Petitioner challenges the actions of the Environmental Protection Agency (EPA) in rejecting portions of the State of Idaho’s implementation plan under the Clean Air Act, 42 U.S.C.A. §§ 1857 , et seq. (Supp. 1977), and substituting therefor its own regulations. See 40 Fed.Reg. 53584 (1975). The rejected portions of the Idaho implementation plan deal with the control of sulfur dioxide (S02) from the Idaho plant of petitioner. The petitioner’s challenge relies on several grounds. While we find in favor of the Administrator on various legal and procedural issues, we remand this proceeding to the EPA for further consideration of the technological feasibility of certain modifications of petitioner’s smelter operations which would be required by the substituted regulations.

History of the Case.

Bunker Hill’s Kellogg, Idaho operations include lead and zinc smelters that emit substantial amounts of S02 gas. Some of the S02 gas presently is vented directly to the atmosphere; the remainder is treated in three acid plants — two servicing the zinc operations, and the third handling the lead smelter. There is no dispute that, as presently operated, Bunker Hill’s Idaho plant is not meeting the federal ambient air quality standards for SO2. The primary issue before us is what modifications Bunker Hill must make in its operations in order to satisfy the requirements of the Clean Air Act. To understand this issue the administrative proceedings culminating in this challenge will be outlined and thereafter the focus will be upon the technological feasibility of the control technology made necessary by EPA’s substituted regulations. Our disposition of this challenge will conclude with a discussion of certain other issues raised by petitioner and with our instructions pertaining to the procedure on remand to the EPA.

On January 3, 1975, the Idaho Department of Health and Welfare (IDHW) adopted a regulation requiring Bunker Hill to “capture” 72 percent of its SO2 emissions; the IDHW had decided, after lengthy hearings, that this percentage emission control was the maximum percentage feasible under currently available technology. This level of emission control would be achieved through a series of specific emission limitations: (i) a 4000 parts per million (ppm) SO2 limitation on the emissions from each of Bunker Hill’s acid plants and from its zinc plant main stack, based on eight-hour averages; (ii) a 100 tons per day limitation on emissions from the lead plant main stack; and (iii) a 1200 tons per week limitation on emissions from the entire smelting complex.

Since even this level of emission control would not meet the ambient air quality standards, the Idaho plan also would require Bunker Hill to supplement its control program with whatever dispersion enhancement techniques prove necessary to meet these standards. Bunker Hill would be held responsible for violations in the Kellogg Valley of any federal ambient air quality standard, although, in the case of the primary annual standard, a 30-day investigation period would be provided for purposes of determining whether other undetected and uncontrolled sources might have been the actual cause of the violation, in which case Bunker Hill would not be held liable. Finally, Bunker Hill would be required to conduct research and development aimed at improving emission control techniques; as these techniques became feasible, Bunker Hill’s emission limitations would be tightened.

The EPA refused to approve Idaho’s proposed emission limitations and instead substituted standards that would guarantee 82 percent control. On the basis of a study commissioned by EPA of Bunker Hill’s operation, together with other evidence, the EPA held that 82 percent, rather than 72 percent, was the maximum level of control technologically and economically feasible for Bunker Hill. The EPA study (prepared for EPA by Mr. Tim Browder, an expert in SO2 control technology — hereinafter referred to as the Browder Study) contended that various efficiency problems presently plaguing Bunker Hill’s control process could be solved by modifying Bunker Hill’s acid plants. In particular, the Browder Study argued that adding sulfur burners to the acid plants would guarantee sustained autothermality.

Under the EPA regulations, SO2 emissions from Bunker Hill’s acid plants would not be allowed to exceed 2600 ppm on a 6-hour average; emissions from the main lead stack would not be allowed to exceed 2000 ppm; and a 680 tons per day limitation would be imposed on the entire Kellogg operation. Since even 82 percent control would not guarantee attainment of the air quality standards, dispersion enhancement techniques would again be required. As under the Idaho provisions, Bunker Hill would be held accountable for any violations of the federal ambient air quality standards; however, the EPA made no provision for a 30-day investigation period. Again, Bunker Hill would be required to carry on a research and development program with the objective of raising over the years the level of constant emission control.

Bunker Hill focuses its attack on the EPA’s determination that Bunker Hill can feasibly achieve 82 percent emission control. First, it argues that the EPA was bound by the contrary conclusion of the IDHW that only a 72 percent control level is feasible. Second, even assuming that the EPA was free to reconsider the question of the maximum feasible control level, Bunker Hill argues that the EPA’s determination was “arbitrary and capricious” and was marred by various procedural irregularities. Bunker Hill also attacks the EPA’s decision to eliminate the 30-day review period provided for in the original Idaho regulations.

Our study of the administrative record cast considerable doubt on whether the sulfur burner system proposed in the Browder Study, a necessary element in EPA’s proposed 82 percent emission control standard, is indeed technologically feasible. In particular, Bunker Hill’s principal expert witness contended that sulfur burners could not track, or be coordinated with, the wide fluctuations in SO2 concentration that plague Bunker Hill’s smelter gases. Mr. Browder failed to respond fully and satisfactorily. However, because of the complexity of the technological issues involved and the importance of clearly establishing the deficiencies in the Administrator’s deliberations before remanding the regulations, we requested a supplemental brief from the EPA clarifying Browder’s sulfur burner proposal; a reply brief was requested from Bunker Hill and both parties were allowed to augment the record. The EPA was also granted permission to file a response to Bunker Hill’s reply brief.

In reviewing administrative regulations, the courts generally are forbidden from conducting a full-fledged and independent evidentiary hearing. See, e. g., Independent Meat Packers Ass’n v. Butz, 526 F.2d 228, 239 (8th Cir. 1975), cert. denied, 424 U.S. 966 , 96 S.Ct. 1461 , 47 L.Ed.2d 733 (1976). Neither can the courts uphold regulations on the basis of post-hoc rationalizations offered by the agency. See, e. g., Local 814, International Brotherhood of Teamsters v. NLRB, 546 F.2d 989, 992 (D.C. Cir. 1976); Bradley v. Weinberger, 483 F.2d 410, 414-15 (1st Cir. 1973). See generally Camp v. Pitts, 411 U.S. 138,142-43 , 93 S.Ct. 1241 , 36 L.Ed.2d 106 (1973); Citizens to Preserve Overton Park v. Volpe, 401 U.S. 402, 420-21 , 91 S.Ct. 814 , 28 L.Ed.2d 136 (1971). But in the often difficult task of reviewing administrative regulations, the courts are not straight jacketed to the original record in trying to make sense of complex technical testimony, which is often presented in administrative proceedings without ultimate review by nonexpert judges in mind. Here, the augmenting materials were merely explanatory of the original record. No new rationalization of the SO2 regulations was offered by the EPA. Instead, the augmenting materials clarified a dispute that we felt was less than clear from the original record and were clearly admissible.

Despite this clarification in dealing with a technical subject with respect to which we are not experts and ever mindful of our limited authority in reviewing the Administrator’s actions, we nonetheless reluctantly are compelled, on the basis of the record as augmented, to conclude that the Administrator has not “exercised a reasoned discretion” in concluding that the Browder sulfur burner proposals, and hence an 82 percent emission control standard, are technologically feasible. See Portland Cement Ass’n v. Ruckelshaus, 158 U.S.App. D.C. 308, 486 F.2d 375, 402 (1973), cert. denied, 417 U.S. 921 , 94 S.Ct. 2628 , 41 L.Ed.2d 226 (1974). We must, therefore, remand to the EPA for further proceedings.

II.

Maximum Feasible Control Technology.

