with whom
joined TAMM and ROBB, Circuit Judges (dissenting): *We are called upon in these cases, consolidated for purposes of argument and deci*443sion, to review regulations promulgated by the Environmental Protection Agency (EPA) pursuant to section 211(c)(1)(A) of the Clean Air Act.1 The regulations provide for the phased reduction of the lead content of motor vehicle gasoline. By 1979 the regulations will permit refiners to manufacture gasoline which has an average lead content of no greater than 0.5 grams of lead per gallon.
In these cases EPA is acting under authority of section 211(c)(1)(A) of the Clean Air Act which permits EPA’s Administrator to prohibit, control, or regulate a fuel or fuel additive if “any emission products of such fuel or fuel additive will endanger the public health or welfare . . . .” EPA’s authority to promulgate these regulations must be distinguished from EPA’s authority and action reviewed in this court’s opinion in Amoco Oil Co. v. EPA.2 In Amoco *444we had before us regulations which prohibited use of leaded gasoline in automobiles fitted with the catalytic converter for controlling exhaust emissions and required retail marketing of at least one grade of unleaded gasoline for the use of such automobiles.3 Those regulations were issued under the authority of section 211(c)(1)(B) of the Clean Air Act which permits regulation, control, or prohibition of a fuel or fuel additive “. . .if emission products of such fuel or fuel additive will impair to a significant degree the performance of any emission control device or system which is in general use, or which the Administrator finds has been developed to a point where in a reasonable time it would be in general use . . . .”4
The petitioners challenge the regulations on two basic grounds. They contend the Administrator incorrectly interpreted section 211(c)(1)(A) and as a result used an improper legal standard in making the determinations upon which the regulations are based. More importantly, they also contend that the evidence in no way supports EPA’s conclusion that the public health is or will be endangered, and that the case against auto lead emissions is a speculative and inconclusive one at best.5 On reargument the dissenting judges conclude that the disagreement with regard to the legal standard reduces itself to semantics, and while the Administrator did not couch his principal determinations in the language of the statute, the language that he did use, properly interpreted in accordance with the statute, did provide a sufficient standard for his determinations. Taking the Administrator’s own phrasing of his statutory authority, we are convinced that the paucity of scientific evidence pointing to any firm conclusion, the gaps in the logic supporting the Administrator’s analysis, aggravated by the faulty procedure behind the promulgation of the regulations and the Third Health Document, render the issuance of the challenged regulations arbitrary and capricious and procedurally infirm.6
I. THE THREE-YEAR HISTORY OF EPA’S EFFORTS TO DEVISE ACCEPTABLE REGULATIONS
A. The Regulations and Supporting Health Documents
The regulations we are reviewing lessen the use of certain lead compounds, known as lead antiknocks, in motor vehicle gasoline.7 These compounds control a phenomenon known as knock — the detonation of gasoline in the cylinder prior to maximum compression. Knocking places a limitation on the efficiency and economy of the internal combustion engine because maximum power is only obtained when the explosion in the cylinder occurs at that point when fuel and air are most compressed. Lead antiknock compounds raise a fuel’s resistance to knock (known as' the octane of the fuel), thus permitting higher compression engines, which in turn are more efficient and economical in their use of fuel. The refiners claim that to achieve the same antiknock qualities without lead additives would require greater amounts of scarce crude oil. In addition, they claim that lead additives provide them with the flexibility to produce gasoline with differing antiknock qualities (differing octane ratings with the minimum of refining and processing equipment.
The final regulations were accompanied by a detailed health position paper, entitled “EPA’s Position on the Health Implications *445of Airborne Lead” [hereinafter referred to as the Third Health Document].8 The document was intended to be an authoritative examination of the “. . . pertinent evidence upon which a decision could be made as to whether or not there is a health justification to regulate lead in gasoline.”9 We shall refer extensively to this document in reviewing the evidence in Part IV below.
The Preamble to the Regulations contains an extended discussion of EPA’s health concern. It focuses separately upon adults and children. While conceding that the available evidence does not show that lead in auto emissions alone adversely affects the health of either adults or children, EPA argues that auto lead emissions make a significant cumulative contribution to human lead exposure. EPA declares that it would be “prudent” to limit this source of lead exposure, because it is a component of overall lead exposure susceptible to human control. Much of the discussion in the Preamble to the Regulations relates to whether airborne lead from auto emissions actually does get into the human body to a significant degree. (That portion of the Preamble that refers to EPA’s health concern is reproduced as an appendix to this opinion.) EPA contends that airborne lead is respired by persons in certain occupational groups and contributes to their high blood lead levels. It is also contended that young children ingest dust and dirt containing lead from auto emissions which contributes to excessive blood lead levels among the young. Petitioners take issue with both forks of EPA’s argument, pointing out that the evidence is inconclusive on both issues.
B. Adverse Nature of Government Expert Scientific Comment
The present section 211 was added to the Clean Air Act on 31 December 1970.10 Section 211(c)(1) authorizes the EPA Administrator to
control or prohibit the manufacture, introduction into commerce, offering for sale, or sale of any fuel or fuel additive for use in a motor vehicle or motor vehicle engine (A) if any emission products of such fuel or fuel additive will endanger the public health or welfare . . . .11
Within a month of the enactment of the Clean Air Act EPA issued an advance notice of proposed rulemaking.12
Almost three years passed before the final regulations were promulgated. This extended gestation period has significance in itself. Implicit in the administrative record generated by this three-year delay is the recognition by EPA that available scientific data did not provide a clear and certain basis for reaching the statutorily mandated conclusion, i. e., that a “fuel additive will endanger the public health or welfare.”
What the gestation period of these regulations would have been without the intervention of this court, we will never know, for the decision to issue the regulations was precipitated by the third in a series of orders of this court in Natural Resources Defense Council v. EPA13 on 29 October 1973, directing the agency to reach a final decision within thirty days. Out came the final regulations on 28 November, accompanied by the Third Health Document as the supporting scientific rationale for the regulations. The history of the regulations is really the history of the EPA Health Documents, a history of EPA’s effort to discover somewhere, somehow, a scientific rationale which would withstand the unanimous criticism of the remainder of the government scientific community.
In February 1972, over a year after the prompt advance notice, the Administrator *446issued a notice of proposed rulemaking under section 211(c)(1)(A).14 In the notice the Administrator stated the following conclusion: “that airborne lead levels exceeding two milligrams per cubic meter, averaged over a period of three months or longer, are associated with a sufficient risk of adverse physiologic effects to constitute endangerment of public health.”15 EPA’s first health document, entitled “Health Hazards of Lead,” was issued contemporaneously. Subsequently, an additional paper entitled “Atmospheric Lead and Public Health” was issued by EPA to bolster further EPA’s conclusion.16
In the ensuing year EPA received a great deal of adverse comment as to its health position, including critical comment from every other department of the Government asked to comment thereon. We regard this of considerable significance, as each of these departments has its own independent scientific staff, and presumably is in that group of “independent” experts to which EPA itself stated (rightly or wrongly) that it would give more credence than to either industry scientists or environmental crusaders.17 Dr. Edward E. David, Jr., Director of the White House Office of Science and Technology, wrote:
. OST has reviewed the medical evidence that supposedly supports the EPA Administrator’s desire to reduce lead content of gasoline. OST finds the Administrator’s position in this regard unsupported by the evidence.
With respect to the medical evidence there is no correlation between airborne lead and lead levels in blood for the range of airborne lead levels encountered in rural or urban communities. . . 18
In “Airborne Lead in Perspective,” the National Academy of Science (NAS) made a thorough study of the scientific literature as to lead in the air and its biological effects [hereinafter referred to as the NAS Report]. After preliminary comparison of the NAS Report and EPA’s health grounds, the Under Secretary of the Interior, M. J. Pécora, gave his evaluation:
After examining the National Research Council report and that of the background information of EPA, it seems that EPA’s conclusions are strongly oriented toward a one-sided view of the report’s findings. In general, the Research Council indicates there is no basis for restriction of lead in gasoline from a health standpoint.19
The Department of Commerce commented:
It therefore appears that EPA’s position that any amount of airborne lead constitutes an endangerment of public health is largely conjectural and not adequately *447premised on available scientific or medical information.20
Perhaps the most biting comment came in a letter from Dr. Merlin K. DuVal, Assistant Secretary for Health, Department of Health, Education and Welfare:
The decision having been made on grounds other than those having to do with hazard to the public health, your staff now wish to explore with us the question of whether or not hazard to the public health could be invoked as a reason for accelerating the implementation date of the primary decision. It was our firm view that there is no firm evidence, at this time, that lead poisoning in humans could be traced per se to the presence of lead in gasoline, indeed, if there were, this would have constituted justification for the elimination of lead as an additive in the first instance.21
As a result EPA was forced back to basics, asking for public comment on virtually every conclusion it had reached as to lead in the air and its relation to man.22 On 4 *448January 1973 a new notice of proposed rule-making was issued, in which EPA repudiated its earlier conclusions and reached the following weakened appraisal:
Though none of the above findings viewed individually and in the context of possible experimental error can be taken as conclusive evidence that airborne lead by itself is a current public health problem, considered together, they do suggest that airborne lead is contributing to excessive total lead exposures among the general urban population. In light of this evidence, the Administrator has concluded that it would be prudent to reduce preventable lead exposures from automobile emitted “airborne” lead to the fullest extent possible.23
This latest EPA position was supported by a Second Health Document.24
Despite these changes in EPA’s position, criticism from other governmental agencies continued. For example, Dr. Charles H. Powell, Assistant Director of the National Institute for Occupational Safety and Health of the United States Public Health Service, evaluated the reproposed regulations and the Second Health Document, as follows:
We have not been persuaded, however, that reduction in environmental lead by its reduction in gasoline will result in a significant improvement in the health of the public. This statement is based on our review of the problem and discussions with responsible scientists in the Health Services and Mental Health Administration (NIOSH and BCEM) and the Food and Drug Administration.
This position has been reconsidered by a review of the report on health effects of airborne lead submitted with your letter. The report is often speculative and is based in significant part on unpublished information, which we have not been able to review. We have been able to review some of the abstracts of the October 1972 meeting in Amsterdam, but this has not been sufficient to allow evaluation of the content of the papers.
A specific point that has been commented on in previous drafts of this paper is the description of lead absorption in greater than normal amounts, as reflected by higher than normal blood lead levels, as asymptomatic lead poisoning. In the present version, this description is attributed to unnamed clinicians and health departments, whose knowledge, reputation, and expertise cannot be evaluated. It invites unfavorable inferences.25
Acting Secretary of the Interior John C. Whitaker stated, “After reviewing the re-proposed regulations, we see no reason to change from our position as detailed in our enclosed analysis. . . ,”26
Secretary Elliot L..Richardson of Health, Education, and Welfare (HEW), in a letter to Senator Tunney, took issue with EPA’s revised health justification for the regulations:
*449Our view is:
(a) there is no firm evidence at this time that lead derived from combusted gasoline is harmful to the health of the general public, though it may, under certain circumstances, contribute to the total body burden of lead and may, as a consequence, reduce somewhat the physiologic reserve against lead poisoning from other sources . . . 27
At this point it is perhaps appropriate to point out that we are not substituting our judgment for that of the Administrator; as members of a reviewing court we are saying that his procedural process was faulty, and that on the data he had available to him any conclusion could be only speculative. We are not disputing a supposed expert’s conclusions on his own ground, but it is readily apparent from the recital of the other government agency comments that we would have plenty of company in our judgment if we did. Our analysis of the flaws in the Administrator’s logic finds powerful support in the unanimity of conclusion of the independent scientific minds concerned throughout the Government outside of EPA itself. For the collective scientific judgment in every department or agency of the Government commenting on the first two EPA Health Documents and proposed regulations is dead against the Administrator’s conclusions — and he did not risk the same fate with the Third Health Document.28
On 28 November 1973 the final regulations here under review were promulgated after an order of this court directed the Administrator to reach a final decision on *450the matter within thirty days.29 These final regulations were accompanied by the Third Health Document, whose subjection to public- — or other government official— comments will be now explored.
C. The Opportunity for Informed Public Comment on the Third Health Document
The Administrative Procedure Act (APA) provides that the notice of rulemaking shall include “either the terms or substance of the proposed rule or a description of the subjects and issues involved.”30 Notice was given at the outset of the three-year rule-making period, and also early in 1973 when in response to comments EPA reproposed the regulations with slight changes. In the fall of 1973, when the change in the regulations from leaded pool averaging to total pool averaging was about to be made, EPA notified interested parties and comments were received on this. As will appear from our detailed analysis later, nearly all of the comment received in October-November 1973, so stressed by the majority opinion, related to the economic impact of the regulations and total pool averaging, not to the health issue involved on this appeal.
Pertinently, the APA also provides in section 553(c):
After notice required by this section, the agency shall give interested persons an opportunity to participate in the rule máking through submission of written data, views, or arguments with or without opportunity for oral presentation. After consideration of the relevant matter presented, the agency shall . . . .”
Pursuant to this statutory provision, ample opportunity was given interested private parties and government agencies to comment on the original proposed regulations, the regulations reproposed with slight modifications, and the first two Health Documents. The last formal period of comment, on the Second Health Document and the reproposed regulations, closed 11 March 1973. It is for the subsequent period that the question arises as to whether the agency gave “interested persons an opportunity to participate in the rulemaking through the submission of written data, views, or arguments . . more precisely, whether an opportunity was afforded to submit written data, views, or arguments specifically directed to the Third Health Document and the new studies EPA was preparing to embrace therein.
Three matters seem beyond dispute:
(1) The Third Health Document was issued simultaneously with the final regulations on 28 November 1973.
(2) No formal period of comment was announced on the Third Health Document, although a “draft Third Health Document” was apparently given at least a limited circulation on 19 October 1973.
(3) The independent scientific minds in the other government agencies, whose devastating critical comments on the first two Health Documents have been discussed above, made no comment on the Third Health Document.
It is highly significant that nowhere in the fourteen-page detailed discussion of the EPA procedures at the close of their opinion do our colleagues maintain that such independent government agency scientific comment was received by EPA, although our colleagues go to great lengths to establish notice in fact and comment by some of the private industry petitioners upon portions of the scientific studies which eventually made up the Third Health Document.
Two matters appear very much in dispute:
(1) The extent to which the Administrator relied upon new scientific data, appearing as EPA sponsored for the first time in the Third Health Document, as the basis for the final regulations.
(2) The extent to which interested private parties, some of whom are petitioners here,
*451(a) were given or acquired actual notice of the existence of the scientific studies,
(b) were given notice as to which studies were to be relied upon by the Administrator in the Third Health Document, and
(c) commented thereon prior to the issuance of the Third Health Document itself.
We set forth the record facts on these two issues as briefly as is consistent with clarity and accuracy:
1. The extent to which the Administrator relied upon new scientific data, appearing as EPA sponsored for the first time in the Third Health Document, as the basis for the final regulations.
The Preamble to the Regulations (attached to this opinion) makes clear that when the Administrator decided, he relied materially upon studies which had not before been subjected to public criticism. In the Administrator’s own words, “A discussion of the four major areas of criticism and a summary of the significant new information received since the regulations were reproposed are provided below.”31 In this Preamble, after the Administrator posed the central evidentiary question — Is there a correlation between air lead levels and blood lead levels? — he then reviewed the critical comment received by EPA on the studies it had been using to support the existence of a correlation, and then notably made no attempt to defend the earlier studies. Instead, he “weighed against these criticisms [new] studies which have shown that airborne lead does contribute significantly to lead exposure in the general population.” 32
The Preamble then launches into a discussion of six studies relating to lead as a hazard to the general public — two pilot isotope studies, “[a]n unpublished study in Japan,” the chamber studies, the Seven Cities study, and the Daines study. These were followed by several studies relied on for the Administrator’s conclusions concerning children in urban areas — the Newark studies (two parts), and the Chicago, Philadelphia, and Rochester (N.Y.) studies.
(a) Pilot Lead Isotope Studies — In the Third Health Document EPA relies on two pilot lead isotope studies by the same authors — Rabinowitz, Wetherill, and Kopple. The first of these studies is an unpublished paper presented to a scientific conference at Raleigh, North Carolina, on 1-2 October 1973.33 The preliminary results of this study were reported to EPA in a letter dated 28 August 1972, which was filed and indexed with the public comments at that time.34 This letter, however, emphasized the preliminary nature of the reported results,35 and there is no indication that EPA received a more complete and final version of the study before its presentation at the October 1973 conference. After the conference, on 4 October 1973, abstracts of the paper were placed in EPA’s public comment file and sent to Ethyl Corporation under an *452agreement resulting from a Freedom of Information Act (FOIA) suit brought by Ethyl.
The second pilot lead isotope study was published in the November 1973 issue of Science magazine.36 An early draft of this study was sent to EPA by one of its authors on 5 May 1973.37 The revised and final draft of this study was not received by EPA, nor sent to Ethyl under the FOIA agreement, nor placed in the public comment file until sometime after 6 August 1973.38 Thus, the two pilot lead isotope studies, in the forms relied upon in the Third Health Document, did not reach Ethyl under the FOIA agreement or anyone else via EPA’s public comment file until at least 4 October 1973 and 6 August 1973 respectively.
(b) Unpublished Study in Japan —Significantly, the cover page of this study bears the following inscription: “Draft/Do Not Cite or Quote.”39 This unpublished study was transmitted to Ethyl under the FOIA agreement and placed in the public comment file on 22 July 1973,40 but it is not even cited or referred to in the Third Health Document. Its absence from the Third Health Document is critical because the purpose of EPA’s Health Documents was notice — notice of the particular items of evidence (i. e., studies) on which EPA intended to base its regulations. Only with such notice would interested parties be able to make meaningful1 and effective comments.
(c) The Chamber Studies — The chamber studies41 were cited in the Second Health Document, but were only available at that time in preprint form. The Third Health Document relies upon the printed report which is dated May 1973.42-
(d)The Seven Cities Study ■ — The Preamble’s discussion of blood lead levels of urban and suburban residénts and between cities probably refers to both the Seven Cities study and the Daines study. The Preamble earlier had referred directly to the Seven Cities study as an example of previous EPA data which had been cited by other experts as contrary to EPA’s own conclusions:
For example, the Seven Cities Study did not show a close correlation between increase in blood lead levels and simultaneous increases in air lead exposures. Blood lead levels were lower among the New York City residents studied than the Philadelphia residents, despite the fact that air lead exposures among the New York residents were actually greater than those in Philadelphia.43
Both the Preamble and the Third Health Document state that “EPA has reanalyzed the Seven Cities study and has found that air lead was a significant, though not the most influential factor affecting blood lead levels.”44 This “Additional Analyses of the Seven City Lead Study” bears a 15 March 1973 date.45
The majority opinion asserts that EPA made the reanalysis public at an EPA committee meeting on 26 February 1973 — thirteen days before, instead of four days after, the close of the formal comment period on the Second Health Document (11 March 1973)46 The majority also seems to attach *453great importance to the fact that “[Representatives of Ethyl attended the meeting, prepared a transcript of it, including critical comments by some of those in attendance, and forwarded the transcript to the Administrator.” 47 Of course, this EPA committee meeting with only one of the petitioners in this lawsuit present did not notify other interested parties (or, for that matter, even Ethyl) that EPA intended to rely on this reanalysis to support its regulations. Such a limited gesture adheres to neither the letter nor the spirit of the APA’s notice provision.
