Donner Hanna Coke Corp. v. Costle

464 F.Supp. 1295 (1979)

DONNER HANNA COKE CORPORATION, Plaintiff,
v.
Douglas M. COSTLE, Administrator of the United States Environmental Protection Agency, Defendant.

No. Civ.-77-232.

United States District Court, W. D. New York.

February 12, 1979.

Hodgson, Russ, Andrews, Woods & Goodyear, Buffalo, N. Y. (Stephen H. Kelly, and Robert B. Conklin, Buffalo, N. Y., of counsel), for plaintiff.

Richard J. Arcara, U. S. Atty., Buffalo, N. Y. (James A. Fronk, Sp. Asst. U. S. Atty., Buffalo, N. Y., Walter E. Mugdan and Stephen A. Dvorkin, U. S. Environmental Protection Agency, New York City, and Doug G. Farnsworth, U. S. Environmental Protection Agency, Washington, D. C., of counsel), for defendant.

CURTIN, Chief Judge.

In this action, the plaintiff seeks judicial review of an administrative order issued by the Administrator of the Environmental Protection Agency [EPA] directing the plaintiff to permit EPA to inspect the plaintiff's coke oven batteries. The EPA has counterclaimed for enforcement of its order of inspection. The case was tried before the court. What follows is the court's findings of fact and conclusions of law, made in accordance with Rule 52 of the Federal Rules of Civil Procedure.

I. BACKGROUND

Donner Hanna is a New York corporation engaged in the business of operating a by-product coke plant. Its coke oven batteries are located in Buffalo, New York, a short distance inland from Lake Erie. Its three batteries are in line, running west southwest to east northeast. Battery A-B is the westernmost battery, battery # 3 is immediately east of battery A-B, and battery # 4 is immediately east of battery # 3. The batteries are black in color and are exposed to wind, sun, snow, and rain.

Each battery consists of approximately 50 ovens measuring about 17 inches wide, 13 feet high, and 32 feet long. Between each oven are heating ducts which permit the heating of a special mixture of coal in the absence of oxygen to very high temperatures (about 2000°) to produce coke. The operation of each oven is cyclical and is performed in established regular order throughout the battery.

On top of each battery is a larry car, which operates on rails and carries coal from a storage facility to the individual oven. The coal is discharged from the larry car into the oven to be "charged" through lidded openings in the top of the oven. After the individual oven is charged, the lids are replaced and the volatile components of the coal driven off by the heating process are removed by a "standpipe" to a "collector main" to a by-product recovery plant.

When the coking cycle is completed (16-17 hours), the coal in the oven has been transformed into coke. At that time the doors on each end of the oven are opened and a ram-like device is inserted from the "pusher side" in order to push the coke out the other end ("coke side") into a railroad car. The railroad car carries the hot coke under a "quench tower" where the hot coke is drenched with water.

Coke oven batteries do not continuously discharge smoke into the atmosphere but rather emit smoke for short periods of time from a large number of discrete sources. They are therefore classified as intermittent sources of air pollution.

When the damp coal is charged into the hot oven from the larry car, emissions may occur at the charging holes, the larry car hoppers, the larry car control system, and the standpipe lid. This group of emissions is referred to as "charging emissions" and can be reduced or eliminated by carefully controlling the sequence of charging and creating a negative pressure in the oven with an aspiration system. Another group of emissions may occur at the doors, lids and standpipes located at each end of the oven when the volatile components of the coal are removed from the oven after charging. A third group of emissions may occur during the pushing operation and is caused by burning coal which has not been completely converted to coke at the time that it is pushed out of the oven. Finally, if there are defects or leaks in the oven walls, volatile materials may escape into the heating ducts, causing emissions from the waste heat stack. Charging and pushing emissions are typically of very short duration and rarely exceed six minutes.

As a result of a state inspection in 1974 indicating that battery A-B was not in compliance with the three-minute rule, Donner Hanna improved its pushing emission controls on battery A-B. At trial, testimony was introduced to the effect that Donner Hanna now uses state of the art pollution control technology on all of its operations and is in good condition compared to other coke plants in the United States. Barnes at 452, 454-56. This testimony was not challenged by EPA.