The importance of the technological feasibility of the Browder sulfur burner proposals has its source in the fact that cases in this and other circuits have established that the national ambient air quality standards must be met, to the maximum extent possible, by constant emission controls (such as acid plants augmented or not by sulfur burners); dispersion enhancement techniques (such as tall stacks) are to be used only in a supplemental role where necessary to meet the air quality standards. See Kennecott Copper Corp. v. Train, 526 F.2d 1149 (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 (6th Cir. 1975), cert. denied, 425 U.S. 934 , 96 S.Ct. 1663 , 48 L.Ed.2d 175 (1976); Natural Resources Defense Council v. EPA, 489 F.2d 390 (5th Cir. 1974), reversed in part on other grounds sub nom., Train v. Natural Resources Defense Council, 421 U.S. 60 , 95 S.Ct. 1470 , 43 L.Ed.2d 731 (1975). The Clean Air Act, as interpreted by these cases, requires pollution sources to achieve the highest emission control level that is technologically and economically feasible. Id. Conversely, the EPA cannot require a level of control technology that is technologically and economically infeasible. Thus, if the Browder sulfur burner proposals are technologically feasible, their use is required because only then will constant emission controls be in use to the maximum extent possible. If not, their use cannot be required.

A. EPA’s Ability to Reexamine State Determinations of Feasibility.

Bunker Hill, however, argues that the technological feasibility of constant emission controls is fixed by the Idaho state plan. Thus, it insists that EPA cannot require the implementation of the Browder sulfur burner proposals because Idaho, by not requiring their implementation, has conclusively established their technological infeasibility. We disagree.

The Clean Air Act requires that the Administrator disapprove any state implementation plan that he “determines” fails to meet any of the requirements of 42 U.S. C.A. § 1857c-5(a)(2) (Supp.1977). One of. these requirements is the maximum feasible use of constant control systems. No argument has been suggested for why, on the one hand, the Administrator can examine some state findings in reviewing a state implementation plan (e. g., a finding that the plan assures the attainment of the various ambient air quality standards) but, on the other hand, be bound by state findings of feasibility or infeasibility. The Clean Air Act requires that the Administrator “determine” that a state plan meets all of the requirements of section 1857c-5(a)(2) before approving the plan, and one of these requirements is the maximum feasible use of constant control systems. Bunker Hill’s reading would reduce EPA’s approval of the Idaho implementation plan to a rubber stamp, a result which, we are certain, Con gress did not intend. To the extent that the recent district court opinion in Kennecott Copper Corp. v. Train, 424 F.Supp. 1217, 1231 (D.Nev.1976) (holding that the Administrator was bound by a state finding of economic infeasibility) suggests to the contrary, we believe it to be in error.

B. Review of the EPA’s Feasibility Findings.

Although EPA is not bound by Idaho’s determination of technological feasibility, its rejection of the state plan must not be “arbitrary, capricious, an abuse of discretion, or otherwise not in accordance with law”, 5 U.S.C. § 706 (1970). See Duquesne Light Co. v. EPA, 522 F.2d 1186, 1192-93 (3rd Cir. 1975), vacated on other grounds, 427 U.S. 902 , 96 S.Ct. 3185 , 49 L.Ed.2d 1196 (1976), South Terminal Corp. v. EPA, 504 F.2d 646 , 655 (1st Cir. 1974). Moreover, from Kennecott Copper Corp. v. Train, 526 F.2d 1149 (9th Cir. 1975), cert. denied, 425 U.S. 935 , 96 S.Ct. 1665 , 48 L.Ed.2d 176 (1976), it is clear that, before requiring 82 percent control, the Administrator was required to find that this level of emission control is economically and technologically feasible for Bunker Hill to achieve. In reaching this determination, the Administrator must have exercised “reasoned discretion.” Portland Cement Ass’n, supra, at 402. The Administrator’s finding must have been “based on a consideration of the relevant factors” and not “a clear error of judgment.” Citizens to Preserve Overton Park v. Volpe, 401 U.S. 402, 415-16 , 91 S.Ct. 814, 823-824 , 28 L.Ed.2d 136 (1971).

The technological feasibility of the EPA standards for Bunker Hill is dependent on a finding that sulfur burners are technologically capable of curing the problems of fluctuating SO2 concentrations facing Bunker Hill. The feasibility of Browder’s sulfur burner proposal could have been established in two ways. First, EPA could have located contemporary examples of sulfur burners being used to cure a similar problem in a similar context. Second, even if the proposed sulfur burner system was not in current use, EPA could have relied on expert testimony demonstrating that sulfur burners under present technology can cure Bunker Hill’s SO2 concentration problem. This second method of establishing technological feasibility, however, carries a significant burden of proof. The demonstration of feasibility must not be based on a “subjective understanding of the problem or ‘crystal ball inquiry.’ ” Essex Chemical Corp. v. Ruckelshaus, 158 U.S. App.D.C. 360, 486 F.2d 427, 433 (1973), cert. denied, 416 U.S. 969 , 94 S.Ct. 1991 , 40 L.Ed.2d 558 (1974). The expert testimony must clearly demonstrate that the technology is available and must answer cogent criticisms of opposing experts. The proposed technology must not be “at a level that is purely theoretical or experimental”. Id. at 433-34 . On the basis of the record before us, we cannot say that the Browder sulfur burner proposal is other than “purely theoretical or experimental.”

1. Bunker Hill’s Problem.

To validate this view it is necessary that the manner in which Bunker Hill operates be described in some detail. Its problem fundamentally is whether, given its operational necessities, the 2600 ppm acid plant limitation can be met. A significant percentage of current SO2 emissions from Bunker Hill’s lead and zinc smelters is controlled through the use of their three acid plants. These acid plants convert SO2 into SO3 and ultimately into sulfuric acid, which can either be sold or neutralized. A necessary element to this catalytic process is heat, which can be provided, at least in part, by circulating back the heat given off by the transformation process itself. The more concentrated the gas being converted the more heat will be given off by the transformation process. If the gas stream being emitted into the acid plant has a sufficient SO2 concentration, then the heat given off by the transformation process will be sufficient in itself to run the transformation process almost indefinitely. This condition of operation is known as autothermality.

At Bunker Hill, there are frequent periods during which the three acid plants cannot operate autothermally. During these periods, either SO2 must be vented from the smelters directly to the atmosphere, or the acid plants must be operated at abnormally low efficiencies (resulting in emissions significantly greater than 2600 ppm). None of Bunker Hill’s three acid plants can achieve autothermality immediately upon start-up of operations because the gas initially emitted from the operations is very low in SO2 concentration. According to Bunker Hill’s principal expert, Mr. J. R. Donovan, the feed gases to all three acid plants are plagued by “relatively large” fluctuations in SO2 concentration, hindering autothermality. Joint Appendix [J.A.] 1131. These difficulties particularly trouble the acid plant servicing the lead sintering process (hereinafter referred to as the “lead-smelter acid plant”). The sintering machine conveyor belt is stopped and started several times a day. During the stopped period, which is typically one or two hours, lead concentrate remaining on the belt continues to smolder, producing a low level of SO2 emission. When the sintering machine is restarted, a strong but widely fluctuating SO2 gas stream is produced that can continue for extended periods of time. J.A. 991-92; Respondent’s First Supplemental Brief, at 2-4.

EPA does not dispute the significance of these efficiency problems, at least with respect to the lead-smelter acid plant. See Respondent’s First Supplemental Brief, at 8-9. However, on the basis of the Browder Study, EPA contends that these problems can be solved and a more efficient operation obtained by various modifications in the acid plants, in particular the addition of sulfur burners.

2. Feasibility of Sulfur Burners.

The Original Record:

The Browder Study concluded that the problem of fluctuating SO2 content could be successfully overcome by the use of sulfur burners. J.A. 742-43, 745. Bunker Hill’s expert, Mr. J. R. Donovan, strongly disagreed in a series of rebuttal comments. In particular, according to Donovan, sulfur burners

In summary, according to Donovan, sulfur burners have been used successfully to upgrade weak SO2 concentrations (with minor fluctuations in strength) but have never been used, and, given current technology, probably cannot be used, to even out wide fluctuations in SO2 streams, the problem presented by the Bunker Hill operations. See J.A. 524, 1131-33.