(e) The Daines Study — The Daines study had been published in Industrial Medicine in October 1972. It and the chamber studies represent the only studies mentioned in the Preamble to appear in equivalent form in both the Second and Third Health Documents.
(f) The Newark Studies —This study was done in two parts by the same investigators. After studying blood levels of children in various areas of Newark, the first paper of May 1973 was unable to relate elevated blood levels to traffic or traffic density, or to any other variable source. The investigators were with EPA in spirit, though, even if not with the facts, for they wistfully reported, “We only wish our results were more positive.”48
•Petitioners strongly attack the Newark study. They point out that in May 1973 “the investigators found no statistically significant correlation between blood lead levels and proximity to traffic or traffic density . . .”49 Yet in November 1973, by “increasing the amount of data in the study through the inclusion of another group of children with blood lead levels less than 40,” the investigators suddenly were able to find the statistically significant correlation that eluded them in May.50 EPA concedes that it was only upon the submission of new data and reanalysis that the study was able to provide support for EPA’s dirt and dust hypothesis by November 1973; however, EPA contends that “[petitioners make no argument why this additional data may not legitimately be included.”51 What EPA conveniently forgets is that the second paper was only in press in November 1973, and had not even been published when the regulations were issued. To our minds, if one pays any deference to primary principles of administrative procedure, this is a powerful argument against EPA’s reliance on this data.
(g)The Chicago, Philadelphia, and Rochester Studies —In both the Preamble and in the Third Health Document itself heavy reliance is placed on the studies from Chicago, Philadelphia, and Rochester (N.Y.), as well as the studies from Newark.52 The results of the Chicago screening,53 the Philadelphia study,54 and one of the two Rochester studies55 were first presented in papers at the October conference iii Raleigh, North Carolina. The other Rochester study,56 like the Newark study above, was in press at the time the regulations and the Third Health Document were promulgated by EPA.
According to the Third Health Document, “studies from Philadelphia, Chicago, and Newark discussed above provide persuasive *454evidence to strongly suggest that sources of lead other than paint, including that resulting from the presence of lead in gasoline, play an important role in childhood lead exposure. These other sources may be especially significant at levels of exposure below overt clinical poisoning.”57 Thus, the data allegedly supportive of the Administrator’s dust and dirt hypothesis on the health hazard to children was not available until late 1973, as set forth in detail above.
These studies were absolutely crucial to the Third Health Document and to the Administrator’s entire conclusions in regard to the lead danger to urban children. Each of these studies — Newark, Philadelphia, Chicago, and Rochester — came to EPA in October-November 1973 after the orders of this court and shortly before promulgation of the regulations on 28 November 1973.
2. The extent to which interested private parties, some of whom are petitioners here, were given or acquired actual notice of the existence of the scientific studies, were given notice as to which studies were to be relied upon by the Administrator in the Third Health Document, and commented thereon prior to the issuance of the Third Health Document itself.
The Administrator argues, and our colleagues confidently assert, “There is nothing in Section 4 [5 U.S.C. § 553] that requires new notice whenever the agency responsibly adopts the suggestions of interested parties.”58 This is a salient point. The Administrator did not adopt the suggestions of the interested parties. His prior conclusions expressed in the First and Second Health Documents and in the regulations were diametrically opposed by all of the independent scientific minds in other government agencies, and by most of the affected industry, as represented by petitioners herein. Rather than adopting the suggestions of interested parties, the Administrator persisted in his conclusions as expressed in the regulations, although he did repropose the regulations (with slight modifications) on 10 January 1973 along with the Second Health Document.
More importantly, by October-November 1973 he was largely shifting his ground from the discredited scientific data of the First and Second Health Documents to the new data embraced in the Third Health Document. This is the crucial time at which informed comment from the best scientific minds in other government agencies and elsewhere should have been sought— unless EPA was irrevocably resolved to promulgate the restrictive regulations on lead which it had originally proposed years before, in spite of the barrage of unanimous critical comment from other government scientific minds, as well as interested outsiders.
With relation to the precise dates on which the scientific information later incorporated in the Third Health Document could possibly have become available to anyone, we summarize what the record shows — or fails to show.
(a) Pilot Lead Isotope Studies and
(b) Unpublished Study in Japan — Since the formal period for comment expired 11 March 1973, and since the final regulations accompanied by the Third Health Document were promulgated on 28 November 1973, the dates when the two pilot lead isotope studies and the unpublished study in Japan were first placed in EPA’s “public file” and transmitted to Ethyl under a FOIA agreement — 4 October 1973, 6 August 1973, and 22 July 1973, respectively — indicate the limited opportunity available for effective comment to be received by EPA. Moreover, even these late dates do not fully reveal the inadequacy of EPA’s compliance with section 553 since (1) filing papers un*455der obscure headings59 in a so-called “public file,” (2) transmitting information to one interested party under a FOIA agreement, and (3) accepting a transcript from a scientific conference can hardly be equated with the public notice and comment mandated by the APA.60 It is not surprising that nowhere in our colleagues’ minute discussion of comments on all topics which were received by the Administrator do they claim that anyone besides Nalco, private or government, directed any comments (other than the conference transcript) to EPA on these three new studies. Yet EPA relies heavily on these studies in the Third Health Document. In fact, these are the first studies mentioned by the Administrator himself in his Preamble to the Regulations.
(c) The Chamber Studies — This study was available in preprint form at the time of the Second Health Document. However, since public access to preprints is generally quite limited, the study probably was not available to the public in a meaningful sense until it was published in May 1973.
(d) The Seven Cities Study — The critical feature here, of course, is the “Additional Analyses of the Seven City Lead Study” of 15 March 1973. Besides the one EPA committee meeting,61 which representatives of Ethyl attended, we do not know what, if any, opportunity for comment on the reanalysis was given other petitioners or interested parties. Since it was the “additional analysis” only which gave any statistically significant support to the Administrator’s conclusions, the pertinent, relevant, and material question here is whether administrative due process was followed regarding this additional analysis after the close of comments on the Second Health Document.62
We conclude that the reanalysis marked a dramatic shift in the EPA’s appraisal of the Seven Cities study — not, as the majority suggests, a mere reconfirmation of the validity of earlier data or a determination “that comments received during a preceding comment period [did] not undermine the validity of a particular piece of evidence.”63 The Preamble accentuates the importance of this shift by its reference to the Seven Cities study itself (before EPA’s reanalysis) as an example of data cited by other experts as contrary to EPA’s conclusions.
(e) The Daines Study — Adequate opportunity for comment was afforded because this study was in the same form in both the Second and Third Health Documents.
(f) The Newark Studies• — The first part gave no support to the Administrator’s conclusions, but in the second paper new and additional data was brought in. The second part was in press in November 1973 and had not even been published when the regulations were issued. It is indisputable that no opportunity for comment by anyone on this important second study was afforded.
*456(g) The Chicago, Philadelphia, and Rochester Studies — Three of these four studies were first presented as conference papers on 1-2 October 1973, and the second Rochester study was still in press at the time the final regulations and Third Health Document issued. If any meaningful opportunity for comment on these studies was afforded, it is not reflected in the record.
Our colleagues attempt to minimize the importance of the Newark, Chicago, Philadelphia, and Rochester studies as a basis for the Administrator’s conclusions, through a reading of the Administrator’s Preamble designed to escape the fact that the Administrator was relying materially on these new, uncommented upon studies. The court’s opinion suggests that “[t]he only clear references to these studies, which bear on the validity of the dustfall hypothesis, occur in a portion of the Administrator’s decision that addresses the question, ‘What new information has become available since reproposal of the regulation and as a result of the additional comment period?’ ”64 Accordingly, our colleagues conclude that “[tjhese additional studies play[ed] no role in the Administrator’s decision to regulate.”65
That may be the “only clear reference” to our colleagues, but open eyes can read more. These studies were not just listed in the “new information” sections; they were discussed in the reliance portion, too. A comparison of the two critical sections of the Preamble reveals that the Administrator’s language describing the evidence he is relying upon (found in subsection E of section II — entitled “Does dust lead contribute to lead poisoning in children?”) is repeated virtually verbatim in section V’s description of “[w]hat new information has become available since reproposal of the regulation ,?”66 Hence, we logically conclude that the Administrator is referring to the same evidence in both places.
We are as well aware as are our colleagues that practical administrative procedure requires that at some point the responsible agency emerge from the decisionmaking process, and act. By act we mean that the agency should take such action as is supported by the evidence in hand, whether that action is to change drastically the petroleum refining industry and ultimately the type automotive engine manufactured in this country, or simply to let this portion of the economy remain unchanged until and if further evidence calls for changes.
We do not contend, as our colleagues suggest, that new information brought to the agency’s attention too late to be made available during the formal comment period must be ignored or that an additional formal comment period is required whenever a review of recently developed information identifies new studies the Administrator finds supportive of his conclusions.67 To the contrary, we only submit that a responsible Administrator would not materially rely on recently acquired, uncommented upon studies — especially when the results of previous studies had been undermined severely by the unanimous criticism of other independent government agencies. While the Clean Air Act does direct the Administrator to consider “all relevant medical and scientific evidence available to him”68 before he issues a regulation, this instruction must be read in context with the procedural requirements of the APA.69 When he “considers” medical and scientific evidence under the Clean Air Act, he “con*457siders” in accordance with the Administrative Procedure Act. The Clean Air Act contains no exemption from the APA.
Indeed, there appears to be little, if any, difference in our position on EPA’s responsibilities under the APA and our colleagues’ position on this subject. We are not attempting to develop a novel procedural theory or “to transmute the three-step process established by § 4 [APA § 553] into a potentially unending and fruitless series of notices, comments, and notices of intent to rely on comments.”70 Our colleagues excuse EPA’s procedural shortcomings because, in their view, all' new studies were made available to the public for comment at least three months before the Administrator reached his decision.71 This, and not the notice and comment requirements of the APA, is the fundamental point of disagreement between the majority and the dissent in this case. As we have made abundantly clear, there was no meaningful opportunity for informed public comment on the Third Health Document. Rather than contending that “the public [must] have yet another opportunity to comment on [the new studies],” 72 we submit only that the public is entitled to one such opportunity if the Administrator intends to rely materially on this new, uncommented upon evidence. In other words, we submit that EPA may not keep secret information important to its decision, i. e., the evidence on which it intends to rely materially. As our colleagues concede,73 the authorities cited at 450-452 nn.78-80 of 176 U.S.App.D.C., at 85-86 nn.77-79 of 541 F.2d infra support this proposition. Final agency action must be explained in terms of a record to which all interested parties have had an opportunity to contribute informed comment.
It is important to understand that we fault EPA not because all evidence “was not specifically delivered to or called to the attention of petitioners when it arrived at EPA,”74 but because certain new studies, on which the Administrator intended to rely materially in the Third Health Document, were never brought to the public’s attention in a manner that made informed public comment possible. EPA gave no notice indicating which studies it intended to embrace (1) by placing every shred of lead-related evidence which it received in a public file or (2) by sending all of this evidence to one interested party pursuant to a FOIA agreement. Absent some form of notice indicating on which studies EPA intended to rely (e. g., the type of notice furnished by the First and Second Health Documents), we are at a loss to understand how interested parties could formulate effective comments. Apparently our colleagues require the interested public to comment on all the evidence received by EPA or to guess which evidence the Agency will embrace and which it will ignore. In our view, the procedure, approved by a majority of this court, affords no opportunity for informed public comment.75
Our position does not conflict with the following, previously stated position of this court:
In order that rule-making proceedings . be conducted in orderly fashion, information should generally be disclosed as to the basis of a proposed rule at the time of issuance. If this is not feasible, as in the case of statutory [or judicial] *458time constraints, information that is material to the subject at hand should be disclosed as it becomes available, and comments received, even though subsequent to the issuance of the rule — with court authorization, where necessary.76
In fact, one page earlier in the opinion containing this quotation Judge Leventhal states, “It is not consonant with the purpose of a rule-making proceeding to promulgate rules on the basis of inadequate data, or on data that, critical degree [sic], is known only to the agency.”77 Our problem with the data described in the Preamble and the Third Health Document does not lie in the fact that this data was too recently acquired or too late, per se. Instead, we base our objections on the identical grounds articulated by Judge Leventhal in Portland Cement, i. e., EPA’s data was “inadequate” (partly because of its inconclusive nature and lack of scientific validity and partly because it had not withstood the test of public scrutiny) and to a “critical degree, [was] known only to the agency.”
When this court issued its order on 29 October 1973 for EPA to act within thirty days on the question of regulating lead additives in gasoline, we did not hint in what manner the Administrator should act, but presumably we intended that the Administrator should act on tested data he had been accumulating for some three years. Bearing in mind the fate EPA’s previous data had suffered at the hands of the knowledgeable scientific community, it was not a responsible choice for the Administrator to resort to the most recently acquired, mostly never previously revealed, and never commented on data as the principal basis for his action.78 It is particularly difficult to square the majority’s analysis of the procedural issues in this case with the following language from a 1974 law review article by Judge Wright:
If a particular rule rests on an extensive analysis of data or on a complex prediction . . . courts should require the agency to operate the three-step procedure of § 553 in a manner sensitive to the empirical complexities at stake . Most important, the agency should not rely on any research methods or data which were not presented to the interested parties for comment or criticism.79
Similarly, in an adjudicatory proceeding the Supreme Court has commented,
A party is entitled ... to know the issues on which decision will turn and to be apprised of the factual material on which the agency relies for decision so that he may rebut it. Indeed, the Due Process Clause forbids an agency to use evidence in a way that forecloses an opportunity to offer a contrary presentation.80
Much of the data described in the Preamble and the Third Health Document was data which had been shielded from adverse comment by the factor of time of acquisition. The fate of previous submissions to public scrutiny is an indication of the type comment this new data might likely have received. While agreeing that at some point the Administrator must cease deciding and act, no valid administrative procedure would countenance launching an enterprise drastically affecting a substantial portion *459of our economy on such a shaky foundation as that relied upon in this case.81
The interested private parties had some fear that just this might occur. On 19 October 1973, following the conference in early October at which several of the studies later relied on by EPA first were revealed, EPA “circulated” a “draft Third Health Document.”82 Exactly to whom this draft was circulated, the record does not reveal. Ten days later came this court’s order, making certain that some action on some basis would be taken by EPA within thirty days. On 19 November 1973 counsel for petitioner E. I. Du Pont requested that “in the event that these regulations . are based upon different scientific and technical arguments and data than those enunciated in the January 10, 1973, statement . [they] be given an opportunity to comment on the new basis for the regulations in writing and at a public hearing prior to their final promulgation.”83
Our colleagues assert that such a request was an effort to evade the mandate of this court that regulations issue within thirty days. We think that petitioner Du Pont recognized the same point that we make here, that if indeed the Administrator had discovered a new basis for the regulations, then responsible, practical administrative procedure demanded that the interested parties be given the same opportunity to *460comment on the new basis as they had on the old. It certainly, as a practical matter, does no good for an agency to propose an action, support it with data which is severely criticized, abandon that supporting data for new data, fail to subject the new data to informed comment, and then promulgate the same proposed regulations on the basis of the new data.84 Nor is it valid thereafter to claim that full opportunity for comment has been afforded.
The court’s opinion also asserts that “[t]he early October preparation and circulation of a draft decision and a draft Third Health Document, both in obviously near final form, beliefs] [any] contention that this court’s 30-day order . . . forced EPA to rush into a decision it was not yet ready to make.”85 Aside from the fact that the record fails to reveal who received copies of these eleventh-hour documents, we submit that it was not at all obvious that they were in “near final form.” The cover page of the draft Third Health Document carried the following inscription bearing on its finality: “DISCLAIMER: THIS DOCUMENT IS A PRELIMINARY DRAFT. IT HAS NOT BEEN FORMALLY RELEASED BY THE AGENCY AND SHOULD NOT BE CONSTRUED TO REPRESENT AGENCY POLICY . . . .”86 Moreover, as to the draft decisions referred to in the court’s opinion, both were internal EPA briefing memoranda, one dated November 1973 and the other dated October 1973.87 Finally, we reject the majority’s assertion that this court’s thirty-day order did not force EPA to rush into a precipitate decision. On this point the shaky evidentiary foundation of the Third Health Document speaks for itself.88
It is instructive to see the result of EPA’s action in circulating in some quarters a draft Health Document containing new data in late October, but affording no formal comment period. According to the certified index to the record, precisely twenty persons, corporations, or entities made twenty-four written submissions to EPA during the sixty-day period up to and including 28 November 1973, the date upon *461which the regulations were promulgated.89 This compares to at least 218 persons, corporations, agencies, or other entities that commented on the regulations on a prior occasion. Of the twenty-four submissions, fully nine were dated 26 November 1973 or 28 November 1973; it is thus virtually impossible for any of them to have been considered by the Administrator. Only seven submissions were dated 18 November or earlier. Only two submissions were from agencies of the federal government, a letter from the Secretary of Transportation dated 26 November 1973 90 and two pages of comments from the Treasury Department dated 19 November 1973.91
The content of these twenty-four submissions is equally illuminating. After carefully examining each of these submissions, we found only one, Naleo’s review of EPA’s health position dated 19 November 1973,92 to be a substantive critique of EPA’s draft Third Health Document. In fact, it was the only submission which made any significant comment on EPA’s health position at all. All but four of the other submissions focused upon the energy or economic costs of the proposed regulations. Even these comments were quite short, ranging from one page (seven of the letters or telexes) to four pages (one of the letters) in length. To the extent any of these nineteen submissions mention the health issue (most do not and the few that do merely state their disagreement), none mention or refer to the draft Third Health Document. Three of the four remaining submissions were from Ethyl’s counsel. One was a letter dated 29 October 1973 calling EPA’s attention to two articles which EPA had so far overlooked;93 one was an unedited transcript of the October conference on low-level lead effects mailed to EPA on 24 October 1973;94 and the third was a letter dated 12 October 1973 requesting EPA to place in its files the individual presentations made at the October conference, rather than just the abstracts.95 The only other health-related submission during this period was the collection of abstracts referred to in Ethyl’s 12 October letter.96 None of these submissions mentions or comments on the draft Third Health Document.97
*462Despite this paucity of comment, our colleagues assert that “[a]ll significant new information developed during the rulemaking in this area on the frontiers of scientific knowledge was made available to petitioners and the public for comment well in advance of issuance of the final regulations on November 28, 1973. Thus both the requirements and the spirit of Section 4 [APA § 553] were complied with.”98 With this conclusion behind them, our colleagues are able to avoid the key procedural issue raised by this record, i. e., whether the Administrator may rely on information not previously made available in any meaningful sense to the petitioners or the public without violating both the letter and the spirit of the APA.99 Apparently, our concept of making information available to the public well in advance of promulgation and our understanding of the compliance requirements of section 553 differ considerably from those of our colleagues.