In September of 1976, EPA attempted to inspect Donner Hanna's coke oven batteries for the purpose of determining compliance with emission standards contained in New York's State Implementation Plan [SIP], 6 N.Y.C.R.R. § 214.3. Donner Hanna refused to allow the inspection because it disputed the reliability of the testing method which EPA proposed to use. This testing method is the focal point of the controversy between the parties.

Under the Clean Air Act, EPA is authorized to inspect sources of air pollution to determine compliance with the Act. 42 U.S.C. § 7414(a)(2)(A).^[1] It is also authorized to order compliance with its inspection requests. Id. § 7413(a)(3). On October 1, 1976, EPA issued an order pursuant to its statutory powers directing Donner Hanna to allow the proposed inspection. At that time and on all later occasions pertinent to this litigation, EPA has made clear that it intended to conduct the proposed inspection in accordance with the testing procedures set forth in "EPA Visible Emission Inspection Procedures (August 1975)" (1975 EPA Guidelines) (Ex. 2), and that it would use the "stopwatch" technique of measuring the duration of emissions. See EPA's Counterclaim ¶ 19; Ogg at 23-32.

After a conference pursuant to 42 U.S.C. § 7413(a)(4) proved unsuccessful, Donner Hanna filed an action in this court seeking judicial review of the October 1, 1976 order. Donner Hanna v. Costle, Civ. No. 76-567. This action was discontinued without prejudice when EPA withdrew its October 1, 1976 order.

On April 12, 1977, EPA issued a new order substantially restating the provisions of the order of October 1, 1976.

On May 23, 1977, a conference hearing was held in the EPA Region II offices, and representatives of Donner Hanna and EPA attended. At the hearing, EPA refused to modify the order of April 12, 1977. It also advised Donner Hanna that EPA intended to seek criminal sanctions against the plaintiff and its officers and employees in the event of noncompliance with the order.^[2]

Immediately thereafter, Donner Hanna filed this action in the district court, seeking a declaratory judgment as to the constitutionality of the proposed inspection under the fourth amendment and judicial review of the April 12, 1977 order under the Administrative Procedure Act. EPA answered the complaint and also asserted a counterclaim pursuant to 42 U.S.C. § 7413(b) seeking a mandatory injunction directing the plaintiff to grant access to its plant. EPA's motion for summary judgment was denied and the case proceeded to trial.

II. JURISDICTION AND SCOPE OF REVIEW

Jurisdiction over the complaint and the counterclaim is alleged under 28 U.S.C. §§ 1331, 1345, and 2201; 5 U.S.C. § 702; and 42 U.S.C. § 7413(b). Although EPA originally objected to the court's jurisdiction to grant the relief requested by Donner Hanna, this objection was withdrawn in open court in order to have this court rule on the testing method. In return, Donner Hanna withdrew its objection to the inspection based on the fourth amendment. The only issue now before the court is the validity of the April 12, 1977 order seeking inspection of the plaintiff's coke oven batteries.^[3]

Although the parties now agree that the court has jurisdiction to review EPA's order, the court recognizes its obligation to engage in an independent inquiry into subject matter jurisdiction. For the reasons stated below, I find that the district court has jurisdiction over EPA's counterclaim under 42 U.S.C. § 7413(b) and that this jurisdiction includes the power to decide whether EPA's proposed testing method is subject to rulemaking requirements.^[4]

As previously mentioned, the Clean Air Act gives EPA the power to enter onto the premises of persons operating emission sources and to sample the emissions in order to determine compliance with emission standards. Id. § 7414(a)(2). Where permission to inspect is denied, the EPA "may issue an order requiring such person to comply" with the inspection request. Id. § 7413(a)(3). If the person "violates or fails or refuses to comply with any order issued under [§ 7414(a)]," EPA may commence a civil action for a permanent or temporary injunction "in the district court of the United States for the district in which the violation occurred or in which the defendant resides or has his principal place of business ...." Id. § 7413(b). These provisions expressly give the court jurisdiction to enforce EPA's order of April 12, 1977.