The original record nowhere contains a reasoned refutation of Donovan’s objections. Nor does it contain anything approaching an adequate demonstration of the feasibility of sulfur burners. Brow der’s principal but inadequate rebuttal to Donovan consisted of a list of 37 acid plants that purportedly use sulfur burners. J.A. 832-33. However, as Bunker Hill pointed out, only three of the plants shown on the list service nonferrous smelters. And there is no indication that any of the acid plants on the list confront the same apparent problems of SO2 feed concentration as Bunker Hill’s acid plants. As Donovan admitted in his own testimony, sulfur burners are used on acid plants; the issue presented by Donovan is whether sulfur burners have been used on a plant similar to Bunker Hill’s plants. See J.A. 561. Browder’s list is meaningless without any discussion of the analogousness of the acid plants listed to the problem confronted.

Bunker Hill also cast considerable doubt on the reliability of the list. According to Bunker Hill, two of the three acid plants listed as servicing nonferrous smelters are nonexistent and the remaining acid plant does not use sulfur burners. J.A. 1109-11. EPA did not reply to this critique.

EPA, however, subsequently supplemented the Browder list with a claimed addition al example of an acid plant utilizing sulfur burners. According to letters from the State of Montana, “[t]wo sulfur burners are being installed in the East Helena Lead Smelter.” J.A. 1016; see also J.A. 1025,1163. However, there is again no indication that the East Helena facility is a relevant analogy; relevance cannot simply be presumed. There is no evidence in the original record that the East Helena smelter is plagued by the same wide fluctuations in SO2 concentration as the Bunker Hill operations. What evidence there is in the original record indicates that sulfur burners are being installed at East Helena for a totally different purpose than the purpose for which the Browder study recommends their installation. According to the owner of the East Helena facility, the sulfur burners are being installed due

And, of course, all must admit that the fact that sulfur burners are being installed at a lead smelter does not prove that the sulfur burners are feasible means of meeting the problems of Bunker Hill’s lead smelter emissions.

Browder also attempted to support his proposal by arguing that various acid plant manufacturers have recommended or installed sulfur burners. Browder claimed that Monsanto had “previously installed sulfur burners in acid plants with fluctuations weak gas [sic]” and that “one is being installed for AMAX in Pennsylvania, at this time.” J.A. 791. But no supporting evidence was presented. And again there is no discussion of whether these purported installations are relevant analogies. Browder also stated his “understanding” that Lurgi, a builder of sintering machines, rec ommends the use of sulfur furnaces “when the Lurgi Sintering Machine is shut down, and to upgrade the gas of low concentrations.” J.A. 790. Even if this “understanding” were to be substantiated, which it is not in the record before us, there is no indication that Lurgi recommends the use of sulfur furnaces to even out large fluctuations in S02 gas as is being proposed by EPA in this instance.

In summary, there was no evidence in the original record demonstrating that sulfur burners are presently utilized by acid plants with similar input problems to those confronted by Bunker Hill. None of the acid plants cited by EPA as using sulfur burners were claimed to be, let alone shown to be, analogous to Bunker Hill’s acid plants. Furthermore, considerable doubt was cast on the validity of many of the purported examples of sulfur burner use. EPA made no attempt through detailed expert testimony to demonstrate that the S02 concentration problems outlined by Donovan could be overcome. Thus, on the basis of the original record, we would be forced to conclude that to require Bunker Hill’s acid plants to meet a 2600 ppm emission limitation would be arbitrary and capricious.

The Augmented Record:

As noted above, because of the apparent deficiencies in the Administrator’s proof of technological feasibility, we requested a supplemental brief from the EPA clarifying and elaborating on Browder’s sulfur burner proposals. The EPA was also permitted to augment the record. EPA’s brief virtually concedes that the East Helena smelter is not a relevant analogy in evaluating the feasibility of Browder’s sulfur burner proposal. See Respondent’s Second Supplemental Brief, at 12-14. And no evidence was presented documenting any case where sulfur burners have been or are being used to cure fluctuation problems similar to those encountered at Bunker Hill. However, EPA introduced a letter from Mr. Browder to the EPA detailing his sulfur burner proposal, including what he asserts are feasible automatic and manual methods of adjusting sulfur burners to even out widely fluctuating SO2 concentrations. Letter from Tim J. Browder to Kenneth A. Lepic, submitted by respondent in augmentation of the record, at 4-7.

Material submitted by Bunker Hill in augmentation of the record, however, casts significant doubt on the feasibility of the specific sulfur burner system proposed in Mr. Browder’s letter to the EPA. According to Mr. J. B. Rinckhoff, a Supervising Process Engineer for Davy Powergas, Inc., Browder’s “proposed control method of measuring fluctuations in SO2 content and gas volume and adjusting sulfur burning rates to produce a more uniform gas stream are impractical and unworkable.” Affidavit of J. B. Rinckhoff, supra, at 5. Moreover, according to Mr. Rinckhoff, the proposed system would have to burn sulfur with a range of from 0 to 100 tons-per-day in order to even out the wide SO2 concentration fluctuations at Bunker Hill. This required range of operation would pose further serious, and possibly unsolvable, problems since (i) “no sulfur nozzle exists capable of covering that range,” (ii) such a range would produce “large amounts of excess heat” that would be difficult if not impossible to remove from the system, and (iii) the furnace specified by Mr. Browder in his letter “is suitable only for a 10 ton-per- day operation.” Id. at 5-6, 7-10. Although the EPA argues in response that the Browder proposal would actually only have to meet a 0 to 25 tons-per-day range, it is impossible to resolve this dispute on the basis of the augmented record before us.

Given these significant and unresolved criticisms of the Browder proposal, a remand to the EPA for further consideration of the feasibility of its standards becomes necessary. Even read in the deferential light required of a reviewing court, the record before us merely establishes that Browder’s sulfur burner proposal might work. As we noted above, this is not enough. The record must establish that the required technology is feasible, not merely possibly feasible. Approval of EPA’s standards on the basis of this record would seriously risk closing down a major smelting operation on the strength of a showing that does no more than demonstrate that the technology on which EPA relies is only “theoretical or experimental”.

3. Extent to Which the 2600 ppm Standard Would Be Met Assuming the Feasibility of the Sulfur Burners.

Our remand provides an opportunity for the EPA to clarify an additional area of uncertainty which would exist even had EPA succeeded in establishing adequately the feasibility of Browder’s sulfur burner proposal. It concerns the percentage of time the proposal would enable Bunker Hill to meet the 2600 ppm standard. According to the original Browder Study, the modifications proposed by the Study, if feasible, would “produce equivalent average stack losses of [1500-2000] ppm for most of the time.” J.A. 729. In later portions of the original record, Mr. Browder translates “most of the time” to mean 85 percent-plus of the time. J.A. 769, 775. It is unclear from these statements, however, whether the Browder Study modifications would guarantee that Bunker Hill could meet the 2600 ppm standard all of the time, as required by the EPA regulations.

Clarifying materials submitted by EPA in augmentation of the record have not resolved this uncertainty. According to an affidavit submitted by Mr. Browder,

Prom this and other statements of Mr. Browder, the EPA concludes that Brow-der’s modifications “would, in a more or less gaussian (or bell-like) distribution, achieve such emissions levels [1500-2000 ppm] at the Bunker Hill acid plants in excess of 85% of the operating time.” Respondent’s First Supplemental Brief, at 28. Thus, “the likelihood of exceeding the 2600 ppm limitation would be less than .01%, excluding prolonged malfunction periods. This is as close to a guarantee of achieving the emission limit as any piece of equipment can be expected to come.” Id.

