This is the third occasion on which this court has been asked to review Environmental Protection Agency (EPA or Agency) regulations promulgated under authority of the Clean Air Act, as amended, 42 U.S.C. § 7401 et seq. (Supp. I 1977) (the Act), and specifically designed to deal with the health problems associated with lead in the ambient air. In Amoco Oil Co. v. EPA, 501 F.2d 722 (D.C.Cir.1974), we upheld regulations prohibiting the sale of leaded gasoline for use in automobiles equipped with "catalytic converter" devices for controlling exhaust emissions and requiring widespread retail marketing of at least one grade of unleaded gasoline. And in Ethyl Corp. v. EPA, 541 F.2d 1 (D.C.Cir.) (en banc ), cert. denied, 426 U.S. 941, 96 S.Ct. 2663, 49 L.Ed.2d 394 (1976), we affirmed an EPA order requiring annual reductions in the lead content of leaded gasoline. In the present consolidated cases we are asked to review EPA regulations establishing national ambient air quality standards for lead. These air quality standards prescribe the maximum concentrations of lead that will be permitted in the air of our country. We must decide whether EPA's Administrator acted within the scope of his statutory authority in promulgating these regulations and, if so, whether the evidence adduced at the rulemaking proceeding supports his final determinations. In addition, we must examine the petitioners' claims that infirmities in the procedures employed by EPA in this rulemaking warrant remand of the regulations to the Agency. Petitioners are the Lead Industry Association, Inc. (LIA), a nonprofit trade association whose 78 members include most of the country's producers and commercial consumers of lead (No. 78-2201), and St. Joe Minerals Corporation (St. Joe) (No. 78-2220).1

Man's ability to alter his environment to achieve perceived goals has undoubtedly made an enormous contribution to his economic and social well-being. This undertaking is not, however, without attendant costs. One of these costs is the toll that these alterations may exact on the environment itself and, in turn, the dangers that this may pose for the public health and welfare. Unfortunately, man's ability to alter the environment often far outstrips his ability to foresee with any degree of certainty what untoward effects these changes may bring. The issues presented by these cases illustrate this sad fact.

Lead's environmental significance is a consequence of both its abundance and its utility. The relative abundance of lead in the earth's crust makes it unique among the toxic heavy metals. EPA's "Air Quality Criteria For Lead" (hereinafter cited as CD) 1-1, Joint Appendix (JA) 1105. And centuries of mining and smelting, and the use of lead in a variety of human activities, have increased the natural background concentration of lead in the environment. Id. But it is only since the industrial age and the use of lead as a gasoline additive that lead has become pervasive. Id. at 1-2 1-3, JA 1106-1107. Today lead is ubiquitous. It is found in almost every medium with which we come into contact food, water, air, soil, dust, and paint, id. 1-1, JA 1105, each of which represents a potential pathway for human lead exposure through ingestion or inhalation. The widespread presence of this toxic metal in the environment poses a significant health risk. Lead is a poison which has no known beneficial function in the body, id. 1-12, JA 1116, but when present in the body in sufficient concentrations lead attacks the blood, kidneys, and central nervous and other systems and can cause anemia, kidney damage, severe brain damage, and death. Id. 1-6 1-9, JA 1110-1113.2

There are three major sources of the body's lead burden. In most people the largest source is diet. CD 7-9, JA 1179.3 Another source, particularly in children, is the habit of placing hands, objects, and materials in the mouth.4 The third major source is the ambient air; airborne lead is deposited in the respiratory tract as a person breathes lead-contaminated air and is subsequently absorbed into the bloodstream. CD 1-5, JA 1108. Once the lead is in the bloodstream its source is immaterial; total lead intake is the sum of the intake from all these sources. The multiplicity of sources of lead intake increases the difficulty of controlling human lead exposure. Much of the protective activity in this area has focused on limiting the amount of lead in the ambient air, the most controllable source of lead exposure. In this country, by far the largest source of lead emissions accounting for 88 percent of total lead emissions according to EPA estimates is the exhaust of motor vehicles powered by gasoline containing lead additives. CD 5-3, JA 1140. Another eight percent of lead emissions is the result of solid waste incineration and combustion of waste oil. Id. Industrial facilities account for the remaining four percent of total lead emissions. Id.

Acting pursuant to authority conferred on it by Congress in the Clean Air Act, as amended, 42 U.S.C. § 7401 et seq., EPA has been involved in regulation of lead emissions almost since the Agency's inception.5 Its initial approach to controlling the amount of lead in the ambient air was to limit lead emissions from automobiles by restricting the amount of lead in gasoline. To this end it promulgated the regulations which we upheld in Amoco Oil Corp. v. EPA, supra, and Ethyl Corp. v. EPA, supra. However, in 1975 the Natural Resources Defense Council, Inc. (NRDC), and others brought suit against EPA claiming that the Agency was required by Section 108 of the Clean Air Act, 42 U.S.C. § 7408, to list lead as a pollutant for which an air quality criteria document would be prepared, and for which national ambient air quality standards should be promulgated under Section 109 of the Act, 42 U.S.C. § 7409. The District Court agreed with NRDC and directed the Administrator to list lead as a pollutant under Section 108 of the Act, by March 31, 1976. Natural Resources Defense Council, Inc. v. Train, 411 F.Supp. 864 (S.D.N.Y.1976). The Second Circuit affirmed, 545 F.2d 320 (2d Cir. 1976), and EPA initiated the proceedings outlined in the statute which are under review here.