Although EPA's right to enter Donner Hanna's plant and its right to use a particular test method are conceptually distinct, under the circumstances of this case the two questions are inextricably connected. Donner Hanna has from the outset been willing to permit EPA to inspect using the "remote" method. EPA states that it seeks permission to inspect for the sole purpose of using its proposed method. EPA's counterclaim, ¶ 19. The government has stipulated to the determination of the proper inspection method by the district court. In its brief filed on June 24, 1977, EPA states:

EPA believes that its right to enter and inspect the Donner-Hanna facility pursuant to § 114 of the Act is entirely independent of the applicability, as a matter of law, of a particular test method. The latter question, the Agency believes, relates strictly to the evidentiary value of any resulting measurements, and like any other evidentiary matter, is therefore not ripe for review until EPA seeks to assert those measurements as evidence against the company. However, raising the issue of ripeness in this connection would seem to be inconsistent with the stipulation between the Agency and Donner-Hanna. Therefore, while the stipulation might not strictly be viewed as governing this issue, as a matter of good faith EPA will not assert a lack of ripeness as a bar to review at this time of the applicability, as a matter of law, of Reference Method 9 as opposed to the company's suggested observation method.

Brief at 26 n. 12.

It would serve no useful purpose to allow the inspection without giving some consideration to EPA's testing method because EPA has made clear its intent to seek criminal sanctions against Donner Hanna if the plant does not permit the inspection on EPA's terms. The manner in which EPA intends to conduct the inspection therefore should be considered in this proceeding.

In connection with the court's jurisdiction, two additional points should be noted. First, one of the defects in EPA's proposed testing method alleged by Donner Hanna is that it was never adopted pursuant to rulemaking procedures, either as part of New York's SIP or as part of EPA's testing methods for new emission sources. See 40 C.F.R. §§ 52.12(c)(1), 60 Appendix A. As a result, the method was never subject to judicial review by the appropriate Court of Appeals as expressly provided in 42 U.S.C. § 7607(b)(1). Thus § 7607(b)(2), which provides that actions reviewable under (b)(1) cannot be reviewed in civil or criminal enforcement proceedings, does not preclude the district court from reviewing the method in enforcement proceedings. West Penn Power Co. v. Train, 522 F.2d 302, 309, 312 (3d Cir. 1975), cert. denied, 426 U.S. 947, 96 S.Ct. 3165, 49 L.Ed.2d 1183 (1976).

Second, several courts have found that preenforcement review of the validity of § 7413 orders is barred by the Clean Air Act. Lloyd A. Fry Roofing Co. v. EPA, 554 F.2d 885 (8th Cir. 1977); West Penn Power Co. v. Train, supra; Getty Oil Co. v. Ruckelshaus, 467 F.2d 349 (3d Cir. 1972), cert. denied, 409 U.S. 1125, 93 S.Ct. 937, 35 L.Ed.2d 256 (1973). But these cases do not preclude judicial review of the counterclaim because the counterclaim was brought under the federal enforcement provisions of the Clean Air Act, 42 U.S.C. § 7413, and can be classified as an enforcement proceeding.

Having found that the court has jurisdiction to review EPA's proposed testing method, the next question is the scope of review. This is governed by the Administrative Procedure Act, 5 U.S.C. § 706. See, e. g., Citizens to Preserve Overton Park v. Volpe, 401 U.S. 402, 413-14, 91 S.Ct. 814, 28 L.Ed.2d 136 (1971); Texas v. EPA, 499 F.2d 289, 296 (5th Cir. 1974), cert. denied, 427 U.S. 905, 96 S.Ct. 3191, 49 L.Ed.2d 1199 (1976). Under § 706, the court must determine whether EPA followed lawful procedures in deciding to use its proposed testing method for inspecting Donner Hanna and whether use of its method would be arbitrary, capricious, an abuse of discretion, or otherwise not in accordance with law.

III. STATUTORY AND REGULATORY BACKGROUND

Under the Clean Air Act, each state must submit to EPA for its approval an SIP providing for the attainment, maintenance and enforcement of national ambient air quality standards. 42 U.S.C. § 7410(a). If a state fails to submit a plan, or if the plan submitted does not meet federal requirements, EPA is authorized to promulgate an implementation plan for the state. Id. § 7410(c)(1).