We agree with EPA that it need not show that its standards can be met 100 percent of the time. Given the possibility of mechanical failure, etc., a showing that the standards can be met 99 percent-plus of the time is sufficient. Furthermore, assuming that Browder’s guarantees are valid, we would agree with EPA that the measure for performance has been met. However, doubts are raised by Bunker Hill as to whether Mr. Browder’s interpretation of his earlier statements is consistent with the proposals that he is making. According to Donald A. McCaughran, a Professor of Biological Statistics at the University of Washington Center for Quantitative Sciences:

Given the possible inconsistency in Mr. Browder’s interpretation of his 85 percent-plus guarantee, the EPA should clarify upon remand the percentage of the time that Browder’s proposals will ensure Bunker Hill’s meeting the 2600 ppm standard.

C. Procedural Irregularities.

Bunker Hill also argues that the proposed EPA standards must be remanded because of various procedural irregularities. Bunker Hill complains that the EPA failed to reveal several important documents in the record until after the EPA had promulgated its questioned regulations, thus denying Bunker Hill the opportunity to inspect and comment. Bunker Hill’s charges on their face are quite serious in light of the fact that most of the complained-of documents deal with the feasibility of sulfur burners — the rebuttal comments of Browder, the list of acid plants purportedly using sulfur burners, and the exhibits showing East Helena’s intention to utilize sulfur burners in its new acid plant.

We believe, however, that our order allowing the parties to submit supplemental briefs on the feasibility issue and to augment the record has cured these procedural errors. Bunker Hill has had an opportunity to comment on the materials that they argue were excluded from their view below and to submit rebuttal materials. Thus, we find it unnecessary to consider further Bunker Hill’s procedural arguments.

However, the EPA should be reminded that “adherence to proper prescribed procedure is the soundest route to a correct substantive result.” Ethyl Corp. v. EPA, 541 F.2d 1 , 94 (D.C. Cir.), cert. denied, 426 U.S. 941 , 96 S.Ct. 2662 , 49 L.Ed.2d 394 (1976) (Wilkey, J., dissenting). See also International Harvester v. Ruckelshaus, 155 U.S.App.D.C. 411 , 478 F.2d 615 , 652 (1973) (Bazelon, J., concurring). Greater attention to the procedural rights of petitioner perhaps might have forestalled the necessity of this appeal.

The 30-Day Investigation Period.

Bunker Hill also objects to EPA’s disapproval of Idaho’s provision for a 30-day investigation period following a violation in the Kellogg Valley of the primary annual ambient air quality standard. Under the Idaho regulations, the period would be used to determine whether other undetected and uncontrolled sources might have been the actual cause of the violation. See p. 1291, supra. EPA’s proposal, on the other hand, would make Bunker Hill liable for any violation of the standard and no formal opportunity to prove that it was not the cause of the violation would be available. We believe EPA’s decision to eliminate the investigation period as reasonably necessary for the EPA’s proper enforcement of the ambient air quality standards represents an exercise of reasoned discretion.

Section 1857c-5(a)(2) of the Clean Air Act requires that the EPA Administrator, before approving any portion of a state implementation plan, determine that the plan will meet the ambient air quality standards; this determination must include the adequacy of enforcement procedures. As noted by the EPA in its disapproval of the 30-day investigation period, it is extremely difficult to determine who is responsible for a violation of the ambient air quality standard where dispersion enhancement techniques are being used to meet the standards. Because of this difficulty,

This strikes us as sensible; therefore, EPA’s decision to eliminate the review period is not arbitrary or capricious.

IV.

Procedure on Remand.

Finally, we turn to the procedure on remand. We begin by noting that our remand is limited to the technological feasibility of the proposed EPA standards. Because we find it necessary for the EPA to consider further the technological feasibili ty of its standards, we do not reach Bunker Hill’s objections as to economic feasibility. After further consideration of the question of the technological feasibility of the proposed EPA standards, EPA can either affirm its present standards or issue new limitations. Whatever standards are set by EPA, however, must be supported by an adequate demonstration of their feasibility. This court will retain jurisdiction and Bunker Hill will be free to appeal the new regulations if it believes they are not supported by the administrative record.

Bunker Hill urges that it should have an opportunity upon remand to cross-examine EPA’s experts on the technological feasibility of Browder’s sulfur burner proposals. We agree. The type of administrative procedure due in a particular case turns on the nature of the issues presented. Marine Space Enclosures, Inc. v. Federal Maritime Comm’n., 137 U.S.App.D.C. 9 , 420 F.2d 577 , 589 n. 36 (1969). Fairness may dictate cross-examination on crucial issues, even though a strict reading of the APA does not. Walter Holm & Co. v. Hardin, 145 U.S.App.D.C. 347 , 449 F.2d 1009, 1016 (1971) ; see Kennecott Copper Corp. v. EPA, 149 U.S.App.D.C. 231 , 462 F.2d 846, 850 (1972) . In particular, cross-examination may be necessary if a “proceeding involves specific issues of critical importance that cannot be adequately ventilated” by normal procedures. Thompson v. Washington, 162 U.S.App.D.C. 39 , 497 F.2d 626 , 641 n. 48 (1973); see O’Donnell v. Shaffer, 160 U.S.App.D.C. 266 , 491 F.2d 59 , 62 (1974). Here, the issue upon remand is of a highly complex and technical nature; cross-examination will help crystalize the varying contentions of the experts and help guarantee that both parties’ experts are responsive to criticisms and counterarguments. Cross-examination of the witnesses will also aid this court in reviewing the new regulations should an appeal be brought.

We also recognize, however, that the present dispute has already lasted more than four years and has seriously delayed implementation of the goals of the Clean Air Act in the Kellogg Valley. All parties are anxious for a quick but fair resolution of the questions involved. An expedited remand and review process, therefore, is required. We instruct the EPA to enter its final order in this matter within six months, after giving Bunker Hill an opportunity to cross-examine its technical experts ánd at least 30 days to comment on the evidence on which EPA intends to rely. If Bunker Hill is dissatisfied with the Administrator’s final action, it shall have ten days to file its objections with this court. An accelerated briefing schedule would then be set.

REMANDED.

On Petition for Rehearing

The Administrator’s petition for rehearing has been carefully considered by this court. We have been mindful during our deliberations of the enactment of the Clean Air Act Amendments of 1977, Pub.Law 95-95, 91 Stat. 685 (1977), subsequent to the filing of our opinion in this case.

We find nothing in the 1977 Amendments or otherwise to alter our position that national ambient air quality standards must be met, to the extent possible by constant emission controls. Whatever doubt there may have been about the correctness of that position was eliminated by the 1977 Amendments. See Section 301, Clean Air Act Amendments of 1977. To the extent a state plan does not require such controls it may be modified by the Administrator to bring it into line with this basic necessity. 42 U.S.C. § 1857c-5(c)(1). We express no opinion with respect to the effect, if any, of section 123 of the Clean Air Act Amendments of 1977 on the operations of the Bunker Hill Company.

In our opinion we indicated that the Administrator could mandate a particular constant emission control when it was technologically and economically feasible. Because we were unable on the basis of the record before us to conclude that the Administrator’s required constant emission control was technologically feasible, rather than “purely theoretical or experimental,” we remanded the case for further consideration of the issue of technological feasibility and did not address Bunker Hill’s objections as to economic feasibility. The Administrator, in his petition for rehearing, urges us to delete economic feasibility as a consideration in determining whether a particular constant emission control is possible. He insists that this deletion is required by the 1977 Amendments. We recognize that this contention is not without force. However, we believe it better at the present time to defer any consideration of this issue. For present purposes it is sufficient to recognize that the’ issue is an open one and that our opinion must be read as modified to the extent such recognition so requires it to be. Our remand, therefore, remains, as before, limited to the technological feasibility of the proposed EPA standards.

The Administrator in his petition for rehearing also contends that our opinion imposes on him the burden of designing and engineering the control technology modification needed by a source to comply with an emission limitation. Moreover, he insists that our opinion suggests that the burden of persuasion is upon him to show that a specific control technology is technologically and economically feasible. In our opinion we do not focus on burden of proof issues. We reviewed the actions of the EPA in rejecting the State of Idaho’s plan in accordance with the usual standards of review employed by courts in reviewing actions by administrative agencies. Such actions must reflect an exercise by the administrator of “reasoned discretion” and not a “crystal ball inquiry.” We held, and on this rehearing do not depart from that holding, that the Browder sulfur burner proposal on the basis of the present record is purely theoretical and experimental. It follows from that holding that the Administrator’s action in rejecting the state plan on the basis of the Browder proposal did not reflect an exercise of “reasoned discretion” and thus was arbitrary and capricious. We said no more than this is our opinion.