The first step toward establishing national ambient air quality standards for a particular pollutant is its addition to a list, compiled by EPA's Administrator, of pollutants that cause or contribute to air pollution "which may reasonably be anticipated to endanger public health or welfare(.)" Section 108(a)(1), 42 U.S.C. § 7408(a)(1). Within twelve months of the listing of a pollutant under Section 108(a) the Administrator must issue "air quality criteria" for the pollutant. Section 108 makes it clear that the term "air quality criteria" means something different from the conventional meaning of "criterion"; such "criteria" do not constitute "standards" or "guidelines," but rather refer to a document to be prepared by EPA which is to provide the scientific basis for promulgation of air quality standards for the pollutant. This criteria document must "accurately 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." Section 108(a)(2), 42 U.S.C. § 7408(a)(2).

At the same time as he issues air quality criteria for a pollutant, the Administrator must also publish proposed national primary and secondary air quality standards for the pollutant. Section 109(a)(2), 42 U.S.C. § 7409(a) (2). National primary ambient air quality standards are standards "the attainment and maintenance of which in the judgment of the Administrator, based on such criteria and allowing an adequate margin of safety, are requisite to protect the public health." Section 109(b)(1), 42 U.S.C. § 7409(b)(1). Secondary air quality standards "specify a level of air quality the attainment and maintenance of which in the judgment of the Administrator, based on such criteria, is requisite to protect the public welfare from any known or anticipated adverse effects associated with the presence of such air pollutant in the ambient air." Section 109(b)(2), 42 U.S.C. § 7409(b)(2). Effects on "the public welfare" include "effects on soils, water, crops, vegetation, manmade materials, animals, wildlife, weather, visibility, and climate, damage to and deterioration of property, and hazards to transportation, as well as effects on economic values and on personal comfort and well-being." Section 302(h), 42 U.S.C. § 7602(h). The Administrator is required to submit the proposed air quality standards for public comment in a rulemaking proceeding, the procedure for which is prescribed by Section 307(d) of the Act, 42 U.S.C. § 7607(d).

Within six months of publication of the proposed standards the Administrator must promulgate final primary and secondary ambient air quality standards for the pollutant. Section 307(d)(10), 42 U.S.C. § 7607(d)(10). Once EPA has promulgated national ambient air quality standards, responsibility under the Act shifts from the federal government to the states. Within nine months of promulgation of the standards each state must prepare and submit to EPA for approval a state implementation plan. Section 110(a)(1), 42 U.S.C. § 7410(a) (1). These state implementation plans must contain emission limitations and all other measures necessary to attain the primary standards "as expeditiously as practicable," but no later than three years after EPA approval of the plan, and to attain the secondary standards within a reasonable period of time. Section 110(a)(2)(A) & (B), 42 U.S.C. § 7410(a)(2)(A) & (B). The Administrator is authorized to extend the deadline for attainment of the primary air quality standards by two years, but thereafter it must be met. Section 110(e), 42 U.S.C. § 7410(e).

III. THE LEAD STANDARDS RULEMAKING PROCEEDINGS

As required by statute, EPA's first step toward promulgating air quality standards for lead was to prepare a criteria document. The Lead Criteria Document was the culmination of a process of rigorous scientific and public review, and thus is a comprehensive and thoughtful analysis of the most current scientific information on the subject. The Lead Criteria Document went through three major drafts, and three separate reviews, including public meetings by the Subcommittee on Scientific Criteria for Environmental Lead of EPA's Science Advisory Board (SAB Lead Subcommittee).6 The Agency reviewed over 280 public comments, most of a sophisticated scientific nature, before it issued the final Criteria Document. Members of the public, industry (including the petitioners in these cases), environmental groups, the scientific community, and state and federal government agencies actively participated in the review of the drafts. Notice of the meetings of the SAB Lead Subcommittee was published in the Federal Register, and the drafts of the Criteria Document which were to be reviewed were available before the meetings. A formal record and a transcript of the proceedings were kept, and a review of the transcript shows that scientists with differing views could and did exchange ideas with each other as well as agency staff, and that all were questioned by the members of the Subcommittee.

EPA released its "Air Quality Criteria For Lead" on December 14, 1977. 42 Fed.Reg. 63076, JA 1480. The document was "prepared to reflect the current state of knowledge about lead specifically, those issues that are most relevant to establishing the objective scientific data base that will be used to recommend an air quality standard for lead that will adequately safeguard the public health." CD 1-1, JA 1105. Accordingly, the Criteria Document examined a large number of issues raised by the problem of lead in the environment. One of these was the effects of lead exposure on human health. The Criteria Document concluded that, among the major organ systems, the hematopoietic (blood-forming) and neurological systems are the areas of prime concern. CD 1-12, JA 1116. Its discussion of the effects of lead on these two organ systems is central to our review of the lead standards.8

The Criteria Document identified a variety of effects of lead exposure on the blood-forming system. We will discuss only the effects that played an important role in the Administrator's analysis. Anemia, which can be caused by lead-induced deformation and destruction of erythrocytes (red blood cells) and decreased hemoglobin synthesis,9 is often the earliest clinical manifestation of lead intoxication. CD 11-7, 11-8, 11-13, JA 1229-1230, 1235. Symptoms of anemia include pallor of the skin, shortness of breath, palpitations of the heart, and fatigability.10 The Criteria Document concluded, after a review of various studies, that in "children, a threshold level for anemia is about 40 u(g) Pb/dl, whereas the corresponding value for adults is about 50 ug Pb/dl." CD 11-13, JA 1235. (The concentration of lead in the blood is measured in micrograms of lead per deciliter of blood ug Pb/dl.)

