On this appeal we review orders of the Nuclear Regulatory Commission granting licenses for low power and full power operations at the Diablo Canyon Nuclear Power Plant in San Luis Obispo County, California.1 Petitioners are groups and individuals who intervened in the licensing proceedings before the Commission; petitioners and their members live and work in the vicinity of the Diablo Canyon plant. Petitioners contend that, in granting licenses to Pacific Gas and Electric Company ("PG & E") for operation of the facility, the Commission violated specific legal requirements of the National Environmental Policy Act,2 the Atomic Energy Act,3 the Administrative Procedure Act4 and numerous regulations promulgated pursuant to those statutes. Respondent Commission and intervenor PG & E characterize petitioners' contentions as fundamental disagreements with the Commission's factual findings. They defend those findings as lying well within the range of agency discretion afforded the Commission under applicable statutes and regulations, and they urge this court to defer to the Commission's scientific expertise in assessing the legality of the licensing actions challenged herein.

After an exhaustive examination of the record in these prolonged and complex proceedings, we conclude that the Commission acted within the parameters of legal discretion in all but two minor respects. The Commission committed technical errors in licensing reactor operators who had been trained on computer simulators, and in denying petitioners a hearing on issues of construction quality assurance when it twice extended the term of Diablo Canyon's license. The Commission has corrected its first error by amending its operator license requirements to recognize experience gained on computer simulators, and there exists no impediment to relicensing the Diablo Canyon operators under the new provision. The second error remains uncorrected, but we conclude that it would serve no discernible purpose to allow petitioners now to introduce the evidence they would have produced had they been properly accorded a hearing. Because that evidence was not material or safety-significant, we conclude that its exclusion calls into question neither the Commission's decision to license Diablo Canyon nor the safe operation of the plant. Under these circumstances we believe a remand would be an empty gesture, and one with which we decline to burden the Commission. We therefore affirm the Commission's decision to allow issuance of low power and full power licenses for the Diablo Canyon Nuclear Power Plant.

While our analysis necessarily critically examines individual decisions of the Commission and its boards, we recognize and respect the great deference due such expert determinations. The Atomic Energy Act of 1954 created a regulatory scheme which is "virtually unique in the degree to which broad responsibility is reposed in the administrative agency, free of close prescription in its charter as to how it shall proceed in achieving the statutory objectives."5 As the Supreme Court has recently reminded us, "the Commission is making predictions, within its area of special expertise, at the frontiers of science. When examining this kind of scientific determination, as opposed to simple findings of fact, a reviewing court must generally be at its most deferential."6

Just as we must not usurp the scientific role properly accorded the Commission by the Atomic Energy Act and other governing statutes, neither are we permitted to question this nation's commitment to the development of nuclear power as a significant power source. On several occasions--often in the course of reversing decisions of this court--the Supreme Court has explicitly recognized Congress's commitment to the careful nurturing of the nuclear power industry. In Vermont Yankee Nuclear Power Corp. v. Natural Resources Defense Council7 a unanimous Supreme Court described the limited mandate under which courts act in deciding questions of nuclear regulation:

Nuclear energy may someday be a cheap, safe source of power or it may not. But Congress has made a choice to at least try nuclear energy, establishing a reasonable review process in which courts are to play only a limited role. The fundamental policy questions appropriately resolved in Congress and in the state legislatures are not subject to reexamination in the federal courts under the guise of judicial review of agency action. Time may prove wrong the decision to develop nuclear energy, but it is Congress or the States within their appropriate agencies which must eventually make that judgment. In the meantime courts should perform their appointed function. [The National Environmental Policy Act] does set forth significant substantive goals for the Nation, but its mandate to the agencies is essentially procedural. It is to insure a fully informed and well-considered decision, not necessarily a decision the judges of the Court of Appeals or of this Court would have reached had they been members of the decisionmaking unit of the agency. Administrative decisions should be set aside in this context, as in every other, only for substantial procedural or substantive reasons as mandated by statute, not simply because the court is unhappy with the result reached.8

On remand from the Vermont Yankee decision this court held that a "zero-release" assumption adopted by the Commission to describe the environmental impacts of radioactive waste storage violated both the Administrative Procedure Act ("APA") and the National Environmental Policy Act ("NEPA").9 Because the Commission had not factored into the licensing process uncertainties underlying its assumption, this court concluded that the assumption was arbitrary and capricious and a violation of NEPA.

The Supreme Court reversed, holding that the Commission had not acted arbitrarily and capriciously in assigning a zero value to the environmental impact of radioactive waste storage.10 The Court again emphasized the limited nature of the judicial role in reviewing such Commission determinations:

We are acutely aware that the extent to which this Nation should rely on nuclear power as a source of energy is an important and sensitive issue. Much of the debate focuses on whether development of nuclear generation facilities should proceed in the face of uncertainties about their long-term effects on the environment. Resolution of these fundamental policy questions, lies, however, with Congress and the agencies to which Congress has delegated authority, as well as with state legislatures and, ultimately, the populace as a whole. Congress has assigned the courts only the limited, albeit important, task of reviewing agency action to determine whether the agency conformed with controlling statutes.11

In People Against Nuclear Energy v. Nuclear Regulatory Commission12 this court held that the Commission violated NEPA when it improperly failed to consider whether the risk of a nuclear accident at the Three Mile Island plant in Pennsylvania might cause harm to the psychological health and community well-being of nearby residents. Again the Supreme Court reversed.13 Emphasizing that "Congress [did not] intend[ ] to extend NEPA as far as the Court of Appeals has taken it,"14 the Court concluded that the 1979 accident at Three Mile Island did not

transform[ ] PANE's contentions into "environmental effects." ... NEPA is not directed at the effects of past accidents and does not create a remedial scheme for past federal actions. It was enacted to require agencies to assess the future effects of future actions. There is nothing in the language or the history of NEPA to suggest that its scope should be expanded "in the wake of" any kind of accident.15

The Supreme Court has recently detailed Congress's historic commitment to development of a viable nuclear power industry. In Pacific Gas & Electric Co. v. State Energy Resources Conservation & Development Commission16 the Court described Congress's intent in enacting the Atomic Energy Act, the Price-Anderson Act and the Energy Reorganization Act of 1974:

There is little doubt that a primary purpose of the Atomic Energy Act was, and continues to be, the promotion of nuclear power. The Act itself states that it is a program "to encourage widespread participation in the development and utilization of atomic energy for peaceful purposes to the maximum extent consistent with the common defense and security and with the health and safety of the public." The House and Senate Reports confirmed that it was "a major policy goal of the United States" that the involvement of private industry would "speed the further development of the peaceful uses of atomic energy." The same purpose is manifest in the passage of the Price-Anderson Act, which limits private liability from a nuclear accident. The Act was passed "in order to protect the public and to encourage the development of the atomic energy industry ...."

* * *

The evident desire of Congress [in enacting the Energy Reorganization Act of 1974] to prevent safety from being compromised by promotional concerns does not translate into an abandonment of the objective of promoting nuclear power. The legislation was carefully drafted, in fact, to avoid any anti-nuclear sentiment. The continuing commitment to nuclear power is reflected in the extension of the Price-Anderson Act's coverage until 1987, as well as in Congress' express preclusion of reliance on natural gas and petroleum as primary energy sources in new power plants, [in the] Powerplant and Industrial Fuel [Use] Act of 1978. It is true, of course, that Congress has sought to simultaneously promote the development of alternative energy sources, but we do not view these steps as an indication that Congress has retreated from its oft-expressed commitment to further development of nuclear power for electricity generation.17

The foregoing passages clearly establish the limited nature of the role courts are to play in reviewing orders of the Nuclear Regulatory Commission. We are to determine whether the Commission's decisions are "arbitrary, capricious, an abuse of discretion, or otherwise not in accordance with law"--not to seize upon the fortuity of appeal to usurp those governmental functions properly accorded the political branches. Mindful of these unequivocal proscriptions against judicial overreaching, we turn to an examination of the orders issued in the present case and the legal deficiencies which allegedly taint them.

Few forms of administrative adjudication are as prolonged and complex as the licensing proceedings for a nuclear power plant, and few nuclear licensing proceedings have produced a record as complicated and massive as that concerning the facility at Diablo Canyon. The parties have documented numerous significant actions taken by the Commission and its boards over the past fifteen years. The great majority of these decisions have little relevance to the issues raised on the present appeal; their chronological recitation would do more to obfuscate the issues with which we must deal than to clarify them. Accordingly, we set forth below only a capsule history of the Diablo Canyon proceedings; those facts bearing on specific issues are set forth in the appropriate sections of our opinion.

The Atomic Energy Commission ("AEC"), predecessor to the Nuclear Regulatory Commission,18 issued construction permits to PG & E for Units 1 and 2 of the pressurized water reactor plant at Diablo Canyon in 1968 and 1970.19 Construction began shortly thereafter, based on the assumption that the nearest significant earthquake fault was a safe distance away.20 Four years later, however, offshore exploration for petroleum revealed the presence of the Hosgri Fault within three miles of the Diablo Canyon site.21 Petitioners, who had intervened in the administrative proceeding, requested that construction at the facility be halted until the implications of the discovery could be assessed.22 The AEC chose instead to authorize the continuance of construction,23 but it ordered an extensive reexamination of the plant's seismic design.24 On 16 June 1981 the Commission's Appeal Board approved the plant's seismic design.25

In June 1980 PG & E applied for a license to load fuel and conduct low-power testing.26 On 21 September 1981 the Commission rejected claims of deficiencies in the quality assurance program at Diablo Canyon; it therefore issued the requested license for Unit 1.27 Less than a week later, on the very eve of fuel loading, it was discovered that blueprints had mistakenly been reversed in the design of the facility's reactor.28 Further investigations by PG & E and the Commission staff uncovered numerous other design errors,29 leading to the Commission's decision on 19 November 1981 to suspend PG & E's fuel loading and low power test license.30 To ensure that a functioning Diablo Canyon would be adequately protected against seismic disturbances, the Commission ordered PG & E, as a condition to reinstatement of the license, to institute an Independent Design Verification Program ("IDVP").31 At the same time, the Commission staff imposed conditions on PG & E's eligibility for a full power license: the additional requirements concerned seismic and other design verification issues.32

The ensuing reanalysis and modification of the plant's design consumed more than 2,000,000 person-hours of professional effort before its completion in October 1983.33 Upon independent review of the IDVP by the Commission staff,34 the Commission progressively reinstated elements of the suspended low power license on 8 November 1983,35 25 January 198436 and 13 April 1984.37 Reinstatement of the license was consistent with the findings of the Appeal Board on 20 March 1984 that "[t]he applicant's verification efforts provide adequate confidence that the Unit 1 safety-related structures, systems and components are designed to perform satisfactorily in service and that any significant design deficiencies in that facility resulting from defects in the applicant's design quality assurance program have been remedied."38

On 10 August 1984 the Commission approved issuance of a license for full power operations at Diablo Canyon;39 before the license could issue, however, petitioners obtained a stay of the Commission's order from this court.40 We heard oral argument on an expedited basis on appeals from both the low power and the full power orders on 30 October. On 31 October we lifted our stay,41 thereby permitting issuance of the full power license and the commencement of operations at Diablo Canyon.42

A. Environmental Effects of a Core Meltdown Accident

The National Environmental Policy Act43 requires that agencies prepare an Environmental Impact Statement ("EIS") detailing the possible environmental impacts of any proposed agency action that may significantly affect the environment.44 Only after an EIS has been prepared may an agency proceed with the action it proposes to take. The agency's obligations under NEPA do not cease once the EIS has been prepared, however: both federal case law45 and regulations46 recognize a continuing duty to supplement EISs which have already become final whenever the discovery of significant new information renders the original EIS inadequate.