The court’s opinion strives strenuously to prove that three of the five petitioners received notice in fact of the Third Health Document, or at least portions thereof. They assert that “petitioner Ethyl was directly furnished with all such documents as a result of a Freedom of Information Act . suit.”100 We do not believe that agency compliance with the APA can rest on individual interested parties filing FOIA suits. If the APA is to avoid becoming a docile paper tiger, it cannot rely on the enforcement teeth of other federal statutes.101
Furthermore, of course, as shown by the dates of many critical documents relied upon by the Administrator (including, inter alia, one which was not even included in the Third Health Document;102 two which were mentioned in the Third Health Document, but had not even been published when the regulations were issued;103 and four which came to light only after the October 1973 conference104) it is clear that Ethyl did not have adequate time to prepare effective comments, even if one assumes that Ethyl knew that EPA intended to rely on these documents when it promulgated the final regulations. In addition to Ethyl, our colleagues point to “comments” from petitioners NPRA and Nalco.105 Re*463garding NPRA’s submissions, our colleagues are not consistent in their assertions, since they seem to agree that these comments related only to the shift from leaded pool averaging to total pool averaging, and not to EPA’s health position.106 Unlike other aspects of the regulations, the shift from leaded pool averaging to total pool averaging did not incorporate and rely on new information. Therefore, this particular change in the regulations managed to avoid the procedural infirmities that plague all other aspects of the regulations, and NPRA’s comments on this change are irrelevant to the procedural issue involved in this case. On the other hand, as we indicated above, Naleo’s comments represent the one and only substantive critique of the draft Third Health Document received by EPA before promulgation of the regulations.
In their attempts to demonstrate that “comments [on the Third Health Document] were received”,107 our colleagues equate the unedited transcript of the October conference, which was submitted by Ethyl to EPA, with comment on the new studies presented at that conference.108 Similarly, they imply that Ethyl’s submission of a transcript from the EPA committee meeting where the Seven Cities reanalysis was first discussed amounted to comment by Ethyl on the reanalysis.109 We simply cannot accept the court’s conclusion that the submission of these unedited transcripts amounts to even a form of comment. Moreover, if these unedited transcripts can be characterized as some form of comment, surely they are not such an effective form that they prove beyond peradventure that EPA afforded the petitioners (not just Ethyl) and all other interested parties a meaningful opportunity for effective comment.
Strangely, our colleagues seem to feel that EPA “went far beyond the measures it was required to take” by placing new information in the so-called “public file” and thereby making it “available for comment and criticism.”110 To the contrary, we submit that (1) placing information in a public file (not to mention indexing that information under obscure headings) and (2) making information “available for comment and criticism” affords interested parties no notice whatsoever that EPA will rely on this information. This intent to rely was only noticed in the Health Documents, on the last of which no effective comment was possible. We can only wonder how many thousands of pieces of paper were “placed in the public file” that EPA never mentioned in any Health Document. Conceivably, under the court’s view of administrative procedure the burden of sifting through hundreds of pounds of poorly indexed paper and guessing which studies will be relied upon can be shifted from an agency to the interested parties through the simple device of a “public file.”
In summary, we find the conclusion undeniable that no effective comment on EPA’s revised health position took place — nor could it have taken place — in the brief period just prior to promulgation, as the court’s opinion would have us believe. The absence of any comment on EPA’s new health data *464(with the sole exception of Nalco) gives rise to the inference that either circulation of the draft Third Health Document was limited to a more receptive audience than previous circulations had addressed or (hat the time between receipt of the draft and promulgation of the regulations was not sufficient to permit effective comment. The total absence of any comment on EPA’s new health position by any government agency, the source of previous unanimous caustic criticism, is most significant.
As indicated by the heading of this opinion subsection, it is not enough under the Administrative Procedure Act merely to make the scientific data “available” to the public and all interested parties. First, the claimed “availability” here boils down to a compelled furnishing of data to one interested party under the Freedom of Information Act and the placing of hundreds of miscellaneous documents — scientific studies, papers, published articles, etc. — in a poorly indexed111 dust bin of a file in the public rooms of the Environmental Protection Agency. Second, there was never any notice — and our colleagues do not and cannot contend that there was — on which data out of the great miscellany EPA would rely until the Third Health Document was published simultaneously with the regulations. Third, the opportunity to comment effectively on the new data on which EPA ultimately relied was farcical, as our detailed discussion above nails down.
Since Nalco was able to prepare and submit a substantive critique of the draft Third Health Document on 19 November 1973, obviously Nalco somehow received actual notice of the existence of the draft Third Health Document and the new evidence therein. (Perhaps Nalco’s representatives were luckier or more adept than the representatives of other interested parties when it came to sifting through poorly indexed documents.) The record, however, does not indicate that any other interested party received notice of the existence of this new evidence, and clearly, no interested party, including Nalco, received any notice that EPA intended to rely on these new studies. As we mentioned earlier, on its cover page, the 19 October draft bore the following notation:
DISCLAIMER:
THIS DOCUMENT IS A PRELIMINARY DRAFT. IT HAS NOT BEEN FORMALLY RELEASED BY THE AGENCY AND SHOULD NOT BE CONSTRUED TO REPRESENT AGENCY POLICY. IT IS BEING CIRCULATED FOR COMMENT ON ITS TECHNICAL ACCURACY AND POLICY IMPLICATIONS.112
Accordingly, Nalco wrote in the introduction to its substantive comments, “It is understood that these documents [the draft Health Document and the draft regulations] are preliminary drafts and have not been formally released by the agency and should not be construed to represent Agency policy.”113 On the basis of previous EPA practice in regard to the First and Second Health Documents, until EPA formally released the Third Health Document as representing the Agency’s position, no one had any notice of EPA’s intent to rely on new evidence.
To emphasize the gross procedural irregularities approved by the majority of this court, in the following chart we capsulize what the record reveals — or fails to reveal:
*465
*466In no previous case, environmental or otherwise, has this court tolerated such “public notice” and “opportunity to comment” as it sanctions here. From the pages of the court’s opinion there seeps the theme that this is an environmental case; hence, the court like the agency need not labor by the usual rules. We recognize no such exemption for this or any other type case; certainly Congress did not give it to us; we know that adherence to proper prescribed procedure is the soundest route to a correct substantive result.
II. THE LEGAL STANDARD
A. The Statute — Its Interpretation
Section 211(c)(1) of the Clean Air Act authorizes the EPA Administrator to
control or prohibit the manufacture, introduction into commerce, offering for sale, or sale of any fuel or fuel additive for use in a motor vehicle or motor vehicle engine (A) if any emission products of such fuel or fuel additive will endanger the public health or welfare . . . ,114
The initial question presented is whether the Administrator properly interpreted this legal standard when he made the factual determinations upon which these regulations must be predicated.
The petitioners contend that the “will endanger” standard requires a determination that lead motor vehicle emissions, in and of themselves, pose a direct and provable health hazard to significant portions of the general public. By contrast, if it can only be demonstrated that lead emissions contribute to a health hazard, the Administrator is limited to acting under section 202,115 which deals with emission standards imposed on new car manufacturers, and sections 107 through 110 of the Clean Air Act,116 which deal with ambient air quality standards.
The Administrator stated his interpretation of the statutory language when he issued the regulations:
It is the Administrator’s view that the statutory language quoted above does not require a determination that automobile emissions alone create the endangerment on which controls may be based. Rather, the Administrator believes that in providing this authority the Congress was aware that the public’s exposure to harmful substances results from a number of sources which may have varying degrees of susceptibility to control.117
As far as this statement goes, we deem it correct. To recognize lead in the ambient air as but one possible source of lead in the human body is merely to recognize undisputed fact. But to argue that, because there are multiple sources of lead, a chain of causation from lead in the air produced by auto emissions to a level of lead in the blood high enough to be dangerous to health need not be established is totally fallacious. We are not clear that the Administrator really argues this, but it is certain that our colleagues, in order to bolster the agency’s position, have throughout their opinion embraced this fallacy.
Our reading of the language of the Act, its legislative history, and the administrative record lead us to the view that section 211(c)(2)(A) requires the Administrator to conclude “after consideration of all relevant medical and scientific evidence available to him” that a fuel or fuel additive causes an emission which causes a significant health hazard to a substantial portion of the general population before he can control or prohibit the use of particular fuels or fuel additives.
Logically, only if the Administrator can say that a fuel additive causes a significant health hazard can he say that controlling or prohibiting such fuel additive would reduce significantly such health hazard. Whether we term the Administrator’s decision one of “assessment of risks,” as our colleagues urge, or whether we emphasize that such decision must be based on “facts,” the caus*467al connection between lead emissions and the harm must be established by relevant scientific and medical evidence.
When we refer to the necessity under the statute of showing that a fuel additive causes a health hazard, we do not mean that the statute requires a showing that the fuel a'dditive (lead) causes a health hazard without consideration of other sources of lead in the human body having their determined effect also. To interpret the statute in this manner would be preposterous on its face, for as we shall discuss later in Part IV, on EPA’s data most of the lead in the human body comes from dietary sources, food and drink. We think that the statute does require that, before the Administrator can prescribe the regulations involved here, he must find that the lead from auto emissions by itself or alone contributes a measurable increment of lead to the human body, and that this measurable increment causes a significant health hazard.
To repeat, only if the Administrator can say that an identifiable measurable increment of lead in the human body is derived from auto fuel additives and that this measurable increment of lead itself (taking into consideration all other sources of lead) causes a significant health hazard, can the Administrator claim that controlling or prohibiting lead would reduce significantly such health hazard.
To us all analysis leads inescapably to this conclusion, yet the court’s opinion endeavors to avoid this by resorting to two obfuscating dichotomies (perhaps the same distinction expressed in two ways). First, the opinion attempts to separate actual from potential harm, and, second, to separate risk from fact. “[T]he ‘will endanger’ standard is precautionary in nature and does not require proof of actual harm before regulation is appropriate.”118 “[T]he power to assess risks, without relying solely on facts, flows inexorably from the nature of the ‘will [en-]danger’ standard.”119 We submit no such separation occurs in the real world, and if the Administrator proceeded on any such fanciful theory, it is an instance of his failure to discern the intent of Congress manifested by the words “will endanger.”
As to the first distinction, there is no distinction possible here between actual and potential, between past and future harm. The Administrator is dealing with a continuing situation. If there can be found potential harm from lead in exhaust emissions, the best (and only convincing) proof of such potential harm is what has occurred in the past (either in 50 years of practical usage or in laboratory experimentation), from which the Administrator can logically deduce that the same factors will produce the same harm in the future. For the court’s opinion to hold that the Administrator can dispense with proof of actual harm, i. e., what has occurred in the past, and can nevertheless somehow determine potential harm, is to grant the plainest license for the wildest speculation. We have always thought scientific conclusions, above all, demanded proof by events recorded and observed.120
The court’s second asserted dichotomy, risks versus facts, is equally indefensible in logic. All true risk assessment is based on facts and nothing else.121 Those professional risk-assessors, the professional sports gambling fraternity, would smile at any other theory. To the extent that hunch and intuition enter into any final decision, these *468are separate factors outside of any scientific risk calculation.
Our colleagues apparently find it necessary to legitimatize the Administrator playing hunches. They assert, “Danger is a risk, and so must be decided by assessment of risks as well as by proof of facts.”122 Of course the Administrator assesses risk— from the facts as he knows them. The question here is how much he knows. To the extent the agency found it necessary to make an “assessment of risks as well as [rely on] proof of facts,” the agency was frankly just speculating. No reviewing court can countenance this. If such agency decision is not “arbitrary and capricious,” what decision could be? It is precisely a devotion to facts, not hunches, that distinguishes the professionals from the amateurs in assessing risks; we deem the Administrator to have been intended by Congress to be a “professional.”
B. Threshold Factual Determination Required
EPA, relying upon our recent opinions in Amoco, supra, and Industrial Union Department, AFL-CIO v. Hodgson,123 argues:
In the present case, as in Hodgson, the EPA must determine what amount of lead it should permit to be used as additives to gasoline. Here, too, “reliable data is not currently available with respect to the precisely predictable health effects of various levels” of lead which the EPA could permit to be used. The Agency’s decision to require a gradual diminution in the permissible amount of lead is, in the last analysis, a policy judgment “concerning the relative risks of underprotection as compared to overprotection.”
The regulations here reflect reasoned judgment on a border area of scientific knowledge and a policy choice in favor of protecting people exposed to automobile exhausts and particularly urban children. This judgment should be upheld.124
In essence, it is argued that the “will endanger the public health . . standard is a delegation of quasi-legislative power to the Administrator and not'a requirement that he reach a reasoned determination purely on the scientific and medical data.
It is manifest that Congress wished the Administrator’s threshold determination under section 211(c)(1)(A) to be a reasoned factual determination based solely on the medical and scientific evidence. This is made quite clear by subsection 211(c)(2)(A) which emphasizes that “[n]o fuel, class of fuels, or fuel additive may be controlled or prohibited by the Administrator pursuant to clause (A) of paragraph (1) except after consideration of all relevant medical and scientific evidence available to him . .” By directing the Administrator to consider “all relevant medical and scientific evidence” before deciding to institute controls, the Congress must have intended that such evidence act as the basis for his determination.
If the intention had been to permit him to make “an essentially legislative policy judgment, rather than a factual determination . '. .,”125 some indication of this should have appeared in the statute. Other sections of thé Clean Air Act closely related to section 211 do contain some indication that a broad discretionary power is being vested in the Administrator. Section 108 authorizes the Administrator to set air quality criteria for each air pollutant “which in his judgment has an adverse effect on public health or welfare . . .”126 Section 202 authorizes the Administrator to set standards for each automobile emission “which in his judgment causes or contrib*469utes to, or is likely to cause or to contribute to, air pollution which endangers the public health or welfare.”127
By the same token, neither section 108 nor section 202 requires the Administrator in his threshold determination to consider either scientific or medical evidence. Section 108 air quality criteria which EPA ultimately promulgated are required, however, to “reflect the latest scientific knowledge useful in indicating the kind and extent of all identifiable effects on public health or welfare which may be expected from the presence of such pollutant in the ambient air, in varying quantities.”128 Section 108 therefore requires a factual determination based on “the latest scientific knowledge” in establishing criteria, while permitting wider discretion in making the original threshold determination as to which pollutants to list.
In conclusion, we believe that the threshold determination whether an emission “will endanger the public health . . . ” does turn and was intended by Congress to “turn crucially on factual issues” and not upon “choices of policy.”
III. SCOPE OF REVIEW
EPA in promulgating these regulations followed the informal rulemaking procedure outlined in the Administrative Procedure Act, 5 U.S.C. § 553.129 Since the Clean Air Act neither provides for review “on the record after opportunity for an agency hearing” nor provides specifically the standard of judicial review, our scope of review of these regulations is limited to a determination of whether “agency action, findings [or] conclusions . . . [are] arbitrary, capricious, an abuse of discretion, or otherwise not in accordance with law . .” or “in excess of statutory jurisdiction, authority, or limitations, or short of statutory right,” or “without observance of procedure required by law.”130
It is well recognized that in reviewing the record we are not to substitute our judgment for that of the Administrator.131 Thus, we are required to affirm agency determinations with which we disagree. However, we are also under an obligation not to act as a rubber stamp of agency rulemaking action. As a result, in reviewing agency rulemaking under section 553, we are obligated to engage in a “substantial inquiry,” and “inquiry into the facts is to be searching and careful . . . .”132 As the Supreme Court has indicated, we “must ‘consider whether the decision [of the administrative agency] was based on a consideration of the relevant factors and whether there has been a clear error of judgment.’ ”133
Certainly a determination made in the absence of any evidence in the record to *470support it would lead a reviewing court to conclude that a “clear error of judgment” had occurred. However, a reviewing court could reach a similar conclusion in the presence of some evidence supporting the agency determination.134 Since a court must review “the entire record,” it may well be that the evidence detracting from the agency’s conclusion is so overwhelming or so persuasive, or the agency’s approach so one-sided, or the decision-making process so flawed, that a reviewing court must conclude that the agency erred in the exercise of its rulemaking power.135 Or, if there is an essential point or element missing in the logical progression toward the conclusion that the agency reaches, then the agency’s action likewise may be arbitrary or capricious, because it is not supported by a logical thought process.
In these consolidated environmental protection cases the thought process by which an agency reaches its conclusion on informal rulemaking resembles a chain. If there is a link missing, then the agency, to reach the conclusion that it did, was required to *471take an arbitrary jump in its logic to reach that conclusion. To illustrate, we have the admitted fact that there is lead in gasoline which goes into the automobile tank. At the other end of the thought progression there is the scientific conclusion that too much lead in the human blood level is harmful — although there seems to be no agreed conclusion as to exactly what that blood level is (for this case, we assume the EPA level correct). Logically, as we view the scientific evidence, the chain of transmittal of the lead in the auto tank to the lead in the bloodstream is vital on the issue before us — whether there is valid reasoning based on evidence to sustain a finding that lead additives will endanger public health. The standard of “will endanger” makes the existence or nonexistence of a scientifically demonstrated chain of transmittal (and not a mere pattern of guesswork) decisive. For if no such scientifically proved chain exists, the Administrator’s decision can only be arbitrary and capricious.
The recent First Circuit case, South Terminal Corporation v. EPA,136 supports our application of the “clear error of judgment” standard of review to the Administrator’s determination here. In South Terminal the court was called upon to review the Metropolitan Boston Air Quality Transportation Control Plan promulgated by the Administrator under section 110137 after the Commonwealth of Massachusetts had failed to submit an acceptable plan to implement the national primary and secondary ambient air quality standards. We quote, at length, the First Circuit’s discussion of the standard of review:
Under § 706, we must determine whether EPA followed lawful procedures in evolving its plan; whether it acted within its statutory authority; and whether the plan is constitutional. If so, we must set aside the plan only if it is found to be “arbitrary, capricious, an abuse of discretion, or otherwise not in accordance with law”. 5 U.S.C. § 706 (2)(A).