The Criteria Document also examined other more subtle effects on the blood-forming system, associated with lower levels of lead exposure. The most pertinent of these "subclinical"11 effects for purposes of these cases is lead-related elevation of erythrocyte protoporphyrin (EP elevation).12 According to the Criteria Document, this phenomenon must, for a number of reasons, be regarded as an indication of an impairment of human health. CD 11-11, JA 1233. First, EP elevation indicates an impairment in the functioning of the mitochondria, the subcellular units which play a crucial role in the production of energy in the body, and in cellular respiration. CD 1-6, 11-11, 11-14, 13-5, JA 1110, 1233, 1236, 1335. Second, it indicates that lead exposure has begun to affect one of the basic biological functions of the body production of heme within the red blood cells.13 Heme is critical to transporting oxygen to every cell in the body. Third, EP elevation may indicate that any reserve capacity there may be in the heme synthesis system has been reduced. CD 11-11, JA 1233. Finally, the Criteria Document noted that lead's interference with the process of heme synthesis in the blood may suggest that lead interferes with production of heme proteins in other organ systems, particularly the renal and neurological systems. CD 11-11, JA 1233. The Criteria Document reported that the threshold for EP elevation in children and women is at blood lead levels of 15-20 ug Pb/dl, and 25-30 ug Pb/dl in adult males. CD 13-8 (Table 13-2), JA 1338 (Table 13-2). While suggesting that some of the initial hematological effects of lead exposure may constitute relatively mild effects at low blood lead levels, the Criteria Document concluded that "they nevertheless signal the onset of steadily intensifying adverse effects as blood lead elevations increase. Eventually, (these) * * * effects reach such magnitude that they are of clearcut medical significance as indicators of undue lead exposure." CD 1-13, JA 1117. The Criteria Document did not identify a particular blood lead level at which regulatory response was appropriate, but it did note with approval the 1975 guidelines issued by the Center For Disease Control, which use elevated EP at blood lead levels of 30 ug Pb/dl as the cut-off point in screening children for lead poisoning. CD 13-5, JA 1335.

The Criteria Document also examined the effects of lead exposure on the central nervous system. Among the most deleterious effects of lead poisoning are those associated with severe central nervous system damage at high exposure levels. The Criteria Document noted that neurological and behavioral deficits have long been known to be among the more serious effects of lead exposure, but it pointed out that there is disagreement about whether these effects are reversible, and about what exposure levels are necessary to produce specific deleterious effects. CD 11-14, JA 1236. Much of the impetus for the debate on these questions has been provided by the continual emergence of new information suggesting that lead exposure levels previously thought to be harmless actually cause significant neurological damage. Id. The more severe neurological effects of high level lead exposure are the clinical syndrome of lead encephalopathy. Early symptoms include dullness, restlessness, irritability, headaches, muscular tremor, hallucinations, and loss of memory. These symptoms rapidly progress (sometimes within 48 hours) to delirium, mania, convulsions, paralysis, coma, and death. Id. at 11-15, JA 1237. The Criteria Document expressed particular concern that the onset of these serious symptoms can be quite abrupt, even in the absence of prior overt or clinical symptoms of disease. Id. at 13-6, JA 1336. After a review of various studies, the Criteria Document concluded that the blood lead threshold for these neurological effects of high level exposure is 80-100 ug Pg/dl in children, and 100-200 ug Pb/dl in adults. Id. at 1-13, 11-18, 11-25, 13-6, JA 1117, 1240, 1247, 1336.

The Criteria Document also went on to consider the evidence on whether lower level lead exposures can affect the central nervous system, particularly in children. It acknowledged that the issue is unsettled and somewhat controversial, but it was able to conclude, after a careful review of various studies on the subject,14 that "a rather consistent pattern of impaired neural and cognitive functions appears to be associated with blood lead levels below those producing the overt symptomatology of lead encephalopathy." CD 1-7, JA 1111. The Criteria Document reported that "(t)he blood lead levels at which neurobehavioral deficits occur in otherwise asymptomatic children appear to start at a range of 50 to 60 ug/dl, although some evidence tentatively suggests that such effects may occur at slightly lower levels for some children." Id.15

In addition to examining the health effects of lead exposure, the Criteria Document also discussed other issues critical to the task of setting air quality standards for lead. One of these issues is the relationship between air lead exposure and blood lead levels a relationship commonly referred to as the air lead/blood lead ratio. The Criteria Document acknowledged that derivation of a functional relationship between air lead exposure and blood lead levels is made difficult by the fact that the relationship is not a linear one; rather, the ratio tends to increase as air lead levels are reduced, CD 12-24, JA 1311. The Document was nevertheless able to conclude, after a detailed examination of the relevant studies, CD 12-22 12-29, JA 1309-1316, that air lead/blood lead ratios fall within a range of 1:1 to 1:2 (ug Pb/m 3 air):(ug Pb/dl blood) at the levels of lead exposure generally encountered by the population, i.e., blood lead levels increase by between 1 and 2 ug Pb/dl of blood for every 1 ug Pb/m 3 of air. (Air lead content is measured in micrograms of lead per cubic meter of air ug Pb/m 3.) CD 12-38, JA 1325. The Criteria Document reported that the studies indicate that the ratio for children is at the upper end of this range or even slightly above it. Id.