An EIS was prepared for the Diablo Canyon plant in 1973; it was supplemented in 1976. In neither the final nor the supplemental EIS was any significant reference made to the possible environmental consequences of a "core melt" accident. Petitioners contend that the failure to discuss such accidents constitutes a violation of NEPA.

At the time the final and supplemental EISs for Diablo Canyon were prepared, the NRC employed a nine-part classification scheme for describing accidents that can occur at nuclear power plants.47 Class One accidents were considered so trivial as not to warrant inclusion in the Commission's Environmental Impact Statements.48 Accidents of Classes Two through Eight were progressively more severe; with limited exceptions, discussion of their potential environmental impacts was required.49 Class Nine accidents were considered the most severe.50 The Class Nine designation represented a catch-all, or residual, category embracing those accidents not accorded lesser status. The most prominent type of Class Nine accident was the "core melt" accident, one in which the core of a plant's nuclear reactor melts and radiation is released into the environment. The consequences of such an accident, should one occur, were conceded to be of potentially catastrophic proportions. At the same time, the probability that such an accident would occur was regarded as so low as to be scientifically and legally insignificant.51 The Commission therefore maintained a policy of not requiring detailed discussion of Class Nine accidents in its Environmental Impact Statements. As the Commission explained,

[d]efense in depth (multiple physical barriers), quality assurance of design, manufacture, and operation, continued surveillance and testing, and conservative design are all applied to provide and maintain the required high degree of assurance that potential accidents in this class are, and will remain, sufficiently remote in probability that the environmental risk is extremely low. For these reasons, it is not necessary to discuss such events in applicants' Environmental Reports.52

This court upheld the Commission's policy regarding Class Nine accidents in Carolina Environmental Study Group v. United States.53 "Viewing the record as a whole," the court stated,

we cannot say that the A.E.C.'s general consideration of the probabilities and severity of a Class 9 accident amounts to a failure to provide the required detailed statement of its environmental impact. That the probability of a Class 9 accident is remote and that its consequences would be catastrophic are undisputed. Neither the A.E.C.'s finding of low probability, nor its methodology or basis for that finding, are challenged here by appellant.

Because each statement on the environmental impact of a proposed action involves educated predictions rather than certainties, it is entirely proper, and necessary, to consider the probabilities as well as the consequences of certain occurrences in ascertaining their environmental impact. There is a point at which the probability of an occurrence may be so low as to render it almost totally unworthy of consideration.... We find nothing in the instant record which would indicate that the A.E.C. findings regarding Class 9 accidents are clearly erroneous or that the A.E.C.'s compliance with NEPA Section 102(2)(C)(i) in this case was inadequate.54

Pursuant to the Commission's policy, neither the final nor the supplemental EIS for Diablo Canyon discussed in detail the environmental impacts of a Class Nine accident.

Two events in the late 1970's led the Commission to consider revising its policy. In July 1977 the NRC commissioned a Risk Assessment Review Group to assess the "achievements and limitations" of a 1975 report which had concluded that the probability of a Class Nine accident was very low.55 In September 1978 the Risk Assessment Review Group issued its findings; it concluded that it was "unable to determine whether the absolute probabilities of accident sequences in [the 1975 study] are high or low, but believes that the error bounds on those estimates are in general, greatly understated."56

The second event which prompted the Commission to reconsider its original policy was the 1979 accident at the Three Mile Island ("TMI") nuclear power plant in Pennsylvania.57 While technically falling within the Commission's Class Nine category, the TMI accident entailed no breach of the reactor containment vessel and no substantial release of radiation into the atmosphere.58 Despite its negligible environmental consequences, however, the accident at TMI was considerably more severe than any previous accident in the history of the American nuclear power industry.

In light of these two developments, the Commission took the precautionary step of reassessing its assumptions regarding the probability of Class Nine accidents. On 13 June 1980 it published a Statement of Interim Policy,59 directing that future EISs include a discussion of "the site-specific environmental impacts attributable to accident sequences that lead to releases of radiation and/or radioactive materials, including sequences that can result in inadequate cooling of reactor fuel and to melting of the reactor core."60 In departing from its practice of not requiring discussion of Class Nine accidents, the Commission noted the need for further study of the matter; it was not prepared to say, however, that the probability of such accidents was scientifically significant. Indeed, in the first sentence of its Statement the Commission referred to "the more severe kinds of very low probability accidents."61 Because the Commission had not yet concluded that its earlier assumption was incorrect, it reaffirmed its faith in final EISs done under the old policy, holding that they need not be revised to conform to the new requirement:

It is expected that these revised treatments will lead to conclusions regarding the environmental risks of accidents similar to those that would be reached by a continuation of current practices, particularly for cases involving special circumstances where Class 9 risks have been considered by the staff, as described above. Thus, this change in policy is not to be construed as any lack of confidence in conclusions regarding the environmental risks of accidents expressed in any previously issued Statements, nor, absent a showing of similar special circumstances, as a basis for opening, reopening, or expanding any previous or ongoing proceeding.62

Only for cases in which "special circumstances" existed was the Statement of Interim Policy to apply retroactively to EISs that had already become final.

At issue on the present appeal is the validity of the Commission's decision to exempt EISs that had already become final from the requirements of the Statement of Interim Policy, and its conclusion that the Diablo Canyon case does not involve "special circumstances" warranting special retroactive application.

Petitioners contend that the Commission's decision to deny the Statement of Interim Policy retroactive effect (except when special circumstances exist) violates the Commission's duty under NEPA to supplement final EISs. We believe that petitioners fundamentally misconceive the nature of that duty. As a number of courts have held, Environmental Impact Statements need not address "remote and highly speculative consequences."63 Under this well-established "rule of reason,"64 agencies need not discuss in detail events whose probabilities they believe to be inconsequentially small.

It would be anomolous to apply a rule of reason to the preparation of final EISs but require that supplemental EISs address remote and highly improbable consequences. This would

lead to the absurd result of a continuing agency requirement to supplement its EIS with consideration of effects that were not significant enough to require preparation of an EIS in the first place. Nothing in NEPA ... supports such a scheme, whereby a supplemental EIS is more easily triggered than an original EIS ....65

As our case law makes clear, this probabilistic rule of reason governs judicial review of both the preparation of an EIS and its supplementation.66 We conclude, therefore, that the Commission was under no obligation to supplement the Diablo Canyon EIS with a discussion of Class Nine accidents if the Commission reasonably believed that such accidents were highly unlikely to occur.

As we have discussed above, the Commission did not conclude in its Statement of Interim Policy that its original assumption regarding Class Nine accidents was scientifically incorrect. Rather, it recognized the need for renewed study of the issue.67 The clear import of the Commission's Statement is that, until such time as its research yields a contrary result, the Commission continues to regard Class Nine accidents as highly improbable events.

We do not consider that conclusion unreasonable. Neither the 1978 study by the Risk Assessment Review Group nor the accident at Three Mile Island established that the probability of a Class Nine accident with significant environmental consequences is anything but very small. The Risk Assessment Review Group concluded that the error bounds of an earlier study were greatly understated.68 As to the probability issue itself, however, the Group was unable to determine whether the estimates of the earlier study were too low or too high.69 Similarly, while the accident at Three Mile Island technically fell within the Commission's Class Nine category, it resulted in no significant release of radiation into the atmosphere.70 Because the environmental consequences of Three Mile Island were scientifically and legally inconsequential, the fact that the accident occurred does not establish that accidents with significant environmental impacts will have significant probabilities of occurrence.

NEPA, therefore, does not require the consideration of Class Nine accidents in future EISs, nor does it require that final EISs be supplemented to take account of the Class Nine risk. The approach adopted in the Statement of Interim Policy--to include discussion of such accidents in future EISs--was a discretionary policy choice of the Commission. Because it need not have imposed upon itself the burden it did, the Commission was perfectly free to deny its new policy retroactive effect. We conclude that the Commission did not violate its obligations under NEPA by declining to supplement the Diablo Canyon EIS with a discussion of the environmental impacts of a Class Nine accident.

Petitioners also contend that the Statement of Interim Policy is invalid under the "worst case analysis" regulation promulgated by the Council on Environmental Quality pursuant to NEPA.71 The Council was given authority, by Executive Order of President Carter in 1978, to direct the manner in which federal agencies comply with NEPA.72 Not only are the CEQ regulations binding on those federal agencies which they affect,73 they are also entitled to substantial deference by the courts.74

The worst-case-analysis regulation sets forth a procedure for dealing with scientific uncertainty in the preparation of Environmental Impact Statements. It provides that:

When an agency is evaluating significant adverse effects on the human environment in an environmental impact statement and there are gaps in relevant information or scientific uncertainty, the agency shall always make clear that such information is lacking or that uncertainty exists.

(a) If the information relevant to adverse impacts is essential to a reasoned choice among alternatives and is not known and the overall costs of obtaining it are not exorbitant, the agency shall include the information in the environmental impact statement.

(b) If (1) the information relevant to adverse impacts is essential to a reasoned choice among alternatives and is not known and the overall costs of obtaining it are exorbitant or (2) the information relevant to adverse impacts is important in the decision and the means to obtain it are not known (e.g., the means for obtaining it are beyond the state of the art) the agency shall weigh the need for the action against the risk and severity of possible adverse impacts were the action to proceed in the face of uncertainty. If the agency proceeds, it shall include a worst case analysis and an indication of the probability or improbability of its occurrence.75

Because the regulation is specifically designed to cope with the problem of scientific uncertainty, its provisions are triggered only when such uncertainty exists. No worst case analysis is required if an agency has carefully studied the potential environmental impacts of a proposed action and has determined, with a reasonable degree of certainty, the probability and consequences of such impacts.76 Where uncertainty exists as to either probability or consequences, however, a worst case analysis is mandatory if the agency decides to proceed with its proposed action. Most notably (and most controversially), the regulation requires such an analysis even where the probability of an impact is assumed to be very small.77

We do not believe, however, that the CEQ regulation governs the present case. The final Environmental Impact Statement for the Diablo Canyon facility was filed in 1973 and supplemented in 1976. The relevant CEQ regulations became effective on 30 July 1979;78 they provide that:

These regulations shall apply to the fullest extent practicable to ongoing activities and environmental documents begun before the effective date. These regulations do not apply to an environmental impact statement or supplement if the draft statement was filed before the effective date of these regulations. No completed environmental documents need be redone by reasons of these regulations.79

Although the Statement of Interim Policy was issued after the CEQ regulations became effective, the final and supplemental EIS for Diablo Canyon was filed significantly before that date. By its own terms the CEQ regulation does not require the reworking of existing EISs. Nor do we find warrant in the language of the rule for applying its provisions to the duty to supplement final EISs. Were we to require such "worst case" supplementation, we would effectively circumvent the principle of finality underlying the quoted provision. We decline to construe the duty to supplement final EISs in a manner which would contravene the explicit language of the regulations.

