petitioners concede that § 169A(g)(2) does not require a state to undertake a cost-benefit analysis in deciding the type of controls to impose, or specify the weight to be accorded to any of the five factors.10 All that is required is that the state “take into consideration” the five listed factors. 42 U.S.C. § 7491(g)(2). Because the statute does not specify how the state should take those factors into consideration, it does not bar EPA from employing a group rather than source-by-source mode of analysis in considering benefits. See Weyerhaeuser Co. v. Costle, 590 F.2d 1011, 1045 (D.C.Cir.1978) (holding that where “Congress did not mandate any particular structure or weight” for the factors EPA is to consider, “it left EPA with discretion to decide how to account for the consideration factors, and how much weight to give each factor”); see also New York v. Reilly, 969 F.2d 1147, 1150 (D.C.Cir.1992) (same).
Other related provisions of the Clean Air Act support EPA’s reading of § 169A(g)(2) as permitting a region-wide assessment. Section 169A(a)(3) directs EPA to undertake a study to “identify the classes or categories of sources ... which, alone or in conjunction with other sources ..., may reasonably be anticipated to cause or contribute significantly to impairment of visibility,” 42 U.S.C. § 7491(a)(3) (emphasis added), and § 169A(b)(l) directs that the regulations promulgated under *20§ 169A take into account the recommendations of that study, 42 U.S.C. § 7491(b)(1). Similarly, § 169B(a)(l) instructs EPA to conduct research “to identify and evaluate sources and source regions of ... visibility impairment.” 42 U.S.C. § 7492(a)(1) (emphasis added); see id. § 7492(a)(2). These provisions not only permit, but again appear to invite a group-BART approach.
The court states that “under EPA’s take on the statute, it is ... entirely possible that a' source may be forced to spend millions of dollars for new technology that will have no appreciable effect on the haze in any Class I area.” Op. at 7. In accordance with the statute, however, EPA has structured the Haze Rule to avoid this result. The Rule creates an evidentiary presumption that, if a source emits pollution into an upwind region from which it can be shown that pollution is transported downwind to a Class I .area, then it “may reasonably be anticipated” that the source “cause[s] or contribute^] to” impairment in the Class I area — and hence that limiting the source’s emissions will reduce that impairment.11 But the presumption is not irrebuttable. To the contrary, the Haze Rule incorporates the exemption provision of § 169A(c)(l), which permits EPA to
exempt any major stationary source from the [BART] requirement of subsection (b)(2)(A) of this section, upon his determination that such source does not or will not, by itself or in combination with other sources, emit any air pollutant which may reasonably be anticipated to cause or contribute to a significant impairment of visibility in any mandatory class I Federal area.
42 U.S.C. § 7491(c)(1); see also 40 C.F.R. §§ 51.303, .308(e)(4). Hence, a source that emits pollution into a source region, but that can show that BART controls are unnecessary because its pollution does not contribute to a significant impairment of visibility in a Class I area, will not have to spend money installing BART controls.12 All that the Haze Rule does is put the burden of proof on the polluter, rather than on the state. Moreover, the statute’s limitation of the exemption to a source that does not by itself “or in combination with other sources” contribute to a significant impairment, 42 U.S.C. § 7491(c)(1), once again invites the collective assessment approach taken by EPA.
*21Finally, one more provision of § 169A deserves repeat mention here. As discussed in Part A above, § 169A(a)(4) instructs EPA “to promulgate regulations to assure reasonable progress toward meeting the national goal” of restoring natural visibility conditions. 42 U.S.C. § 7491(a)(4). Yet EPA’s findings indicate that it will not be possible “to assure reasonable progress” if the statutory interpretation announced today prevails: it is simply not practicable to determine, as the court’s interpretation requires, how much a particular “source is contributing to visual impairment in downwind Class I areas,” or the degree of improvement in visibility in such areas “that would result from [a particular] source’s installing and operating” BART controls. Op. at 7; see supra notes 4, 5. Indeed, EPA explained that it “avoided inclusion of any approach in the regional haze rule that required the assessment of the visibility improvement attributed to an individual source because” the National Academy of Sciences had determined that such an approach was “doomed to failure.” Resp. to Pets, for Recons, of Regional Haze Rule 16 (Jan. 10, 2001) (J.A. at 17) (quoting National Academy of ScienCes, National ReseaRch Council, PROTECTING VISIBILITY IN NATIONAL Parks and Wilderness AREas 7-8, 196-99 (1993) (J.A. at 362, 456-57)). We should not lightly assume that Congress enacted a statute that makes it impracticable to achieve the same statute’s stated goal. There certainly is nothing in the language of the Clean Air Act that requires us to adopt such a self-defeating construction.