In the following parts of this opinion we deal first with the procedural objections to the plan and later with the constitutional ones. In between we consider statutory objections and, most difficult of all, those objections addressed to the merits of the plan. The last objections, it is clear, are outside our province unless they show that EPA’s decision was not based on consideration of relevant factors or else included a “clear error of judgment”. Overton Park, supra, 401 U.S. at 416, 91 S.Ct. 814 [28 L.Ed.2d 136]. We are not empowered to substitute our judgment for that of the agency.5
The questions about the plan on review are of two types: the rationality of EPA’s technical decisions (such as its determinations of local photochemical oxidant and carbon monoxide levels and the amount of reductions required to meet national standards), and the rationality of EPA’s “control strategy”, that is, the measures adopted to reduce emissions. The former present peculiar difficulties for nonexperts to evaluate. Yet “[our] inquiry into the facts is to be searching and careful”, id., and we must assure ourselves as best we can that the Agency’s technical conclusions no less than others are founded on supportable data and methodology and meet minimal standards of rationality. See Section III infra.
Assuming EPA’s technical determinations are reasonably based, we must decide whether the selected controls are arbitrary or capricious. In so doing, we must bear in mind that Congress lodged with EPA, not the courts, the discretion *472to choose among alternative strategies.6 Unless demonstrably capricious — such as much less costly but equally effective alternatives were rejected or the requisite technology is unavailable- — the Administrator’s choices may not be overturned.
Appropriately enough, Part III of the opinion is entitled, Whether EPA Committed a Clear Error of Judgment in Computing the Need for Emission Reductions. The First Circuit’s emphasis on adequate data and appropriate methodology is fully in accord with our interpretation of the “arbitrary and capricious” standard.
In the case at hand, it seems obvious that the Administrator made “clear error[s] of judgment.” This conclusion is based upon a “searching and careful” inquiry into the facts underlying the determination that airborne lead will endanger the public health. At several points in the Administrator’s reasoning we have found little or no evidence to support his conclusions, and at several points we have noted clear errors of a substantial nature in the Administrator’s analytical and evaluative methodology and EPA’s decision-making process. Several vital links in the chain are unsupported; for the Administrator to leap to the conclusion he did can only be termed arbitrary and capricious. We submit that “a clear error of judgment” has occurred,139 as our concluding analysis of the evidence and methodology demonstrates.
IV. THE EVIDENCE
A. This Court’s Task
It is of crucial importance to recognize that this court’s task on review of agency rulemaking is twofold. One of these tasks is universally acknowledged. We must explore the evidentiary record to determine whether the statements and conclusions of facts have an adequate basis in the underlying evidence. However, an equally important role is our review of agency analysis in order to determine whether it is principled and reasonable. Properly understood, our role is not to review directly the evidence, but to review the agency’s treatment and analysis of the evidence. Of course, in so doing we will defer to the agency’s expertise and experience in the subject matter of the decision.
In our view, the court’s treatment of the evidence neglects this second, equally important function of a reviewing court. Our colleagues’ extended review of the evidence (in some respects much more extensive than EPA’s own Third Health Document) at several points supplies reasons, distinctions, or modes of analysis that are otherwise absent from the Administrator’s discussion. Just as agency counsel cannot supply a post hoc rationalization for agency determinations, neither can a reviewing court.140
*473In order to avoid slighting either of our reviewing functions, our treatment of the evidence will track closely the Administrator’s discussion and analysis of the evidence in the Preamble. To the extent necessary, the Administrator’s discussion will be amplified by the Third Health Document. Rather than roving through the record looking for isolated snippets of evidence, only from these major starting points will we make the necessary comparisons with the underlying evidence.
B. “A small but significant portion of the urban adult population . . . [is] over-exposed to lead."
1. The 40 ug Demarcation
The Administrator introduces his analysis of the evidence with the following statement:
Environmental lead exposure is a major health problem in this country. A small but significant portion of the urban adult population and up to 25 percent of children in urban areas are overexposed to lead. The lead exposure problem is caused by a combination of sources including food, water, air, leaded paint, and dust. The aggregate contribution of lead from all these sources poses a significant threat to health.141
Although the Preamble does not define what is meant by “overexposed to lead,” Chapter Three (and to a lesser extent Chapter Four) of the Third Health Document explores the issue. The Document notes that blood lead levels, measured usually in terms of ug/100 g of blood (micrograms per hundred grams), are generally considered an appropriate index by which to assess the total lead body burden. The general population carries blood lead levels of between 10 and 30 ug/100 g.142 As exposure to environmental lead increases, blood lead levels increase slowly and in a nonlinear fashion. As a result, blood lead levels are a satisfactory index under constant exposure conditions but are less reliable under changing exposure conditions.143 Clinical symptoms of lead poisoning usually do not appear until blood lead levels reach 80 to 100 ug/100 g or higher.144
The bulk of the discussion in Chapter Three relates to those studies which have attempted to identify the health effects associated with blood lead levels below those associated with lead poisoning. On the basis of those studies, EPA’s staff tentatively concluded that “subclinical changes” may be associated with blood lead levels in the 40 to 60 ug/100 g range. The document is careful to emphasize that a 40 ug/100 g blood level “does not represent a sharp demarcation between health and disease.” It rather represents a level above which it would be “prudent” to prevent further lead exposure.145
The court’s opinion states that petitioners are contesting the setting of 40 ug/100 g as *474a precautionary level.146 This is not the case. Petitioners’ arguments have simply sought to emphasize the precautionary nature of the 40 ug level. They especially seek to emphasize EPA’s own conclusion that no known functional injury or clinical effects occur in the 40 to 60 ug range and that whatever metabolic effects occur are “slight” and without clinical significance.147 With these qualifications stated, petitioners seem willing to concede for its limited purposes the 40 ug level as the line between normal and elevated blood lead levels.148 Their central position, which we explore below, is that EPA lacks a scientific basis for asserting the existence of blood lead levels in excess of 40 ug among a significant portion of the general adult population.
2. Over-Exposure in the General Adult Population
Again we must turn directly to the Third Health Document, since the Preamble does not further amplify the basis for concluding that lead levels are indeed elevated in a significant portion of the population. As regards adults, the Health Document discusses the issue in the equivalent of one page of text and in two charts.149 The Health Document begins by acknowledging that “due to limitations” on prior studies, “it is impossible to make precise estimates of the number of persons having blood lead levels within a given range.”150 These limitations include problems with the chemical procedure which analyzes lead in blood and deficiencies in quality control at laboratories. As the document concedes, “[reproducible results even within the same laboratory are not always obtained.”151
Given the limitations on the underlying data, EPA’s staff was still willing to conclude that certain groups, principally ones “occupationally exposed to automobile exhaust,” “frequently or characteristically” have elevated blood lead levels.152 Petitioners vigorously attack this conclusion on two levels. First, they point out that EPA’s data only supports a conclusion that certain occupational groups have higher blood lead levels as a result of occupational exposure. It does not support a conclusion that a significant portion of the general urban population, not occupationally exposed to lead, has elevated levels. In fact, petitioners emphasize that the only comprehensive study of the general urban adult population, the Seven Cities study,153 found only three persons (or 0.15%) out of 1,935 persons tested had blood levels of 40 ug or higher.154 This study was supported and directed in part by EPA and was completed recently (in 1972). Therefore, the study is particularly credible.
In addition, petitioners point to the study of Azar, et a I.,155 published in 1972, in which *475five groups of thirty subjects each were tested. One group consisted of Los Angeles cab drivers exposed to air lead levels of 6.1 ug/m3. Despite these very high air lead levels, blood readings in this group averaged a normal 24.6 ug (±4.5 ug for one standard deviation).156 Other groups had blood lead levels averaging from 13.8 ug to 22.4 ug. Of all 150 persons tested, only one subject showed a blood level over 40 ug and he was believed to drink moonshine whiskey.157
Second, petitioners attack the Administrator’s reliance upon data derived from occupational categories exposed to unusual quantities of automotive exhaust, such as those working in parking structures, traffic tunnels, and closed garages, as well as those exposed to lead from sources other than the air, such as service station attendants and garage mechanics.158 They question, implicitly at least, EPA’s picking and choosing among the available data, choosing data in which subjects are occupationally exposed to lead and rejecting data based upon samples drawn from the general urban adult population. In effect, they argue that nation-wide restrictions upon lead additives must not be based upon merely precautionary blood levels in those occupationally exposed to unusually high amounts of lead.
We agree in substantial measure with petitioner’s position. EPA has presented no evidentiary basis for concluding that a significant portion of the general urban adult population has elevated blood lead levels. Table VII-1 in the Third Health Document amply demonstrates this fact.159 The Azar study of taxi drivers and office workers from Los Angeles, taxi drivers from Philadelphia, and persons from Barksdale, Wisconsin, and Starke, Florida, found no one (after excluding the imbiber of moonshine) with a level of 40 ug or higher. A sample of 55 “Women Living Near Freeways” uncovered one person at 40 ug or above.160 Two other composite urban samples of 423 and 833, respectively, found only 0.8% and 2.7% of the subjects with precautionary levels.161
Unexplained is the absence from Table VII-1 of the results of the Seven Cities study — 1,935 persons tested and at most three with elevated blood levels. The court’s opinion declares that “[petitioners rely heavily on the results of the so-called Seven Cities Study, which found a very small percentage of adults with elevated (in excess of 40 ug) blood levels. . . The Administrator, on the other hand, finds serious methodological flaws in the Seven Cities Study that limit its usefulness, 38 Fed.Reg. 33735 . . . .”162 We can understand the majority’s anxiety to explain EPA’s rejection of the Seven Cities study; however, the Administrator never discusses the studies on which he does rely for his conclusion that a significant portion of the urban adult population is over-exposed to lead. The discussion cited by the court relates to use of the Seven Cities study in an entirely different context, the presence or absence of a correlation between air lead levels and blood lead levels. Significantly, the methodological flaw being referred to by the Administrator is the fact that “. other sources of lead influencing blood lead levels were not adequately considered in the blood lead-air lead comparisons.”163 Although such a flaw might arguably affect the study’s reliability as to the existence of a correlation, it is quite irrelevant as regards the overall blood levels measured in the course of the study, which negate the hypothesis of over-exposure in the general adult population.
*476The Seven Cities study clearly presents the second major defect in the Administrator’s approach. The Administrator chose among the available data without any explanation as to why — why he was relying upon certain studies and rejecting others. No explanation is offered in either the Preamble or the three relevant paragraphs in the Third Health Document. This is not a case of a “clear error of judgment”; it is rather a case of a clear absence of judgment. On this score, our colleagues assert,
Petitioners argue that the negative conclusions of the Seven Cities Study and several other studies should outweigh the positive indications of [certain other studies relied upon by EPA] . . . . The Administrator disagreed, and we cannot fault his conclusion.164
Thus, the court concludes that ,“[t]he problem here is one of choosing among the items of evidence.”165 We respectfully disagree. The problem here is really one of choosing among the items of evidence and explaining why l The Administrator disagreed with the negative conclusions of the Seven Cities study and several other studies, but never said why he disagreed. This is why we can fault his conclusion.
The court, however, again steps into the breach to offer its analysis of the studies:
[S]tudies of occupational groups are often particularly valuable in acting as an early warning system of possible effects on the public at large. The intensive exposure in particular occupations essentially accelerates the effects of long-term exposure of the general public and provides an identifiable, highly exposed, and readily accessible group of research subjects. See D. Clark & B. MacMahon, Preventive Medicine 26 (1967) .166
This rationalization of the Administrator’s decision is purely post hoc. There is no basis for it in the Third Health Document.
The above is illustrative of the amazing lengths this court will go to to produce a decision for some uncertain, ill-defined, supposed environmental benefit. This court not only commands the Administrator to produce a decision by a certain date, but rewrites his decision to fill in the gaps when his logic falters.
The court, in addition, attempts to draw support from the fact that in the Reserve Mining case some of the evidence indicating the possibility of absorption of asbestos particles by the general public came from studies of asbestos workers. This reliance is entirely misplaced.' We do not here suggest that the Administrator cannot seek to reach a conclusion about the general public on the basis of studies of occupationally exposed individuals. Rather, we believe, that, if he has so concluded, he has not adequately explained how he did it. By contrast, the Eight Circuit undertook an extensive discussion of asbestos fiber exposure, relating occupational exposure to exposure by the general public through the ambient air. There is considered analysis on this precise issue in both the Eighth Circuit’s opinion and the district court’s opinion.167
*477In any event, the fact that such evidence proved relevant in Reserve Mining does not mean it would be relevant in this case. The issue of relevance is one for scientific inquiry and analysis. Such analysis was undertaken by the various witnesses, including especially the court-appointed expert, in Reserve Mining. Such analysis is completely absent from the Third Health Document. This court cannot now repair that defect. We cannot perform this district court or administrative agency function.
A crucial aspect of the Administrator’s decision is deficient both because of its analysis and because of the absence of an evidentiary basis. On the record before us, we fail to see how the Administrator found an endangerment to a significant portion of the general urban adult population.
C. “[AJbsorption of air lead does contribute to total lead exposure and when added to lead from other sources . results in total exposure that is excessive.”
In promulgating the regulations, the Administrator stated the basic factual question to be resolved as follows: “Is there a correlation between air lead levels and blood lead levels?”168 Acknowledging that the data failed to produce “consistent correlations,” the Administrator nevertheless was able to reach the ultimate conclusion that auto lead emissions contributed significantly to excessive blood lead levels in adults. In our view, the record indicates that the Administrator acted arbitrarily in choosing among the data relating to a possible correlation between air and blood lead levels.
In the Preamble the Administrator explicitly relies upon the following studies to support the conclusion that there is a correlation between air lead levels and blood lead levels: the pilot lead isotope studies,169 the unpublished study in Japan,170 the chamber studies,171 and the Daines study.172 Specifically rejected is the Seven Cities study173 on the ground (previously noted) that “other sources of lead influencing blood lead levels were not adequately considered in the blood lead-air lead comparisons.” In neither of the two epidemological studies relied upon by the Administrator, the Japanese study and the Daines study, is there any indication that other sources of lead were controlled. And yet, the Administrator in the Preamble, in an obvious reference to the Daines study, declares that “[w]hen comparable groups with similar lead intakes were studied,” blood lead was higher among persons living in urban areas near highways.174
This is no minor slip on the Administrator’s part. His entire argument in reply to criticism about the way in which he picked and chose among the various studies is that “[sjtudies which have come to contrary conclusions have generally failed to take into account the influences of other sources of lead on blood lead levels in people being studied.”175 The record indicates that even the studies relied upon by the Administrator failed in this same respect.
*478With regards to the Seven Cities study, the Administrator feebly attempts to gloss over his consistently selective approach with the following equivocal comment:
[I]n the Seven Cities Study, urban-suburban differences in blood leads between comparable groups were consistently found which at least in part reflect differences in air lead exposure.176
The court’s opinion, however, well recognizes the gravity of the Administrator’s error and as a result offers for our edification a meatier, amplified version of this explanation as to why the Administrator embraces favorable studies possessing the same deficiency as the studies he rejects:
The [epidemiological] studies before the Administrator [e. g., the Seven Cities study] were of large groups of people; correlations were sought between blood level and exposure to lead in the ambient air. The studies were confounded, however, by the multiple sources of lead. Since diet accounts for a major portion of the body lead burden, an individual’s blood lead level varies not only according to his exposure to lead in the ambient air, but according to his daily dietary intake of lead. Wide variations in dietary lead intake, which are common, can completely mask the effects of air lead absorption. Nonetheless, none of the epidemiological studies could control or measure dietary lead intake. This uncertainty in the data severely limited the usefulness of. the broadly conceived epidemiological studies, and led the Administrator to rely instead on data limited to situations in which dietary exposure could roughly be termed constant.
Following this rationale, the Administrator focused on the consistent relationship found between air and blood lead levels within particular metropolitan areas, rather than on the lack of such a relationship between areas. . . ,177
No matter whether the court or the Administrator himself first came up with this explanation for EPA’s inconsistent and selective approach to the evidence, it passes our understanding how anyone can find dietary control a problem in a comparison between greater Philadelphia and greater New York (thus justifying rejection of that data), but of no importance in a comparison between Scarsdale and Harlem (thus justifying reliance on urban-suburban data from greater New York). The art of reconciling total inconsistencies has soared to new heights when this court can seriously conclude (1) that “dietary lead [can] be assumed relatively constant”178 between the most affluent and the least affluent neighborhoods of greater New York City and (2) that “[t]he Administrator treated all the evidence in a consistent and rational manner.” 179 The Administrator has thus utterly failed to present us with a reasoned and principled analysis of the evidence.180 In fact, the Third Health Document does not even discuss the studies and analyses running counter to EPA’s position; as a result, it is not strange that it finds no need to distinguish or reason them away.
*479To criticize EPA’s overall analysis as arbitrary should not be construed as approval of each of the studies used by the Administrator to support his position. Petitioners have presented substantial criticisms of the methodology of those studies. They also point out that EPA has drawn inferences or conclusions from many of the studies which their authors were unwilling or unable to draw. We feel no need to discuss these criticisms in detail in light of our overall conclusion.
D. The Contribution of Auto Lead Emissions to Blood Levels in Children — The Speculative Nature of EPA’s Determination
As a second subsidiary reason for adopting these regulations, EPA contends that contamination of dust and dirt by lead from automotive emissions is a significant source of lead exposure to children. This conclusion is based on the following chain of reasoning:
a. High lead concentrations in dust and dirt are prevalent in urban areas.
b. In most circumstances, lead from exhausts and not lead paint or lead from stationary sources is the primary source of lead in urban dust and dirt.
c. Children, especially those between the ages of one and three, eat non-food objects, including dust and dirt. This phenomenon is known as pica.
d. As a result of ingesting leaded dust and dirt, children “can be expected to absorb some of the lead into their bodies.”
An alternative argument is also made that excessive lead levels are found among children who do not come from areas where peeling paint is common and therefore their lead levels must be accounted for by auto lead emissions.181
Neither the Administrator in promulgating these regulations nor EPA’s scientists in the Third Health Document have explicitly stated that auto lead emissions are a significant source of lead to children. In fact, the Administrator states: “Ingestion of peeling paint has long been recognized as the primary cause of clinical lead poisoning in children.”182 The Administrator has quite frankly stated that the chain of reasoning outlined above is a hypothesis not yet completely proved:
Currently, the contention that lead contamination of dust and dirt by automotive emissions is a significant source of lead exposure is a hypothesis consistent with information provided by a variety of studies. However, at this time, not all links in the argument have been established beyond dispute and no single study has collectively inter-related all steps in the exposure process to conclusively inter-related all steps in the exposure process to conclusively [sic] prove or disprove the hypothesis. Despite the existing uncertainties, comments received from the majority of scientists not affiliated with industrial or environmental groups support the contention that dust is an important source of exposure.183
The conclusion drawn by EPA’s scientists, after reviewing all the evidence submitted in the proceedings, is even more tentative:
In conclusion, clinical experience indicates that leaded paint is primarily responsible for the great majority of overt clinical lead toxicity in children. There is, however, sufficient data to strongly suggest that sources of lead other than paint play an important role in childhood lead exposure. These other sources may be especially significant at levels of exposure below overt clinical poisoning.