New York submitted an SIP in accordance with these provisions in 1972, and it was approved by EPA. Both the adoption of the SIP and its approval by EPA followed rulemaking procedures and could have been subjected to judicial review. The portion of the New York SIP relevant to this action provides as follows:

Smoke emissions.

(a) After December 31, 1974, or such later date as determined by an order of the commissioner, no person shall operate a by-product coke oven battery which emits a smoke equal to Ringlemann No. 1 or 20 percent opacity.

(b) Such person who operates a by-product coke oven battery shall be allowed an emission of smoke from the battery of greater than Ringlemann No. 1 or 20 percent opacity if such emission continues for a period or periods aggregating no more than three minutes of any consecutive 60 minute period.

6 N.Y.C.R.R. § 214.3.

This regulation, referred to as the "three-minute rule," does not specify the method to be used in measuring smoke opacity nor does any other part of the SIP.

The three-minute rule is known as an "aggregate opacity" regulation because it allows emissions from the coke oven battery in excess of the regulatory opacity limitation (20% opacity) for a period or periods aggregating no more than three minutes out of any consecutive 60 minutes. By contrast, some opacity standards prohibit any emissions in excess of a certain opacity limitation, for any amount of time. This sort of standard is typically applied to industrial sources which have smoke stacks and which emit smoke fairly steadily and without great variance in the opacity level.

Where the SIP does not specify testing procedures, federal regulations allow EPA to determine compliance by means of "appropriate procedures and methods prescribed in Part 60 of this chapter ..." 40 C.F.R. § 52.12(c)(1). Part 60 sets forth the new source performance standards established by EPA for newly constructed or modified sources of air pollution. Appendix A of Part 60 contains certain "Reference Methods" for determining compliance with the new source performance standards. One of these methods is Method 9, which establishes a procedure by which human observers can determine the opacity of emissions from stationary sources.

Method 9 was originally promulgated in December, 1971. 36 Fed.Reg. 24,895 (Dec. 23, 1971). In response to the court's direction in Portland Cement Association 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), it was revised in November, 1974 in an effort to increase its accuracy. 39 Fed.Reg. 39,873 (Nov. 12, 1974). Revised Method 9 requires the determination of compliance with opacity standards to be based on an average of 24 consecutive readings taken at 15-second intervals. A finding of violation can be made only if the average of 24 observations exceeds the applicable emission standard by 7.5%. The observer must be positioned in such a manner that the sun is oriented in the 140° sector to his back, his line of vision is perpendicular to the plume direction, and no more than one plume is in his line of vision. The observer must have a clear view of the plume and must estimate opacity at its darkest point. If visible water vapor is present with the smoke, opacity cannot be estimated until the water vapor has dissipated.

In addition to setting forth a method for measuring opacity, Method 9 establishes a program for training observers to correctly associate an observed contrast with opacity and for certifying observers who successfully complete the training program. Observers are not trained or certified for the use of the stopwatch technique, discussed below. Ogg at 86.

The 1975 EPA Guidelines (Ex. 2), which EPA seeks to use along with the stopwatch technique to determine compliance with the three-minute rule, are not contained in Part 60 of the regulations and were not promulgated by formal rulemaking procedures. They represent EPA's attempt to "adapt" Method 9 to the testing of various types of stationary sources of air pollution, including coke oven batteries.

The 1975 EPA Guidelines translate the general positioning requirements of Method 9 into specific procedures applicable to coke oven batteries. Thus, the Guidelines suggest that particular contrasting backgrounds be used for viewing the various types of emissions produced by coke oven batteries. They also specify where observers should stand in order to obtain a "clear and unobstructed" view of the various emissions. Ex. 2 at 25-27.