The Administrator’s petition for rehearing raises many interesting additional issues regarding the effect of the 1977 Amendments which we decline to pursue. To do so in the present setting of this case would involve our rendering an advisory opinion. This we cannot do.

The Administrator’s petition for rehearing, except as indicated herein, is denied. The sixth-month period within which the EPA is required to enter its final order in this matter shall commence with the date this order is filed.

. The Clean Air Act provides for pollution control through a joint effort of the state and federal governments. Under the Act, it is the responsibility of each state to formulate implementation plans (such as the one under review here) that guarantee the attainment within the state of ambient air quality standards set by the EPA. 42 U.S.C.A. § 1857c-5(a)(2)(A) (Supp.1977). The Administrator is directed by the Act to set two types of ambient air quality standards. “Primary” ambient air quality standards are to be set “the attainment and maintenance of which in the judgment of the Administrator, . . . and allowing an adequate margin of safety, are requisite to protect the public health.” 42 U.S.C.A. § 1857c-4(b)(1) (Supp.1977). Generally, the Administrator will state the primary standards both in terms of a yearly maximum average and in terms of a maximum concentration of pollutant during a limited number of hours that cannot be exceeded more than once per year. The Administrator is also to set “secondary” standards “the attainment and maintenance of which in the judgment of the Administrator . . . [are] requisite to protect the public welfare from any known or anticipated adverse effects associated with the presence of such air pollutants in the ambient air.” 42 U.S.C.A. § 1857c-4(b)(2) (Supp.1977). In most cases, although not all, the secondary standards will be higher and more difficult to attain than the primary standards. This is reflected in the provisions of the Act requiring the primary standards to be met as expeditiously as practicable and in no case later than three years after approval of the state implementation plan, but requiring secondary standards to be met only at some “reasonable time” thereafter. 42 U.S.C.A. § 1857c-5(a)(2)(A) (Supp.1977). In the case of S02, the national primary ambient air quality standards are “80 micrograms per cubic meter (0.03 ppm) — Annual arithmetic mean” (the “primary annual standard”) and “365 micrograms per cubic meter (0.14 ppm)— Maximum 24-hour concentration not to be exceeded more than once per year.” The national secondary standard for S02 is “1,300 micrograms per cubic meter (0.5 ppm) maximum 3-hour concentration not to be exceeded more than once per year.” See 40 C.F.R. §§ 50.4 , 50.5. . There are two basic forms of pollution abatement techniques. The first, “emission controls,” limit the amount of SO2 emitted from the pollution source; thus, an “emission limitation” constricts the amount of pollutant that a source can actually emit to the atmosphere. It is this form of abatement technique that is referred to when one speaks of “capturing” a certain percentage of SO2 emissions. The second set of abatement techniques, “dispersion enhancement techniques,” do not reduce the concentration of pollutants emitted, but instead increase the dispersion of the pollutants through the atmosphere away from high-concentration areas and toward lower-concentration areas. . As discussed infra, polluting sources must use the maximum level of emission control that is economically and technologically feasible in meeting the federal ambient air quality standards. Dispersion enhancement techniques are to assume a secondary role in attaining the standards, to be used only where the maximum emission control still will not guarantee that the standards will be met. See 41 Fed.Reg. 7450; cases cited infra at pp. 1292-1293. Thus, a key issue in this case is what level of emission control is the maximum feasible under current technology. . There are two basic types of dispersion enhancement techniques. “Supplementary control systems” involve staggering the hours of operation of the polluting facilities, with the facilities operating more extensively when meteorological conditions are favorable to dispersion, and sometimes even temporarily closing when meteorological conditions are unfavorable. The other major dispersion technique is tall stacks, which by emitting pollutants at higher altitudes disperse the pollutants over a wider area. . As elaborated more fully in note 1 supra, the “primary annual standard,” set nationally by the Administrator of the EPA, is a limit on the yearly average SO2 concentration in any given region “the attainment and maintenance of which ., allowing an adequate margin of safety, [is] requisite to protect the public health.” Idaho does not provide for a 30-day review period with respect to violations of the other ambient air quality standards set by the Administrator, discussed supra at note 1. . As defined infra at p. 1295, an acid plant operates “autothermally” when the heat necessary to efficiently convert the SO2 into SOs (and ultimately into sulfuric acid) is provided by the transformation process itself. . While Bunker Hill objected in proceedings before the EPA to the proposed change in the averaging period for the 2600 ppm standard from 8 hours to 6 hours, this attack has not been pursued on appeal. . In an effort to better understand the opposing contentions of the parties on the technological feasibility of the EPA regulations, the court requested both sides to submit supplemental briefs detailing the technical issues involved and the portions of the record relevant to the issues’ resolution. These briefs are cited as petitioner’s and respondent’s “First Supplemental Briefs.” . The specific questions posed to the EPA for clarification are listed in note 26 infra. . Cf. Independent Meat Packers Ass’n, supra, at 239 (“the court in conducting the plenary review mandated by Overton Park should limit its inquiry to the administrative record already in existence supplemented, if necessary, by affidavits, depositions, or other proof of an explanatory nature”). . The IDHW reached its conclusion that 72 percent control was the maximum feasible level of control before the Browder Study was prepared. As elaborated at pp. 1291-1292, supra, the Browder Study, which recommends a variety of changes in Bunker Hill’s operations, served as the foundation for the EPA’s finding that 82 percent emission control is feasible. The principal holding and logic of the IDHW is found in the testimony of Mr. Lee Stokes, an administrator of the Environmental Services Division of the IDHW, during a hearing on the implementation plan before the EPA. According to Mr. Stokes, both the federal government and the State “require best available technology. Both of us recognize that Bunker Hill has installed the best equipment available for controlling S02 emissions; neither of us is asking that different equipment be installed at this time. This equipment, tested over the past year, averages 72 per cent control of emissions. Neither we nor EPA has any specific suggestions for improving this operation. Nonetheless EPA is requiring 85 per cent control from this equipment. We feel that it is senseless to require something that, at this time, is impossible.” Joint Appendix 408. The Browder Study has now provided “specific suggestions for improving” Bunker Hill’s operation. However, while never having formally evaluated the Browder Study, the IDHW has indicated no interest in reconsidering its proposed implementation plan in light of the Study; furthermore, Idaho has filed an amicus curiae brief in this case urging this court to hold that the IDHW’s original 72 percent finding is binding on the EPA. See Brief of the State of Idaho. . Cf. Union Electric Co. v. EPA, 427 U.S. 246 , 96 S.Ct. 2518 , 49 L.Ed.2d 474 (1976). While the Court notes that “[pjerhaps the most important forum for consideration of claims of economic and technological infeasibility is before the state agency formulating the implementation plan”, id. at 266 , 96 S.Ct. at 2529, it goes on to hold: “Economic and technological factors may be relevant in determining whether the minimum conditions are met. Thus, the Administrator may consider whether it is economically or technologically possible for the state plan to proceed more rapidly than it does. If he determines that it is, he may reject the plan as not moving ‘as expeditiously as practicable’ in achieving primary standards or as failing to attain secondary standards within a ‘reasonable time.’ ” Id. at 264-65 n.13, 96 S.Ct. at 2529. . The district court in Kennecott Copper Corp. presented neither judicial precedent nor statutory argument for why the Administrator was bound by the state determination. The court merely asseverated “that the State of Nevada was the proper authority to make [a finding of economic infeasibility] and that it is binding on the Administrator of the EPA.” 424 F.Supp. at 1231 . . If, upon remand, the EPA finds the 2600 ppm acid plant limitation to be feasible, the 2000 ppm main lead stack limitation and the 680 tons plant-wide limitation would appear to be valid. Assuming the feasibility of sulfur burners, as discussed below, a concomitant to the feasibility of the 2600 ppm standard, all of the strong stream SO2 emissions from the lead smelter would be fed to the acid plant, leaving only the untreated S02 gas from the sintering machine weak stream, the blast furnace and the lead refinery to be discharged through the main stack. Calculations made by the EPA show that the maximum emissions expected from these sources during normal operation would be only 1812 ppm, well within the 2000 ppm standard. Joint Appendix 835-37, 846-47. The 680 ton limit merely incorporates the other two standards along with a reasonable estimate of the amount of S02 produced by the lead and zinc plants. Joint Appendix 840-47. . The necessary heat can also be partly supplied by the use of external heat sources. Bunker Hill, for example, presently uses a fossil fuel fired preheater to help increase the efficiency of their acid plants. Browder’s suggestions for increasing catalytic efficiency, however, concentrate on improving autother-mality, discussed infra, and improving the efficiency of heat exchange. See J.A. 727-30. See also note 36 infra. . Before submitting this case for our consideration, the parties jointly prepared a condensed version of the administrative record that the parties agree contains all portions of the original record relevant to our determination. This condensed record is referred to throughout this opinion as the Joint Appendix (J.A.). . Both counsel for Bunker Hill and counsel for the EPA initially focused on the feasibility of the 2600 ppm limitation as applied to the lead-smelter acid plant. Both parties recognize that the lead-smelter acid plant presents the greatest problems in meeting a 2600 ppm standard. Zinc roasting is a continuous process that does not yield the same stopping-starting problems as lead smelting. J.A. 795, 989; Affidavit of J. B. Rinckhoff, submitted by petitioner in augmentation of the record, at 12. It is conceivable that, upon remand, the EPA could find that the 2600 ppm limitation is feasible as applied to the zinc-roaster acid plants but not as applied to the lead-smelter acid plant. However, we remand the issue of feasibility to the Administrator with respect to all three of the acid plants. Donovan’s criticisms of the sulfur burner proposal were not directed solely to the lead-smelter acid plant but to all of the acid plants. See J.A. 524, 1131-33. According to another Bunker Hill expert, “the same sulfur furnace control problems are posed in [all three] cases.” Affidavit of J. B. Rinckhoff, supra, at 12. Browder and the EPA failed to adequately address Donovan’s criticisms both with respect to the lead-smelter acid plant and with respect to the zinc-roaster acid plants. This failure left the Administrator’s decision to apply a 2600 ppm standard to each of the acid plants without a rational basis. . The following passages represent the extent of the Browder Study’s discussion of sulfur burners. Nowhere in these passages does Browder attempt to demonstrate the burners’ feasibility. “Several solutions can be used to improve the operation of the gas heat exchangers, and reduce the preheating of the SO2 gas [in the zinc-roaster acid plants]. They are as follows: “(d) Install sulfur burning capability near the converter system which will add increased temperature, and gas strength to the gas flowing to the converter. This being the preferred and least expensive method.” J.A. 734. “Also, the [lead-smelter acid] plant would require some additional source of sulfur values at the preferred location near the present acid plant converter in order to fully utilize the combustion heat of the sulfur (as a fuel), level out the fluctuation of the gas strength, and have an available source of sulfur dioxide when the sintering plant is not in operation. The addition of a sulfur combustion furnace adjacent to the converter would have a payout period of probably less than two years, since a saving could be realized in the long time maintenance and retubing of the heat exchangers. “If the stack gas S02 content is to be substantially reduced for this plant, it can be achieved by different methods. The reduction of sulfur dioxide in the stack from the [lead-smelter acid] plant can be reduced to 1500-2000 ppm of sulfur dioxide by the addition of a converter stage, the addition of heat exchanger surface, and the addition of a sulfur combustion furnace to upgrade, and steady out the sulfur dioxide content being fed to the plant’s converter. These changes and additions are all required to achieve this stack condition.” J.A. 730. “The [lead-smelter acid] plant and cold gas heat exchanger are operating on gas which may not exceed 4.5 percent S02 for extended periods of time. This gas also fluctuates from 3 to 5 percent at times, and at other times, there is no flow at all from the sintering machine . . .. [A] continuous operating sulfur combustion furnace should be installed to keep the acid plant hot and in operation when the sintering machine is shut down. An acid plant which is allowed to cool down, soon deteriorates due to corrosion of the condensed sulfuric acid attacking the plant equipment. “[The lead-smelter acid plant’s] converter . . . will probably only yield 96 percent conversion, provided the gas strength is continuously maintained at about 4.5 percent, and the flow and inlet temperatures are maintained more constant. This can only be accomplished by adding a sulfur combustion furnace, and the related sulfur handling equipment. The addition of the sulfur furnace will alleviate the fluctuating S02 content since additional sulfur values can be added to smooth out the converter feed gas. This will also add heat to the system. The addition of a sulfur combustion furnace and auxiliary equipment will be a less expensive investment than additional cold heat exchanger capacity.” J.A. 742-43. . Two of the three nonferrous-smelter acid plants on the list apparently operate on zinc roasters. As noted earlier, zinc roasters are not likely to have the same magnitude of S02 concentration problems as lead smelters. See note 17 supra. Thus, they would not appear to be comparable to Bunker Hill’s lead-smelter acid plant. While these two acid plants might presumably be comparable to the two Bunker Hill acid plants servicing zinc roasters, this has not been shown in the record before us. Nor do we feel that such comparability can be presumed. According to Donovan, zinc roasters are of two varieties: fluid bed roasters and flash roasters. Flash roasters, which are used at Bunker Hill, have more severe S02 concentration problems than fluid bed roasters. J.A. 493. There is no indication what type of roaster is utilized by the two zinc-roaster plants on Browder’s list. For similar reasons, we do not feel free to presume that the final nonferrous-smelter acid plant on Browder’s list, which operates apparently on a lead smelter, is comparable to Bunker Hill’s situation. No such comparability has been shown in the record before us. For the dangers inherent in assuming the comparability of acid plant operation solely because of similarities in smelters, see the discussion' of the East Helena smelter infra in the text. . According to Donovan, “almost all plants which burn sludge acid, spent acid and/or H2S gas [which represent a majority of the acid plants on the list] typically burn sulfur simultaneously in the same furnace, which is a completely different concept than the scheme proposed by the EPA.” J.A. 1131. The EPA made no attempt in the record to defend its list. . Neither did the EPA respond to Donovan’s claim that “there are obviously some errors [in the list] concerning the plants for which I have knowledge. In some cases, the feedstock is now completely different from that quoted under ‘Type of Process,’ in other cases, the plant has either been shut down, modified or trans- . ferred to other ownership.” J.A. 1130. . This impression is confirmed by augmenting material submitted by Bunker Hill. In particular, according to the engineer responsible for the process design of the East Helena acid plant and sulfur furnace, “the East Helena plans in no way demonstrate the technological feasibility of Mr. Browder’s recommendation and do not support that recommendation.” Affidavit of J. B. Rinckhoff, supra note 17, at 12. See note 27 infra. . The EPA-sponsored study of the East Helena lead smelter, which formed the basis for the EPA regulations with respect to East Helena, implies that sulfur burners are not state-of-the-art, at least with