AMERICAN PETROLEUM INSTITUTE and National Petrochemical & Refiners Association, Petitioners, v. U.S. ENVIRONMENTAL PROTECTION AGENCY and Carol M. Browner, Administrator, U.S. Environmental Protection Agency, Respondents. Valero Energy Corporation, Intervenor, 198 F.3d 275

Opinion for the Court filed by Circuit Judge STEPHEN F. WILLIAMS. STEPHEN F. WILLIAMS, Circuit Judge: Automobile engines emit volatile organic compounds (“VOCs”), which together with nitrogen oxides (“NOx”) form ozone. Reformulated gasoline (“RFG”) can reduce VOCs emissions compared to levels associated with ordinary gasoline; but it costs more, and there is some concern about the nation’s current RFG production capacity. See 42 U.S.C. § 7545(k)(6)(B) (1994) (creating special provisions that operate in the event of insufficient capacity).

In guiding the efforts of the Environmental Protection Agency to limit ozone concentrations, Congress in the Clean Air Act (the “Act”) authorized limited reliance on RFG. It directed that RFG should initially be mandatory in the nine worst ozone areas with populations over 250,000, with any area later classified as “Severe” to be added as well. See 42 U.S.C. § 7545(k)(l), (5), (10)(D). And it also provided for “opt-in,” i.e., for election by a state to demand an EPA ban on the sale of non-RFG in specified areas.

But Congress carefully limited the eligible areas: Upon the application of the Governor of a State, the Administrator shall apply the prohibition [on the sale of non-reformulated gasoline] in any area in the State classified ... as a Marginal, Moderate, Serious, or Severe Area.... Act § 211(k)(6)(A), 42 U.S.C. § 7545(k)(6)(A) (emphasis added). There is a fifth category, “Extreme,” to which only Los Angeles belongs; there was no need to include that category because it was automatically covered by Congress’s direct mandate.

Apart from Los Angeles, most areas of the United States that are not in “attainment” for EPA’s ozone standards belong to one of the four specified categories. But for several reasons — mainly the interaction between Congress’s (1) rules limiting the ability of a nonattainment area to break into the broad sunlit uplands of attainment, and (2) provisions governing the treatment of missing data — an area may be in “nonattainment” but not in any of the four specified classes.

In interpreting the opt-in provision EPA decided that Congress meant to include not only “Marginal, Moderate, Serious, or Severe” areas, but also any other areas that either were currently out of attainment or had ever been. The American Petroleum Institute (“API”) has petitioned for review of the rule, arguing that the agency exceeded its statutory authority; we agree. The Act requires EPA to establish and periodically revise a primary national ambient air quality standard (“NAAQS”) for each air pollutant that the agency identifies as meeting certain criteria. See 42 U.S.C. §§ 7408-7409.

The primary NAAQS for each pollutant is the maximum concentration “requisite to protect the public health” with “an adequate margin of safety.” 42 U.S.C. § 7409(b)(1). In 1979 the EPA administrator set a primary NAAQS for ozone at 0.120 parts per million (“ppm”), averaged over intervals of one hour. See 44 Fed.Reg. 8202. That level was upheld by this court in American Petroleum Inst. v. Costle, 665 F.2d 1176 (D.C.Cir.1981), and remains in effect today. In approaching this case the most critical distinction is between “nonattainment” and “design value” as measures of compliance.

A maximum concentration, without more, leaves open the question of how often an area’s hourly reading can exceed 0.120 ppm without causing the area to be out of compliance. Congress adopted EPA’s answer to this question. It decreed, “by operation of law,” that each area’s attainment status would be based on the regulatory standards “in effect immediately before November 15, 1990.” 42 U.S.C. § 7407(d)(1)(C). Under those standards (also still in effect), an area is allowed no more than one day a year in which its maximum hourly ozone concentration is greater than 0.120 ppm.

But the exact formula is more complicated because it recognizes that many areas will not have data for every hour of the year; it therefore uses estimates to fill this gap. See 40 CFR pt. 50, App. H. The formula generates an “expected number of days per calendar year with maximum hourly average concentrations above 0.12 parts per million,” id. § 50.9(a), and if the expected number of exceedances for a three-year period is greater than one, the area is in nonattainment.