C
The industry petitioners’ second attack on the Haze Rule marches under the banner of states’ rights, but in this case that banner is a false flag. The Rule gives states great leeway to make the BART determinations required by the Clean Air Act, reserving to EPA no more authority than Congress conferred upon the agency. Moreover, as discussed above, the industry petitioners’ insistence that both EPA and the states are barred from using group-BART principles will impose an enormous unfunded mandate on the states — requiring them to engage in lengthy, expensive, and likely fruitless studies to trace pollutants from specific sources into specific Class I areas.13 It is not surprising, therefore, that only a single state has enlisted under the petitioner’s banner. Five others have filed briefs in support of EPA, while the balance remain silent.
The industry petitioners attack, as unlawfully constraining state authority, both the provision of the Haze Rule that concerns which sources are subject to BART requirements, and the provision that concerns the kind of BART controls that must be installed on subject sources. With respect to the former, the petitioners emphasize § 169A’s declaration that “each major stationary source ... which, as determined by the State ... emits any air pollution which may reasonably be anticipated to cause or contribute to any impairment of visibility” in a Class I area, is subject to BART requirements. 42 U.S.C. § 7491(b)(2)(A) (emphasis added). With respect to the latter, they stress that § 169A requires that each subject source install “the best retrofit technology, as determined by the State,” 42 U.S.C. § 7491(b)(2)(A), and that “in determining best available retrofit technology the State ... shall take into consideration” the five factors discussed in Part B above, id. § 7491(g)(2) (emphasis added). By direct*22ing the states to employ a group-BART analysis in making these determinations, the industry petitioners contend, and the court agrees, that EPA has unlawfully constrained the states’ decisionmaking authority. Op. at 7-8.
The Haze Rule, however, does not contravene the statutory commands italicized above. Under the Rule, it is the state and not EPA that determines which specific sources emit pollution that “may reasonably be anticipated to cause or contribute to” impairment, and hence are subject to BART requirements. All that EPA has done, as explained in Part B, is reasonably interpret that phrase to include sources that emit pollution into upwind regions from which pollution is transported to national parks. It is still the state that must determine both that the source emits covered pollutants, and that the region into which the source emits such pollutants is one from which emissions may reasonably be anticipated to be transported to downwind parks. See 40 C.F.R. § 51.308(e)(l)(ii); Regional Haze Regulations, 64 Fed.Reg. at 35,739-^11; Br. for EPA at 43. Similarly, it is still the state that must take into consideration the five statutory factors and the state that must then determine the best available retrofit technology for a particular source. All that EPA has done, again ■ as explained in Part B, is reasonably interpret the fifth of those factors to require the state to analyze the degree of anticipated improvement on a group basis. See Regional Haze Regulations, 64 Fed.Reg. at 35,741.
Moreover, the Clean Air Act expressly delegates to EPA the authority to make these kinds of judgments. As already noted, § 169A directs EPA to promulgate regulations to assure reasonable progress toward meeting the national goal of restoring natural visibility. 42 U.S.C. § 7491(a)(4). It further instructs that those regulations shall “provide guidelines to the States ... on appropriate techniques and methods for implementing” the section’s provisions, including the provisions. governing which sources are subject to BART requirements and the kind of BART controls that should be imposed. Id. § 7491(b)(1). The section likewise directs EPA to “require each applicable implementation plan for a State ... to contain such emission limits, schedules of compliance and other measures as may be necessary to make reasonable progress toward meeting the national goal” of restoring natural visibility. Id. § 7491(b)(2). Similarly, the next section of the Act, § 169B, orders EPA to “carry out [its] regulatory responsibilities” under § 169A by promulgating “criteria for measuring ‘reasonable progress’ toward the national goal.” 42 U.S.C. § 7492(e)(1). These provisions give EPA ample authority to promulgate guidelines requiring states to use group-BART principles to determine both the sources that are subject to BART requirements and the kinds of controls those sources must install.