3. "Special Circumstances"

Petitioners contend that the EIS for Diablo Canyon is deficient in one final respect. In its Statement of Interim Policy, the NRC recognized an exceptional class of cases--those involving "special circumstances"--for which the Statement's provisions would be given retroactive effect. Petitioners contend that the presence of the Hosgri fault within three miles of the Diablo Canyon site constitutes just such special circumstances as was envisioned by the statement.80

The special circumstances exception had precedent in the Commission's pre-Statement practice. Although the Commission had generally not required consideration of the environmental impacts of core meltdown accidents prior to adoption of the Statement, in certain exceptional cases the Commission had included a discussion of such accidents as a matter of discretionary policy.81 The Commission's Statement cites three such exceptional cases: the Clinch River Breeder Reactor Plant, "a liquid metal cooled fast breeder reactor very different from the more conventional light water reactor plants for which the safety experience base is much broader";82 the early site review for the Perryman facility, a site in an area of high population density;83 and the application of Offshore Power Systems to manufacture floating nuclear power plants, a proposal which entailed "potentially serious consequences associated with water (liquid) pathways leading to radiological exposures if a molten reactor core were to fall into the water."84

The Commission's decision in the Black Fox case,85 issued only three months prior to the Statement of Interim Policy, provides further insight into the meaning of "special circumstances." In Black Fox the Commission identified "exceptional case[s] that might warrant additional consideration: higher population density, proximity to man-made or natural hazard, unusual site configuration, unusual design features, etc."86 Such cases were those in which "the environmental risk from such an accident, if one occurred, would be substantially greater than that for an average plant."87

The parties initially disagree as to whether the special circumstances exception speaks only to cases in which the consequences of an accident, if one occurred, would be unusually severe, or whether the exception speaks to exacerbated probabilities as well. Despite the silence of the Statement of Interim Policy and the ambiguous language of the Black Fox opinion, we believe that petitioners are correct in espousing the latter view. Both the probability and the magnitude of a nuclear accident contribute to its risk; unusual factors that exacerbate that risk demand special consideration, regardless of whether they make an accident more likely or more catastrophic.

It is clear, however, that such special circumstances do not exist in the present case. Petitioners rely solely on the fact that Diablo Canyon is located near the Hosgri Fault; they contend that the concomitant earthquake risk greatly enhances both the probability and the expected magnitude of a nuclear accident. There are three possible ways in which petitioners' argument might be said to have merit, but we believe that none is applicable to the Diablo Canyon plant. First, the existence of the Hosgri Fault might make a nuclear accident relatively more likely. After extensive studies and retrofitting of the plant, however, the Commission has determined that Diablo Canyon's seismic design is more than adequate to guard against such a horrific possibility;88 significantly, petitioners have not challenged this conclusion on appeal. We must assume, therefore, that the likelihood that an earthquake will trigger a nuclear accident at the facility is so small as to be rated zero.89 Second, "special circumstances" might exist if there were a real danger that a core meltdown and an earthquake would occur--by coincidence--simultaneously. As we discuss in the following section, the Commission has determined that the chance of such a bizarre concatenation of events occurring is extremely small.90 Not only is this conclusion well supported by the record evidence, it accords most eminently with common sense notions of statistical probability. The third manner in which "special circumstances" might be said to exist is if a core meltdown were capable of triggering an earthquake. At oral argument petitioners disclaimed any reliance on this theory, and wisely so. We have been apprised of no studies suggesting that seismic disturbances may be precipitated by a nuclear accident, and we seriously doubt that any exist.

We conclude, therefore, that the Commission did not violate NEPA by declining to discuss the possible environmental impacts of a Class Nine accident in the EIS for Diablo Canyon.B. Earthquakes and Emergency Planning91

NRC regulations require, as a prerequisite to issuance of an operating license for a nuclear power reactor, that "the state of onsite emergency preparedness provide[ ] reasonable assurance that adequate protective measures can and will be taken in the event of a radiological emergency."92 By virtue of this provision, enacted in the aftermath of the accident at Three Mile Island, emergency planning is made an integral element of the licensing decision.

In its 1981 San Onofre decision,93 the Commission held that its regulations

do not require consideration of the impacts on emergency planning of earthquakes which cause or occur during an accidental radiological release. Whether or not emergency planning requirements should be amended to include these considerations is a question to be addressed on a generic, as opposed to case-by-case, basis.94

Relying on the San Onofre decision, the Commission's Appeal Board declined to require consideration of the potential complicating effects of an earthquake on emergency planning at the Diablo Canyon facility.95 In December 1983 the Commission itself summarily declined review of the Appeal Board's decision.96

Further study of the matter by the NRC staff,97 however, prompted a Commission promise to address whether to reverse its former policy and allow consideration of the effects of seismic events on emergency planning "under the circumstances" of the Diablo Canyon case.98 Comments were submitted by the NRC staff, petitioners and PG & E in response to specific questions posed by the Commission.99 After considering the responses, the Commission issued a decision on 10 August 1984 reaffirming its holding in San Onofre.100 While it promised to initiate a rulemaking "shortly" to resolve the earthquake issue on a generic basis,101 "the issuance of a full-power operating license need not be delayed until the conclusion of any such proceeding."102 In so holding, the Commission explained in detail the factors which weighed against individualized consideration of the issue for the Diablo Canyon facility:

The focus of the emergency planning controversy among the parties is on the possible need to consider the contemporaneous occurrence of an earthquake and radiologic release from the plant. For the earthquakes up to and including the Safe Shutdown Earthquake (SSE), the seismic design of the plant was reviewed to render extremely small the probability that such an earthquake would result in a radiologic release. While a radiologic release might result from an earthquake greater than the SSE, the probability of occurrence of such an earthquake is extremely low. In addition, as the NRC staff noted in its January 13, 1984 memorandum to the Commission on the generic subject of earthquakes and emergency planning, for those risk-dominant earthquakes which cause very severe damage to both the plant and the offsite area, emergency response would have marginal benefit because of its impairment by offsite damage. Thus, the Commission agrees with the NRC staff's conclusion that the expenditure of additional resources to cope with seismically caused offsite damage under those circumstances is of doubtful value considering the modest benefit in overall risk reduction which could be obtained.

There remains only the possibility of a contemporaneous occurrence of both a radiologic release from the plant caused by an event other than an earthquake, and an earthquake that would complicate emergency response. NUREG-0654 does call for some consideration of site-specific adverse or emergency conditions on emergency response. In prior cases such frequently occurring natural phenomena as snow, heavy rain, and fog have been considered. With one exception, the focus has always been on frequently occurring natural phenomena. The Commission believes, based on the information provided by the parties, that earthquakes of sufficient size to disrupt emergency response at Diablo Canyon would be so infrequent that their specific consideration is not warranted.

The Commission's view that it need not give specific consideration to the complicating effects of earthquakes on emergency planning in this case is bolstered by the following consideration. Specific consideration has been given in this case to the effects of other relatively frequent natural phenomena. The evidence includes the capability of the emergency plan to respond to disruptions in communication networks and evacuation routes as a result of fog, severe storms and heavy rain. In the extreme, these phenomena are capable of resulting in area-wide disruptions similar to some of the disruptions which may result from an earthquake. Testimony in the Diablo Canyon record indicates that adverse weather conditions such as the effect of heavy fog could increase evacuation time to approximately 10 hours. Thus, while no explicit consideration has been given to disruptions caused by earthquakes, the emergency plans do have considerable flexibility to handle the disruptions caused by various natural phenomena which occur with far greater frequency than do damaging earthquakes, and this implicitly includes some flexibility to handle disruptions by earthquakes as well.103

The Commission also concluded that the Diablo Canyon site did not present "special circumstances" within the meaning of 10 C.F.R. Sec. 2.758 that would require individualized consideration of the earthquake issue.104 The Commission noted the decision of its Appeal Board that the site was not one of high seismicity, despite its proximity to the Hosgri Fault.105 At most, the Commission concluded, the region surrounding the Diablo Canyon facility was characterized by "moderate seismicity."106

Petitioners challenge the Commission's decision on essentially three grounds. First, they contend that the earthquake issue is "material" to the licensing decision; as such, this court's decision in Union of Concerned Scientists v. Nuclear Regulatory Commission107 would preclude its exclusion from the scope of the licensing proceeding. Second, petitioners assert that the earthquake issue is inappropriate for resolution through generic rulemaking: because the risk of seismic disturbance is significant for only two reactor sites in the United States (Diablo Canyon and San Onofre), petitioners contend that the issue should be addressed in the context of individual licensing proceedings. Finally, even if rulemaking were the proper avenue for agency resolution of the earthquake issue, petitioners contend that the failure of the Commission to commence rulemaking proceedings before approval of the Diablo Canyon license has effectively denied them an opportunity to participate in the decisionmaking process.

The first contention raised by petitioners is by far the most significant, for the materiality or immateriality of the earthquake issue determines the extent to which petitioners possess a right to demand particularized agency consideration of the issue. The Commission has determined, both in San Onofre and in its 10 August decision, that the potential complicating effects of a seismic event on emergency planning are immaterial. We therefore confront a situation considerably different from that which this court was required to address in Union of Concerned Scientists. In that case we held that section 189(a) of the Atomic Energy Act108 guarantees interested parties the right to a hearing on material issues; thus, an issue recognized by the NRC in its own regulations as material was found to require inclusion in hearings held under section 189(a).109 Here we are faced with precisely the opposite situation, for the Commission has twice held the earthquake issue to be immaterial.

The Commission's finding of immateriality does not close our inquiry, however, for otherwise the Commission could escape its responsibilities under section 189(a) by labelling any scientific or environmental factor "immaterial." Ultimate resolution of contested agency actions properly lies within the province of the courts. Because courts lack the scientific and technical expertise of regulatory agencies, however, the judicial role on review is limited to determining whether the agency has complied with relevant statutory and regulatory requirements. We must determine in the present case whether the NRC's finding of immateriality is "arbitrary, capricious, [or] an abuse of discretion";110 in doing so, we must accord great deference to decisions, such as the one at issue here, whose subject-matter lies "at the frontiers of science."111 Applying these well-established principles of administrative law, it becomes obvious that the NRC's decision to treat seismic issues as immaterial for purposes of individual licensing proceedings does not transgress the boundaries of agency discretion. The Commission's decision was based on three well-articulated scientific conclusions, none of which warrant characterization as arbitrary or capricious. First, the Commission found that the probability of a radiological accident caused by an earthquake was extremely low.112 The plant has been designed to withstand earthquakes of a magnitude up to and including that of a hypothetical "Safe Shutdown Earthquake" ("SSE"); for Diablo Canyon, the SSE has been calculated to lie at 7.5 on the Richter scale.113 Moreover, because the SSE is chosen with reference to the region's maximum earthquake potential, the probability that an earthquake greater than the SSE will occur is so low as to be legally insignificant.114 Second, the Commission considered the danger that an earthquake of significant magnitude would occur contemporaneously with a radiological accident precipitated by some other cause. The Commission concluded that, because Diablo Canyon is not located in an area of high seismicity, the probability that such a coincidence would occur is also extremely low.115 Finally, the Commission noted that the emergency plan for Diablo Canyon already possesses sufficient flexibility to deal with most disruptions that would be caused by an earthquake.116 While clearly not intended to be a complete justification for the Commission's decision, this final conclusion is reasonable with regard to a wide variety of potential disruptions. Many of the emergency planning problems that could be caused by an earthquake--the closure of roads, the disruption of communications, etc.--are already taken into account in Diablo Canyon's emergency plans because they could also be caused by other, more frequently occurring natural phenomena.117 We conclude that the Commission did not exceed its discretion in finding that the complicating effects of a seismic event on emergency planning was not material for purposes of individual licensing proceedings.

Having found that the earthquake risk was immaterial, the Commission was also clearly justified in deciding to treat the issue as a generic problem to be resolved through rulemaking rather than adjudication. Petitioners challenge the Commission's decision on the ground that the earthquake risk affects only two nuclear plants, Diablo Canyon and San Onofre, both of which are located on the California coast. Rulemaking, they contend, is appropriate only for those issues which affect a large number of nuclear plants across the country, while specialized issues (such as the earthquake risk) should be dealt with in the context of individual licensing proceedings.