Lead in the air, and particularly lead in dust, are ubiquitous sources of lead which may well be important contributing factors to the problem. Exposure to dirt and dust sufficiently contaminated by lead could reduce significantly the quantity of additional lead exposure required to produce clinical poisoning in a child *480with other sources of exposure. Though exposure to lead contaminated dirt and dust from automobile exhaust, alone, has not been shown to be responsible for cases of overt lead poisoning, automotive lead has been related to undue lead absorption in children. At this time, it would be prudent to decrease the potential air and dust lead exposure. It should be recognized that further studies are necessary to better quantitate the sources of lead contamination in dust and dirt and the magnitude of the contribution that leaded dust and dirt make to both subclinical and clinical lead overexposure.184
The Administrator has implied that the large incidence of excessive lead exposure among children “known to not reside in homes where peeling lead based paint can be found . . . indicate[s] that in some circumstances dust lead is an important factor and at times may be the primary factor contributing to excessive lead exposure associated with subclinical if not clinical effects.” 185 There is literally nothing in the record which accounts for these high lead levels among children who live outside of “the central city environment.” Assuming that lead based paint was not the explanation, neither EPA’s scientists nor the studies they evaluated could offer an alternative explanation, much less tie the excessive exposure to auto emissions. The Third Health Document identified “lead in food, water, pottery, toys, pencils, solder, etc.” as possible explanations for excessive lead exposure in those cases where lead paint was not a problem.186 It also noted that “[f]urther information is required to arrive at a definitive assessment of the significance of . [lead contaminated dust and dirt as a] source of lead contamination.” 187
Probably the best overall conclusion on EPA’s dust and dirt hypothesis was made in one of EPA’s own internal memorandums by two members of the Office of Planning and Evaluation to Dr. Kenneth Bridbord of the Office of Research and Monitoring:
The reaction of this Office to the balance of Section VI, The Mechanism of Dustfall Lead Exposure, is an ambivalent one. We basically have no objections to the content of this Section; that is, we feel that it meets satisfactory standards of objectivity and quality as a scientific report in itself. On the other hand, if the purpose of this section is to establish a direct line between lead-contaminated dust and dirt from automotive exhausts and cases of lead poisoning and to thereby support the proposed lead regulations, then we find it lacking.
The report’s conclusion concerning the bio-availability of lead from ingested contaminated dust and dirt is not at issue here. However, we should like to emphasize that none of the studies cited directly implicated lead from automotive exhausts as being responsible for the harming of any human being.188
Implicit in EPA’s position is the premise that lead .from auto emissions falls to the ground and mixes with the dust and the dirt at those places where children are likely to play or be. Petitioners concede that lead from auto emissions does fall to the ground, e. g., the center of a six-lane highway. However, they contend it does not fall in those places where significant numbers of children usually spend their time.189 They contend that lead paint and, in some circumstances, emissions of stationary sources account for lead in the dust and the dirt that children are likely to eat. Finally, they emphasize that lead paint (not auto emission lead) is the culprit pointed to by the available evidence.
*481We agree with petitioners’ argument, but not, as our colleagues assert, because we refuse “to recognize that city children play regularly in city streets.”190 Rather, we reject the Administrator’s dust-fall hypothesis because we refuse to recognize that preschool children (the only children in which pica is a common phenomenon)191 play in heavily trafficked city streets. In other words, we submit that (1) the older the child, the less likely it is thyat he or she will ingest the dust and dirt found in a city street, (2) the younger the child, the less likely it is that he or she will be playing in a city street — especially a heavily-trafficked city street, and (3) significant amounts of lead from auto emissions are only found in heavily trafficked city streets.
Recognizing the weakness of their position, our colleagues attempt to discount the importance of the Administrator’s dustfall hypothesis,
While the [dustfall] hypothesis is admittedly not proved as fact, we need not decide whether it would be sufficient by itself to support the low-lead regulations, for it is offered only in support of the evidence already presented.192
Since we, unlike our colleagues, would not sustain the low-lead regulations on the basis of “the evidence already presented” (i. e., the evidence discussed in sections IV. B. and C. supra), we cannot avoid deciding whether the Administrator’s dustfall hypothesis is sufficient by itself to support the regulations. Based on the three common-sensical assumptions stated above, we conclude that it is not even more likely than not that lead dustfall from automobile emissions is swallowed by preschool children with pica. Until, and if, the Administrator can at least establish this causal connection, we should reject his dustfall hypothesis.193
Based in part upon EPA’s own assessment of the evidentiary record, it is undeniable that the dust and dirt hypothesis continues to be a speculative explanation of high blood lead levels among certain chil*482dren. There is simply no evidence in the record which directly proves (used in its scientific sense) the hypothesis. EPA has thus had to argue that its hypothesis is consistent with the evidence. The fact is that there may be some evidence in the record that may be consistent with the hypothesis, but there is also a mass of evidence (as EPA acknowledges) that is inconsistent. Once again, EPA has not even gone so far as to explain why it is relying on the “consistent” evidence and not relying on the “inconsistent” evidence. This portion of the Administrator’s determination is by any definition “arbitrary and capricious.”
V. CONCLUSION
The regulations on fuel additives promulgated by the Environmental Protection Administrator should be held invalid: assuming the correctness of the Administrator’s interpretation of section 211(c)(1)(A) of the Clean Air Act, his analysis reflected a clear error of judgment upon the available evidence. Under the standards of review in 5 U.S.C. § 706(2)(A), (C), and (D), the regulations should be set aside.194
There is an uncomfortably large body of evidence, relied upon by EPA to support these regulations, which came into its possession just days before final promulgation. Further, EPA’s practice of accumulating a mass of scientific documents of all types in its “public information” file, without designating on which information it would rely until the promulgation of the Third Health Document simultaneous with the regulations, is by no means the “notice” required by the Administrative Procedure Act. Thus, interested parties had no reasonable opportunity to question the methodology of the investigators; EPA counsel can thus point to the absence of rebuttal evidence and imply that all was well in the later studies. In light of the fact that EPA had been gathering evidence on airborne lead for three years, and that after each of the previous public comment periods EPA had been forced to reconsider its health position, the fact that so many of the studies that EPA uses to support its position were reported to the Agency less than sixty days prior to promulgation of the regulations, must necessarily give pause to this court.
The Administrative Procedure Act (5 U.S.C. §§ 706 and 553)195 contemplates review of agency determinations after, at a minimum, an opportunity for public and governmental agency comment on all the evidence in the record. When this court mandated a decision within thirty days on a matter the Administrator had had under consideration for three years, there was no suggestion that the Administrator go to totally new, untested, and uncommented upon data as a basis for his decision. The rationale behind such an order could only have been that the court believed the Administrator already not only had the necessary data in hand, but also had afforded the opportunity for informed public and other governmental department comment required in valid agency rulemaking pursuant to section 553. If the Administrator believed that the data in the First and Second Health Documents, along with the searching criticisms received, would not support the decision he was about to make, then he could not lawfully make that decision. Nor could he make it on the basis of new and *483untested data in the Third Health Document.196
Turning from the procedure to the substantive basis of the decision, the supporting evidence, and the analytical rationale applied by the agency, EPA falters on both prongs of its argument in regard to the impact of lead emissions on the general public health and on children. We find no plausible showing that lead in the air makes a “significant contribution to elevated blood lead levels” in either the general population or among children. The Preamble to these regulations itself states, “It is generally agreed that food is the major source of lead to the general population.” This is backed by the conclusion of Dr. Carl Shy of EPA, that on the basis of the Seven Cities study197 only three percent of the differential in blood lead levels between those who lived in urban areas and those who lived in the suburbs “can be accounted for by the air lead gradient” between the two areas.198 As for children, the Preamble states, “Ingestion of peeling paint has long been recognized as the primary cause of clinical lead poisoning in children.” As we pointed out at the outset,199 only if auto lead emissions can be shown to contribute significantly to blood lead levels can it logically follow that a reduction or elimination of auto lead emissions would contribute significantly to solving the problem of lead in the human body.
We judges may be, as we are sometimes reminded, without scientific background or access to expertise, but we think this brand of administrative agency action should be readily apparent — and equally abhorrent— to any appellate judge. To detect and set aside agency action based on such shoddy foundations, not to engage in rival scientific calculations or substitute judgment, is the function of a reviewing court.
We appreciate the quandry in which the Administrator found himself after three years of pondering this question. Beset on all sides by those urging the Administrator to take action, to ban or limit drastically the use of lead in gasoline, confronted by the arguments of those who asserted that such action was completely unjustified, attempting in the First Health Document and in the Second Health Document to postulate a scientific basis for the action he had tentatively decided to take only to have the scientific bases of his action riddled by the analyses of scientists in all other departments of the Government commenting thereon, and confronted finally by an order of this court that he act one way or another in a period of thirty days, the Administrator found himself with the question “whether ... to take arms against a sea of troubles, and by opposing end them.” 200 So the Administrator acted, in reliance primarily on data only partially analyzed by his own staff and not commented on according to the prescribed procedure or by the most knowledgeable parties. The result was, Hamlet-like, a blind stab through a curtain of ignorance, inflicting anguish, but in our judgment not rationally solving any problem.
It should be understood that our view would not prohibit EPA from regulating lead additives under section 211(c)(1)(A) at some point in the future, should evidence *484sufficient to meet the statutory standard be developed and should action appear appropriate. On the present record, however, we cannot agree that EPA has properly decided that lead additives in gasoline will cause emissions which “will endanger the public health.”
APPENDIX
Title 40 — Protection of Environment
Chapter I — Environmental Protection Agency
Part 80 — Regulation of Fuels and Fuel Additives
Control of Lead Additives in Gasoline
Health implications of airborne lead — Introduction. The issue concerning the contribution of automobile lead exhausts to the country’s lead exposure problem is complex and controversial. In order to complete a fair assessment of this problem, EPA has made a concentrated effort to obtain and review all the medical and scientific evidence. The Agency has repeatedly requested information and comments from the medical and scientific communities as well as the general public. Since the reproposal of the regulations, information gathered through the comment period on the reproposed regulations, earlier comment periods on the originally proposed regulations, and surveys of relevant studies by EPA personnel have been thoroughly reviewed and evaluated by a task force of EPA medical experts and scientists. A paper entitled “EPA’s Position on the Health Implications of Airborne Lead” sets forth in detail the Agency’s evaluation that there is a health basis for reducing the use of lead in gasoline. A copy of this paper is available from the Publications Section, Environmental Protection Agency, 401 M Street SW, Room 238W, Washington, D.C. 20460.
General summary of health issue. Environmental lead exposure is a major health problem in this country. A small but significant portion of the urban adult population and up to 25 percent of children in urban areas are over-exposed to lead. The lead exposure problem is caused by a combination of sources including food, water, air, leaded paint, and dust. The aggregate contribution of lead from all these soürces poses a significant threat to health. However, it is extremely difficult to determine what percentage of the problem each separate environmental factor contributes. Since there are additive sources whose importance varies considerably among individuals it is likewise difficult to determine what impact would be achieved by partial or total reduction of lead from any source. Should the lead in all sources be reduced, however, it seems clear that the situation would be substantially improved. Leaded gasoline is a source of air and dust lead which can be readily and significantly reduced in comparison to these other sources. It is also one of the few lead sources not yet subject to any controls other than EPA’s lead-free gasoline regulations.
Lead from gasoline accounts for approximately 90 percent of airborne lead, total lead additive usage being well over 200,000 tons a year. Lead from stationary sources and deteriorating leaded paint from buildings, combined with lead from gasoline cause high lead levels in dirt and dust. Of these sources, lead from gasoline is the most ubiquitous source of lead found in both the air and the dirt and dust in urban areas. Human exposure to this lead takes place by inhalation and by ingestion of dirt and dust contaminated by air lead fallout. Since exposure to lead among the general population is widespread, it is reasonable that efforts be made to reduce preventable sources of lead exposure including lead emissions resulting from lead in gasoline.
Many of those disagreeing with the re-proposed regulations based their comments on EPA’s failure to show sufficient evidence of adverse health effects specifically caused by the use of lead additives in gasoline. While most agree that the combustion of leaded gasoline causes an increase in the amount of lead in the environment, they do not believe that lead in gasoline represents a sufficient endangerment to health or a *485sufficient risk to the environment to warrant promulgation of controls. The arguments against the position set forth in EPA’s ■ reproposed regulations include the following: (1) EPA has failed to show a clear correlation between lead levels in the air and those in the blood of exposed individuals; (2) lead from dust and dirt does not represent a significant threat to body burden of lead; (3) leaded paint is the primary cause of childhood lead poisoning and lead in gasoline does not play an important role in and lead poisoning or excessive lead exposure; (4) lead in food and water and not airborne lead are the principal sources of lead to the general population.
A discussion of the four major areas of criticism and a summary of the significant new information received since the regulations were reproposed are provided below.
I. Is there a correlation between air lead levels and blood lead levels? A portion of the comments received were critical of EPA’s reproposed regulation on the basis that consistently strong correlations have not been found between air lead and blood lead levels. The conclusion expressed by many comments is that except for persons whose occupations bring them in close contact with environmental lead, exposure to airborne lead does not contribute to increased blood lead levels and does not pose a significant threat to health.
These comments cite several studies which did not demonstrate a strong correlation between air lead and blood lead levels. For example, The Seven Cities Study did not show a close correlation between increase in blood lead levels and simultaneous increases in air lead exposures. Blood lead levels were lower among the New York City residents studies than the Philadelphia residents, despite the fact that air lead exposures among the New York residents were actually greater than those in Philadelphia. Also cited as evidence against EPA’s position is the observation that despite significant increases in the use of lead in gasoline in recent years there have been no discernible increases in blood lead levels of populations so exposed.
Residential differences in blood lead levels have also not always corresponded to differences in air lead exposures. For example, studies of primitive populations, as well as studies of rural U.S. populations, have shown that the blood lead levels in some of these groups are as high or higher than those of persons living in industrial areas, even though the air lead levels in those rural areas should have been much lower. A comparison between London day and night taxi drivers has also shown no significant differences in blood lead levels but did find differences in exposure to carbon monoxide suggesting that despite the possibility that air lead exposure in the day may have been higher than at night, this was not reflected in blood lead increases. However, differences in smoking intensity, as well as actual differences in air lead exposure between groups, could explain these results and neither were measured.
In summary, a number of comments have criticized EPA’s position on the basis that there is not a good correlation between air lead exposure and blood lead levels.
The Agency has weighed against these criticisms studies which have shown that airborne lead does contribute significantly to lead exposure in the general population. For example, using a pilot lead isotope approach, preliminary data show that airborne lead at 2 ug/m3 can contribute as much as Vs to total lead exposure in man. This result is consistent with data concerning the deposition of lead particles in the pulmonary tract and the absorption of such particles into the blood stream.
An unpublished study in Japan similar to the Seven Cities Study, but which has not yet been completely analyzed, has preliminarily demonstrated that airborne lead exposures below 2 ug/m3 affect blood lead levels.
Chamber studies in carefully controlled environments, have shown significant increases in blood lead of men exposed to air lead slightly greater than 3ug/m3.
Differences in the blood lead levels between urban and suburban residents in the *486same geographic area have been found. When comparable groups with similar lead intakes from other sources besides air were studied, blood leads were consistently higher in urban areas and near highways where air lead concentrations were greatest. Thus while correlations between blood lead levels and air leads at lower exposure levels are not always good, the evidence indicates that air lead does contribute to general population lead exposure.
Failure to find consistent correlations does not in the Administrator’s judgment invalidate the above conclusions. Studies which have come to contrary conclusions have generally failed to take into account the influence of other sources of lead on blood lead levels in people being studied. In the Seven Cities Study, for example, these other sources of lead influencing blood lead levels were not adequately considered in the blood lead-air lead comparisons. EPA has re-analyzed the Seven Cities Study and has found that air lead was a significant, though not the most influential factor affecting blood lead levels. Further, in the Seven Cities Study, urban-suburban differences in blood leads between comparable groups were consistently found which at least in part reflect differences in air lead exposure.
In summary, absorption of air lead does contribute to total lead exposure and when added to lead from other sources such as food and water results in total exposure that is excessive. Thus, the partial removal of lead from the air will help to reduce the degree of excess lead exposure which currently exists among adults and children in the United States.
II. Does dust lead contribute to lead poisoning in children? Many comments received by the Agency express the viewpoint that the primary cause of lead poisoning in children is ingestion of lead-based peeling paint. Investigations of cases of clinical lead poisoning in children have repeatedly demonstrated peeling leaded paint as the major source of exposure. Since peeling leaded paint has consistently been observed in the environment of lead poisoned children, many commentors thought it unlikely that lead in dust and dirt could make a significant contribution to this problem. They also point out that lead in dust could be caused by peeling or erosion of leaded paint in or near a home.
One commentor cites X-ray studies of the abdomen among children with lead-poisoning as showing paint chips in the majority of instances. Another argues that differences in blood lead levels between Black and Puerto Rican children could not be explained by exposure to different quantities of lead in dust. Further, studies have shown that animals do not absorb lead from dust as readily as they absorb lead from paint.
Commentors have criticized the Agency for considering that the El Paso Study supports the dustfall hypothesis related to lead in gasoline. In the El Paso Study, children living near a leadsmelter were examined for blood lead levels and for sources of lead in their environment. These results showed that children living nearest the smelter had the highest blood lead levels and that dust lead was a probable major cause. Many commentators, however, considered the El Paso Study applicable only to stationary lead sources and not to lead in gasoline which is different in particle size and chemical composition from smelter-emitted lead.
EPA recognizes the importance of leaded paint as a source of lead exposure for children and that it is the primary cause of clinical lead poisoning. However, based on the evidence available to it, EPA does not believe that leaded paint is the only significant source of lead contributing to excessive lead exposures in children. The Agency’s position is that numerous sources contribute to childhood exposure including lead in food, water, air, dust, and dirt as well as paint. Among these sources, contaminated dust and dirt from motor vehicles exhausts are believed to be important exposure routes.
Currently the contention that lead contamination of dust and dirt by automotive emissions is a significant source of lead exposure is a hypothesis consistent with *487information provided by a variety of studies. However, at this time, not all links in the argument have been established beyond dispute and no single study has collectively inter-related all steps in the exposure process to conclusively prove or disprove the hypothesis. Despite the existing uncertainties, comments received from the majority of scientists not affiliated with industrial or environmental groups support the contention that dust is an important source of exposure. This is based on the following evidence:
A. Environmental sampling in a number of cities has demonstrated the ubiquitous presence of lead contaminated dust in urban areas. These measurements were taken inside and outside of buildings including homes and schools. Dust lead measurements outside homes commonly ranged from 0.1 to 0.5 percent lead by weight. Measurements well in excess of 0.5 percent have also been recorded. Inside homes, samples were found to contain lead contents ranging from 0.05 to 0.2 percent and in some instances as high as 0.5 percent. Current Federal regulations have already established that lead concentrations in paint in excess of 0.5 percent represent a definite hazard to children and serious consideration is being given to reducing the allowable level to 0.06 percent. In testimony before the United States Senate, Dr. Merlin Du-Val, at the time Assistant Secretary of Health and Scientific Affairs at HEW, commented on an appropriate safe level for lead in paint:
Based on information now available to us, we are satisfied that it is technologically feasible, and desirable from a health viewpoint to more toward the .06 percent standard recommended by the American Academy of Pediatrics.