' My colleagues contend that the Conference Report on the 1977 Clean Air Amendments reinforces their view that the Haze Rule impermissibly constrains state authority. Op. at 8. But that report is a weak reed upon which to rest a Chevron step one claim regarding the Act’s plain meaning. As the court recounts, the report merely states that the conference “agreement clarifies that the State, rather than the Administrator, identifies the source that -impairs visibility,” and that in determining the appropriate BART controls for such a source, “the state shall determine what constitutes ‘best available retrofit technology’ ... in establishing emission limitations on a source-by-source basis.” H.R. Conf. Rep. No. 95-564, at 535 (1977), U.S.Code Cong. & AdmimNews 1977 at pp. 1502, 1536. The report tells us nothing more about the referenced “agreement” than can be gleaned from these *23quotations, and the quotations themselves do little more than restate the statutory-language. Moreover, as noted above, the Haze Rule is consistent with these quotations: under the Rule, it is the state rather than EPA that identifies the sources that impair visibility, and it is the state that determines the best available retrofit technology for each such individual source. All that the group-BART provisions of the Rule do is effectuate EPA’s authority to “provide guidelines to the states” for making these determinations regarding particular sources. 42 U.S.C. § 7491(b)(1).14
As the Clean Air Act repeatedly declares, restoring natural visibility to national parks and wilderness areas is a “national” goal. See id. § 7491(a)(1), (a)(4), (b)(2), (b)(2)(B); id. § 7492(e)(1). It is not surprising, therefore, that while the Act leaves many determinations regarding particular sources to the states, it grants EPA authority to establish national guidelines for the kind of analysis the states must employ in making those determinations.15 Under the statute, those guidelines must “assure ... reasonable progress toward meeting the national goal” of restoring natural visibility. Id. § 7491(a)(4). Because EPA has reasonably determined that group-BART principles are necessary to provide such assurance, the provisions of the Haze Rule that incorporate those principles are a permissible exercise of the agency’s delegated power.
D
In sum, there is nothing in the language, structure or history of the Clean Air Act that bars EPA from promulgating the group-BART provisions of its Haze Rule. To the contrary, those provisions represent “a reasonable interpretation of an ambiguous statute,” and therefore must be given effect by this court. Christensen v. Harris County, 529 U.S. 576, 586, 120 S.Ct. 1655, 1662, 146 L.Ed.2d 621 (2000) (citing Chevron, 467 U.S. at 842-844, 104 S.Ct. at 2781-83). Accordingly, I respectfully dissent from the court’s decision to strike down those provisions.
. Section 169A declares the national goal to be "the prevention of any future, and the remedying of any existing, impairment of visibility in mandatory class I Federal areas." 42 U.S.C. § 7491(a)(1). As the court holds today, agency regulations that aim to accomplish these objectives "will of necessity aim to achieve a state of natural visibility.” Op. at 10.
. See, e.g., Congressional Research Service, Regional Haze: EPA’s Proposal to Improve Visibility in National Parks and Wilderness Areas 2 (1997) (J.A. at 242); National Academy of Sciences, National Research Council, Protecting Visibility in National Parks and Wilderness Areas 7-8, 196-99 (1993) (J.A. at 362, 456-57) [hereinafter "NAS Report"].
. See Regional Haze Regulations, 64 Fed.Reg. 35,714, 35,718 (July 1, 1999). The court does not dispute the reasonableness of, or support for, the latter proposition. Cf. Appalachian Power Co. v. EPA, 135 F.3d 791, 802 (D.C.Cir.1998) (noting that "computer'models are a useful and often essential tool for performing the Herculean labors Congress imposed on EPA in the Clean Air Act,” and that "their scientific nature does not easily lend itself to judicial review” (internal quotation marks omitted)); id. at 814 (“[0]ur consideration of EPA’s use of computer models proceeds with considerable deference to the agency’s expertise.”).
. See NAS Report at 240-41 (J.A. at 478) ("It would be extremely time-consuming and expensive to try to determine the percent contribution of individual sources to haze one source at a time.”); Regional Haze Regulations, 64 Fed.Reg. at 35,740 ("[Establishing the contribution from one particular source to the problem of regional haze would require lengthy and expensive studies and pose substantial technical difficulties.”).
. See NAS Report at 2 (J.A. at 359) ("During transport, the emissions from many sources mix together to form a uniform, widespread haze known as regional haze.”); id. at 20 (J.A. at 368) ("[T]he extent to which [source-specific] techniques can be used in attributing visibility impairment is uncertain, as is their usefulness in estimating the effect that different control strategies might have on visibility.”); id. at 25-26 (J.A. at 370-71) ("Efforts to decide whether a particular source is contributing to regional haze have thus far encoun*17tered grave obstacles. Studies designed to estimate the effect of a particular source on surrounding visibility are expensive, and-the results can be uncertain and controversial.”). To take just one example, ‘‘the efforts to trace the contribution of the Navajo Generating Station to haze in the Grand Canyon National Park took several years and cost millions of dollars without leading to quantitatively definitive answers.” Id. at 7 (J.A. at 361).