As a general proposition petitioners' analysis has merit: common issues should ideally be resolved through rulemaking, while unusual issues are best resolved through adjudication. This simple proposition, however, is incapable of being reduced to mathematical terms. One cannot say, for example, that an issue which affects 15% of the nuclear power plants in the country should be resolved generically, while one which affects only 10% of the nation's plants is inappropriate for generic resolution. Determination of the best decisionmaking method must be made, in the first instance, by the agency; our review of that decision is circumscribed by the "arbitrary and capricious" standard.118 Particularly when agencies are dealing with probabilities--here, a risk which may or may not materialize--considerable deference should be accorded the agency's decision. Even if it were currently believed that Diablo Canyon and San Onofre were the only nuclear facilities in America located in areas of significant seismicity, future research might reveal the existence of faults near other facilities. Indeed, the Hosgri Fault itself was unknown until the early nineteen-seventies. Because seismology remains a difficult and inexact science, we hesitate to conclude, as petitioners would have us do, that the earthquake risk is unique to Diablo Canyon and San Onofre.

Petitioners' final contention focuses on the Commission's delay in instituting even rulemaking proceedings to address the earthquake risk. The Commission's failure to institute such proceedings, petitioners argue, has effectively denied interested parties the right to present evidence on the seismic risk at Diablo Canyon. But we have already concluded that the Commission did not err in finding the earthquake issue immaterial for purposes of the Diablo Canyon proceeding; petitioners therefore did not possess the right they now assert. At most, their claim resolves to a contention that the Commission has unreasonably delayed in instituting rulemaking proceedings.

The Commission itself has admitted that it should have acted sooner in initiating a rulemaking. As it stated in its 10 August 1984 decision,

[w]e previously indicated in San Onofre that this matter would be considered on a generic basis. Some time ago the NRC staff advised us that, in its view, generic consideration was not necessary. However, we were diverted from this issue by the press of other important Commission business, and we took no action in response to that advice. In retrospect, since we disagree with the NRC staff's view, we should have acted sooner and initiated rulemaking. The need to address this issue in this case has again focused our attention on this matter. By this order we are indicating our desire to initiate rulemaking shortly, and directing the NRC staff to give priority attention to the matter.119

Taking into account the time period in question, as well as the fact that the Commission's delay was occasioned in large part by the recommendation of its staff that generic consideration was not necessary, we conclude that the delay in instituting rulemaking proceedings was not so unreasonably long as to have violated the agency's obligations under the Administrative Procedure Act or the Atomic Energy Act.

Before it may issue an operating license for a nuclear power plant, the Commission is required by its regulations to find that "the facility will operate in conformity with the application as amended, the provisions of the Act, and the rules and regulations of the Commission";120 "[t]here is reasonable assurance ... that the activities authorized by the operating license can be conducted without endangering the health and safety of the public";121 and "[t]he applicant is technically ... qualified to engage in the activities authorized by the regulations."122 Essential to these findings is a determination that the plant's reactor operators have been properly trained and licensed. In its 10 August 1984 decision123 the Commission concluded that the operators at Diablo Canyon were properly qualified.124 Petitioners now challenge that determination as contrary to the Commission's own rules for the licensing of reactor operators.125

NRC regulations require that applicants for the position of reactor operator pass both a written examination and an operating test.126 These requirements may be waived, but only if the applicant "[h]as had extensive actual operating experience at a comparable facility within two years prior to the date of application."127 The NRC regulations also provide for the administration of a simulated operating test to applicants, but only if the applicant "has had extensive actual operating experience at a comparable reactor."128

The Commission has consistently construed its regulations to allow experience on a computer simulator to suffice in the absence of "actual operating experience."129 The Commission's rationale for doing so turns on an alleged inadequacy in the regulatory language: according to the Commission, great advances in computer simulation render anachronistic a literal reading of the rule.130 To correct the perceived anachronism, the Commission recently amended its regulations to permit experience on simulators to suffice in the absence of actual operating experience.131 That amendment came, however, long after reactor operators at Diablo Canyon had been licensed; many of the operators had received only computer-simulated training.132 Petitioners now challenge the qualifications of those operators who did not possess actual operating experience at the time of their licensing.

It is quite obvious that the Commission's "long-standing practice" was in direct contravention of the regulatory provision. It is true, of course, that "[w]ords ... do not have absolute and constant referents,"133 and that "a judicial belief in the possibility of perfect verbal expression ... is a remnant of a primitive faith in the inherent potency and inherent meaning of words."134 But if words in a regulatory provision may be said to retain any meaning at all, we are convinced that "actual operating experience at a comparable reactor" cannot be said to encompass experience on a computer simulator. Courts must defer to the interpretation administrative agencies accord their governing statutes and regulations,135 but such deference is appropriate only so long as the agency's interpretation does no violence to the plain meaning of the provision.136

We would therefore be required to find Diablo Canyon's reactor operators unqualified under NRC regulations, were it not for the recently-promulgated amendment to those regulations. The amendment addresses precisely the issue raised on this appeal: simulated experience is now considered sufficient to ensure that reactor operators are qualified to perform their crucial and demanding role. For us to remand the issue to the Commission would serve no discernable purpose: the Commission would clearly be justified in relicensing the Diablo Canyon operators under the amended regulation. We conclude, therefore, that the Commission practice under which Diablo Canyon's operators were licensed was contrary to the plain meaning of the Commission's own rule, but that the option of relicensing the operators effectively precludes the necessity of remanding the issue to the Commission.

D. The Right to a Hearing Under Section 189(a) of the Atomic Energy Act

The Atomic Energy Act grants to interested persons the right to a hearing in certain types of nuclear regulatory proceedings. Section 189(a) provides that

[i]n any proceeding under this chapter, for the granting, suspending, revoking, or amending of any license or construction permit, or application to transfer control, and in any proceeding for the issuance or modification of rules and regulations dealing with the activities of licensees, and in any proceeding for the payment of compensation, an award or royalties under Sections 2183, 2187, 2236(c) or 2238 of this title, the Commission shall grant a hearing upon the request of any person whose interest may be affected by the proceeding, and shall admit any such person as a party to such proceeding.137

Petitioners contend that the Commission violated section 189(a) by refusing to grant them a hearing when it lifted a summarily-imposed suspension of Diablo Canyon's license for low power operations, and when it amended that license to extend its term.138 The Commission defends its denial of petitioners' request for a hearing on essentially two grounds. It first contends that the actions it took in the low power proceeding were not the types of actions for which Congress intended a hearing to be accorded under section 189(a). Neither the lifting of a license suspension nor the extension of a license term, it argues, automatically triggers the procedural protections of the Act. If section 189(a) is found to apply, however, the Commission contends that petitioners were in fact accorded a hearing by virtue of their ongoing attempts to reopen the record in the full power proceeding. Because the issues petitioners sought to raise in connection with the lifting of the license suspension and the extension of the license term involved questions of design and construction quality assurance, the Commission contends that it was justified in referring petitioners to their contemporaneous efforts to reopen the full power record with regard to those very same issues. Petitioners were ultimately successful in obtaining reopening of the record on design quality assurance, but they were unable to persuade the Commission to reopen on matters of construction quality assurance. In Section E below we conclude that the Commission was correct in denying petitioners' request to reopen on the construction quality assurance issues.

Our analysis of petitioners' present claims under section 189(a) may be briefly summarized. We agree with the Commission that the lifting of a license suspension does not fall within any of the enumerated categories of Commission action for which a hearing must be held. We agree with petitioners, however, that the extension of the term of a license constitutes a license "amendment" within the meaning of section 189(a). Petitioners were therefore entitled to a hearing when PG & E sought two extensions of its low power operating license. We conclude that the Commission satisfactorily discharged its obligation to provide such a hearing on issues of design quality assurance when it referred petitioners to the reopened proceedings on that issue. The Commission violated section 189(a), however, when it denied petitioners a hearing on issues of construction quality assurance. The crucial distinction between construction and design issues stems from the fact that the full power record was never reopened on construction issues, as it was on matters of design. Because the Commission's criteria for reopening a closed record are higher than the criteria for obtaining a hearing under section 189(a), the mere fact that a party can seek reopening is not a sufficient substitute for the hearing rights guaranteed by section 189(a). In light of the unique circumstances of the present case, however, we do not believe that judicial relief for the Commission's procedural error is appropriate. Because we conclude in Section E that petitioners were not entitled to reopen the record on construction quality assurance, we must also conclude that none of the evidence petitioners would have presented on that issue, had they been accorded a hearing under section 189(a), would have been material or safety-significant. The Commission's violation of section 189(a) therefore, does not call into question the decision to license Diablo Canyon or the safe operation of the plant, and it would be pointless for us to enjoin such operations as a result.

Our analysis of petitioners' claims under section 189(a) necessarily begins with a determination of whether the types of action taken by the Commission come within the ambit of the statutory provision.

As an initial proposition, petitioners suggest that all actions by the Commission which bring about a "significant change" in the licensing status of a nuclear facility come within the mandate of section 189(a). That provision is quite specific, however, in its enumeration of agency actions which trigger the hearing requirement. No fewer than eight such types of action are listed, and among them no mention is made of actions which effect "any significant change."

An examination of the section's legislative history confirms the specificity and precision with which the section was drafted. When first proposed, the Atomic Energy Act of 1954 included no provision for hearing rights.139 After considerable discussion140 a new section 181 was incorporated in the substitute bill introduced in the House;141 it provided that "[u]pon application, the Commission shall grant a hearing to any party materially interested in any 'agency action.' "142 The new provision was faulted, however, for being "too broad, broader than it was intended to be [be]."143 Senator Hickenlooper offered an amendment to restrict the hearing right to specific categories of agency action.144 The result, a new subsection within section 189,145 was the substantive equivalent of the modern provision. The Act's legislative history, therefore, reveals the intent of Congress to guarantee hearings rights in certain classes of agency action, but not in others.146

We are aware that our decision in Sholly v. Nuclear Regulatory Commission147 may be read to suggest a contrary conclusion. In Sholly this court reversed orders of the NRC which had permitted the venting of radioactive gas into the atmosphere from the damaged Three Mile Island nuclear plant.148 The petitioners in Sholly contended that the Commission's Venting Order constituted an amendment to the Three Mile Island operating license, and that section 189(a) therefore required that a hearing be held. The NRC disputed the characterization of the order as a license amendment; it contended that the order merely reinstated a preexisting authority already accorded the plant's licensee.149 This court sided with petitioners, concluding that "adequate venting of the reactor building might not [have been] possible under the existing license authority.... Authority for venting--in this case the June 12 Venting Order--therefore had to come in the form of a license amendment."150 The court clearly regarded the Venting Order as such an amendment. It stated later in the opinion that

the Venting Order appears as an amendment to the February 11 amendment to TMI-2's operating license. Because the June 12 Venting Order modified the February 11 order, and granted the licensee authority to do something that it otherwise could not have done under the existing license authority, the Venting Order was a license amendment within the scope of section 189(a).151

In dictum, however, the court suggested an unnecessarily broad interpretation of section 189(a):

Our reading of the Venting Order is also supported by Congress' intent in enacting section 189(a). By requiring a hearing upon request whenever a license is "grant[ed], suspend[ed], revok[ed], or amend[ed]," Congress apparently contemplated that interested parties would be able to intervene before any significant change in the operation of a nuclear facility. Whatever the Venting Order is called, it certainly was such a change.152

Because the court had repeatedly characterized the Venting Order as a license amendment, and because such amendments are among those agency actions for which a hearing is explicitly guaranteed by section 189(a), the court's suggestion that "any significant change" will trigger the hearing right was wholly unnecessary for resolution of the case. Significantly, the court cited no legislative history to support its interpretation of the provision. We therefore regard Sholly as inadequate precedent for the proposition that any significant change in the licensing status of a nuclear power plant triggers the procedural protections of section 189(a).