In 1990 Congress also introduced, for ozone, a refinement based on how far each nonattainment area was from attainment status, establishing different dates for compliance according to the severity of the existing violations. See 42 U.S.C. § 7511(a)(1). To group areas according to the various deadlines, Congress used a concept already in use by EPA, known as “design value,” and once again adopted EPA’s method for calculating this number. See id. (“The design value shall be calculated according to the

interpretation methodology issued by the Administrator most recently before November 15, 1990.”). Much like the calculation of attainment, EPA’s method for determining design values also excuses one exceedance per year (e.g., the first three exceedances in a three-year period have no effect on the design value). But it has no mechanism for generating data to fill gaps in monitoring: Here, the design value is simply the fourth-highest daily maximum ozone concentration in an area over three consecutive years for which there are sufficient data. See American Trucking Ass’ns v.

EPA 175 F.3d 1027, 1046 n. 6 (D.C.Cir.1999) (“ATA”), modified on reh’g, 195 F.3d 4 (D.C.Cir.1999); EPA, The Clean Air Act Ozone Design Value Study: Final Report 1-3 to 1-5 (1994). In § 181(a)(1) of the Act Congress used design value to create five categories of nonattainment, with varying compliance deadlines for each category: Marginal 0.121 to 0.138 ppm Moderate 0.138 to 0.160 ppm Serious 0.160 to 0.180 ppm Severe 0.180 to 0.280 ppm Extreme 0.280 ppm and above See 42 U.S.C. § 7511(a)(1) tbl.1.

But because Congress treated missing data differently for purposes of design value and attainment status, while every area with a known design value above 0.120 ppm is in nonattainment, an area may be in nonat-tainment even though its design value is 0.120 ppm or below. EPA called such areas “submarginal.” See 56 Fed.Reg. at 56,697/2 (1991). Similarly, § 107(d)(1)(C)(

i) of the Act, 42 U.S.C. § 7407(d)(1)(C)(i), requires areas designated nonattainment under portions of the previous standards, see 42 U.S.C. § 7407(d)(1)(C) (adopting provisions of the Clean Air Act Amendments of 1977, § 103, Pub.L. No. 95-95, 91 Stat. 685, 687-88), to remain so classified because of inadequate data. EPA called such areas “incomplete data areas.” 56 Fed.Reg. at 56,697/3; cf. 42 U.S.C. § 7511e (allowing areas that can demonstrate compliance with the ozone NAAQS for the years 1987-89 to have a special, “transitional,” status).

The key issue here is the application of the RFG program to these two types of areas, “submarginal” and “incomplete data.” * * * In the disputed rule, EPA stated that any “area currently or previously designated as a nonattainment area for ozone under 40 CFR 50.9 ... or any time later, may be included on petition of the governor.” 40 CFR § 80.70(k); 63 Fed.Reg. at 52,104. Largely because of the divergence between the concepts of nonattainment and design value, this rule swept into “opt-in’’ a variety of areas not belonging to the four categories specified by Congress-Marginal, Moderate, Serious or Severe.

We assess the validity of the rule under the familiar two-step process in Chevron U.S.A. Inc. v. NRDC, 467 U.S. 837, 842-43 & nn.9 & 11, 104 S.Ct. 2778 (1984). Chevron requires us to determine whether Congress spoke “to the precise question at issue.” Id. at 842, 104 S.Ct. 2778. It is hard to imagine how Congress could have done so more clearly. Acting within a universe where nonattainment and the four categories overlap but are distinct, Congress chose the four categories. If it meant to express “nonattainment,” its wording was not merely a long-winded but a positively obtuse way of doing so.

As we said in Michigan Citizens for an Independent Press v. Thornburgh, 868 F.2d 1285 (D.C.Cir.), aff'd by an equally divided court, 493 U.S. 38, 39, 110 S.Ct. 398, 107 L.Ed.2d 277 (1989), if Congress makes an explicit provision for apples, oranges and bananas, it is most unlikely to have meant grapefruit. Id. at 1293. Despite the text, EPA argues that the scope of § 211(k)(6) is ambiguous, thus opening the door to “reasonable”

interpretations by EPA. It notes that § 181(a)(1) of the Act states that “[e]ach area designated nonattainment for ozone ... shall be classified at the time of such designation ... as a Marginal Area, a Moderate Area, a Serious Area, a Severe Area, or an Extreme Area based on the design value for the area.” 42 U.S.C. § 7511(a)(1).

From this it infers that, despite the different methods for calculating design value and attainment status, Congress thought that no nonattainment area would be classified as other than Marginal, Moderate, Serious, Severe, or Extreme, and thus the reference to the first four categories in § 211(k)(6) was Congress’s way of making the RFG program available to all nonat-tainment areas. Because § 211(k)(6) does not prohibit the inclusion of nonattainment areas with design values below 0.121 or areas whose design values are unknown, EPA argues, it is at least ambiguous as to whether they may join.