. EPA, Resp. to Pets, for Recons, of Regional Haze Rule 16 (Jan. 10, 2001) (J.A. at 17) (quoting NAS Report at 7 (J.A. at 361)); see also NAS Report at 240 (J.A. at 478) ("The committee doubts ... that such attributions could be the basis for a workable visibility protection program.”).
. See Appalachian Power, 135 F.3d at 801-02 ("Our analysis is guided by the deference traditionally given to agency expertise, particularly when dealing with a statutory scheme as unwieldy and science-driven as the Clean Air Act.”); see also Husqvarna AB v. EPA, 254 F.3d 195, 199 (D.C.Cir.2001).
. Under the Haze Rule, the state must establish the first condition directly and the second through the application of computer modeling techniques. See Regional Haze Regulations, 64 Fed.Reg. at 35,740, 35,741; supra note 3.
. See Regional Haze Regulations, 64 Fed.Reg. at 35,741 ("EPA interprets the language 'from the use of such technology' to refer to the application of BART level controls to all sources subject to BART.”).
. Reply Br. for Industry Pet’rs at 8 ("Industry Petitioners agree ... that states are free to determine the weight and significance to be assigned to each of the CAA § 169A(g)(2) factors.”); see Op. at 7; cf. American Textile Mfrs. Inst., Inc. v. Donovan, 452 U.S. 490, 510, 101 S.Ct. 2478, 2491, 69 L.Ed.2d 185 (1981) ("When Congress has intended that an agency engage in cost-benefit analysis, it has clearly indicated such intent on the face of the statute.”); Central Ariz., 990 F.2d at 1542 n. 10 (holding that "Congress has not required 'cost-benefit' analysis in the [Clean Air] Act”).
. The court does not dispute the reasonableness of this presumption. See American Iron & Steel Inst. v. EPA, 115 F.3d 979, 1000 (D.C.Cir.1997) (holding that it is reasonable for EPA to presume that if a pollutant is present in fish tissue at a level exceeding that set by regulation, then any facility "that contributes a pollutant to a body of water [in which the fish swims] ... has the reasonable potential to contribute to that exceedence”); see also Baltimore Gas & Elec. Co. v. Natural Res. Def. Council, Inc., 462 U.S. 87, 103, 103 S.Ct. 2246, 2255, 76 L.Ed.2d 437 (1983) (holding that a reviewing court must be "at its most deferential” when the agency is "making predictions, within its area of special expertise, at the frontiers of science”); American Trucking Ass'ns, Inc. v. EPA, 175 F.3d 1027, 1055 (D.C.Cir.1999) ("[W]e have expressly held that EPA’s decision to adopt and set air quality standards need only be based on reasonable extrapolations from some reliable evidence.” (internal quotation marks omitted)), rev’d on other grounds sub nom. Whitman v. American Trucking Ass'ns, Inc., 531 U.S. 457, 121 S.Ct. 903, 149 L.Ed.2d 1 (2001).
. The court correctly notes that under this exemption, it is EPA rather than the state that determines whether a source has made the required showing. EPA, however, does not rely on the exemption to answer the state-authority issue discussed in Part C below, but rather to counter the petitioners’ claim that the Haze Rule fails to provide a source with the opportunity to demonstrate that it makes no appreciable contribution to visibility impairment in a Class I area. Br. for EPA at 29-30, 32.
. See supra notes 4, 5; Br. for Maine, et al. at 10 (protesting that to adopt the industry petitioners' interpretation of § 169A(g)(2) ''would impose staggering and costly administrative burdens” on the states).
. The court states that the "agreement” referred to in the report was an agreement to reject the provisions of an earlier House bill. As there may have, been many reasons for rejecting that bill, the "[rjejection of [the] proposed legislation during the course of enactment provides a hazardous basis from which to determine legislative intent,” GAO v. GAO Pers. Appeals Bd., 698 F.2d 516, 525 n. 52 (D.C.Cir.1983), and a particularly hazardous foundation for a Chevron step one claim. In any event, the most the court can divine regarding the content of the agreement is that it was to insert language clarifying that the states were to "determine whether a source contributes to visibility impairment and, if so, what BART controls should be applied to that source." Op. at 8. As noted in the text, the Haze Rule leaves both determinations in the hands of the states.
. Cf. Appalachian Power Co. v. EPA, 249 F.3d 1032, 1047 (D.C.Cir.2001) (holding that a state’s development of its implementation plan under CAA § 110 is not “free of extrinsic legal constraints,” including EPA's reasonable construction of CAA § 126).