Petitioners' analysis, however, does not depend solely on the existence of an implied residual category under section 189(a); petitioners also contend that the actions at issue here--the lifting of the license suspension and the extension of the license term--come within the categories of agency action specifically enumerated in the section. As we have discussed above, what legislative history there exists suggests that Congress intended the provisions of the section to be construed quite literally. If a particular form of Commission action does not fall within one of the eight categories set forth in the section, no hearing need be granted by the Commission.

Section 189(a) makes no mention of the lifting of a license suspension. Petitioners have suggested no reason for construing such action as an "amendment," and our examination of the section suggests none. The lifting of a suspension does nothing to alter the original terms of a license; indeed, it removes a significant impediment to the enforcement of those terms. Nor do we believe that Congress intended the lifting of a license suspension to be subsumed within the statutory category of license suspensions. If it had, then we should also conclude that license revocations are implicitly included in the category of license grants; but Congress found it necessary to list revocations as a separate category.153 Because none of the actions specified in section 189(a) may be said to include the lifting of a license suspension, we conclude that such action does not give rise to the right to a hearing.154

We reach the opposite result with regard to the two requests for extension of Diablo Canyon's license term. This court held in Brooks v. Atomic Energy Commission155 that the extension of the term of a construction permit constitutes an "amendment" within the meaning of section 189(a).156 We see no reason for treating extensions of an operating license differently; if anything, concerns of public safety are more strongly implicated by the extension of an operating license, simply because a licensed plant is much nearer completion than one which has only been granted a construction permit.

Respondents attempt to distinguish Brooks in two respects, but we find neither persuasive. First, respondents argue that Brooks applies only to extensions undertaken for substantive purposes; the extensions of the Diablo Canyon license, they contend, "merely changed the time at which various activities--each of which was specifically authorized by the 1981 low power license--might be conducted, and not the nature or duration of the activities."157 We believe, however, that the reference to "amendments" in section 189(a) means all amendments, and not just those which effect a substantive change in a plant's status. We have already explained why we decline to create a residual category in section 189(a) for all actions which bring about a substantive change in the licensing status of a plant. The logic of that analysis now leads us to conclude that substantiality should not be imported as a limitation on the categories specifically listed in the section.158 Had Congress intended substantiality to play a role in the hearing context it could have said so; Congress's silence leads us to conclude that even nonsubstantive amendments come within the section's purview.

Respondents also contend that Brooks should not apply to cases in which there exists an ongoing licensing proceeding to which parties seeking a hearing under section 189(a) may be referred.159 As we discuss below, we believe it proper for the Commission to have directed petitioners to pursue their claims regarding design quality assurance in the reopened proceedings on that issue. That conclusion, however, goes not to the issue of whether there has been an "amendment" within the meaning of section 189(a), but rather to the sufficiency of the Commission's attempt to comply with that section once it has been found to apply.

Having concluded that the extension of Diablo Canyon's license term was an amendment under section 189(a), and that petitioners were therefore entitled to a hearing on issues raised by that extension, we now turn to the sufficiency of the Commission's response. Petitioners sought to present evidence concerning alleged deficiencies in design quality assurance and construction quality assurance;160 we have found no suggestion in the briefs or the record that petitioners sought to raise other issues. We may therefore confine our inquiry to determining whether petitioners received a legally sufficient hearing on the design and construction quality assurance issues.

The Commission declined to grant petitioners a hearing in the immediate context of the license extension requests. Instead, it referred petitioners to the ongoing license proceeding and, specifically, to their efforts to reopen the full power record on issues of design and construction quality assurance.161 The record was ultimately reopened on the design quality assurance issue;162 with regard to construction, however, the Commission's boards held that petitioners had failed to meet the prescribed standards for reopening.163

We believe the Commission was justified in referring petitioners to the reopened proceedings on design quality assurance. In the reopened proceedings, petitioners were afforded a full and complete opportunity to present new evidence on the design issue. If anything, the hearing they thereby received was more comprehensive and entailed greater procedural protections than would have been required under section 189(a). We conclude that the Commission satisfied the requirements of section 189(a) insofar as petitioners sought to raise issues of design quality assurance.

We reach a different conclusion with regard to the construction quality assurance issue. Here, too, the Commission referred petitioners to their ongoing efforts to reopen the closed full power record; on construction issues, however, the Commission subsequently concluded that petitioners had failed to satisfy its reopening criteria. As we discuss in Section E, we believe the denial of the request to reopen was proper in light of the Commission's high standards for reopening. Precisely because of the stringency of those criteria, however, we cannot conclude that the opportunity to seek reopening was an adequate substitute for the hearing guaranteed petitioners as a matter of right under section 189(a). In order to obtain reopening, petitioners were required to show that they possessed new evidence which was timely; material, in the sense that it would have resulted in a different outcome had it been known earlier; and safety-significant.164 None of these three criteria applies to requests for a hearing under section 189(a). Under the latter provision parties need only show that their "interest may be affected" by a proceeding to bring about one of eight specified types of Commission action.165 At most, parties must show that a particular issue is "material" in order to prevent its exclusion from a hearing under section 189(a); this much our decision in Union of Concerned Scientists v. Nuclear Regulatory Commission establishes.166 But the material issue requirement implicit in section 189(a) is significantly different from the material evidence requirement of the Commission's reopening criteria. In most cases, as here, the latter requirement will impose a substantially more onerous burden on parties than the former.167

Nor does the fact that a four day "Mini-Hearing" was held to consider petitioners' reopening request on construction quality assurance issues168 alter our conclusion that the Commission violated section 189(a). The sole purpose of the hearing was to determine whether petitioners had met the Commission's reopening criteria; petitioners were allowed no generalized inquiry into other subjects, as they undoubtedly would have had under section 189(a).

We therefore conclude that the Commission denied petitioners their statutorily-prescribed rights to a hearing on issues of construction quality assurance in the proceedings held pursuant to PG & E's two requests to extend the term of its low power license. We must now determine what relief, if any, is appropriate to remedy the agency's procedural deficiency.

Our conclusion that the Commission acted unlawfully in denying petitioners a hearing itself constitutes one particularly powerful form of judicial relief: it clarifies the scope of section 189(a), thereby making it less likely that the Commission will violate that provision again. At the time the Commission extended PG & E's license, there existed a paucity of federal case law explicating the relationship between section 189(a) and the reopening option.169 The Commission's legal error, therefore, may well be attributable to the lack of clear judicial guidance and not to a deliberate desire to deny petitioners their rights. Our holding today that consideration of a request to reopen the record does not satisfy the requirements of section 189(a) should preclude such Commission error in the future. In the unlikely event the Commission repeats its mistake, this court would have no choice but to presume bad faith on the part of the Commission and act accordingly.

We believe, however, that additional relief in the present case would serve no meaningful purpose. Our conclusion proceeds from two factors which, taken together, establish that neither the original licensing nor the safe operation of the Diablo Canyon plant is called into question by the Commission's violation of section 189(a). First, the failure to accord petitioners a hearing occurred in the context of the low power proceedings. Low power testing at the facility has now been completed. While we do not hold that the issues raised by the issuance of the low power license have thereby become moot, we conclude that a remand on the low power issue would serve no practical purpose. A second, and more substantive, reason for denying further relief is that the evidence petitioners would have introduced on the construction quality assurance issue had they been accorded a hearing was neither outcome-determinative nor safety-significant. When petitioners sought to reopen the full power record on the very same evidence, the Commission concluded that their evidence failed to satisfy the reopening criteria of materiality and safety-significance;170 we affirm that conclusion in Section E. The Commission defines as material only that evidence which, if known at the time of the original decision, would have led to a different substantive result;171 "material" therefore means "outcome-determinative." By definition, the introduction of petitioners' (immaterial) evidence would not have changed the Commission's decision to license the Diablo Canyon facility. Moreover, because the evidence was not safety-significant, its introduction would not have called into question the safe operation of the plant. A remand to the Commission directing it to consider the same evidence now would accomplish no more than its original introduction would have. We decline to require the Commission to engage in a procedural exercise whose outcome we already know would be adverse to petitioners.

E. Reopening the Record on Construction Quality Assurance

On several occasions petitioners sought to reopen the record on issues of construction quality assurance. In each case the Commission's boards denied petitioners' requests. Our role in reviewing those decisions is a limited one; as we recently stated in Mobil Oil Corp. v. Interstate Commerce Commission,172 "[w]here as here the agency has taken final action on a matter that is peculiarly within its realm of expertise, we will not require the agency to reopen its proceedings except upon a clear showing of abuse of discretion or of extraordinary circumstances."173 The Commission has established standards for determining when a closed record should be reopened; a failure by the Commission to respect its own guidelines would constitute just such a "clear showing of abuse of discretion" as was contemplated by the Mobil Oil decision.

Under Commission practice, reopening is required when new evidence is shown to be timely, safety significant, and sufficiently material to have changed the result initially taken.174 A party moving to reopen must show that its new evidence is "strong enough, in the light of any opposing filings, to avoid summary disposition."175 Applying these criteria, we conclude that the Commission and its boards did not err in denying petitioners an opportunity to reopen the Diablo Canyon proceedings on issues of construction quality assurance.

Petitioners first moved to reopen the record on the construction issue in April 1977.176 Their request was denied on timeliness grounds by a licensing board of the NRC on 25 May 1977: Because the time for the specification of issues had expired in the spring of 1974, petitioners raised the issue on which they sought reopening three years too late.177 Petitioners now attempt to justify their delay, but we find their explanations ultimately unconvincing. Petitioners assert that

[t]he right to submit a contention on this complex issue at an earlier date had little practical value, because Petitioners were unable to obtain counsel and an expert on this issue until 1976. Moreover, their 1977 application was based substantially on information and reports that became available well after the original contentions were finalized in 1974.178

The inability to retain counsel and experts before expiration of the prescribed time period constitutes no legal justification for petitioners' delay. If it did, the principle of finality underlying filing deadlines, statutes of limitations and other legal time bars would be rendered subordinate to the financial and practical vicissitudes of individual litigants. Nor does the fact that petitioners' motion to reopen was based "substantially " on information that became available after the prescribed deadline render their delay excusable. Petitioners must be held accountable for their failure even to raise the construction issue by the 1974 deadline.

Despite the untimeliness of petitioners' request, the licensing board held an on-the-record proceeding on the construction issue in 1977.179 The board solicited presentations by the NRC staff and PG & E; petitioners were accorded a limited right to cross-examine adverse witnesses, but they were allowed no independent right to present evidence or engage in discovery.180 Objecting to the board's procedures, petitioners declined to engage in cross-examination. On this appeal petitioners continue to assert that they were illegally denied full hearing rights in 1977. Because petitioners' reopening request was properly barred by the board on timeliness grounds, however, any opportunities for participation accorded petitioners were over and above those to which they were entitled. The board did not obligate itself to provide a full array of procedural protections merely by holding a hearing that it was not required to hold in the first place.