Finally, the Criteria Document also examined the distribution of blood lead levels throughout the population, concluding that there is a significant variability in individual blood lead responses to any particular level of air lead exposure. It further found that this variability is consistent and predictable, and that the application of established statistical techniques to the distribution of individual blood lead levels would make it possible to predict what proportion of the population would be above or below any particular blood lead level at a given level of air lead exposure.16 The Criteria Document looked into the question whether any sub-groups within the population are particularly vulnerable to the effects of lead exposure. It concluded that preschool-age children and pregnant women are particularly sensitive to lead exposure, the latter mainly because of the risk to the unborn child. CD 13-11 13-14, JA 1341-1344.

Simultaneously with the publication of the Lead Criteria Document on December 14, 1977, the Administrator proposed a national primary ambient air quality standard for lead of 1.5 ug Pb/m 3 monthly average. 42 Fed.Reg. 63076, JA 1480. He also proposed that the secondary air quality standard be set at the same level as the primary standard because the welfare effects associated with lead exposure did not warrant imposition of a stricter standard. 42 Fed.Reg. 63081-63082, JA 1485-1486. In the preamble to the proposed standards the Administrator explained the analysis EPA had employed in setting the standards.

The Administrator first pointed out that a number of factors complicate the task of setting air quality standards which will protect the population from the adverse health effects of lead exposure. First, some sub-groups within the population have a greater potential for, or are more susceptible to the effects of, lead exposure. Id. at 63077, JA 1481. Second, there are a variety of adverse health effects associated with various levels of lead exposure. Id. Third, the variability of individual responses to lead exposure, even within particular sub-groups of the population, would produce a range of blood lead levels at any given air lead level. Id. at 63079, JA 1483. Fourth, airborne lead is only one of a number of sources of lead exposure and the relative contribution from each source is difficult to quantify. Id. at 63080, JA 1484. Finally, the relationship between air lead exposure and blood lead levels is a complex one. Id. at 63079, JA 1483.

In response to the first problem the Administrator began by noting that protection of the most sensitive groups within the population had to be a major consideration in determining the level at which the air quality standards should be set. And he determined that children between the ages of 1 and 5 years are most sensitive to the effects of lead exposure both because the hematologic and neurologic effects associated with lead exposure occur in children at lower threshold levels than in adults, and because the habit of placing hands and other objects in the mouth subjects them to a greater risk of exposure. Id. at 63077-63078, JA 1481-1482. Next, the Administrator examined the various health effects of lead exposure and proposed that EP elevation should be considered the first adverse health effect of lead exposure because it indicates an impairment of cellular functions, and should be the pivotal health effect on which the lead standards are based. Id. at 63078, JA 1482. Accordingly, he proposed that the air lead standards be designed to prevent the occurrence of EP elevation in children. In order to accomplish this, and to address the problem of variable responses to lead exposure, the Administrator selected 15 ug Pb/dl, the lowest reported threshold blood lead level for EP elevation in children, as the target mean population blood lead level.17 He reasoned that setting the target mean population blood lead level at the lowest reported threshold blood lead level for EP elevation would ensure that most of the target population would be kept below blood lead levels at which adverse health effects occur. Id. at 63078, JA 1483. The Administrator also discussed the alternative approaches of basing the standard on more severe effects such as anemia, or attempting to decide the actual level of EP elevation which represents an adverse effect on health, and then making an adjustment to allow a margin of safety. Id. He specifically invited comments on these alternative approaches. Id. Finally, the Administrator outlined another approach to calculating the target mean population blood lead level involving the use of statistical techniques discussed in the Criteria Document. Id.18

Having selected a target mean population blood lead level, the Administrator's next step was to allow for the multiplicity of sources of lead exposure. He thus had to estimate the amount of blood lead that should be attributed to non-air sources. The Administrator admitted that any amount he selected could be no more than a theoretical national average, and on the basis of the evidence available he proposed that the lead standards should be based on the general assumption that 12 ug Pb/dl of blood lead should be attributed to non-air sources. Id. at 63080-63081, JA 1484-1485. Given the target mean population blood lead level of 15 ug Pb/dl and the assumed contribution from non-air sources of 12 ug Pb/dl, the maximum allowable contribution from ambient air is 3 ug Pb/dl. The final step in his analysis was to determine what air lead level would prevent the ambient air contribution to blood lead levels from exceeding 3 ug Pb/dl. This step required determining the relationship between air lead exposure and blood lead levels, i.e., the air lead/blood lead ratio. On the basis of the information in the Criteria Document, the Administrator selected a ratio of 1:2 as appropriate for calculating the effect of air lead exposure on blood lead levels in children. Id. at 63079, JA 1483.

Thereafter, calculation of the air quality standard was a mathematical exercise as shown in the following table.

1.                            Target mean blood lead level         15 ug Pb/dl
        2.                            Assumed non-air contribution        -12 ug Pb/dl
                                                                          -------------
        3.                            Allowable air contribution          =3 ug Pb/dl
        4.                            Permissible air lead concentration
                                      given assumed air lead/blood lead
                                      ratio
                                      3 ug Pb/dl X 1 ug Pbm 3 air         =1.5 ug Pb/m
                                                                            3
                    ----------------  2 ug Pb/dl blood

The Administrator concluded, on the basis of available information, that the averaging period for the lead standard should be a calendar month. Id. at 63081, JA 1485.