The 1975 EPA Guidelines do not specify whether averaging or direct unaveraged measurements should be used in testing coke oven emissions. According to the testimony of Robert N. Ogg, the chief EPA engineer responsible for stationary source compliance in Region II (New York and the Virgin Islands), inspectors in this region are instructed to use the "stopwatch" technique for timing coke oven emissions. Ogg at 22-25. Rather than averaging opacity readings, the inspector views the emission continuously and records the total time the emission equals or exceeds 20% opacity. Two stopwatches are used: one to record violation times and the other to record the time of day when the violation begins and ends. During a period noted as a violation, only three readings are required: the first is at the time that the emission equals or exceeds 20% opacity, the second is the highest reading during a period of violation, and the third is when the violation ends (i. e., the opacity of the emission falls below 20%). The inspection team, which consists of a minimum of three observers, views the battery for one hour and then totals the emissions exceeding 20% opacity. Overlapping observations are excluded by referring to the exact time of day at which the observations were made. A violation is found if the total exceeds the three minutes allowed in Part 214 of the SIP. Ogg at 23-32.

The stopwatch technique was apparently devised by regional EPA officials in order to address the particular problems of monitoring coke oven compliance with New York's three-minute rule. It is used to time emissions which are too short in duration to use 15-second averaging. As the record stands in this case, the stopwatch technique has not been officially adopted by EPA. None of the inspection manuals or other exhibits introduced into evidence describe the details of the technique. The inspection method which Donner Hanna claims should be used in lieu of EPA's proposed method is referred to as the "remote" or "off-site" method.^[5] It was used by the New York State Department of Environmental Conservation in inspecting Donner Hanna in 1974 after Donner Hanna objected to the use of topside observers. It differs from EPA's proposed method in that observers are not placed on top of the battery to observe charging and topside leak emissions but rather stand on the catwalk approximately 200 feet away from the battery. The remote method apparently uses the stopwatch technique to measure the duration of emissions. The parties agree that it is much easier for Donner Hanna to meet the state emission standard using the remote testing method than using the 1975 EPA Guidelines.^[6]

The issue in this case is whether EPA can test Donner Hanna's compliance with the three-minute rule by using the 1975 EPA Guidelines in conjunction with the stopwatch technique. Donner Hanna maintains that it cannot for a number of reasons. First, it argues that EPA's proposed testing method is not merely an interpretation or adaptation of Method 9 because Method 9 does not apply to intermittent sources, such as coke oven batteries. Second, Donner Hanna maintains that the proposed testing method must be formally adopted in order to comply with the regulatory scheme of the Clean Air Act and with the due process clause of the Constitution. Finally, the plaintiff contends that the proposed testing method is arbitrary and unreliable and that EPA should use the "remote" method in inspecting coke oven batteries.

EPA contends that Method 9 provides a basic framework for determining opacity but does not purport to set forth specific testing procedures for each major type of air pollution source. Coke oven batteries, EPA argues, present various problems which must be solved on a case-by-case basis due to the complexity of the industrial processes involved, the numerous types of emissions encountered, their widely varying durations and other characteristics, and (in the case of the New York SIP) the lack of a promulgated set of procedures for determining compliance with regulatory standards. EPA's position is that it is entitled to develop a method applicable to intermittent emissions and to enforcement of a state SIP containing a time exception, and that it may do so without resort to rulemaking and without formal revision to Method 9. The Agency characterizes its proposed testing method as an adaptation or interpretation of Method 9 and argues that its interpretation is entitled to great deference.

IV. FINDINGS

I have carefully considered the positions of both parties in light of the evidence produced at trial and make the following findings.

First, I find that Revised Method 9, which requires averaging of 24 consecutive readings, is not an appropriate method or procedure for determining compliance with an aggregate-type standard such as New York's three-minute rule. By EPA's own admission, utilization of the averaging technique in Method 9 would preclude the aggregation of the durations of all emissions in excess of 20% opacity observed during a given 60-minute period. This fact is acknowledged in the preamble to revised Method 9:

In developing this regulation we have taken into account the comments received in response to the September 11, 1974 (39 FR 35852) notice of proposed rulemaking which proposed among other things certain minor changes to Reference Method 9. This regulation represents the rulemaking with respect to the revisions to Method 9.