Petitioners next sought reopening of the record--on both construction and design quality assurance issues--in June 1982, after the discovery of widespread errors in the plant's design. The Appeal Board, recognizing the potential seriousness of the design deficiencies, reopened the record on design quality assurance.181 Petitioners assert, however, that the acknowledged deficiencies in design quality assurance necessarily raised questions concerning the construction quality assurance program.182 As support for this proposition petitioners rely in large part on a March 1983 memorandum of the NRC staff, which stated that the breakdown in design "raises (implicitly at least) questions regarding the adequacy of these [construction] programs."183 The staff suggested that some construction issues might profitably be incorporated into the Informal Design Verification Program ("IDVP") then in progress. In fact, a number of such construction issues were voluntarily added to the IDVP by PG & E. No need for modifications in the construction program were found, however, and the staff subsequently concluded that construction at Unit 1 was acceptable.184 We would be hard pressed to conclude on the basis of these facts that the March 1983 memorandum establishes a causal link between deficiencies in design and deficiencies in construction quality assurance. The Commission is not bound by tentative suggestions contained in memoranda from its staff, particularly where the staff subsequently concludes that its concerns lack an identifiable factual basis.

Petitioners refiled their application seeking to reopen construction quality assurance issues in May 1983.185 Alleging the existence of "significant new evidence," petitioners proffered (1) expert affidavits and testimony; (2) the statement and testimony of a former quality control manager of one of Diablo Canyon's principal contractors; (3) documentation of "actual" construction defects; and (4) a 1977 audit by the Nuclear Services Corporation, which concluded that a principal piping contractor at Diablo Canyon had violated all eighteen of the Commission's quality assurance standards.186

The Appeal Board held a four day "Mini-Hearing" on petitioners' allegations; no discovery was allowed, but all parties were permitted to cross-examine the others' affiants.187 As a consequence of evidence adduced at the Mini-Hearing, the Appeal Board denied petitioners' motion to reopen the record, first by summary order of October 1983188 and then in a memorandum decision of 19 December 1983.189 Petitioners challenge that decision on a number of grounds, but we find none of their contentions convincing. The Board found, for example, that one of petitioners' principal witnesses, Richard Hubbard, should not be regarded as an expert for purposes of construction quality assurance.190 Hubbard himself admitted that he had never worked on construction quality assurance at a nuclear power plant site;191 most of his allegations of construction deficiencies were based on the assumption that such deficiencies must exist because design deficiencies had been discovered.192 Because the board cannot be said to have acted unreasonably in according Hubbard's testimony limited weight, we see no reason to mandate a contrary result. As we said recently in Carstens v. Nuclear Regulatory Commission, "[s]ince the Board is obviously best situated to assess the credibility and demeanor of witnesses, this court must defer to that judgment so long as it is reasonable."193 Similarly, the Board was justified in discounting the 1977 NSC audit of Diablo Canyon's principal piping contractor. The NRC staff and PG & E concluded that the deficiencies alleged therein had already been corrected, and the Appeal Board explicitly found that "the deficiencies identified by NSC in 1977 did not evidence a significant or systematic failure of the quality assurance program."194 These conclusions clearly lay within the parameters of agency expertise, and it is to these conclusions that we are obligated by our case law to defer.

Petitioners' latest attempt to reopen the record on construction quality assurance issues came in February 1984, when it filed lengthy motions accompanied by numerous affidavits and supporting documents.195 On 23 May the Appeal Board informed petitioners that they would be required to demonstrate the individual significance of their various allegations.196 Petitioners conceded that few of the alleged deficiencies were individually significant;197 they contended, however, that their allegations established a pattern of safety-significant deficiencies.198 In its decision of 28 June, however, the Board explicitly found that petitioners had not demonstrated "a breakdown of the applicant's quality assurance program that raises legitimate doubt that the facility can operate safely."199 Only one issue--the possible improper use of A307 hardware bolts with heads removed as studs--was found to warrant further consideration,200 and that issue was satisfactorily resolved in a later proceeding.201 Moreover, in denying petitioners' motion to reopen, the Appeal Board noted the unorganized and sometimes repetitious nature of petitioners' allegations.202 While the Board did not discuss in great detail the inadequacies of those allegations, it did explain why detailed discussion was not necessary.203 Once again, in light of the Appeal Board's careful consideration of petitioners' evidence, we discern no basis for setting aside that body's findings.

Petitioners also assert that significant questions remain unanswered as to the design quality assurance program at Diablo Canyon. Petitioners successfully obtained reopening of the record with regard to design quality assurance in April 1983.204 Following fifteen days of adjudicatory hearings,205 the Appeal Board found in March 1984 that the plant's design verification program established the adequacy of the unit's design.206 The Board concluded that

[t]he applicant's verification efforts provide adequate confidence that the Unit 1 safety-related structures, systems and components are designed to perform satisfactorily in service and that any significant design deficiencies in that facility resulting from defects in the applicant's design quality assurance program have been remedied. Accordingly, we conclude that there is reasonable assurance that the facility can be operated without endangering the health and safety of the public.207

Despite the unequivocal and well-supported208 conclusions of the Appeal Board, petitioners continue to seek reopening of the record on issues of design quality assurance. In February 1984 petitioners filed lengthy motions, accompanied by over 1,000 pages of affidavits and other documents, alleging the continued existence of design deficiencies at the facility.209 On 28 June 1984 the Appeal Board held that petitioners had failed to satisfy the Commission's reopening requirements.210 The Board stated that it had

carefully examined each of the joint intervenors' charges with their supporting materials and the responses of the applicant and the staff. Our scrutiny of the motions leads us to conclude that the joint intervenors have failed to present new evidence of any significant safety issue that could have an effect on the outcome of the licensing proceeding. Among other things, the movants have not presented evidence that establishes uncorrected design or construction errors that endanger safe plant operation. Nor have they demonstrated that there has been a breakdown of the applicant's quality assurance program that raises legitimate doubt that the facility can operate safely.211

On 10 August 1984 the Commission itself concluded that petitioners' allegations concerning design quality assurance were insufficient to preclude issuance of a full power license.212

On this appeal petitioners rely primarily on alleged deficiencies in the design of the piping system at Diablo Canyon. They contend that the Appeal Board and the Commission gave insufficient weight to allegations by former NRC Staff Inspector Isa Yin that an "apparent Q[uality] A[ssurance] breakdown [exists] in the areas of [small bore and large bore] piping design control."213

The record reveals that Inspector Yin's concerns were accorded serious and thoughtful consideration by the Commission and its staff. The staff formed the Diablo Canyon Peer Review Group, a body composed of experienced staff members and consultants, to investigate Inspector Yin's concerns.214 After meeting with the Inspector and representatives of PG & E, the Peer Review Group formulated seven conditions for inclusion in the low power license as prerequisites to issuance of a full power license.215 The seven conditions dealt specifically with the plant's piping system; they required PG & E to analyze and resolve a range of potential problems in the design of that system.216 After a thorough review that consumed over two years of professional study, both the Peer Review Group and the Advisory Committee on Reactor Safeguards found that PG & E had sufficiently resolved the issues so as to permit issuance of the full power license.217 At a public Commission meeting held in early August 1984, Inspector Yin expressed his disagreement with the Peer Review Group's findings.218 In approving the full power license for Diablo Canyon, the Commission specifically discussed Inspector Yin's continued opposition, ultimately concluding that "the collective judgments of the Peer Review Group and ACRS are deserving of more weight than the views of Mr. Yin."219

The concerns expressed by Inspector Yin deserved serious and deliberate consideration by the NRC; they were not entitled, however, to blind adoption or unquestioning acceptance. For the Commission to have ignored the recommendations of its Peer Review Group would have been to elevate the unsubstantiated objections of one dissident staff member over the considered conclusions of a majority of his peers; it is the views of the one dissident staff member on which petitioners rely.

If it was correct for the Commission to have approved Diablo Canyon's full power license despite the continued objections of Inspector Yin, it was eminently correct for the Appeal Board to have denied petitioners' request to reopen the record on the same grounds. The Board did not ignore petitioners' allegations, it rejected them. In doing so, it stated with specificity its reasons for concluding that the plant's piping system met adequate design criteria:

[A] number of the allegations focus on deficiencies in the methodology, practices, and quality assurance associated with the computer design of small bore (less than 2"' diameter) pipe supports. The staff also found the number of errors occurring in this type of calculation to be higher than expected. A staff-imposed license condition required the applicant to redo all computer-based small bore pipe support calculations--including additional physical effects not addressed in the original analyses. We note that the result of this program, with the reanalysis of all but 15 of 357 supports completed, shows that all of the supports meet design criteria, and no modifications are necessary. Thus, errors in the small bore pipe support computer calculations, though numerous, have had no effect on the design adequacy of the supports.220

In short, both the Commission and its Appeal Board accorded petitioners' design allegations full and complete consideration. We are unable to say that their conclusions exceeded the boundaries of scientific expertise within which the agency is properly accorded the discretion to act.

To bolster their case petitioners moved to supplement the administrative record and the record on appeal with "transcripts and related documents"221 from a meeting of the Nuclear Regulatory Commission. We deny petitioners' motion.

The evidentiary basis for this motion is comprised of correspondence, authored immediately before oral argument, which charges the Commission with various improprieties in its decision to deny petitioners a full-dress, on-the-record hearing to consider the possible complicating effects of an earthquake occurring simultaneously, or nearly so, with a nuclear emergency at the Diablo Canyon nuclear power plant. On 26 October 1984, four days before oral argument, Representative Richard Ottinger, chairman of the House Subcommittee on Energy Conservation and Power, wrote the chairman of the Commission, asserting that his review of confidential transcripts of a Commission meeting and related documents showed that the Commission's "brief seriously misstates the Commission's actions on the case."222 He accordingly called upon the Commission chairman to "correct these statements to the court and release the transcripts and related documents."223 Three days later, on the eve of oral argument, these charges were reiterated by Commissioner James Asselstine, the Commissioner who had dissented from the NRC's decision to license Diablo Canyon for operation at both low and full power and from its decision to deny petitioners an adjudicatory hearing on the effect of an earthquake upon emergency planning. Commissioner Asselstine "agree[d] entirely with the descriptions of the Commission's deliberations ... contained in Chairman Ottinger's ... letter to the Commission."224

NRC Chairman Nunzio Palladino, responding to another Congressman's request to make public the transcripts and documents, denied the charges of impropriety:

It goes without saying that the Commission strongly disagrees with Commissioner Asselstine's criticisms of its Diablo Canyon decision process. The Commission believes that there have been no abuses and rejects categorically Commissioner Asselstine's speculation as to the motives of the majority in reaching its decision. Simply because Commissioner Asselstine did not prevail is not a valid reason for him to castigate the majority, impugn their motives, and use his disagreement as a basis to urge release of the transcripts to the public. As a matter of fact, his statements are disturbing because this kind of disclosure of preliminary exchanges of views and thoughts of Commissioners and their advisors is destructive of the collegial exchange of Commissioner views and frustrates the testing of preliminary adjudicatory positions in closed meetings before they become final.225

Petitioners style this motion as one to supplement the record. As a practical matter, however, they first seek disclosure, under the Sunshine Act,226 of the predecisional transcripts and related documents of the Commission's deliberations; only then do they seek to include these disclosures in the administrative record and the record on appeal. Because these heretofore confidential transcripts and documents are not properly a part of the record of these proceedings, we have no occasion to consider petitioners' Sunshine Act claims.227 This appeal is not the proper forum for resolving satellite discovery requests.