EPA seems to think that the possibility that Congress was unaware of the nonat-tainment-design value divergence suggests that, had it been aware, it might have wanted EPA to allow nonattainment areas with incomplete data or design values below 0.121 ppm to require RFG. There are two problems here: the assumption of congressional ignorance is farfetched, and even if correct would not get EPA where it wants to go. In the normal case Congress is assumed to be conscious of what it has done, especially when it chooses between two available terms that might have been included in the provision in question. See BFP v.

Resolution Trust Corp., 511 U.S. 531, 537, 114 S.Ct. 1757, 128 L.Ed.2d 556 (1994) (“[I]t is generally presumed that Congress acts intentionally and purposely when it includes particular language in one

section of a statute but omits it in another.”). Sometimes (e.g., where the pieces of legislation are not closely linked in either codification or time of enactment) this assumption may be a stretch, justifiable in part because its effect is to push toward coherent

interpretations of law. See West Virginia Univ. Hosps., Inc. v. Casey, 499 U.S. 83, 100, 111 S.Ct. 1138, 113 L.Ed.2d 68 (1991). But here the assumption that Congress was aware of the law is sound: the divergence between nonattainment and design value is the direct product of distinctive

definitions explicitly adopted by Congress. To suppose that Congress was ignorant of the divergence is to impute sleepwalking to the legislators. We have already explained how Congress expressly adopted differing formu-lae. These formulae generate the two problematic categories at issue here. First, as we recognized in ATA the stringent criteria for changing an area’s designation from nonattainment to attainment ensures that there can and will be nonattainment areas with design values below 0.121 ppm (submarginal areas).

ATA, 175 F.3d at 1047; 42 U.S.C. § 7407(d)(3)(E) (noting that an area cannot be redesignated to attainment status until it shows compliance with the relevant NAAQS and that the improvement in air quality is due to permanent and enforceable reductions in emissions). Second, “incomplete data areas” have no hope of leaving nonattainment until they generate enough data to prove that they comply with the ozone NAAQS. See id. Thus, they must remain in nonattainment, but can secure the RFG option if they generate data placing them in the four eongressionally specified categories.

Quite sensibly, the literal reading of § 211(k)(6)(

A) provides RFG as an option when the need is clear, and only then. But even the ignorance assumption, were it true, would not support EPA’s inference. Having used words of art to describe areas eligible for opt-in, a hypothetically ignorant Congress would likely have assumed that if some areas turned up partly resembling the areas it specified' — areas out of attainment but less clearly so — they would not be subject to RFG in the absence of new congressional action.

More specifically, even if Congress had thought that, as of 1990, all nonattainment areas under the 0.120 ozone NAAQS would have a recorded design value of at least 0.121 ppm, it knew that the formula for nonattainment status (unlike the fixed values for design value) was likely to change over time. EPA has a continuing obligation to review and revise the NAAQS every five years, see 42 U.S.C. § 7409(d)(1); ATA 175 F.3d at 1049, and to redesignate attainment status accordingly, see 42 U.S.C. § 7407(d)(1)(B).

In ATA, we noted that Congress had locked the categories of § 181(a)(1) into place, presumably to avoid having its ozone enforcement scheme administratively overridden by EPA as a result of such revision. 175 F.3d at 1049-50. The same is true here. By basing the opt-in provisions in § 211(k)(6) on the statutorily imposed categories in § 181(a)(1), Congress could limit the scope of the RFG program to areas that clearly fall within the categories of its choosing.

On this record we are reluctant even to mention the legislative history. “[W]e do not resort to legislative history to cloud a statutory text that is clear.” Ratzlaf v. United States, 510 U.S. 135, 147-48, 114 S.Ct. 655, 126 L.Ed.2d 615 (1994); see also Sutton v. United Air Lines, Inc., — U.S. -, -, 119 S.Ct. 2139, 2146, 144 L.Ed.2d 450 (1999); United States v. Bost, 87 F.3d 1333, 1336 (D.C.Cir.1996). But it scarcely helps EPA. The conference report observed, “States could elect to have the [RFG] requirements apply in other cities with ozone pollution problems.” H.R. Conf. Rep.

No. 101-952, at 336, reprinted in 1 A Legislative History of the Clean Air Act Amendments of 1990, at 1449, 1786 (1993) [hereinafter Leg. Hist.]. But this is said simply to distinguish the statute’s mandate of RFG for specified regions, and certainly does not claim that every other city with any ozone pollution would qualify for opt-in. The floor debates add little clarity.