The determination of compliance with applicable opacity standards will be based on an average of 24 consecutive opacity readings taken at 15 second intervals. This approach is a satisfactory means of enforcing opacity standards in cases where the violation is a continuing one and time exceptions are not part of the applicable opacity standard. However, the opacity standards for steam electric generators in 40 CFR 60.42 and fluid catalytic cracking unit catalyst regenerators in 40 CFR 60.102 and numerous opacity standards in State implementation plans specify various time exceptions. Many State and local air pollution control agencies use a different approach in enforcing opacity standards than the six-minute average period specified in this revision to Method 9. EPA recognizes that certain types of opacity violations that are intermittent in nature require a different approach in applying the opacity standards than this revision to Method 9. It is EPA's intent to propose an additional revision to Method 9 specifying an alternative method to enforce opacity standards. It is our intent that this method specify a minimum number of readings that must be taken, such as a minimum of ten readings above the standard in any one hour period prior to citing a violation. EPA is in the process of analyzing available data and determining the error involved in reading opacity in this manner and will propose this revision to Method 9 as soon as this analysis is completed. The Agency solicits comments and recommendations on the need for this additional revision to Method 9 and would welcome any suggestions particularly from air pollution control agencies on how we might make Method 9 more responsive to the needs of these agencies.

39 Fed.Reg. 39,873 (Nov. 12, 1974) (emphasis added). See also Ex. 26 at 7-8. Since the 1974 revision, there has been no additional revision of Method 9 to make it applicable to intermittent sources or to opacity standards with time exceptions. It is therefore clear as EPA admits that Method 9 itself cannot be used to test Donner Hanna's compliance with the three-minute rule.

EPA nevertheless argues that its proposed testing method is merely an interpretation or adaptation of Method 9. In other words, the agency argues that it can adopt the applicable parts of Method 9, reject the inapplicable parts, and substitute the stopwatch technique for six-minute averaging.

For a number of reasons, I find EPA's position untenable. First, it is internally inconsistent. On the one hand, EPA admits that Method 9 cannot be used because the averaging technique cannot be applied to coke oven batteries. On the other hand, at trial it characterizes the rejection of averaging as merely an interpretation of or insignificant deviation from Method 9. The preamble recognizes, however, that the averaging of 24 readings at 15-second intervals is central to its reliability. Since EPA's proposed testing method does not include one of the central features of Method 9, it cannot purport to represent an interpretation or adaptation of Method 9.

Second, a testing method which contains an averaging technique is fundamentally different from a testing method which aggregates unaveraged readings. Under the stopwatch technique, a single reading exceeding 20% opacity would count toward the three minutes allowed under the New York opacity standard, thereby increasing the probability that the source will be found to be in violation of the standard. If, however, readings taken at fifteen-second intervals are averaged over six minutes, a single high reading would not necessarily contribute toward a finding of a violation because it could well be offset or at least reduced by lower readings.

The difference between averaging and straight aggregation is exacerbated where human observers are used to make the readings, because averaging makes it possible to assess the observers' general patterns of accuracy and to reduce the impact of occasional erroneous opacity estimates. Government's Post-Trial Brief at 11. EPA's own witness admitted that comparing the stopwatch technique to six-minute averaging was "mixing apples and oranges." Ogg at 32.

In an attempt to demonstrate that the stopwatch technique and Method 9 were not significantly different in terms of quantifying coke oven emissions, EPA presented at trial an analysis of emission data, which compared observations made at 15-second intervals with observations made in accordance with the stopwatch technique. Ex. 29. Mr. Hopkins, who performed the analysis, testified that there was no significant difference between the two types of readings. Because of several serious flaws in the analysis, however, it cannot be accepted in support of EPA's position. First and most important, the readings taken over 15-second intervals were never averaged as contemplated by Method 9. As a result, the analysis did not compare Method 9 to the stopwatch technique and cannot be used to demonstrate the insignificance of averaging. In addition, as EPA acknowledges, the data base used for the analysis was small, it was gathered for other purposes, and it was not in a form which was useful for purposes of this litigation. Finally, the analysis did not take into account a number of significant variables, such as weather conditions and observer continuity.

A final reason for rejecting EPA's proposed testing method as an adaptation of Method 9 is EPA's failure to justify its adaptation. With the exception of Hopkins' analysis, no studies were introduced supporting the reliability of the stopwatch technique or its relationship to averaging. Although the agency was granted an adjournment in the trial in order to call an expert witness to rebut the testimony of Donner Hanna's expert, Dr. Ensor, it failed to do so. It also failed to call the author of the 1975 EPA Guidelines, Kenneth B. Malmberg.