B. As was stated above, high lead concentrations in dust are prevalent in urban areas. It is not clear in all instances, which sources are contributing most to this contamination. Comments received by the agency point out that high lead levels in some cases may be caused by the chipping or peeling of leaded paint from interior and exterior surfaces. EPA agrees that this is true. In other cases, the lead dust content is clearly the result of lead emission from stationary sources such as smelters. However, EPA believes an important and the most ubiquitous source of lead in dust is the exhaust of automobiles using leaded gasoline. Annually, over 200,000 tons of lead are used as additives in gasoline. The vast majority of this lead is emitted into the environment. Although significant amounts of lead remain airborne for extended periods of time, evidence indicates that a large quantity of the exhaust lead rapidly settles to the ground within several hundred feet of the source. Measurements of lead in dust and soil further indicate that lead content decreases with increased distance from the roadway. It has also been found that dust lead levels in homes near heavily traveled roadways are significantly higher than in comparable homes located along side streets. It should be noted that the majority of studies reporting high levels of lead in dust and dirt did not associate sources of peeling leaded paint or stationary lead sources with the lead dust measurements. Accordingly, the Agency believes that in most circumstances lead from automobile exhaust is the primary source of lead in dust and soil in urban areas.
C. The general environment of urban children commonly includes dirt and dust contaminated with lead. A large percentage of children, especially between the ages of one and three years, are known to ingest non-food objects in their mouths. It has been demonstrated that children living in high dust lead environments have greater quantities of lead on their hands than children living in less contaminated environments. The existence of leaded dust on the hands of urban children has been highlighted by the common occurrence of inadvertent lead contamination of finger prick blood lead specimens taken from these children.
D. Children who ingest leaded dust and dirt can be expected to absorb some of the lead into their bodies. Though it is difficult to determine the precise amount of lead *488that would be absorbed, animal experiments suggest that appreciable quantities of this lead, whether from smelters, paint or gasoline exhaust, are absorbed. Further, it has also been shown that at least some children residing in environments heavily contaminated by leaded dust and dirt absorb enough to suffer from subclinical and even clinical effects of lead overexposure. This was particularly true in the case of El Paso, mentioned above. Though the lead source was a smelter, animal studies indicate that lead in dust due to leaded gasoline would be absorbed in quantities comparable to that emitted by the smelter. Another study from Charleston, South Carolina indicates that children residing in homes near high soil lead concentrations had a greater frequency of lead poisoning than children residing in less contaminated areas. This study suggests that lead from soil was absorbed, although it is not clear what sources were primarily responsible for those high soil lead levels. It should be further noted that instances such as those above, coupled with known high levels of lead in dirt and dust, indicate that children could easily ingest enough lead by this route to be significant.
E. Various studies indicate that cases of lead poisoning and significant over-exposure are not always associated with urban home environments in which sources of peeling or chipping leaded paint were observed. These studies include children residing primarily in inner city areas. Admittedly children may be exposed to peeling or chipping leaded paint in environments away from their own homes. However, since several recent studies indicate that up to 60 percent of children with excessive lead exposure are known to not reside in homes where peeling lead based paint can be found, it is unlikely that peeling paint exposure away from the homes accounts totally for this difference. Furthermore, extension of blood lead screening programs outside of slum areas indicates that the lead exposure problem is found in children residing in higher income areas where peeling paint is not frequent and exposure to this source away from the home is less likely. In conjunction with these findings, residence near roadways have been found to contain higher quantities of lead than those measured away from the road. Findings such as these indicate that in some circumstances dust lead is an important factor and at times may be the primary factor contributing to excessive lead exposure associated with subclinical if not clinical effects.
F. Clinical symptoms resulting from very high lead exposure in children are known to be associated with permanent neurologic damage. It has also long been suspected, but not proven beyond doubt that lead exposures below those sufficient to cause clinical symptoms in children are also harmful. In particular it has been observed that physiologically significant biochemical changes occur in children with excessive exposures below clinical toxicity and it has been proposed that these changes are reflective of subclinical changes that precede overt disease. Recently available scientific information, though far from completely resolving this issue, supports the view that adverse effects due to lead in children are not confined only to situations in which overt clinical symptoms of lead poisoning occur. Included in these findings are increased subtle neurological impairments among children more highly exposed to lead below levels known to cause clinical disease.
III. Will a reduction of lead in gasoline reduce the incidence of clinical lead poisoning in children? Ingestion of peeling paint has long been recognized as the primary cause of clinical lead poisoning in children. This position has been expressed in many comments received by the Agency including those from several noted authorities in the field of lead poisoning. For this reason, numerous comments have questioned the need to reduce lead in gasoline on the basis that this action would have little if any impact on reducing the incidence of clinical lead poisoning in children.
While EPA recognizes the importance of leaded paint as a source of lead for children and has supported governmental efforts to reduce this risk, the findings of several *489studies suggest that lead poisoning can develop in the absence of significant sources of leaded paint. Though this possibility-does not affirm that reducing lead in gasoline will reduce the incidence of lead poisoning in children it indicates that lead in gasoline may, in conjunction with other nonpaint sources, contribute to the development of lead poisoning. Whatever the impact this reduction may have upon clinical lead poisoning, the action will significantly reduce lead exposure among children.
EPA is also concerned about the probability that children exposed to lead at levels below those associated with clinical poisoning are also being adversely affected. Several effects identified as subclinical lead effects include impairment of fine motor functions, and altered behavior.
It is noteworthy that in a significantly large percentage of excessive lead exposure cases (up to 50 percent in some instances) peeling lead based paint in the home cannot be identified as a source of the exposure. Thus, while leaded paint is recognized as the major cause of childhood lead poisoning, it is not clear that leaded paint is singly responsible for the large degree of excess childhood lead exposure in this country.
IV. Excess lead exposure among the general population could result from a combination of lead sources, not one of which by itself is sufficient to be a problem. Under these circumstances, would it not be preferable to formulate a control strategy based upon reducing lead levels among those sources that contribute the most to this total exposure? It is generally agreed that food is the major source of lead to the general population. A World Health Organization expert committee reports that according to the results of total diet studies in the industrialized countries, the total intake of lead from food generally ranges from 200-300 ug per person per day. WHO further states that based upon available data, these levels are similar to those found in the past 30-40 years and that no upward trend in lead levels in food is evident.
This information suggests that the level of lead in food has remained relatively constant in recent times. Though lead in food would certainly contribute to total lead exposure for the general population, lead in food is probably not the source that is most readily reduced in the event that total exposure to lead is excessive. According to WHO, “Any increase in the amount of lead derived from drinking water or inhaled from the atmosphere will reduce the amount that can be tolerated in food. The lead in air is probably the contribution that is most accessible to action for reducing the total body burden of lead, especially where this fraction is large compared with that absorbed from food.”
V. What new information has become available since reproposals of the regulation and as a result of the additional comment period? The majority of comments addressed the evidence presented by EPA in support of its proposed regulation and did not introduce new evidence. The number of comments received were approximately evenly divided between those in favor and those against. The bulk of comments critical of EPA’s health position was submitted by industry or industry affiliated scientists. Independent scientists who commented, not affiliated directly with the industry or environmental groups, were in favor of the regulation by approximately 2/1. Most favorable comments, though often from scientists knowledgeable in the field of lead, provided testimonial support rather than new evidence. Most new data that either was presented in comments or which subsequently became available to EPA does support the need to reduce lead emissions from automobiles. Among these latest data are the following:
(1) Studies of subclinical lead effects in children continue to suggest that fine motor function and behavior are affected. Though this issue is not completely resolved, the new data emphasize the potential subclinical risk.
(2) It has been reaffirmed that high dust lead levels, up to 1% lead content, have been found in children’s play areas, inside schools and in homes.
*490(3) New evidence reaffirms that high dust lead levels can be caused by leaded gasoline. A recent study in Rochester, New York, demonstrates that high dust lead levels in homes are not always associated with peeling paint and that house dust lead levels are higher in urban than suburban homes. A study in Vermont has shown that higher concentrations of lead in house dust are found in homes located near busy roads compared to homes on sidestreets. This latter point is consistent with the previously known fact that air lead fallout decreases with increased distance from roadways. A study by EPA in New York City indicates that higher household dust and soil lead levels are found in areas with greater dust lead fallout from the air as compared to areas with little lead fallout.
(4) Young children living in homes with high dust lead contents have been found to have more lead on their hands than children in homes with low dust lead content. This finding provides an important link in the dust fall lead hypothesis. The finding is consistent with observations that finger prick blood-lead specimen taken from children are routinely contaminated by lead that is present on the fingers.
(5) Studies continue to indicate that a high degree of exposure to environmental lead is not confined to inner city areas. Cases of over-exposure continue to be reported from areas in which leaded paint would not be expected to be the predominant factor.
(6) Studies from Newark, New Jersey, observed that the frequency of lead poisoning and undue lead exposure is doubled among children living close to major roadways compared to children living farther away.
******
38 Fed.Reg. 33734-33737 (6 December 1973); App. 2-5.
. 42 U.S.C. § 1857f — 6c(c)(l)(A) (1970).
. 163 U.S.App.D.C. 162, 501 F.2d 722 (1974). In the Amoco catalytic converter case we were dealing entirely with un leaded gasoline and the necessity for unleaded gasoline to be used to prevent destruction of the platinum catalytic converter. The Congress and EPA directed unleaded gasoline to be used, not because it had found that leaded gasoline in any quantity was dangerous to human health, but because it had been scientifically established that leaded gasoline would destroy essential elements of the catalytic converter in a very few thousand miles of use. Now this court is concerned with the possible impact of leaded gasoline on human health. This is a matter decided by neither EPA nor this court in Amoco, nor was it decided by Congress when it enacted the legislation.
Unlike the basic facts in Amoco, we have here a situation in which leaded gasoline must necessarily be used in all pre-1975 cars and leaded gásoline may very well be used in many new cars manufactured without the platinum catalytic converter. Some of the parties before us have argued strenuously that the universality of use of the catalytic converter is by no means as certain today as when EPA promulgated its regulation. Other engineering devices to avoid the air pollution of the hydrocarbons, i. e., PNA carcinogenic poisons, may come on the market, and the number of new cars using leaded gasoline increase. Be that as it may, the catalytic converter is a device to take out the PNA carcinogenic elements from gasoline; it is not in any way a device to take out lead.
. Those regulations appear at 38 Fed.Reg. 1254 (10 Jan. 1973) and are reproduced in full as an Appendix to the Amoco opinion.
. 42 U.S.C. § 1857f — 6c(c)(l)(B) (1970).
.Petitioners also raise other objections to the regulations which we do not need to reach. See note 194 infra.
. 5 U.S.C. § 706(2)(A) & (D) (1970).
. The regulations appear at 40 C.F.R. § 80 (1975).
. Doc. 7, Appendix to the Briefs [hereinafter App.] at 27 (28 Nov. 1973).
. Id. at 30.
. Clean Air Act Amendments of 1970, Pub. L.No. 91-604, § 9(a) (31 Dec. 1970).
. 42 U.S.C. § 1857f — 6(c)(1) (1970).
. 36 Fed.Reg. 1486 (30 Jan. 1971).
. No. 72-2233 (D.C.Cir., 29 Oct. 1973) (Wright & McGowan, JJ.)
. 37 Fed.Reg. 3882 (23 Feb. 1972), App. at 23.
. Id.
. Doc. 12, App. at 276 (11 April 1972).
. Our colleagues chide us for characterizing the views of government agencies besides EPA as “independent” scientific appraisals. Yet the majority is willing to accept a view from HEW they deem somewhat favorable to their position because HEW “shares with EPA responsibility for and scientific expertise in the areas of public and environmental health . . . .” Court’s opinion at 419 n.98 of 176 U.S.App.D.C., 47 n.98 of 541 F.2d. We submit that other departments of the Government (e. g., the Departments of Interior, Commerce, and Transportation) share this same responsibility. When Congress created EPA it did not intend for that agency to become the Government’s sole spokesman on environmental concerns.
. Memorandum from Dr. Edward E. David, Jr., Office of Science and Technology, to Donald E. Crabill, Chief, Natural Resources Programs Division, Office of Management and Budget, 1 Nov. 1972, App. at 2471.
. Letter from M. J. Pécora, Under Secretary of .the Interior to George P. Schultz, Director, Office of Management and Budget, 14 Feb. 1972, App. at 2501.
In the preface to its report the NAS panel stated that it had “decided to consider biologic effects of lead not necessarily attributable directly to atmospheric sources and not necessarily at levels of exposure as low or as prolonged as those related to general ambient air. Such consideration was necessary because lead attributable to emission and dispersion into general ambient air has no known harmful effects.” NAS Report, Doc. 14, App. at 314.
. Letter from W. N. Letson, General Counsel, Department of Commerce, to Casper W. Weinberger, Director, Office of Management and Budget, 2 Nov. 1972, App. at 2464.
. Letter from.Merlin K. DuVal, M.D., Assistant Secretary of Health (HEW), 17 Nov. 1972, App. at 2481.
. The Environmental Protection Agency is attempting to gather additional health effects information by asking various members of the medical and scientific community to respond to the following questions. All other interested parties are also invited to address these questions:
(1) In the light of any criticisms you may have of the Goldsmith Hexter approach and the Environmental Protection Agency’s use of a regression equation based upon it (see Figure 3-3 of “Airborne Lead in Perspective.” National Academy of Sciences, 1972, and Table 7 of the “Health Hazards of Lead,” Environmental Protection Agency, revised April 11, 1972, which was corrected in “Corrections and Additions to Health Hazards of Lead,” April 27, 1972), what are the permissible uses and limitations in its application for obtaining reasonable estimates of blood lead levels as a function of air lead exposures?
(2) How accurate a reflection is blood lead of lead body burden? What is the effect of elevated blood leads upon lead body burden? Can small increments in blood lead be expected to result in a significant lead body burden elevation? From a public health point of view, is it permissible to allow slight increases in lead body burdens among the general population when this increment can be prevented? Can the pool of body lead stored in the bone be viewed as totally “physiologically inert”? It is known that chelation therapy of children with elevated blood leads can result in acute clinical symptoms of lead poisoning as a result of mobilizing lead from bone. Is there any evidence that subtle metabolic changes could also mobilize this lead pool under other conditions?
(3) The Environmental Protection Agency has relied upon the National Academy of Sciences (NAS) Report (Appendix C, p. 249, footnote “a”) for estimates of daily respired air by an average adult in its own calculations in Table 7 of the “Health Hazards” paper. How accurate are these estimates of pulmonary physiology (a) that an adult male breathes 23 cubic meters of air per day, (b) that 30 percent of respired lead particles will be retained, and (c) that nearly 100 percent of retained lead particles will be absorbed? Is there additional evidence available in this area besides that which is cited in the NAS Report?
(4) What is an appropriate safety factor for extrapolating industrial threshold limit values (TLV) to the general population? Should such an extrapolation to the general population even be permitted? The proposed TLV for lead is due to be revised to 150 ug./m.3 for a 40-hour week. On a weekly basis, this corresponds to breathing air continually at between 35-40 ug./m.3 of lead. If TLV’s can be extrapolated to the general population, what would be an appropriate safety factor for this purpose so that all groups, including those most susceptible to lead will be protected?
(5) In regard to the dustfall lead theory (p. 139 of the NAS Report): How much of a hazard is dustfall lead to children prone to pica? The Environmental Protection Agency’s calculations indicate that continued ingestion of even small amounts of lead contaminated dusts and dirt containing as much as 0.25-0.35 percent lead could theoretically result in dangerously elevated blood leads among children, or could contribute significantly to additional unnecessary lead burdens in children with other known lead exposures (such as lead paint). Will the Environmental Protection Agency’s proposed 60-65 percent of reduction of leaded auto emissions significantly reduce the risk of this potential contamination?
(6) Although lead paint has traditionally been considered the prime causal factor in childhood lead poisoning, how effective *448would reductions in other known environmental sources of lead exposure (such as dustfall) be in helping to reduce the risk of undue lead exposure among children also exposed to peeling lead paint? How clear is it that all lead poisoning in children is, in fact, caused only by lead paint? Since many years are required to solve the lead paint problem, would the risk of undue lead absorption and possible lead poisoning not be reduced by also decreasing airborne lead and consequently lead in dust?
(7) What is the consequence upon the environment in general of allowing large quantities of lead to be expelled into the atmosphere from motor vehicle exhausts? Does this environmental contamination pose any direct or indirect threat to man?
37 Fed.Reg. 11786, 11787 (14 June 1972), App. at 21.
. 38 Fed.Reg. 1258, 1259 (10 Jan. 1973), App. at 16.
. Doc. 9, App. at 158 (29 Nov. 1972).
. Letter from D. Charles H. Powell to Dr. Kenneth Bridbord, Acting Chief, Health Effects Branch, Office to Research and Monitoring, EPA, 2 Mar. 1973, App. at 2477.
. Letter from Acting Secretary John C. Whitaker to Hon. William D. Ruckelshaus, Administrator of EPA, 26 Feb. 1973, App. at 2494.
. Letter from Elliot L. Richardson, Secretary of Health, Education, and Welfare to Senator John V. Tunney, 29 Jan. 1973, App. at 2479. Admittedly, HEW did soften its position on EPA’s regulations when Richardson’s successor, Caspar Weinberger, became Secretary of HEW. In a letter to David Shoenbrod of Natural Resources Defense Council, Inc., Secretary Weinberger stated,
A final Departmental position, summarizing [HEW’s] technical comments, was submitted to EPA prior to the final rule making. The conclusions stated in the letter to EPA were substantially the same as those expressed in the letter to Senator Tunney [from Secretary Richardson. App. at 2479.] . . .
The position of the Department on the removal of lead from gasoline is that it should be removed because it provides no known direct benefit for human health and may contribute to the body burden of lead, but this should not be done unless it is clear that the health consequences associated with the alternatives to lead in gasoline are indeed less hazardous.