The public comment period ran from December 14, 1977 to March 17, 1978, and public hearings on the proposed standards were held on February 15 and 16, 1978. 43 Fed.Reg. 46246, JA 2948. The comments on the proposed standards were sharply divided. The comments submitted by the lead industry and its experts uniformly opposed the proposed standards, and many endorsed a standard of 5 ug Pb/m 3, the standard proposed in the discredited first draft of the Criteria Document, see note 7 supra, as adequate to protect the public health. 43 Fed.Reg. 46248, JA 2950. On the other hand, environmental groups, medical experts, and state, local, and federal agencies either endorsed the proposed standards or called for even stricter standards. Id. None of the comments seriously questioned the selection of children between the ages of 1 and 5 years as the target population group, or the estimate of a contribution from non-air sources of 12 ug Pb/dl. The major areas of controversy were the Administrator's choice of EP elevation as the pivotal adverse health effect and his conclusion that the threshold blood lead level for EP elevation in children is 15 ug Pb/dl, the selection of an appropriate air lead/blood lead ratio, the appropriate allowance for an adequate margin of safety, and the averaging time period for the standards. Id.19

A number of comments challenged the selection of EP elevation as the pivotal adverse health effect, insisting that EP elevation merely indicates a biological change or response which is in no way harmful to health,20 and in addition they criticized the Administrator's determination that the blood lead threshold for EP elevation in children is 15 ug Pb/dl.21 These comments suggested that a decrease in hemoglobin levels, which begins at blood lead levels no lower than 40 ug Pb/dl, should be the pivotal adverse health effect on which the standards are based.22 Other experts, however, agreed with the Administrator's conclusion that EP elevation must be considered an adverse health effect of lead exposure, and argued that using EP elevation as the pivotal adverse health effect would, in addition, allow an adequate margin of safety in protecting against the more serious health effects associated with higher levels of lead exposure.23 Finally, several industry experts appeared to indicate a preference for the lognormal statistical procedures that the Administrator had, in the proposed standards, suggested as an alternative method for determining the target mean population blood lead level.24

The Administrator promulgated the final air quality standards on October 5, 1978, prescribing national primary and secondary ambient air quality standards for lead of 1.5 ug Pb/m 3, averaged over a calendar quarter. 43 Fed.Reg. 46246, JA 2948. Although the final standards were the same as the proposed standards (with the exception of the change in the averaging period from 30 to 90 days), the Administrator arrived at the final standards through somewhat different analysis. The preamble to the final standards reveals that the comments on the proposed standards had led the Administrator to reconsider his analysis. In particular, he seemed to feel that legitimate questions had been raised concerning the health significance of the early stages of EP elevation and about the threshold blood lead level for this condition. 43 Fed.Reg. 46248, 46253, JA 2950, 2955. The Administrator's reexamination focused on two key questions: (1) What is the maximum safe individual blood lead level for children? and (2) what proportion of the target population should be kept below this blood lead level? Id. at 46249, 46252-46253, JA 2951, 2954-2955. Addressing the first issue required a review of the health effects of lead exposure discussed in the Criteria Document. The Administrator concluded that, although EP elevation beginning at blood lead levels of 15-20 ug Pb/dl is potentially adverse to the health of children, only when blood lead concentration reaches a level of 30 ug Pb/dl is this effect significant enough to be considered adverse to health. Id. at 46253, JA 2955. Accordingly, he selected 30 ug Pb/dl as the maximum safe individual blood lead level for children. Id. The Administrator based this choice on three mutually supporting grounds. First, it is at this blood lead level that the first adverse health effect of lead exposure impairment of heme synthesis begins to occur in children. Second, a maximum safe individual blood lead level of 30 ug Pb/dl would allow an adequate margin of safety in protecting children against more serious effects of lead exposure anemia, symptoms of which begin to appear in children at blood lead levels of 40 ug Pb/dl, and central nervous system deficits which start to occur in children at blood lead levels of 50 ug Pb/dl. Third, the Administrator reasoned that the maximum safe individual blood lead level should be no higher than the blood lead level used by the Center for Disease Control in screening children for lead poisoning 30 ug Pb/dl. Id.

Having determined the maximum safe individual blood lead level for the target population, the Administrator next focused on the question of what percentage of children between the ages of 1 and 5 years the standard should attempt to keep below this blood lead level. According to the 1970 census, there are approximately 20 million children under the age of 5 years in the United States, 12 million of them in urban areas and 5 million in inner cities where lead exposure may be especially high. The Administrator concluded that in order to provide an adequate margin of safety, and to protect special high risk sub-groups, the standards should aim at keeping 99.5% of the target population below the maximum safe individual blood lead level of 30 ug Pb/dl.25 Id. at 46253, 46255, JA 2955, 2957. The next step in the analysis was to determine what target mean population blood lead level would ensure that 99.5% of the children below the age of 5 years would be kept below the maximum safe individual blood lead level of 30 ug Pb/dl. Using the lognormal statistical technique he had alluded to in the proposed standards,26 he calculated that a target mean population blood lead level of 15 ug Pb/dl (the same number as in the proposed standards, but arrived at through different analysis), would accomplish this task.27 Id. at 46253, 46254, JA 2955, 2956. Thereafter, the Administrator used the same estimate of the contribution from non-air sources, 12 ug Pb/dl, and the same air lead/blood lead ratio, 1:2, that he had used in calculating the proposed standards,28 to compute the final ambient air quality standards for lead. The result was an ambient air quality standard of 1.5 ug Pb/m 3, the same as the proposed standard. Id. at 46254, JA 2956. The Administrator did, however, change the averaging period for the standards from one calendar month to one calendar quarter, id. at 46255, JA 2957, because he felt that this change would significantly improve the validity of the data to be used in monitoring the progress toward attainment of the standards without rendering the standards less protective. Id.