EPA argues that an agency's judgment in interpreting its own regulations is normally entitled to judicial deference. This is an accurate statement of the law. See, e. g., Train v. NRDC, 421 U.S. 60, 87, 95 S.Ct. 1470, 43 L.Ed.2d 731 (1975); Udall v. Tallman, 380 U.S. 1, 4, 85 S.Ct. 792, 13 L.Ed.2d 616 (1964). But from this it does not follow that EPA is entitled to deference in this case. A well-established corollary to the above principle is that no deference is in order where the agency's interpretation is plainly erroneous or inconsistent with the regulation. See, e. g., Bowles v. Seminole Rock Co., 325 U.S. 410, 413-14, 65 S.Ct. 1215, 89 L.Ed. 1700 (1975).

Here, EPA's position at trial is inconsistent with its published position in the preamble to Method 9. Moreover, the evidence introduced at trial made clear that the test procedures which EPA intended to use were not adopted by EPA at its higher levels but were more akin to on-the-spot determinations made in the field by EPA regional inspectors. See, e. g., Ogg at 22-24; 33-34; Bloom at 38-41. At the same time that the regional inspectors were "adapting" Method 9 to coke ovens, EPA commissioned a study of the application of Method 9 to intermittent sources and to emission standards containing time exceptions. Ex. 5. Performed by Research Triangle Institute and dated July, 1976, the study concluded that there was insufficient data available to assess the accuracy of Method 9 when applied to a standard such as New York's containing a time exception. In order to make this assessment, data must be collected "on single minute opacity averages over one hour periods." Ex. 5 at 19. The data which was available indicated that accumulation of noncontiguous high readings, such as would be necessary in determining compliance with a time exception emission standard, would significantly increase the risk of error. Ex. 5 at 17-18. In light of these circumstances, I find that EPA's interpretation is not entitled to judicial deference and, in the absence of supporting evidence, cannot stand.

The cases cited by EPA in support of judicial deference are readily distinguishable. McLaren v. Fleischer, 256 U.S. 477, 41 S.Ct. 577, 65 L.Ed. 1052 (1921), involved a practical construction of a federal statute which was adopted by the Secretary of the Department of Interior, had been in effect long before the controversy arose, and was later converted into a regulation. Udall v. Tallman, supra, involved the Secretary of Interior's interpretation of an executive order. The interpretation was a matter of public record and had been applied on a number of prior occasions. In Train v. NRDC, supra, the administrative decision was made by the top officials in EPA and was publicized. In none of these cases was the agency's published position plainly inconsistent with the administrative interpretation upheld by the Court.

Having found that EPA's proposed testing method is not an interpretation of Method 9, I now turn to the question of whether rulemaking is required before EPA can use its proposed testing method in determining whether coke oven batteries are in compliance with applicable emission standards. I find that under the Administrative Procedure Act and EPA's own regulations, rulemaking is necessary.

It is undisputed that the method of determining compliance with an emission standard can affect the level of performance required by the standard, even though the standard itself has not changed. See Ex. 12; Portland Cement Association v. Ruckelshaus, supra at 400-01. The performance standard for smoke opacity, which was promulgated by the state and approved by EPA in New York's SIP, cannot be changed without following rulemaking procedures under state law. 42 U.S.C. § 7410(a)(3)(A). Similarly, EPA's regulations, including new source performance standards and reference methods, cannot be changed without following rulemaking procedures. 42 U.S.C. § 7607(d); Detroit Edison Co. v. E.P.A., 496 F.2d 244, 249 (6th Cir. 1974). Enforcement officials cannot circumvent the rulemaking requirements of the Clean Air Act by making substantial changes in testing methods without notice and a hearing. The importance of developing an objective method of testing opacity has been recognized in Portland Cement, and the clear implication of that decision is that opacity tests are subject to rulemaking requirements.

Apart from these considerations, EPA's own regulations require rulemaking. 40 C.F.R. § 52.12(c) instructs EPA in determining compliance with the Act to use the test procedures specified in the SIP or, if none are specified, the test procedures and methods applicable to new source performance standards. Regardless of which test methods are used, both would have been subjected to rulemaking, either at the state or the federal level. Moreover, the preamble to revised Method 9 specifically recognizes that further rulemaking is necessary in order to develop a testing method for intermittent sources or emission standards containing a time exception.