We recognize at the outset that "[a]lthough this court has sanctioned supplementation of the record in certain circumstances ... the practice decidedly is the exception not the rule."228 In discharging their obligation to monitor agency action, courts review a record compiled by the agency and containing its rationale and supporting findings, accompanied by record evidence; they do not ordinarily study predecisional transcripts of deliberations within an agency.229 This foundational rule is codified in section 2112(b) of title 28, which provides that

[t]he record to be filed in the court of appeals in such a proceeding [to review agency orders] shall consist of the order sought to be reviewed or enforced, the findings or report upon which it is based, and the pleadings, evidence, and proceedings before the agency, board, commission, or officer concerned, or such portions thereof ....230

Effectuating the statutory mandate, rule 16(a) of our own rules of appellate procedure provides that "[t]he order sought to be reviewed or enforced, the findings of report on which it is based, and the pleadings, evidence and proceedings before the agency shall constitute the record on review in proceedings to review or enforce the order of an agency."231

The principle that judges review administrative action on the basis of the agency's stated rationale and findings, and our correlative reluctance to supplement the record, is well-established. The Supreme Court set the tone early on in Securities & Exchange Commission v. Chenery Corp.232 In an opinion rendered three years before passage of the Administrative Procedure Act,233 the Court reasoned that "[s]ince the decision of the Commission was explicitly based upon the applicability of [certain principles] ... its validity must likewise be judged on that basis."234 Over forty years of judicial experience in the review of administrative action demonstrate the wisdom of this practice. In its 1983 decision in Motor Vehicle Manufacturers Association v. State Farm Mutual Automobile Insurance Co.,235 the Supreme Court reiterated the "well-established" rule "that an agency's action must be upheld, if at all, on the basis articulated by the agency itself."236 This circuit has following the Court's command.237 Recently, in Kansas State Network, Inc. v. Federal Communications Commission,238 we concluded that "[i]n general, an agency's action should be reviewed based upon what it accomplishes and the agency's stated justifications"--"[w]here an agency has issued a formal opinion or a written statement of its reasons for acting, transcripts of agency deliberations at Sunshine Act meetings should not routinely be used to impeach that written opinion."239

Precedent aside, judicial reliance on an agency's stated rationale and findings is central to a harmonious relationship between agency and court, one which recognizes that the agency and not the court is the principal decision-maker. Were courts cavalierly to supplement the record, they would be tempted to second-guess agency decisions in the belief that they were better informed than the administrators empowered by Congress and appointed by the President. The accepted deference of court to agency would be turned on its head: the so-called administrative state would be replaced with one run by judges lacking the expertise and resources necessary to discharge the function they had arrogated unto themselves.240

Inclusion in the record of documents recounting deliberations of agency members is especially worrisome because of its potential for dampening candid and collegial exchange between members of multi-head agencies. While public disclosure stifles debate to some extent, judicial disclosure would suppress candor still further since off-hand remarks could turn out to have a legal significance they would not have if barred from the record on review.

Notwithstanding the weight of precedent and policy favoring exclusion of deliberative documents from the record, petitioner has catalogued three occasions on which supplementation is appropriate and asserts that each is applicable on this motion. Assuming, without deciding, that these predecisional transcripts of agency deliberations are available for inclusion in the record, we find petitioners' grounds inadequate to warrant supplementation of this record.

The first exception claimed by petitioners permits supplementation when there is "such failure to explain administrative action as to frustrate effective judicial review."241 The record before us contains no such defect. As explained in detail above, the Commission has adequately stated the basis of its decision and provided factual findings sufficient to support it.

Moreover, if "the Commission lacked a basis in the record for its assumptions"242--as petitioners allege--then the proper course is to reverse or vacate its decision and remand for further administrative proceedings.243 Additional transcripts would not facilitate our review and, as explained above, if the Commission's judgment indeed lacked a factual foundation, our function is emphatically not to supply one.244 Even more, contrary to what petitioners seem to believe, judges are not historians charged with isolating the "true" basis for an agency's decision when its ostensible justification proves unconvincing. We review administrative action for compliance with applicable law, nothing more. Nor was the Commission majority required to accept the advice of some members of their legal and technical staff (and no inference of bad faith can be derived from their failure to do so): The Commissioners are appointed by the President to administer the agency, the agency's staff is not.

Petitioners' second objection--that the Commission relied on extra-record evidence favorable to its position245--harbors the same defect as their first contention: This court's obligation to disregard such evidence fully protects petitioners' right to meaningful judicial review. Of course, this remedy works only if the evidence we disregard is favorable to the Commission's position, as is alleged here; supplementation might be required if petitioners made a prima facie showing that the agency excluded from the record evidence adverse to its position or that the agency's stated rationale is but a pretext masking its true basis of decision. Although the correspondence does not raise this problem,246 a similar concern is presented in petitioner's third point--that the Commission acted in bad faith.

The Supreme Court has declared that "where there are administrative findings that were made at the same time as the decision ... there must be a strong showing of bad faith or improper behavior before such inquiry [into the mental processes of administrative decisionmakers] may be made."247 Petitioners assert that the Commission denied it a hearing not because the "earthquake/emergency preparedness" issue is not "material," as required under our case law,248 but because it did not want to delay the start of power production at Diablo Canyon.249

Petitioners' allegations, while serious, have not been documented with the requisite degree of specificity to warrant either supplementing the record or reviewing the transcripts in camera. The ease with which charges of "bad faith" could be leveled, combined with the inordinate burden resolution of such claims would entail for courts, persuade us to decline petitioners' invitation to review the transcripts and to supplement the record.250

Contrary to the dissent's characterization, the allegations of dissenting Commissioner Asselstine and Representative Ottinger are conclusory. Their accusations are made with no documentation, support, or explanation beyond their own ipse dixit. What passes for proof of "bad faith" on this motion are inferences--inferences based on the speculation and conjecture of a dissenting Commissioner and a chairman of a House oversight subcommittee who have "read between the lines" of the Commission's deliberations prior to its final decision.251 But statements made as part of the collegial exchange of Commission deliberations do not necessarily represent the Commissioners' final beliefs. Petitioners' characterization of the probity of these transcripts must be further discounted by the likelihood that the authors of the letters read as much into the transcripts as they read in them. Cognitive dissonance applies with special force to "what is one of the most sensitive and difficult issues of our time: the safety of nuclear power."252

A great many parties could make similar accusations of impropriety in future litigation. On their view of the law, petitioners, and other parties to future proceedings of this sort, could successfully move for supplementation as long as one sympathetic Commissioner, Congressman, or investigative reporter was willing to attest that, based on his reading of the transcript, the agency's decision was rendered in "bad faith." Allegations of an "unbalanced presentation," "insufficient attention to arguments of the agency's staff"--indeed, virtually any accusation of irregularity in the agency's decisionmaking process--would require discovery and supplementation of the record.253

The ease with which accusations of impropriety can be made must be compared to the difficulty judges would face investigating such charges--difficulties the dissent ignores. Courts do not have a limitless capacity to review documents in camera on the off-chance that something might turn up. "In camera inspection of disputed documents places very burdensome demands on federal trial courts,"254 and even greater demands on appellate courts, which, unlike the district courts, do not have experience with discovery and factfinding on a day-to-day basis. These demands are exacerbated still further when the affected documents may be voluminous and concern technical regulatory issues such as those raised by the licensing of nuclear power plants. Moreover, every facially valid accusation would have to be accorded this burdensome review. At least when the allegations of bad faith are ostensibly proven by transcripts and memoranda, determination of bad faith requires examination of those documents in camera, effectively prejudging the result in favor of the party moving for supplementation of the record.

In addition to the burden on appellate tribunals, over-eagerness by courts to review documents in camera and to supplement the administratively-created record fails to respect the autonomy of administrative agencies and the "strong presumption" that "when administrative officials purport to decide weighty issues within their domain they have conscientiously considered the issues."255

Given the conclusory nature of petitioners' documentation and the significant costs of routine in camera review, we find their showing insufficient to warrant supplementation of the record or in camera review. Supplementation, along with logically antecedent review, can be justified only when the accusation and accompanying documentation are much more specific.

The Diablo Canyon proceedings have lasted fifteen years, involved scores of decisions by the Commission and its boards, produced thousands of pages of administrative record, and consumed countless professional years of research and study. At virtually every stage of the proceedings petitioners have challenged Commission actions with great tenacity and determination. The present appeal represents the culmination of petitioners' efforts to bring to light any legal errors the Commission may have committed. After careful consideration of the issues raised by petitioners, we affirm the legality of the Commission's action--with two minor exceptions. We conclude that the Commission erred in construing its operator license requirements to recognize training performed on computer simulators. The discrepancy between the Commission's practice and the literal language of its regulation, however, has been corrected by a recent amendment of the provision. Because the operators at Diablo Canyon can now be licensed under the amended rule, remand of the issue to the Commission would serve no useful purpose. Similarly, we conclude that a remand to allow the Commission to correct its violation of section 189(a) on issues of construction quality assurance is not necessary: the evidence petitioners would have presented had their rights to a hearing not been denied was neither material nor safety-significant. Petitioners' evidence therefore does not call into question the initial decision to grant Diablo Canyon an operating license or the safe operation of the facility.

With two minor exceptions, then--neither of which warrants judicial relief--the Commission acted well within the bounds of legal discretion accorded it by its governing statutes and regulations. These protracted proceedings may at last come to a close: we conclude that Diablo Canyon is ready to begin generating electric power for the citizens of California.

Affirmed.

WALD, Circuit Judge, concurring in part and dissenting in part:

I concur in the majority's opinion except for Part III-2 relating to the complicating effects of an earthquake on emergency planning and Part IV relating to the Petitioners' Motion to Supplement the Record. While I too am mindful of the Supreme Court's "unequivocal proscriptions against judicial overreaching" in Nuclear Regulatory Commission ("NRC" or "Commission") actions, see Maj.Op. at 1296, I do not find in those proscriptions a directive for this court to abandon, or even to weaken, its statutory duty to provide thorough and meaningful judicial review of the Commission's actions in this important area of nuclear safety. Indeed, it is difficult to identify any area where effective judicial review of agency action is more crucial than in the area of nuclear power regulation; even a small lapse in rational decisionmaking may have the most profoundly devastating effects on public health and safety. In my view, the Commission's decision not to include consideration of earthquakes in emergency planning for Diablo Canyon exhibits a substantial lapse in rational decisionmaking.

The fundamental issue is whether the Commission acted arbitrarily and capriciously in excluding consideration of the complicating effects of an earthquake on nuclear emergency response plans1 for Diablo Canyon. The Commission concluded that the failure to plan specifically for an earthquake occurring simultaneously with a nuclear emergency is not a material issue upon which affected parties have a right to a hearing under section 189(a) of the Atomic Energy Act (the "Act"), 42 U.S.C. Sec. 2239(a). See Maj.Op. at 1307. It is true, of course, that the Commission has broad authority to decide what matters are relevant to its licensing decision. See Siegel v. Atomic Energy Comm'n, 400 F.2d 778, 783 (D.C.Cir.1968); see also Bellotti v. NRC, 725 F.2d 1380, 1383 (D.C.Cir.1983) (NRC may define scope of Sec. 189(a) hearing except that "[p]ublic participation is automatic with respect to all [Sec. 189(a) ] Commission actions that are potentially harmful to the public health and welfare."). But, at the same time, this court has stressed the acutely serious nature of the Commission's duty to account to Congress and the public for its handling of "one of the most sensitive and difficult issues of our time: the safety of nuclear power." Philadelphia Newspapers, Inc. v. NRC, 727 F.2d 1195, 1203 (D.C.Cir.1984). As a reviewing court, we cannot blink the immense gravity of the stakes in this case--the safety of a nuclear power plant persistently beleaguered by unforeseen calamities since its very inception more than ten years ago.2 The extraordinary combination of calamitous circumstances displayed by the record of this plant not only invites--it compels--extreme hesitation in upholding the Commission's decision not to consider the complicating effects of earthquakes on emergency response plans for Diablo Canyon, even if a similar decision would not be considered arbitary or capricious with respect to a less plagued facility.