True, there are statements of the authors of the provision in question, and sponsors of the amendments generally, to the effect that “any” or “all” non-mandated ozone nonattainment areas could join the RFG program; but none shows enough attention to the problem presented here to overcome the plain language of the text. See Senate Debate on the Clean Air Act Amendments of 1990 Conference Report, reprinted in 1 Leg. Hist, at 731, 1024; House Debate on the Clean Air Act Amendments of 1990 Conference Report, reprinted in 1 Leg. Hist, at 1177, 1266; House Debate on H.R. 3030, reprinted in 2 Leg. Hist, at 2667, 2690.

Interestingly, all the statements contain inaccuracies on another issue, asserting that the area makes the election, not the state or governor, a position with no support in the statute. The colloquial language of debate is at best a rough guide to the intricacies of technical statutory wording. Similarly, the Senate and House committee reports that list the likely classification of nonattainment areas under § 181(a)(1) do not show that Congress meant to base RFG participation on attainment status. Both lists classify the areas according to their design values.

The Senate list does not even mention attainment status, and is thus no evidence at all of congressional determination that it should control. See S.Rep. No. 101-228, at 35-37 (1989), reprinted in 5 Leg. Hist, at 8375-77. Nor does the House Report speak of nonattainment. It uses the looser term “areas violating the ozone NAAQS” and then lists areas by design value. H. Rep. No. 101-490, at 230-32, reprinted in 2 Leg. Hist, at 3254-56.

EPA would have us believe that the list demonstrates acceptance of its view that Congress really meant nonattainment because the table includes two areas, Jacksonville, FL and Waldo Co., ME, as likely to be classified as marginal areas despite their design values of only 0.120 ppm. But the likely explanation is that the compiler mistakenly thought that because 0.120 was the cut-off point for “Marginal,” an area with exactly that reading should be so classified.

Moreover, these lists can shed no light on the proper classification of areas that lack sufficient data to calculate a design value or even to confirm their nonattainment status. EPA next argues that even if the text is clear, this case presents one of the rare instances “in which the literal application of a statute will produce a result demonstrably at odds with the intentions of its drafters.” R.G. Johnson Co. v. Apfel, 172 F.3d 890, 895 (D.C.Cir.1999) (internal quotations omitted).

The agency appears to find absurdity because under the direct reading of the statute it denies opt-in to some areas with “continuing ozone problems.” But the argument assumes away all trade-offs. Given the acknowledged cost and supply drawbacks associated with RFG, it seems entirely sensible to confine opt-in to areas experiencing nonattainment with the comparative clarity implied by belonging to one of the four specified categories. We see no absurdity.

EPA offers a special argument under which it could reach back into history to allow opt-in for an area that once was — but is no longer — classified as Marginal, Moderate, Serious, or Severe.

The statute allows opt-in for “any area in the State classified ... as a Marginal, Moderate, Serious, or Severe Area.” As a matter of sheer linguistic possibility, either of two explicit phrases could be understood to precede the word “classified”: it could read (1) “any area in the state [that is presently] classified ... as a Marginal, Moderate, Serious, or Severe Area,” or (2) “any area in the state [that has ever been] classified as a Marginal, Moderate, Serious, or Severe Area.” EPA favors the second reading, but it seems utterly implausible.

If an area is in attainment, its historical design value has no relationship to its need for RFG. If it is in nonattainment, but lacks sufficient data to be classified under § 181(a)(1), then RFG will be an option if, in the process of generating sufficient data to prove itself in attainment, it is shown to have a design value of 0.121 ppm or above. See 42 U.S.C. § 7407(d)(3)(E). In § 211(k)

(6) Congress provided for opt-in only for areas classified as Marginal, Moderate, Serious or Severe. It meant what it said. Accordingly, API’s petition for review is Granted. . In 1997 the EPA adopted a revised ozone NAAQS of 0.08 ppm averaged over an eight-hour period. See 62 Fed.Reg. 38,856 (1997). But in American Trucking Ass’ns v. EPA, 175 F.3d 1027, 1038 (D.C.Cir.1999) (“ATA"), modified on reh’g, 195 F.3d 4 (D.C.Cir.1999), we granted a petition for review of that order and remanded to the EPA with instructions to provide an intelligible principle guiding its

interpretation of the relevant sections of the Act. See ATA, 175 F.3d át 1038-40. Thus, the 0.12 ppm, 1-hour standard remains in place. . In the rulemaking the EPA expressed its belief that areas in nonattainment for the new, more stringent ozone NAAQS, would be allowed to opt into the RFG program. See 63 Fed.Reg. 52,094, 52,101 (1998). The issue is temporarily moot in the light of ATA, but on its face such a claim seems even less well-founded than EPA's core position.