The significance of rulemaking cannot be underemphasized. It gives parties affected by a decision an opportunity to participate in the decision-making process and forces EPA to articulate the bases for its decisions. See Buckeye Power, Inc. v. EPA, 481 F.2d 162, 170-73 (6th Cir. 1973). These procedures tend to produce more objective testing methods. Portland Cement, supra. It also enables aggrieved parties to seek judicial review under the Clean Air Act. 42 U.S.C. § 7607(b); Citizens to Preserve Overton Park v. Volpe, supra, 401 U.S. at 416, 91 S.Ct. 814.

In finding that rulemaking is necessary in order to develop an appropriate testing method for coke oven batteries, I do not intend to suggest that regional representatives of EPA should not be given some leeway in adapting formally promulgated guidelines to local conditions. It must be emphasized that the finding in this case is premised on my conclusion that the agency, in "adapting" Method 9, strayed so far from the original substance and intent of Method 9 that it in effect created a new and different method, not subject to the scrutiny of rulemaking procedures.

Since the remote method has not been subjected to rulemaking proceedings and is not part of New York's SIP, Donner Hanna is not entitled to an order requiring EPA to use the remote method in testing smoke opacity. But as an interim measure pending completion of rulemaking on a testing method for coke ovens, and in order to enable EPA to continue its enforcement efforts, EPA may use the remote method at the Donner Hanna facility.

At trial, Donner Hanna introduced considerable evidence challenging the objectivity of the 1975 EPA Guidelines. It called an expert witness, Dr. Ensor, who conducted a study of Method 9 and the 1975 EPA Guidelines (Ex. 26), to testify at length about his findings. Dr. Ensor concluded that the EPA Guidelines for the application of Method 9 to coke ovens "result in serious compromises which cause significant errors overestimating the opacity and duration of coke oven emissions." Ex. 26 at 8. The primary difficulty involves the positioning of the observers as specified in the Guidelines, which affects the background against which the plume is observed and consequently the accuracy of the estimate. An additional source of unreliability concerns the 15% margin of error allowed in the observer training school and its effect on the reported duration of emissions observed continuously rather than at fifteen-second intervals. Although the trial was adjourned for several weeks in order to give EPA an opportunity to respond to Dr. Ensor's report, EPA decided not to attempt to rebut the testimony or report.

Although Donner Hanna's evidence on this point was persuasive, I find that it would be inappropriate at this stage to make a determination as to the validity of the 1975 EPA Guidelines since this question is the proper subject of rulemaking proceedings and is reviewable only by the Court of Appeals. 42 U.S.C. § 7607(b)(1).

Donner Hanna is directed to prepare a proposed judgment on notice to defendant.

So ordered.

NOTES

[1] The Clean Air Act was amended in 1977 and the provisions were renumbered. Although the parties' briefs refer to the old section numbers, references throughout this decision shall be to the new numbers.

[2] This fact is alleged in the Complaint (¶ 29) and admitted in EPA's Answer (¶ 9).

[3] See order of September 21, 1977, denying EPA's motion for summary judgment.

[4] As discussed in greater detail infra, EPA's proposed testing method had what amounts to two separate components. First, the 1975 EPA Guidelines specified observer positioning and background requirements. Second, the stopwatch technique, rather than an averaging technique, was used to time the duration of pushing and charging emissions. Unless otherwise indicated, references in this opinion to EPA's proposed testing method are intended to encompass both components.

[5] At trial, the parties disagreed about whether the "remote" method had been endorsed by the state and could be referred to as the "state" method. Since they agreed that it had not been promulgated by the state in accordance with rulemaking procedures, see discussion infra, I find it unnecessary to make a finding of fact on this issue.

[6] The State conducted a second inspection of Donner Hanna in 1977 using both the remote method and EPA's proposed method. Exhibit 12 summarizes the observations and compares the results. Over six 60-minute periods, the remote method resulted in one finding of violation and the EPA method in five.