On December 8, 1978 LIA petitioned EPA for reconsideration and a stay of the lead standards. JA 2980-3000. The Administrator denied the petition on February 2, 1979. JA 3001-3007. These petitions for review of the lead standards regulations followed. Before examining the petitioners' challenges to the regulations, we consider the limits of our reviewing function.

The scope of judicial review of the Administrator's decisions and actions is delineated by Section 307(d) of the Act, 42 U.S.C. § 7607(d). We must uphold the Administrator's actions unless we find that they were: (1) "arbitrary, capricious, an abuse of discretion, or otherwise not in accordance with law"; (2) "contrary to constitutional right, power, privilege, or immunity"; (3) "in excess of statutory jurisdiction, authority, or limitations, or short of statutory right(.)" Section 307(d)(9), 42 U.S.C. § 7607(d)(9). In addition, we may set aside any action found to be "without observance of procedure required by law," if (i) the failure to follow the prescribed procedure was arbitrary or capricious, (ii) the procedural objection was raised during the public comment period, or there were good reasons why it was not, and (iii) the procedural errors "were so serious and related to matters of such central relevance to the rule that there is a substantial likelihood that the rule would have been significantly changed if such errors had not been made." Id. Section 307(d)(8), 42 U.S.C. § 7607(d)(8).

These statutory provisions and a considerable body of case law demonstrate that our role as a reviewing court is limited. The "arbitrary and capricious" standard of review is highly deferential, and presumes agency action to be valid. Citizens to Preserve Overton Park, Inc. v. Volpe, 401 U.S. 402, 415, 91 S.Ct. 814, 823, 28 L.Ed.2d 136 (1971); Ethyl Corp. v. EPA, supra, 541 F.2d at 34. Moreover, the reviewing court may not substitute its judgment for the agency's, Citizens to Preserve Overton Park, Inc. v. Volpe, supra, 401 U.S. at 416, 91 S.Ct. at 823, and must affirm the agency's decision if a rational basis for it is presented. Bowman Transportation, Inc. v. Arkansas-Best Freight Systems, Inc., 419 U.S. 281, 290, 95 S.Ct. 438, 444, 42 L.Ed.2d 447 (1974); United States v. Allegheny-Ludlum Steel Corp., 406 U.S. 742, 749, 92 S.Ct. 1941, 1946, 32 L.Ed.2d 453 (1972). Of course a reviewing court does not serve as a mere rubber stamp for agency decisions. Rather, the function of judicial review is to ensure that agency decisions are "based on a consideration of the relevant factors." Citizens to Preserve Overton Park, Inc. v. Volpe, supra, 401 U.S. at 416, 91 S.Ct. at 824.

In addition, the court must undertake a "substantial inquiry" into the facts, one that is "searching and careful." Id. at 415, 416, 91 S.Ct. at 823; Ethyl Corp. v. EPA, supra, 541 F.2d at 34. In cases such as the ones we have before us, cases which involve complex scientific and technical questions, conducting a "substantial inquiry" into the facts may require the court to delve into the scientific literature. The purpose of this scrutiny of the evidence in the record is to educate the court. As we pointed out in Ethyl Corp. v. EPA, supra, the court "must understand enough about the problem confronting the agency to comprehend the meaning of the evidence relied upon and the evidence discarded; the questions addressed by the agency and those bypassed; the choices open to the agency and those made." 541 F.2d at 36. Only by doing this can the court "penetrate to the underlying decisions of the agency, to satisfy itself that the agency has exercised a reasoned discretion, with reasons that do not deviate from or ignore the ascertainable legislative intent." Greater Boston Television Corp. v. FCC, 444 F.2d 841, 850 (D.C.Cir.1970), cert. denied, 403 U.S. 923, 91 S.Ct. 2229, 29 L.Ed.2d 701 (1971).29 However, it is appropriate to sound some notes of caution about the limits of this exercise. First, we would be less than candid if we failed to acknowledge that we approach the task of examining some of the complex scientific issues presented in cases of this sort with some diffidence.30 More important, we stress that our review of the evidence is not designed to enable us to second-guess the Agency's expert decisionmaker. Ethyl Corp. v. EPA, supra, 541 F.2d at 36. Congress has entrusted the Agency with the responsibility for making these scientific and other judgments, and we must respect both Congress' decision and the Agency's ability to rely on the expertise that it develops. Market Street Railway v. Railroad Commission, 324 U.S. 548, 559-561, 65 S.Ct. 770, 776-77, 89 L.Ed. 1171 (1945); Ethyl Corp. v. EPA, supra, 541 F.2d at 36. As we explained in Ethyl Corp. v. EPA :

(A)fter our careful study of the record, we must take a step back from the agency's decision. We must look at the decision not as the chemist, biologist or statistician that we are qualified neither by training nor experience to be, but as a reviewing court exercising our narrowly defined duty of holding agencies to certain minimal standards of rationality. "Although (our) inquiry into the facts is to be searching and careful, the ultimate standard of review is a narrow one." * * * We must affirm unless the agency decision is arbitrary or capricious.