respect to acid plants operating on lead smelters: “Since intermittent operation of the sinter machine cannot be avoided and since the conventional acid plant operates more efficiently and with less maintenance problems on a continuing basis, it appears that some method should be introduced to provide auxiliary S02 source when the sinter machine is not producing. This approach is somewhat beyond the scope of this study and cannot be considered ‘current state of the art’ but should be considered to minimize the intermittent S02 supply problem.” J.A. 958-59. Another portion of the study cites a telephone conversation with a Monsanto representative to the effect that the use of sulfur burners “during sinter machine shutdown and also to provide additional heat, is a technically feasible method of minimizing acid plant shutdown. However, the turndown ratio of a sulfur plant is probably 4 to 1. This means that the sulfur burner must be continued in operation at all times or the sulfur will solidify and startup problems of the sulfur burner will be encountered.” J.A. 1011. On the basis of this evidence, the study concludes that “[¡Ignition and turndown ratio (4:1 maximum in some cases) of the sulfur are possible problem areas unless [the sulfur burner] can be operated at a continuous fixed level.” J.A. 959 (emphasis added). These views are confirmed in the affidavit of Mr. J. B. Rinckhoff, who designed the East Helena acid plant and sulfur furnace. According to Mr. Rinckhoff, “the sulfur burner operation Mr. Browder has proposed is not workable, it is no more than an idea, and very complex technical problems would have to be solved for it to be workable. Accordingly, Mr. Browder’s proposal cannot be regarded as technically feasible.” Affidavit of J. B. Rinckhoff, supra note 17, at 10. . Relevantly, Mr. Donovan, who as noted above contends that Browder’s sulfur burner proposal is infeasible, is in charge of acid plant technology for Monsanto. J.A. 516. . While formally not in the record before us and therefore not entering into our decision, a telegram from Lurgi to Bunker Hill was produced by counsel for petitioner stating that sulfur burners are not a suitable response to the S02 fluctuation problems confronted by Bunker Hill’s lead-smelter acid plant. The letter reads in part: “The installation of a sulphur burner downstream of a sinter machine would increase the S02 level of the gas, but would not necessarily compensate for the S02 fluctuations caused by the conditions indicated above. We are not familiar with a suitable control instrument which would properly regulate an S-burner in conjunction with the S02 level as found in sinter machine off-gas.” Petitioner’s First Supplemental Brief, opposite p. 31. . The EPA was requested to respond to the following questions: “1. In what manner will the proposed sulfur burners be adjusted to accommodate S02 gases of fluctuating strengths emitted from the sin-tering process? “2. If no adjustment of the burners is considered necessary, why? “3. Are the sulfur burners installed, or being installed, at the East Helena Lead Smelter designed to accomplish essentially the same task that the burners recommended by Mr. Browder are to accomplish? “4. If not, in what way do the East Helena plans tend to demonstrate the technological feasibility of Mr. Browder’s recommendation? “5. Are the fluctuating strengths of S02 gases in the East Helena smelter as great as those encountered at Bunker Hill? “6. If not, why are the East Helena plans relevant to the issue of whether the Administrator’s regulations are technologically feasible? “7. Will the feed gases from the zinc roasters pose the same problem for sulfur burners of fluctuating S02 concentration as the feed gas from the lead smelter?” In view of confusion in the original record as to whether the Browder Study recommendations, even if feasible and implemented, would ensure Bunker Hill’s meeting the 2600 ppm standard, see infra pp. 1301-1302, an eighth question was also posed to the Administrator: “8. What is the basis for the assumption that Mr. Browder’s ‘most of the time,’ which to him means 85 percent-plus of the time, should be read in the context of a more or less gaus-sian distribution?” . “The ASARCO-East Helena sulfur furnace will . . . not be automatically controlled. . [I]t is not the current intent to have the sulfur furnace supplement the sintering machine exhaust gas during routine operation of the sintering machine. . . . [Ujnlike Bunker Hill where continuous operation of the sulfur furnace ... is needed to offset design deficiencies in the acid plant, the ASAR-CO-East Helena sulfur furnace will only have to be activated . . . during those episodes when the sinter machine is shut down or undergoing a period of start up or shut down. “In summary, the sulfur furnace at ASARCO is slightly different from the one Mr. Browder is recommending for Bunker Hill in that the ASARCO sulfur furnace will be manually con trolled and will not be used to iron out S02 fluctuations under routine operating conditions.” Respondent’s Second Supplemental Brief, at 13-14. Testimony by the engineer responsible for the process design of the East Helena acid plant and sulfur furnace confirms that East Helena’s plans are inapposite in determining the feasibility of Browder’s recommendations. According to J. B. Rinckhoff, “the sulfur furnace being installed at the East Helena lead smelter is not designed to accomplish the task that the sulfur furnaces recommended by Mr. Browder for the Bunker Hill complex would have to accomplish. The sulfur furnace facilities to be installed in the East Helena plant have been designed to provide a means of continuing acid plant operation when the sinter machine is shut down weekly for maintenance and major equipment outages during periods of very cold weather. “No facilities are being provided for operating the sulfur furnace at any intermediate rates, and the system has nothing whatsoever to do with reducing S02 fluctuations or total S02 emissions. “Accordingly, the answer to the Court’s fourth question is that the East Helena plans in no way demonstrate the technological feasibility of Mr. Browder’s recommendation and do not support that recommendation.” Affidavit of J. B. Rinckhoff, supra note 17, at 12. . According to the most recent submittals of Mr. Browder, the “ASARCO-Tacoma copper smelter” uses sulfur burners “during all phases of fluctuation of SO2 concentration, including frequent times when little or no S02 gas is fed to the acid plant from the metallurgical operations.” Letter from Tim J. Browder to Kenneth A. Lepic, submitted by respondent in augmentation of the record, at 7. Furthermore, according to Browder, although “ASARCO-Taco-ma is a copper smelter, it is a custom smelter, therefore, [sic] experiences substantial S02 fluctuations, the same as Bunker Hill.” Id. However, there is no evidence that ASARCO-Tacoma’s sulfur burner system is designed to even out (let alone succeeds in evening out) the fluctuations in its S02 gas stream. Nor is any evidence presented as to the exact magnitude of the S02 fluctuations other than that they are “substantial.” The sulfur burners are apparently adjusted manually, id., a process of adjustment which may not “provide the quick, highly sensitive response needed to even out fluctuations in the S02 gas stream” like those confronted at Bunker Hill. Affidavit of J. B. Rinckhoff, supra note 17, at 10. According to Mr. Rinckhoff, to his knowledge, “the sulfur furnace idea proposed by Mr. Browder has not been applied in any operation with the widely variable S02 gas strengths that are present at lead smelters, such as Bunker Hill’s and ASARCO’s East Helena facilities.” Id. at 4. . Failure to remove the excess heat from the system would “result in severe damage to the acid plant. . . [The added heat] would not only destroy the absorption efficiency but would soon boil the absorbing acid and cause severe mechanical damage to the system.” Affidavit of J. B. Rinckhoff, supra note 17, at 6 & 8. . An affidavit was also introduced from Mr. Donovan concurring in the criticisms of Mr. Rinckhoff. Affidavit of J. R. Donovan, submitted by petitioner in augmentation of the record, at 2. . According to the EPA, Mr. Rinckhoff s statement that the Browder system would have to meet a 0 to 100 tons-per-day range was founded on a mistaken assumption that the acid plant would not utilize the heat generated from the sulfur burner in maintaining autothermality. According to the Agency, “the capacity of the Browder furnace need only be lh the capacity of a furnace . which does not utilize the heat of sulfur combustion”; thus, the Browder system' need only meet approximately a 0 to 25 tons-per-day range. Response to Petitioner’s Second Supplemental Brief on Technical Issues, at 6-7. However-, it is not clear that Rinckhoff’s statement is based on such an assumption. We read