Letter dated 7 Aug. 1973, App. at 2507 (emphasis in original). We do, not however, read Secretary Weinberger’s comments as a repudiation of HEW’s “fínal Departmental position” (emphasis added). Indeed, Sec etary Weinberger appears to be quoting from Secretary Richardson’s earlier letter when he uses the phrase “may contribute to the body burden of lead.” (emphasis in original.) Additionally, Weinberger’s other rationale for lead removal — that lead “provides no known direct benefit to human health” — clearly is not a sufficient basis for EPA’s low lead regulations. Thus, we con-elude that Weinberger’s letter to NRDC did not alter HEW’s prior conclusion that “there is no firm evidence at this time that lead derived from combusted gasoline is harmful to the health of the general public.”
. With reference to the majority’s notes 53 and 77, the distinction between the position of the writer of this opinion and his position in Environmental Defense Fund, Inc. v. EPA (Coahoma), 160 U.S.App.D.C. 123, 128, 489 F.2d 1247, 1252 (1973), is obvious. In Coahoma we had a mass of evidence, often conflicting, on which the Administrator and his aides as the experts were entitled to draw their conclusions. For a court to dispute those conclusions, when there was a mass of evidence to sustain them (or indeed possibly to sustain opposite conclusions), would be for the court to assume an expertise that is properly that of the Administrator. We did not do so, and we thus sustained the Administrator.
In the instant case, as members of a reviewing court we are saying that the Administrator’s process was faulty, that the logic of his conclusion has great gaps in the chain, and that the evidence to support his conclusion is totally insufficient. “The agency must articulate a ‘rational connection between the facts found and the choice made.’ ” Bowman Transp. Inc. v. Arkansas-Best Freight Sys., Inc., 419 U.S. 281, 285, 95 S.Ct. 438, 442, 42 L.Ed.2d 447, 456 (1974). The agency’s choice here is only a guess, not an expert judgment drawn from conflicting evidence. We are thus dealing here with procedure, logic, and substantiality of evidence, not attempting to dispute a supposed expert’s conclusions from a mass of evidence.
. Natural Resources Defense Council v. EPA, No. 72-2233 (D.C.Cir., 28 Oct. 1973).
. 5 U.S.C. § 553(b)(3) (1970).
. 38 Fed.Reg. 33735, App. at 3.
. Id
. Doc. 113, App. at 678.
. Doc. 875. See also EPA letter of 2 Oct. 1975. It is interesting to note the comment of an EPA attorney explaining why he did not advise the court of this letter’s existence at an earlier date: “The letter was filed and indexed with the public comments, under ‘California University’ and I overlooked it before.” EPA letter of 2 Oct. 1975. If EPA had such difficulty locating this document, how can it be argued that this letter was “available” to the public in any meaningful sense? This is indicative of the extraordinary efforts to which the majority opinion is forced to resort in attempting to find compliance with required administrative procedures.
.Dr George W. Wetherill, the author of the letter to EPA, cautioned,
. We have by no means completed analysis of materials obtained in this study . . [M]ore accurate data will be forthcoming . . . . I expect that these results are not peculiar to the individual whom we have studied. Of course, this conclusion requires verification.
Doc. 875. See generally EPA letter of 2 Oct. 1975.
. Doc. 113, App. at 704.
. Doc. 470. See also EPA letter of 26 Sept. 1975.
. See EPA letter of 26 Sept. 1975.
. Doc. 462, App. at 1092.
. EPA letter of 26 Sept. 1975.
. Doc. 85, App. at 596.
. App. at 107 (reference 17).
. 38 Fed.Reg. 33735, App. at 3.
. Id.; App. at 89.
. Doc. 228.
. Court’s opinion at 423 & n.114 of 176 U.S.App.D.C., at 51 & n.114 of 541 F.2d.
. Id. at 423 n.114 of 176 U.S.App.D.C., at 51 n. 114 of 541 F.2d.
. Doc. 257, App. at 957.
. PPG/DuPont Supplemental Brief at 40.
. Id. at 40-41.
. Brief for EPA in Nos. 73-2268 & 73-2269 at 32.
. 38 Fed.Reg. 33736 (subsection E of section II), App. at 4.
. Doc. 117, App. at 718. See App. at 139 (reference 38).
. Doc. 103, App. at 662. See App. at 135 (reference 14).
. Doc. 118, App. at 738.
.Id., App. at 720.
. App. at 130.
. Court’s opinion at 420 of 176 U.S.App.D.C., at 48 of 541 F.2d (emphasis added).
. See note 34 supra.
. See infra at 461-466 of 176 U.S.App.D.C., at 89-94 of 541 F.2d.
. See supra at 452 of 176 U.S.App.D.C., at 80 of 541 F.2d.
. The fact that in its reanalysis EPA “relied primarily on a finding reported in the original study” (Court’s opinion at 423 of 176 U.S.App.D.C., at 51 of 541 F.2d), does not change the fact that only the reanalysis makes the original finding statistically significant. In a lengthy footnote explaining the intricacies of statistical analysis, the majority opinion attempts to play down the difference between the statistically insignificant results of the original Seven Cities study and the statistically significant results attributed to the same study by EPA’s reanalysis. Court’s opinion- at 423 n.112 of 176 U.S.App.D.C., at 51 n.112 of 541 F.2d. Regardless of the mathematical percentages involved, we submit that reasoned decision-making requires scientists and administrators to rely only on observed relationships which can be statistically confirmed, i. e., results which “the data support as showing a real effect, as opposed to . . . result[s] that might readily arise from sampling variation.” Id., quoting from F. Mosteller, et aI., Probability With Statistical Applications 307 (2d ed. 1970).
.Court’s opinion at 423 of 176 U.S.App.D.C., at 51 of 541 F.2d.
. Court’s opinion at 423-424 of 176 U.S.App. D.C., at 51-52 of 541 F.2d (footnote omitted).
. Id. at 424 of 176 U.S.App.D.C., at 52 of 541 F.2d (footnote omitted).
. Compare 38 Fed.Reg. 33736, App. at 4 with 38 Fed.Reg. 33737, App. at 5.
. See Court’s opinion at 424 & n.117 of 176 U.S.App.D.C., at 52 & n.117 of 541 F.2d.
. Section 211(c)(2)(A), 42 U.S.C. § 1857f-6c(c)(2)(A) (1970).
.See note 135 infra.
. Court’s opinion at 425 n.117 of 176 U.S.App.D.C., at 53 n.117 of 541 F.2d (emphasis in original).
. Id. at 421 n.102, 424, & 425 n.119 of 176 U.S.App.D.C., at 49 n.102, 52, & 53 n.119 of 541 F.2d.
. Id. at 424 of 176 U.S.App.D.C., at 52 of 541 F.2d (emphasis added).
. Id. at 424 n.117 of 176 U.S.App.D.C., at 52 n.117 of 541 F.2d.
. Id. at 421 n.102 of. 176 U.S.App.D.C., at 49 n.102 of 541 F.2d.
. Thus, our chart at 465 of 176 U.S.App.D.C., at 93 of 541 F.2d infra is not, as our colleagues assert (Court’s opinion at 421 n.102 of 176 U.S.App.D.C., at 49 n.102 of 541 F.2d) “fundamentally misleading.”
. Portland Cement Ass’n v. Ruckelshaus, 158 U.S.App.D.C. 308, 327, 486 F.2d 375, 394 (1973), cert. denied, 417 U.S. 921, 94 S.Ct. 2628, 41 L.Ed.2d 226 (1974).
. Id. at 326, 486 F.2d at 393.
. See id. at 326 & n.67, 486 F.2d at 393 & n.67; Mobil Oil Corp. v. FPC, 157 U.S.App.D.C. 235, 248 n.39, 483 F.2d 1238, 1251 n.39 (1973). See generally K. Davis, Administrative Law Treatise § 15.10 at 402 (1958).
. Wright, The Courts and the Rulemaking Process: The Limits of Judicial Review, 59 Cornell L.Rev. 375, 383 n.34 (1974) (emphasis added).
. Bowman Transp., Inc. v. Arkansas-Best Freight Sys., 419 U.S. 281, 288 n.4, 95 S.Ct. 438, 443 n.4, 42 L.Ed.2d 447, 457 n.4 (1974).
. Can anyone imagine a responsible corporate executive launching a new enterprise with important economic consequences on data as flimsy and counter-indicative as EPA had here? (And private enterprise is supposed to take risks.) Or, a responsible government administrator — in any other field except one such as the environment, where emotional devotion to a “cause” is high — issuing regulations with as portentuous effect as here on the same factual certainties that EPA had? We find these considerations relevant to the “arbitrary and capricious” standard we think should be applied here.
. Doc. 141. The court’s opinion describes the draft Third Health Document as “substantially identical to the final draft” and goes on to explain,
Chapters I, II, V, VII, and VIII [the “Summary and Conclusions”] of this draft . differ only in minor wording changes from the final Third Health Document. Chapters III, IV, and VI of the final version contain a few additional inconsequential paragraphs and references. ...
Court’s opinion at 425 & n.121 of 176 U.S.App.D.C., at 53 & n.121 of 541 F.2d. Ráther than quibble over the degree of similarity or dissimilarity between the draft and final document, we will simply describe with more specificity what the court refers to as “a few additional inconsequential paragraphs and references.” Significantly, the three sections with the more extensive revisions bear these far from inconsequential titles: III. “Health Aspects of Lead Exposure”; IV. “Can an Acceptable Lead Body Burden be Defined?”; and VI. “Lead Exposure from Dustfall.” The following chart roughly indicates the amount of revision performed on each of these sections:
SECTION III. Paragraphs: 4 new or revised paragraphs compared to 24 original paragraphs.
References: 3 new references added to 31 original references (all three dated 1-2 Oct. 1973).
SECTION IV. Paragraphs: 4 new or revised paragraphs compared to 12 original paragraphs.
References: 5 new references added to 10 original references (four dated 1-2 Oct. 1973; one dated Mar. 1973).
SECTION VI. Paragraphs: 13 new or revised paragraphs compared to 26 original paragraphs.
References: 6 new references added to 41 original references (one in print; one in prepublication draft; others dated 28 Aug. 1973, Dec. 1972,1972, and May 1971)"
There is nothing in the Administrator’s decision (i. e., the Preamble) or the Third Health Document that indicates that EPA would have made the same decision without the new supportive evidence indicated above. Unless the Administrator makes' it clear that he considered this additional evidence “inconsequential,” we cannot excuse his failure to make it available for public comment. An agency may not keep secret evidence on which it intends to rely to support its decision.
.EPA denied this request by a letter dated 4 December 1973, after the regulations were promulgated. On 21 November the Lead Industries Association teletyped the Administrator “Final lead in gasoline regulations now being circulated represent substantial changes in both the lead phase down schedule and supporting documentation. The major changes deserve public comment regarding their validity.” (emphasis in original.)
. The medical director of Nalco, the only one to comment on the new EPA health data circulated 19 October 1973, knew exactly what the agency was up to: “The present EPA health' document is one of many final or draft documents which we have had the opportunity to review. In each instance the document has presented a different basis for the same summary and conclusions.” Doc. 821 at 44.
Compare the action of the EPA here with its action in Environmental Defense Fund, Inc. v. EPA (Coahoma), 160 U.S.App.D.C. at 128, 489 F.2d at 1252. See note 28 supra. There the Administrator accumulated a mass of data, much of it conflicting, pondered it, and drew his conclusion. Here the Administrator advanced proposed regulations, saw his supporting data riddled by other scientific minds, shifted twice the supporting rationale, and three years later came out with virtually the identical regulations he had advanced three years earlier. This court respected the Administrator’s decision in Coahoma, but here the Administrator’s action calls logically for a different result.
. Court’s opinion at 425 n.121 of 176 U.S.App.D.C., at 53 n.121 of 541 F.2d.
. Doc. 141 (emphasis in original).
. Docs. 485 & 486, App. 1474-1502, 1502-16.
. See our discussion of the evidence relied upon by EPA infra at 470-480 of 176 U.S.App.D.C., at 98-108 of 541 F.2d. Recent history would indicate that the prime mover behind implementation of the Clean Air Act has not been Congress or EPA, but the courts — specifically this court:
The agency wrote the controversial clean-air regulations in 1973 only after U.S. Court of Appeals here [Washington, D.C.] ordered it to do so. Subsequently, too, the EPA readily granted cities such as Los Angeles more time to meet the standards. Environmentalists went to court to force the EPA to enforce the deadline.
The agency even seems to have felt compelled to apologize for its displays of vigor. In announcing the auto-pollution rules, including the requirement for drastic gas rationing in Los Angeles, William Ruckelshaus, who was the EPA administrator, blamed the courts: “I know this is controversial, but I am under a court order and this is the only demonstrable way to meet the national clean-air standards.”
House, Lost in a Smog Bank, Wall Street Journal, 16 Jan. 1976, at 4, col. 3.
. Interestingly, the majority opinion cites one other submission — a one page letter from the Acting Deputy Assistant to the Secretary of Interior — to support its position that “comments [on the draft of the Administrator’s decision and the draft Third Health Document] were received and acted upon." Court’s opinion at 425 & n.122 of 176 U.S.App.D.C., at 53 & n.122 of 541 F.2d (footnotes omitted and emphasis added). We are curious how EPA “acted upon” or considered this letter, which severely criticized EPA’s current draft of the regulations, when the letter bears a 7 December 1973 date. We can only surmise that this comment had no profound effect on the 28 November 1973 regulations.
. Secretary Brinegar raised serious questions as to the economic impact of the contemplated regulations: “. . . in view of the emergency that we now face it would be unwise to undertake actions that could work in the opposite direction [i.e., to decrease gasoline supplies], as does the lead reduction schedule. . . . Do not the extraordinary recent changes in energy supply, demand, and price • relationships force us to reappraise and re-analyze such a far reaching decision?”
He mentioned the health problem only to query: “With future gasoline usage now likely
to fall significantly, . . . is it not necessary to re-calculate future total lead emissions and their possible health hazards?” Doc. 1269, App. at 2651.
. Doc. 1084, App. at 2577. Like the Secretary of Transportation’s letter, the Treasury Department’s comments were addressed to the economic impact of the regulations, and not to EPA’s health position. EPA’s draft Third Health Document is not referred to in either agency’s communications.
. Doc. 821.
. Doc. 1090.
. Doc. 433.
. Doc. 1089.
. Doc. 1092.
. Our colleagues infer from the date of Ethyl’s second letter to EPA, 29 October 1973, that before this letter was written Ethyl must have received a copy of the 19 October draft Third Health Document (through the FOIA agreement or otherwise). Court’s opinion at 425 n.123 of 176 U.S.App.D.C., at 53 n.123 of 541 F.2d. The record is silent on this point, and the letter from Ethyl’s counsel (Doc. 1090) does not even imply, much less state, that Ethyl had any knowledge of the 19 October draft.
. Court’s opinion at 421 of 176 U.S.App.D.C., at 49 of 541 F.2d (emphasis added and footnote omitted). In a similar vein, the court’s opinion concludes,
The record in this case clearly demonstrates EPA fully satisfied the requirements of administrative due process. In fact, EPA’s efforts to elicit informed comment on its proposed action went far beyond the measures it was required to take. All health-related documents, including internal EPA policy memoranda, were made public upon receipt, and comments on the documents were accepted until the date of final promulgation.
Id. at 425 of 176 U.S.App.D.C., at 53 of 541 F.2d (footnotes omitted).
. The court devotes merely the following one-sentence footnote to the most significant procedural issue in this case:
Since all material information was made available for comment by petitioners and the public, it is not necessary for us to determine whether the Administrator may rely on information not previously made available.
Court’s opinion at 421 n.101 of 176 U.S.App.D.C., at 49 n. 101 of 541 F.2d.
. Id. at 425 n.118 of 176 U.S.App.D.C., at 53 n.118 of 541 F.2d.
. See a discussion on the limitations of the FOIA as a discovery tool in informal rulemaking in 85 Yale L.J. 38, 69-70, 84-86 (1975).
. The unpublished study in Japan, Doc. 468, App. at 1092.
. The second paper of the Newark studies, Doc. 95, App. at 626; one of the Rochester studies, Doc. 118, App. at 720.
. The first pilot lead isotope study, Doc. 113, App. at 678; the Chicago screening, Doc. 117, App. at 718; the Philadelphia study, Doc. 103, App. at 662; one of the Rochester studies, Doc. 118, App. at 738.
. See Court’s opinion 420-121 nn.99 & 100 of 176 U.S.App.D.C., 48-49 nn. 99 & 100 of 541 F.2d, citing Doc. 556, App. at 1962 & Doc. 1259 (NPRA); court’s opinion at 422 n.106 & 108 of 176 U.S.App.D.C., at 50 n.106 & 108 of 541 F.2d, citing Doc. 821 (Nalco).
. Compare id. at 421 n.100 of 176 U.S.App.D.C., at 49 n.100 of 541 F.2d with id. at 420 n.99 of 176 U.S.App.D.C., at 48 n.99 of 541 F.2d. The only other NPRA submission cited in the court’s opinion, Doc. 1259, deals only with the economic analysis prepared for EPA by Bonner and Moore Associates, Inc.
. Id. at 425 of 176 U.S.App.D.C., at 53 of 541 F.2d.
. Id. at 422 n.106 of 176 U.S.App.D.C., at 50 n.106 of 541 F.2d (“Ethyl . . . submitted comments on [the first pilot lead isotope] study in the form of a transcript of the October conference. Doc. 433.”)
. Id. at 423 & n.114 of 176 U.S.App.D.C., at 51 & n.114 of 541 F.2d. See our previous brief remarks on this transcript supra at 453 of 176 U.S.App.D.C., at 81 of 541 F.2d.
. See, e. g., Court’s opinion at 421 & 425 of 176 U.S.App.D.C., at 49 & 53 of 541 F.2d.
.See note 34 supra. Describing how EPA typically assembles a record, William F. Pederson, Jr., an attorney with EPA’s Office of General Counsel, has stated,
At every stage . . . the tendency of the record is to increase in length and become less organized. Factual materials, documents describing the development of policy, and documents included simply to quiet suspicions of bad faith will be jumbled in together.
85 Yale L.J. 38, 70 (1975). These unwieldy and disorganized records plague not only the interested parties, but also the courts.
. Doc. 141 (emphasis in original).
. Doc. 821 at 90.
. 42 U.S.C. § 1857f-6c(c)(l) (1970).
. 42 U.S.C. § 1857Í-1 (1970).
. 42 U.S.C. §§ 1857c-2 through 1857c-5 (1970).
. 38 Fed.Reg. 33734 (1972), App. at 2.
. Court’s opinion at 389-390 of 176 U.S.App.D.C., at 17-18 of 541 F.2d (footnote omitted).
. Id. at 392 of 176 U.S.App.D.C., at 20 of 541 F.2d.
. Poets and politicians concur: “All our past acclaims our future” (Swinburne); “I know no way of judging of the future but by the past” (Patrick Henry).