Initially, I believe that the denial of Petitioners' Motion to Supplement the Record with the transcripts of Commission meetings directly relating to the August 10, 1984, issuance of the full power license, see Maj.Op. at 1323, without first conducting an in camera examination of the transcripts, is an abdication of this court's responsibility to provide meaningful judicial review of the Commission's actions. The petitioners have not presented this court with mere speculative allegations of possible error on the part of the Commission. On the contrary, the petitioners have submitted a letter from Congressman Richard Ottinger, the Chairman of the House Subcommittee on Energy Conservation and Power, one of the NRC's oversight committees, stating that the transcripts and related documents, which he has read, reveal improper and possibly illegal actions by the Commission. Specifically, Chairman Ottinger states that the transcripts and related documents show that the Commission acted despite knowledge that there was insufficient evidence in the record to support its decision, that it relied on material outside the record to support its decision, and that it seriously misrepresented its actions in its brief to this court.3 Chairman Ottinger's allegations were repeated by Commissioner Asselstine, a member--albeit a dissenting one--of the NRC itself.4 I, for one, do not take such serious accusations by responsible officials so lightly that I can justify refusing even to look at the transcripts of the Commission proceedings to see if there is in fact a basis for the accusations.

While I do agree that a court should not routinely grant motions to supplement the record based on discontented parties' general allegations of misconduct, the unprecedented nature of the specific allegations of misconduct lodged by the chairman of one of the NRC oversight committees and one of the commissioners of the NRC, viewed against the disturbing background of the plant's troubled safety history, seem to me clearly sufficient to meet the special exceptions criteria allowing further inquiry by this court. Courts have recognized that supplementation of an administrative record may be justified, in the interest of effective judicial review, where there are credible accusations that an agency has relied on material outside the record or has acted improperly or in bad faith. See Public Power Council v. Johnson, 674 F.2d 791 (9th Cir.1982) (citing cases); see also Citizens to Preserve Overton Park, Inc. v. Volpe, 401 U.S. 402, 420, 91 S. Ct. 814, 825, 28 L. Ed. 2d 136 (1971) (bad faith or improper behavior exception); Natural Resources Defense Council, Inc. v. Train, 519 F.2d 287, 291-92 (D.C.Cir.1975) (exception for material relied on by agency but not submitted as part of the record). Even if the Commission's position that supplementation of the record is not warranted here were to prove ultimately correct, I see no reason not to take the middle course, i.e., an in camera examination by the court of the transcripts and related documents before making the ultimate determination to deny Petitioners' Motion to Supplement the Record. I simply believe the stakes are so high and the nature and source of the accusations sufficiently credible as to make a "business as usual"--"see no evil, hear no evil, confront no evil"--attitude on our part untenable, if we are to provide meaningful judicial review.

My view that it is incumbent upon this court to conduct an in camera examination of the transcripts is reinforced by what I consider to be the very thin support in the record for the Commission's decision not to consider the complicating effects of earthquakes on emergency response plans for Diablo Canyon. The Commission's reliance on largely unsupported conclusions and undocumented support in the record renders its decision a precarious exercise of discretion, at best. Even under the most deferential standard of review used by the majority, the Commission's decision teeters on the line between reasoned decisionmaking and arbitrary and capricious agency action. Indeed, under any verbal formulation of review, our task of making a meaningful assessment of why the Commission refused to explicitly consider earthquakes in emergency response planning is greatly impeded on a practical level here by the Commission's failure to specify the particular evidence in the record supporting the inferences it drew and the conclusions it reached. This court has previously stated:

For a reviewing court to perform [its] task, it is imperative that the Commission articulate the critical facts upon which it relies .... Similarly, when the Commission finds it necessary to make predictions or extrapolations from the record, it must fully explain the assumptions it relied on to resolve unknowns and the public policies behind those assumptions .... Only if the Commission observes these minimum standards can we be confident that missing facts, gross flaws in agency reasoning, and statutorily irrelevant or prohibited policy judgments will come to a reviewing court's attention.

Columbia Gas Transmission Corp. v. Federal Energy Regulatory Comm'n, 628 F.2d 578, 593 (D.C.Cir.1979) (footnote omitted). In my view, the NRC has not met "these minimum standards."

The Commission cites three justifications for its decision not to include consideration of the complicating effects of earthquakes on response to a radiological emergency at Diablo Canyon: the low probability that an earthquake will initiate a radiological release; the low probability that an earthquake will occur contemporaneously with an unrelated radiological release; and the flexibility of the general emergency plans developed for other natural phenomena to handle any disruptions caused by an earthquake. See CLI-84-12, 20 N.R.C. ---- (1984), reprinted in Joint Appendix (Supplement) ("J.A.S.") at 251. In my opinion, only the first of these conclusions is adequately supported by findings in the record.

A critical issue throughout the licensing proceedings, and recognized as such by the Commission, has been the adequacy of the seismic design of Diablo Canyon to withstand an earthquake, a matter to which the Commission has given considerable attention. See LBP-79-26, 10 N.R.C. 453 (1979); ALAB-644, 13 N.R.C. 903 (1981), petition for review denied, CLI-82-12A, 16 N.R.C. 7 (1982); see also CLI-84-12, J.A.S. at 257 ("The resources, time, and attention devoted to seismic design in this case have been unprecedented ....") Consequently, the Commission has adequate support on the record for its conclusion that it is unlikely that an earthquake will cause a radiological release.5 The Commission's second and third conclusions, however, appear to be based--in the main--on unexplained extrapolations from a massive and, for practical purposes, an incomprehensible mound of data, as well as what appear to be purely intuitive assumptions with no evident basis in the record.

The Commission cites no specific record support for its conclusion that it is very unlikely that an earthquake will occur contemporaneously with an unrelated radiological accident. See CLI-84-12, J.A.S. at 255 (vaguely alluding to information supplied by the parties); id. at 258 (citing ALAB-644 for the conclusion that Diablo Canyon is an area of moderate rather than high seismicity). The Commission, in its reply to Petitioners' Motion to Supplement the Record, additionally cites to decisions pertaining to the likelihood that an earthquake might initiate a radiological release to support its quite different conclusion that there is little chance the two events might occur contemporaneously. See Respondents' Opposition to Petitioners' Motion to Supplement the Record at 5 (citing LBP-79-26, 10 N.R.C. 453 (1979); ALAB-644, 13 N.R.C. 903 (1981)). Although it is possible that the record in these decisions contains sufficient data on earthquakes and radiological release accidents for the Commission to determine the mathematical probability of a contemporaneous occurrence of the two events, that was not the issue before the Commission or decided by the Commission in these cases. Consequently, the Commission has not, in any of the cited decisions, elucidated the path of reasoned decisionmaking it followed in arriving at its conclusion in this case that the probability of a contemporaneous occurrence of an earthquake and a radiological accident is very low. We are asked, and the majority agrees, to take it on faith alone that the data is there. In addition, however, a conclusion as to the low probability of the coincidence of the two events is only the first step in any rational analysis required of the Commission before it can conclude that no consideration of earthquakes is necessary.

The Commission claims that the probability of a coincidental occurrence is too low to warrant its inclusion among the other natural events for which emergency planning is required. To make this determination the Commission would need to engage in a comparative analysis of the respective probabilities of the coincidental occurrence of a radiological accident and those natural phenomena required to be considered versus the probability of the coincidental occurrence of an earthquake and a radiological accident.6 The Commission cites to nothing in the record which indicates that it took this essential second step.

The Commission's third conclusion that the emergency planning for other natural phenomena retain enough flexibility to handle any disruptions which might be caused by an earthquake appears to be based on pure conjecture. While it might be that emergency plans designed to deal with the disruption of evacuation routes or communication lines are equally effective regardless of whether the disruption is caused by a hurricane or by an earthquake, it is equally plausible that an earthquake may cause some unique problems not encompassed by the emergency plans for other natural phenomena. The Commission cites nothing in the record to suggest that the Commission considered the possibility that an earthquake might produce unique complicating effects requiring special emergency planning. This issue should have been explicitly addressed in any rational decisionmaking process.

Finally, I must confess, on a common sense level, I am completely baffled by the Commission's paradoxical stance with respect to the materiality of earthquakes. The Commission has considered earthquake associated risks of paramount importance with respect to the design of the Diablo Canyon plant, yet with respect to emergency planning the Commission asserts that earthquake associated risks are too unlikely to warrant consideration. Compounding the illogic of the Commission's position is the fact that the nature of emergency planning is to develop plans to respond to unlikely but possible emergencies.7 Certainly if earthquake associated risks are significant enough to require redesigning the plant to meet seismic design criteria twice that of most other plants, see supra at 1332-33 & n. 5, they are significant enough to require emergency planning.8

In sum, the Commission's decision not to include consideration of the complicating effects of an earthquake on emergency planning for Diablo Canyon rests in substantial part on predictions or extrapolations purportedly derived from the record. Yet the Commission's decision asserts only the Commission's ultimate conclusions with no systematic articulation of the factual data and assumptions relied upon or the path of analysis followed. This state of the record, together with the serious charges levelled by Commissioner Asselstine and Chairman Ottinger, who based on their personal familiarity with the transcripts of the proceedings concluded that the Commission acted in knowing disregard of the insufficiency of the record to support its decision, leads me to seriously question the ability of this court to provide meaningful judicial review of the Commission's action regarding earthquakes and emergency planning. In my view, the answer is ineluctable: There is no way that this court can be confident that the Commission's decision is not based upon "missing facts, gross flaws in ... reasoning, and statutorily irrelevant or prohibited policy judgments." Columbia Gas Transmission, 628 F.2d at 593.

As the majority correctly points out, if effective judicial review is frustrated by an agency's failure to provide adequate explanation of its action or if its action lacks a sufficient basis in the record, the proper course for the court is to remand to the agency for further proceedings. See Maj.Op. at 1326. This is the course that I would follow here.9 I would vacate the Commission's decision not to include consideration of earthquakes in emergency planning for Diablo Canyon and remand to the Commission for further proceedings. Since, based on the foregoing analysis, I do not believe that the Commission can set forth a rational basis for excluding consideration of the complicating effects of an earthquake on emergency response at Diablo Canyon, the Commission on remand should be required to include consideration of earthquake complications in emergency planning not merely re-address the question of whether earthquake complications warrant consideration. In my view, as long as the Commission instituted such proceedings immediately and completed them in an expedited manner, a stay of the operating license pending completion of the proceedings would not be required. But I believe the strong public interest at stake here--ensuring the safety of those working at or living in the vicinity of the Diablo Canyon nuclear power plant--compels the Commission to address forthrightly how the emergency plans for a radiological accident at Diablo Canyon will take account of any simultaneous seismic activity. Any thoughtful inhabitant of this planet during the past century has surely learned to expect the unexpected, to anticipate the unthinkable, and to recognize the imperative need to plan for saving human life in the event of a mass disaster. I believe that the Commission's outright refusal to make explicit provision in emergency response plans for an earthquake in a nuclear plant within three miles of a major, active fault in California is by definition an arbitrary and capricious act. I therefore dissent from the majority's affirmance of the Commission's decision in this respect.