Rinckhoff’s statement of the sulfur furnace’s necessary range as founded on Browder’s plan for the sulfur burners to increase the S02 concentration from a minimum of 0.3 percent S02 (when the sinter machine exhaust SO2 concentration exceeds the autothermal design concentration) to a maximum of 5.5 percent S02 (when the sinter machine is shut down). See Affidavit of J. B. Rinckhoff, supra note 17, at 7-10; Letter from Tim J. Browder, supra note 28, at 5-6 and enclosure 4. It should also be noted that EPA’s response does not meet Rinckhoff’s final objection that Browder’s system as specified “is suitable only for a 10 ton-per-day operation.” Affidavit of J. B. Rinckhoff, supra note 17, at 10. . Bunker Hill also argues that the EPA regulations are economically infeasible. Given that it is impossible on the basis of the record before us to determine what equipment changes ar.e technologically feasible and will be mandated by the EPA on remand, we do not reach the issue of economic infeasibility. However, we do note that a finding of economic feasibility must be made by the Administrator before promulgating the regulations in issue. See Kennecott Copper Corp. v. Train, 526 F.2d 1149, 1156 (9th Cir. 1975), cert. denied, 425 U.S. 935 , 96 S.Ct. 1665 , 48 L.Ed.2d 176 (1976). . As discussed infra note 35, it is not necessary for EPA to show that its standards can be met 100 percent of the time, so long as EPA has provided for the periods when the standards cannot be met by promulgating upset provisions. . In the initial stages of EPA’s consideration of the feasibility of a 2600 ppm standard, the Agency published a list of ten acid plants purportedly meeting a 2000 ppm standard. J.A. 436-37. However, none of the lead smelters on the list actually meet even a 2600 ppm stan dard. While the zinc roasters on the list do meet a 2600 ppm standard, it is not clear that they are relevant analogies for Bunker Hill’s zinc operations. The zinc roasters on the list are fluid bed roasters; Bunker Hill operates flash roasters. And according to Bunker Hill, “a fluid bed roaster is a much more stable process and produces a much higher strength sulfur dioxide gas — and therefore one more amenable to processing in an acid plant — than a flash roaster.” J.A. 493. Bunker Hill further notes that the chart is based on “grab tests” rather than on averaging periods, and that it is on averaging periods that Bunker Hill will be examined. Apparently, the only continuous stack emission tests conducted on any of the listed plants were conducted on the Kennecott plants. Using four-hour averaging periods, the Kennecott plants (which show a maximum grab reading of only 2553 ppm) exceeded 3000 ppm 18 times during a two-month period. EPA’s rebuttal to Bunker Hill’s attack on its list is not convincing and appears to misunderstand the nature of the attack. See J.A. 553, 796. In any case, Mr. Browder, EPA’s expert, would appear to admit that unless all of his proposed modifications, including sulfur burners, are feasible and made, Bunker Hill will not be able to meet EPA’s standards. E. g., J.A. 730. . Of course, it will typically be necessary for the EPA to promulgate upset provisions that excuse the pollution source from liability for technical “violations” of the standards where the “violations” are beyond the source’s control. To hold the source liable for “violations” that even the EPA admits will occur one percent or less of the time would be to require more than available and feasible technology. See FMC Corp. v. Train, 539 F.2d 973 (4th Cir. 1976); Essex Chem. Corp. v. Ruckelshaus, 158 U.S.App.D.C. 360 , 486 F.2d 427 (1973), cert. denied, 416 U.S. 969 , 94 S.Ct. 1991 , 40 L.Ed.2d 558 (1974); Portland Cement Ass'n v. Ruckelshaus, 158 U.S.App.D.C. 308 , 486 F.2d 375 (1973), cert. denied, 417 U.S. 921 , 94 S.Ct. 2628 , 41 L.Ed.2d 226 (1974). . The EPA argues that there is no inconsistency, that the present “low” emissions, which are really the result of Bunker Hill’s bypassing the acid plant during some periods when autother-mality cannot be reached, will indeed be eliminated under Browder’s proposal. Response to Petitioner’s Second Supplemental Brief on Technical Issues, at 10-13. EPA may well be correct. We merely note that the record before us is less than clear on this point and urge the EPA to clarify the point on remand. . In addition to these documents, Bunker Hill claims that it was also denied knowledge of and access to an Agency memorandum discussing the capital costs to Bunker Hill of making the proposed Browder changes (J.A. 568-70), an Agency memo discussing the figures and data in the Technical Support Document (J.A. 571-90), rebuttal comments of an EPA official as to the feasibility of a 2000 ppm standard (J.A. 794-801), a 21-page report on Bunker Hill’s financial condition (J.A. 802-31), and a study of the sulfur market (J.A. 861-913). . EPA originally defended its failure to make the documents available to Bunker Hill on several levels. First, EPA argued that, in an informal rulemaking, an agency need only place relevant documents in a public access file the existence and location of which is repeatedly published; it is not necessary to affirmatively reveal each document to interested parties. See Ethyl Corp. v. EPA, 541 F.2d 1 , 49 n. 102 (D.C. Cir.), cert. denied, 426 U.S. 941 , 96 S.Ct. 2662 , 49 L.Ed.2d 394 (1976). Even assuming that this is true, but see id. at 91-94 (dissenting opinion of Wilkey, J.), EPA did not show that such an alternative procedure was followed here. Indeed, according to Bunker Hill, petitioner was affirmatively led to believe that all the documents upon which EPA intended to rely had been sent to it. Second, EPA noted that Bunker Hill did have an opportunity to examine some of the withheld documents before it filed its Petition for Reconsideration. EPA argued that this opportunity cured any earlier procedural irregularities. However, Portland Cement Ass’n v. Ruckelshaus, 158 U.S.App.D.C. 308 , 486 F.2d 375 (1973), cert. denied, 417 U.S. 921 , 94 S.Ct. 2628 , 41 L.Ed.2d 226 (1974), cited by EPA in support of its position, emphasizes that rule-making proceedings should generally “be conducted in orderly fashion.” Id. at 394 . Material documents and information can in exceptional cases be released subsequent to the issuance of the regulation but only where prior release “is not feasible, as in case of statutory [or judicial] time constraints”. Id. EPA made no such showing of infeasibility here. It is doubtful that an agency will give rebuttal comments the same consideration when they are made in support of a Petition for Reconsideration as when they are made against an initial promulgation of a rule. . Current EPA regulations require, in ali cases, that “before granting a source the right to use [dispersion enhancement techniques], the source must agree to assume liability for violations of the national standards in the vicinity of the source.” No 30-day investigation period is provided for. 41 Fed.Reg. 7452 (1976). Idaho itself dispenses with the 30-day investigation period with respect to all ambient air quality standards other than the primary annual standard. J.A. 378. . See note 32 supra. . Several courts have held that only informal he'árings, without a right of cross-examination, are required by the APA before the promulgation of regulations replacing portions of state implementation plans. See Indiana & Michigan Electric Co. v. EPA, 509 F.2d 839 (7th Cir. 1975); South Terminal Corp. v. EPA, 504 F.2d 646 (1st Cir. 1974); Anaconda Co. v. Ruckelshaus, 482 F.2d 1301 (10th Cir. 1973); Buckeye Power, Inc. v. EPA, 481 F.2d 162 (6th Cir. 1973). However, this “is not an area that may rightly be approached in terms of absolute rigidity of requirement.” Walter Holm & Co. v. Hardin, 145 U.S.App.D.C. 347 , 449 F.2d 1009, 1015 (1971). While cross-examination may not be required as a matter of automatic right before every disapproval or approval of a state implementation plan, particular issues or situations may dictate that cross-examination be allowed. This is such a situation. There are strong rationales, discussed in the text, for granting cross-examination. And the limited number of parties involved on remand insures that cross-examination will not be administratively burdensome. See generally Clagett, Informal Action — Adjudication—Rule Making: Recent Developments in Federal Administrative Law, 1971 Duke L.J. 51 , 67-80. See also Note: The Judicial Role in Defining Procedural Requirements for Agency Rule Making, 87 Harv.L.Rev. 782, 801-02 (1974). . Similar expedited procedures have been utilized in previous cases involving EPA regulations. See, e. g., National Renderers Ass’n v. EPA, 541 F.2d 1281 (8th Cir. 1976); CPC International Inc. v. Train, 540 F.2d 1329 (8th Cir. 1976); CPC International Inc. v. Train, 515 F.2d 1032 (8th Cir. 1975); South Terminal Corp. v. EPA, 504 F.2d 646 (1st Cir. 1974).