541 F.2d at 36-37 (citations and footnotes omitted; brackets in original).

It is also important to note that although the pertinent sections of the Clean Air Act outline the policy objectives to be sought and the procedural framework to be followed in promulgating ambient air quality standards, Congress left the formulation of the specific standards to EPA's Administrator. This task presents complex questions of science, law, and social policy under the Act. The record is lengthy approximately 10,000 pages and it is highly technical. The Administrator's task required both "a legislative policy determination and an adjudicative resolution of disputed facts." Mobil Oil Corp. v.

1 (5, 6) These are conceptually distinct types of decisions, and it is important that we keep this in mind in reviewing the Administrator's decisions. See Industrial Union Dep't, AFL-CIO v. Hodgson, 499 F.2d 467, 474-475 (D.C.Cir.1974). Where factual determinations were necessary the Administrator often had to make decisions in the face of conflicting evidence. In some instances this merely required that he draw conclusions from the evidence in the record. In reviewing these conclusions we can examine the record to ascertain whether there is substantial evidence in the record when considered as a whole which supports the Administrator's determinations. Id. at 474. Other questions involved in the standardsetting process, however, are at the very "frontiers of scientific knowledge."31 Consequently, the information available may be insufficient to permit fully informed factual determinations. In such instances the Administrator's decisions necessarily had to rest largely on policy judgments. Policy choices of this sort "are not susceptible to the same type of verification or refutation by reference to the record as are (other) factual questions." 499 F.2d at 475. While we will indeed scrutinize such judgments carefully, we must adopt a different mode of judicial review. Id.; accord, Nat'l Asphalt Pavement Ass'n v. Train, 539 F.2d 775, 783-784 (D.C.Cir.1976); Automotive Parts & Accessories Ass'n, Inc. v. Boyd, 407 F.2d 330, 336, 338 (D.C.Cir.1968). As this court has previously stated:

What we are entitled to * * * is a careful identification by the (Administrator), when his proposed standards are challenged, of the reasons why he chooses to follow one course rather than another. Where that choice purports to be based on certain determinable facts, (he) must, in form as well as substance, find those facts from evidence in the record. By the same token, when the (Administrator) is obliged to make policy judgments where no factual certainties exist or where facts alone do not provide the answer, he should so state and go on to identify the considerations he found persuasive."

Industrial Union Dep't, AFL-CIO v. Hodgson, supra, 499 F.2d at 475-476. In short, "(t)he paramount objective is to see whether the agency, given an essentially legislative task to perform, has carried it out in a manner calculated to negate the dangers of arbitrariness and irrationality in the formulation of rules for general application in the future." Automotive Parts & Accessories Ass'n, Inc. v. Boyd, supra, 407 F.2d at 338.

Finally, although we may set aside the Administrator's decisions if we find that he exceeded his authority under the statute, we note that EPA's construction of the Clean Air Act has been accorded considerable deference by the courts. Union Electric Co. v. EPA, 427 U.S. 246, 256, 96 S.Ct. 2518, 2525, 49 L.Ed.2d 474 (1976); Train v. Natural Resources Defense Council, Inc., 421 U.S. 60, 75, 95 S.Ct. 1470, 1479, 43 L.Ed.2d 731 (1975); Ethyl Corp. v. EPA, supra, 541 F.2d at 12 n.16.32 Where different interpretations of the statute are plausible, so long as EPA's construction of the statute is reasonable we may not substitute our own interpretation for the Agency's. Train v. Natural Resources Defense Council, Inc., supra, 421 U.S. at 75, 95 S.Ct. at 1479.33 "(T)he construction of a statute by those charged with its execution should be followed unless there are compelling indications that it is wrong(.)" Red Lion Broadcasting Co. v. FCC, 395 U.S. 367, 381, 89 S.Ct. 1794, 1802, 23 L.Ed.2d 371 (1967); accord, Beal v. Doe, 432 U.S. 438, 97 S.Ct. 2366, 53 L.Ed.2d 464 (1977). Deference to the Administrator's interpretation is particularly appropriate in construing a statute that invests him with a considerable amount of discretion. Unless it can be shown that the Administrator's construction of the statute is plainly unreasonable, we must uphold his interpretation. Motor & Equipment Manufacturers Ass'n, Inc. v. EPA, 627 F.2d 1095, at 1106 (D.C.Cir.1979).

Thus mindful of our restricted role, we turn to consider petitioners' claims. Petitioners posit three basic questions for decision. First, did the Administrator exceed his authority under the statute in promulgating the lead standards? Second, were key elements in the Administrator's analysis arbitrary or capricious? Third, do alleged procedural shortcomings in the lead standards rulemaking warrant a remand of the regulations to EPA?

The petitioners' first claim is that the Administrator exceeded his authority under the statute by promulgating a primary air quality standard for lead which is more stringent than is necessary to protect the public health because it is designed to protect the public against "sub-clinical" effects which are not harmful to health. According to petitioners, Congress only authorized the Administrator to set primary air quality standards that are aimed at protecting the public against health effects which are known to be clearly harmful. They argue that Congress so limited the Administrator's authority because it was concerned that excessively stringent air quality standards could cause massive economic dislocation.