.In all games of chance in which theoretically the factor of human control is not present, e. g., roulette, dice, the fall of cards, the risk of a given event occurring or not occurring can be calculated with mathematical certainty. In those games in which human or animal skill is presumed to predominate, e. g., football, basketball, horse racing, the facts as to the physical conditions and mental attitude of the contestants are avidly sought by the professional risk-assessors; it is on the basis of known facts that the initial odds are fixed.
. Court’s opinion at 396 of 176 U.S.App.D.C., at 24 of 541 F.2d (emphasis added).
. 162 U.S.App.D.C. 331, 499 F.2d 467 (1974).
. Reply Brief at 44-45.
. Industrial Union Dep’t, AFL-CIO v. Hodgson, 162 U.S.App.D.C. 331, 339, 499 F.2d 467, 475 (1974).
. 42 U.S.C. § 1857c-3(a)(l)(A) (1970) (emphasis added).
. 42 U.S.C. § 1857f-l(a)(l)(1970) (emphasis added).
. 42 U.S.C. § 1857c-3(a)(2) (1970) (emphasis added).
. Section 553(c) provides:
After notice required by this section, the agency shall give interested persons an opportunity to participate in the rule making through submission of written data, views, or arguments with or without opportunity for oral presentation. After consideration of the relevant matter presented, the agency shall incorporate in the rules adopted a concise general statement of their basis and purpose. When rules are required by statute to be made on the record after opportunity for an agency hearing, sections 556 and 557 of this title apply instead of this subsection.
. 5 U.S.C. § 706(2)(A), (C), & (D) (1970).
. Calcutta E. Coast of India and E. Pakistan/U.S.A. Conference v. FMC, 130 U.S.App.D.C. 261, 264, 399 F.2d 994, 997 (1968).
. Citizens to Preserve Overton Park, Inc. v. Volpe, 401 U.S. 402, 415, 416, 91 S.Ct. 814, 824, 28 L.Ed.2d 136, 153 (1971).
. Id. (emphasis added). This statement of the scope of review was reaffirmed very recently by a unanimous Supreme Court in Bowman Transp., Inc. v. Arkansas-Best Freight Sys., Inc., 419 U.S. 281, 284-86, 95 S.Ct. 438, 441-42, 42 L.Ed.2d 447, 454-56 (1974). See also Portland Cement Ass'n v. Ruckelshaus, 158 U.S.App.D.C. 308, 335, 486 F.2d 375, 402 (1973): “We cannot substitute our judgment for that of the agency, but it is our duty to consider whether ‘the decision was based on a consideration of the relevant factors and whether there has been a clear error of judgment.’ Citizens to Preserve Overton Park v. Volpe, 401 U.S. *470402, 416, 91 S.ct. 814, 824, 28 L.Ed.2d 136 (1971). Ultimately, we believe, that the cause of a clean environment is best served by reasoned decision-making.”
. The Supreme Court made this quite clear in Bowman Transp., Inc. v. Arkansas-Best Freight Sys., Inc., 419 U.S. 281, 284 & 285, 95 S.Ct. 438, 441 & 442, 42 L.Ed.2d 447, 454 & 455 (1974) (footnote omitted):
The District Court properly concluded that, though an agency’s finding may be supported by substantial evidence, based on the definition in Universal Camera Corp. v. NLRB, 340 U.S. 474, [71 S.Ct. 456, 95 L.Ed. 456], it may nonetheless reflect arbitrary and capricious action. There seems, however, to be agreement that the findings and conclusions of the Commission are supported by substantial evidence. The question remains whether, as the District Court held, the Commission’s action was “arbitrary, capricious, an abuse of discretion, or otherwise not in accordance with law” as provided in [5 U.S.C. § 706(2)(A) ].
Under the “arbitrary and capricious” standard the scope of review is a narrow one. A reviewing court must “consider whether the decision was based on a consideration of the relevant factors and whether there has been a clear error of judgment. . . Although this inquiry into the facts is to be searching and careful, the ultimate standard of review is a narrow one. The court is not empowered to substitute its judgment for that of the agency.” Citizens to Preserve Overton Park v. Volpe, supra, [401 U.S.] at 416 [91 S.Ct. 814 at 824, 28 L.Ed.2d 136]. The agency must articulate a “rational connection between the facts found and the choice made.”
The lack of a connection between facts found and choices made is the Agency failure we find here.
. The following exchange during the debate on the Clean Air Act typifies the congressional understanding that this court should review for clear errors of judgment:
Mr. ROGERS of Florida. . We have said that if there is a component part of the gasoline or if there is an additive which the facts show will affect the public health, and medical and scientific facts show this, or will prevent the emission standards from being met, the Secretary may act as to the component part or that additive.
Mr. WAGGONNER. What appeals are made available to the manufacturer of an engine or the producer of a fuel, if they take issue with the findings of the Secretary of Health, Education, and Welfare?
Mr. ROGERS of Florida. They have the Administrative Procedure Act.
Mr. WAGGONNER. And only the Administrative Procedure Act?
Mr. ROGERS of Florida. And an appeal to the court, from the Administrative Procedure Act.
Mr. WAGGONNER. Not in all cases do people h.ave the right of appeal to the court under the Administrative Procedure Act. In some instances under the Administrative Procedure Act the decisions are final, when they render a decision.
Mr. ROGERS of Florida. I believe they would in this instance.
Mr. STAGGERS. I think the gentleman from Louisiana was worried about the fact that this would be a hard ruling made in the courts. Of course, these matters come up under the Administrative Procedure Act, and then they have recourse to the courts. If they feel, after all of the hearings have been laid before them, that these were not the facts and that the ruling was wrong, the court will have the final decision.
116 Cong.Rec. (Part 14) at 19230-31 (House) (10 June 1970) (emphasis added).
. 504 F.2d 646 (1st Cir. 1974).
The wisdom of the plan in the ordinary sense is outside our province. “We inquire into the soundness of the reasoning by which the [Agency] reaches its conclusions only to ascertain that the latter are rationally supported.” United States v. Allegheny-Ludlum Steel Corp., 406 U.S. 742, 749, 92 S.Ct. 1941, 1946, 32 L.Ed.2d 453 (1972).
. 42 U.S.C. § 1857c-5(c) (1970).
“Looking to the future, and commanded by Congress to make policy, a rule-making agency necessarily deals less with ‘evidentiary’ disputes than with normative conflicts, projections from imperfect data, experiments and simulations, educated predictions, differing assessments of possible risks, and the like. The process is quasi-legislative in character, and one will search it in vain for those intermediate ‘findings’ of fact which mark the midway point in an adjudicator’s linear march from raw evidence to single, ultimate conclusion.” Amoco Oil Co. v. EPA, [163 U.S.App.D.C. 162] 501 F.2d 722, at 734-35 (1974).138
. 504 F.2d at 655-56 (emphasis added).
. We find our colleagues do not differ materially with us on the standard for review. The court’s opinion concludes, “[I]n the context of ‘arbitrary and capricious’ review, we shall reverse for a ‘clear error of judgment’ only if the error is so clear as to deprive the agency’s decision of a rational basis.” Court’s opinion at 406-407 n.74 of 176 U.S.App.D.C., at 35 n.74 of 541 F.2d. We believe here that the error of the Administrator was so clear as to deprive the agency’s decision of a rational basis (as discussed in Part IV infra) and that the procedures of EPA were so faulty as to be “without observance of procedure required by law” (as discussed in Part I. C. supra).
.In a footnote the court does pay lip service to this well-settled principle of judicial restraint: “Of course, [the rational] basis must be expressed by the agency itself and not supplied by the court. SEC v. Chenery Corp., 332 U.S. 194, 196 [67 S.Ct. 1575, 1577, 91 L.Ed. 1995, 1999] (1947).” Court’s opinion at-n.73 of 176 U.S.App.D.C. at 34 n.73 of 541 F.2d (emphasis added). Accord, Environmental Defense Fund, Inc. v. EPA, where Judge Leventhal speaks for this court,
The interests at stake here are too important to permit the decision to be sustained on the *473basis of speculative inference as to what the Administrator’s findings and conclusions might have been regarding benefits. Sound principle sustains the practice of vesting choice of policy with the Administrator. Its corollary is that the specific decision must be explained, not merely explainable, in terms of the ingredients announced by the Administrator as comprising the Agency’s policies and standards.
150 U.S.App.D.C. 348, 359, 465 F.2d 528, 539 (1972) (emphasis added). Here, too, the interests at stake are too important to permit EPA’s decision to be sustained on the basis of speculative inference as to why the Administrator might have embraced certain studies and rejected others. “[T]he necessary articulation of reasons must appear in the preamble to the promulgated rule or in some other document of equally formal standing.” 85 Yale L.J. 38, 71 (1975), citing Associated Indus. of N.Y. State, Inc. v. United States, 487 F.2d 342, 353 (2d Cir. 1973); Dry Color Mfrs. Ass’n v. United States Dep’t of Labor, 486 F.2d 98, 105-07 (3d Cir. 1973); Kennecott Copper Corp. v. EPA, 149 U.S.App.D.C. 231, 234, 462 F.2d 846, 849 (1972).
. 38 Fed.Reg. 33734, App. at 2.
. Third Health Document, IV-1, App. at 69.
. Third Health Document, IV-1 to -2, App. at 69-70.
. Third Health Document, III — 2, App. at 55.
. Third Health Document, III — 11, App! at 64.
. Court’s opinion at 410-411 of 176 U.S.App.D.C., at 38-39 of 541 F.2d.
. See Third Health Document, Table III — 1, App. at 65.
. Ethyl Supplemental Brief at 41-42; Nalco Supplemental Brief at 30-31; PPG/Du Pont Supplemental Brief at 15-16; NPRA Supplemental Brief at 5-6.
. Third Health Document, VII-1 to -2, Tables VII-1 & VII-2, App. at 142-43 & 145-46. By comparison, the court devotes four and one half pages of text and an eighteen page appendix to its discussion of the topic.
. Third Health Document, III — 1, App. at 142.
. Id.
. Third Health Document, VII-1 to -2, App. at 142-43.
. Tepper, L. B., and Levin, L. S., “A Survey of Air and Population Lead Levels in Selected American Communities” (the Seven Cities study), Dept, of Environmental Health, College of Medicine, U. Cincinnati, Ohio (EPA Contract PN 22-68-28) (December 1972), Doc. 132, App. at 840.
. See excerpts from International Lead Zinc Research Organization, Inc. (ILZRO) submission of 9 March 1973 at 29, Doc. 836, App. at 2386. ILZRO was one of the joint sponsors of the study along with EPA.
. Azar, A., et a/., “Relationship of Community Levels of Air Lead and Indices of Lead Absorption,” at 581-594, Proceedings of International Symposium on Environmental Health Aspects of Lead, Amsterdam (2-6 October 1972), Doc. 20, App. at 387.
. App. at 392.
. App. at 399.
. See Third Health Document, VII-2, App. at 143.
. App. at 145.
. Id.
. Id.
. Court’s opinion at 412 of 176 U.S.App.D.C., at 40 of 541 F.2d (footnotes and citations omitted).
. 38 Fed.Reg. 33735, App. at 3.
. Court’s opinion at 413 of 176 U.S.App.D.C., at 41 of 541 F.2d.
. Id. at 412 of 176 U.S.App.D.C., at 40 of 541 F.2d.
. Id. at 413 of 176 U.S.App.D.C., at 41 of 541 F.2d.
. See Reserve Mining Co. v. EPA, 514 F.2d 492, 511-514 (8th Cir. 1975); United States v. Reserve Mining Co., 380 F.Supp. 11, 39-54 (D.Minn.1974) (supplemental memorandum). By our comparison, we do not mean to imply, as our colleagues assert, that there are no differences between Reserve Mining and the instant case. See Court’s opinion at 413 n.90 of 176 U.S.App.D.C., at 41 n.90 of 541 F.2d. We maintain only that both cases must be explained for the conclusions reached. Specifically we reject the majority’s conclusion that in the instant case the Administrator was not faced with a situation which “required that the propriety of reaching a conclusion about the public from occupational studies be explained.” Id. In Judge Leventhal’s words, the Administrator’s decision “must be explained, not merely explainable.” Environmental Defense Fund, Inc. v. EPA, 150 U.S.App.D.C. 348, 359, 465 F.2d 528, 539 (1972). See note 140 supra. We search in vain for this explanation in the Administrator’s decision (i. e., the Preamble) or the Third Health. Document.
. 38 Fed.Reg. 33735, App. at 3 (emphasis in original).
. Rabinowitz, M., et al., “Lead Metabolism in the Normal Human: Stable Isotope Studies,” 182 Science 725 (November 1973), Doc. 113, App. at 704; Rabinowitz, M., et al., “Studies of Human Lead Metabolism Using Stable Isotope Tracers,” paper presented at EPA-NIEHS Conference on Low Level Lead Toxicity, Raleigh, North Carolina (1-2 October 1973), Doc. 113, App. at 678.
. Tsuchiya, K., et al., “Study of Lead Concentrations in Atmosphere and Population in Japan,” draft, undated, Doc. 468, App. at 1092.
. Knelson, J. H., et al., “Kinetics of Respiratory Lead Uptake in Humans,” Proceedings of the International Symposium on Environmental Health Aspects of Lead, Amsterdam (2-6 October 1972), Doc. 85, App. at 596.
. Daines, R. H. et al., “Air Levels of Lead Inside and Outside of Homes,” 41 Industrial Medicine 26 (October 1972), Doc. 39, App. at 465.
. See note 153 supra.
. 38 Fed.Reg. 33735, App. at 3.
. Id.
. Id. The Third Health Document devotes all of one sentence to the task of supporting the Administrator’s comment:
Based upon consistent differences between average blood lead levels in urban and suburban dwellers, the authors [of the Seven Cities study] concluded, “It is probable that these observations partially reflect lead absorption from ambient atmospheres differing in lead concentration . . . but that factors other than the atmospheric lead level are of relatively greater importance in determining the blood lead levels in population groups.”
App. at 89.
. Court’s opinion at 414 of 176 U.S.App.D.C., at 42 of 541 F.2d (footnote omitted and emphasis added). Accord, Court’s “Appendix A” at 433 & n.15 of 176 U.S.App.D.C., 61 & n. 15 of 541 F.2d.
. Court’s “Appendix A” at 433 of 176 U.S.App.D.C., at 61 of 541 F.2d.
. Court’s opinion at 415 of 176 U.S.App.D.C., at 43 of 541 F.2d (emphasis in original).
. Likewise, the majority opinion fails to provide us with an acceptable analysis, but, unlike the Administrator, our colleagues may be excused since it is not their proper role to explain the actions of the Administrator. See note 140 supra.
.38 Fed.Reg. 33736 (6 December 1973), App. at 4.
.Id.
.Id. at 33735-36, App. at 3^.
. Third Health Document, VI-21 to -22, App. at 131-32.
. 38 Fed.Reg. 33736 (6 December 1973), App. at 4.
. Third Health Document, VI-13, App. at 123.
. Id.
. Doc. 352, App. at 1057 (emphasis added).
. Reply Brief for Petitioners in Nos. 73-2268 and 73-2269 at 17.
. Court’s opinion at 417 of 176 U.S.App.D.C., at 45 of 541 F.2d.
. The NAS Report, on which the majority opinion relies, contains the following definition of “pica”:
The term “pica” has been defined as the repetitive ingestion of things that are not food (e. g., string, dirt, paper, putty, paint, clay and cigarette butts). It is important in a variety of accidental poisonings in preschool children. The behavioral biologic factors responsible for this age-related activity are not understood. Lourie and his associates have stressed disturbed parent-child relations as a factor that can intensify pica in the child, whereas Neumann has suggested that it may be a vestigial instinct. Lourie and associates and Sobel found that the habit begins at the age of about 12 months, that it may persist until the age of 3-5 years, and that it occurs in at least 50% of the children in both middle-class and poverty groups. Thus, pica is apparently a rather common behavioral activity in normal preschool children, and it may become intensified in response to stress.
NAS Report at 133 (footnotes omitted and emphasis added).
. Court’s opinion at 415 of 176 U.S.App.D.C., at 43 of 541 F.2d.
. The majority opinion states that “the Reserve Mining court held, a reasonable hypothesis supported by evidence is a sufficient basis for regulating under the ‘will endanger’ standard . . ..” Id. at 419 n.96 of 176 U.S.App.D.C., at 47 n.96 of 541 F.2d (emphasis added). We see no inconsistency between this statement of the Eighth Circuit’s holding in Reserve Mining and our position in the instant case. We reject the Administrator’s dustfall hypothesis because it is not a reasonable hypothesis and because its causal connection is not supported by the evidence. Since we are not familiar with the record (i. e., the scientific evidence) relied upon by the Reserve Mining court, we cannot say how that evidence compares with the studies relied upon by the Administrator in this case. Our colleagues, however, assert that the Administrator’s hypothesis here was “undoubtedly . . more supported] in studies already made than the hypothesis that justified regulation in Reserve Mining,” and that in Reserve Mining the evidence did not establish the causal connection between the allegedly offensive pollutant (asbestos fibers) and the potential health hazard (cancer). Id. at 418 of 176 U.S.App.D.C., at 46 of 541 F.2d. If the Eighth Circuit did, indeed, rest its decision on a less reasonable, less supported, hypothesis than the one involved in this case, we can only conclude that this case and Reserve Mining will soon have one more thing in common; they both will be wrongly decided.
. Petitioners have raised other objections to these regulations which we do not reach:
(1) Did the EPA properly consider other technologically or economically feasible means of achieving emission standards under section 202 of the Clean Air Act?
(2) Did EPA adequately consider such alternative means of regulation as air quality criteria and state plans pursuant to sections 107 through 110?
(3) Was the EPA required to file an environmental impact statement or its functional equivalent?
(4) Has the EPA given adequate consideration to the adverse health consequences of lead reduction as required by section 211?
(5) Is the lead reduction schedule provided for in the regulations arbitrary and capricious?
. Sections 10(e) and 4 of the APA, respectively.
..It is significant that the First Circuit in South Terminal Corp. v. EPA quoted the same portion of International Harvester (re the effect of comments on rulemaking) as does the court, yet in evaluating the degree of reliance EPA was entitled to place on an ambient air report, stated, “Moreover, it was published-after the plan was announced and interested parties have not had an opportunity to criticize the findings.” 504 F.2d 646, 664 (1st Cir. 1974). And later, in discussing the whole broad issue of the adequacy of EPA air measurements, the First Circuit said, “[T]he record as now constituted, made without specific consideration of the mentioned objection, leaves us in doubt as to whether or not there is a rational basis for EPA’s estimate . .” 504 F.2d at 666.
. See note 153 supra.
. Transcript of Lead Portion of Meeting of EPA Hazardous Materials Advisory Committee, Washington, D.C., 26 February 1973, Doc. 325, App. at 991-92.
. See supra at 467 of 176 U.S.App.D.C., at 95 of 541 F.2d.
. Hamlet, Act III, scene 1, line 57 ei seq.