In developing this argument St. Joe contends that EPA erred by refusing to consider the issues of economic and technological feasibility in setting the air quality standards for lead. St. Joe's claim that the Administrator should have considered these issues is based on the statutory provision directing him to allow an "adequate margin of safety" in setting primary air quality standards. In St. Joe's view, the Administrator must consider the economic impact of the proposed standard on industry and the technological feasibility of compliance by emission sources in determining the appropriate allowance for a margin of safety.34 St. Joe argues that the Administrator abused his discretion by refusing to consider these factors in determining the appropriate margin of safety for the lead standards, and maintains that the lead air quality standards will have a disastrous economic impact on industrial sources of lead emissions.

This argument is totally without merit. St. Joe is unable to point to anything in either the language of the Act or its legislative history that offers any support for its claim that Congress, by specifying that the Administrator is to allow an "adequate margin of safety" in setting primary air quality standards, thereby required the Administrator to consider economic or technological feasibility. To the contrary, the statute and its legislative history make clear that economic considerations play no part in the promulgation of ambient air quality standards under Section 109.

Where Congress intended the Administrator to be concerned about economic and technological feasibility, it expressly so provided. For example, Section 111 of the Act, 42 U.S.C. § 7411, directs the Administrator to consider economic and technological feasibility in establishing standards of performance for new stationary sources of air pollution based on the best available control technology. See Nat'l Asphalt Pavement Ass'n v. Train, supra, 539 F.2d 775; S.Rep.No.91-1196, 91st Cong., 2d Sess. 416 (1970).35 In contrast, Section 109(b) speaks only of protecting the public health and welfare.36 Nothing in its language suggests that the Administrator is to consider economic or technological feasibility in setting ambient air quality standards.37

The legislative history of the Act also shows the Administrator may not consider economic and technological feasibility in setting air quality standards; the absence of any provision requiring consideration of these factors was no accident; it was the result of a deliberate decision by Congress to subordinate such concerns to the achievement of health goals. Exasperated by the lack of significant progress toward dealing with the problem of air pollution under the Air Quality Act of 1967, 81 Stat. 485, and prior legislation, Congress abandoned the approach of offering suggestions and setting goals in favor of "taking a stick to the States in the form of the Clean Air Amendments of 1970 * * *." Train v. Natural Resources Defense Council, Inc., supra, 421 U.S. at 64, 95 S.Ct. at 1474; see Union Electric Co. v. EPA, 427 U.S. 246, 256-257, 96 S.Ct. 2518, 2525, 49 L.Ed.2d 474 (1976). Congress was well aware that, together with Sections 108 and 110, Section 109 imposes requirements of a "technology-forcing" character. Id. at 257, 96 S.Ct. at 2525; Train v. Natural Resources Defense Council, Inc., supra, 421 U.S. at 91, 95 S.Ct. at 1487; Ethyl Corp. v. EPA, supra, 541 F.2d at 14. The Senate Report on the 1970 Amendments declared:

The protection of public health as required by the national ambient air quality standards * * * will require major action throughout the Nation. Many facilities will require major investments in new technology and new processes. Some facilities will need altered operating procedures * * *. Some may be closed.

In the Committee discussions, considerable concern was expressed regarding the use of the concept of technical feasibility as the basis of ambient air standards. The Committee determined that 1) the health of people is more important than the question of whether the early achievement of ambient air quality standards protective of health is technically feasible; and, 2) the growth of pollution load in many areas, even with application of available technology, would still be deleterious to public health.

Therefore, the Committee determined that existing sources of pollutants either should meet the standard of the law or be closed down, and in addition that new sources should be controlled to the maximum extent possible to prevent atmospheric emissions.

S.Rep.No.91-1196, supra, at 2-3. It is difficult to reconcile these statements of legislative intent with St. Joe's claim that Congress wanted the Administrator to consider economic and technological feasibility in setting air quality standards. The "technology-forcing" requirements of the Act "are expressly designed to force regulated sources to develop pollution control devices that might at the time appear to be economically or technologically infeasible." Union Electric Co. v. EPA, supra, 427 U.S. at 257, 96 S.Ct. at 2525.

Furthermore, St. Joe's attempt to find a mandate for the Administrator to consider economic or technological feasibility in the Act's "adequate margin of safety" requirement is to no avail. The Senate Report explained the purpose of the margin of safety requirement:

Margins of safety are essential to any health-related environmental standards if a reasonable degree of protection is to be provided against hazards which research has not yet identified.

S.Rep.No.91-1196, supra, at 10. We are unable to discern here any congressional intent to require, or even permit, the Administrator to consider economic or technological factors in promulgating air quality standards. And when Congress directs an agency to consider only certain factors in reaching an administrative decision, the agency is not free to trespass beyond the bounds of its statutory authority by taking other factors into account. American Overseas Airlines, Inc. v. CAB, 254 F.2d 744, 748 (D.C.Cir.1958). A policy choice such as this is one which only Congress, not the courts and not EPA, can make. Indeed, the debates on the Act indicate that Congress was quite conscious of this fact. For example, Senator Muskie, one of the prime architects of the Act, in speaking about the automobile emission standards and the automobile industry, noted:

* * * I think that we have an obligation to lay down the standards and requirements of this bill.

I think that the industry has an obligation to try to meet them. If, in due course, it cannot, then it should come to Congress and share with the Congress the representatives of the people the need to modify the policy.

1 Legislative History of Clean Air Act Amendments of 1970 at 232 (Senate Debate on S. 4358, Sept. 21, 1970) (hereinafter Legis.Hist.). See also id. at 236-240.38 In the same manner, if there is a problem with the economic or technological feasibility of the lead standards, St. Joe, or any other party affected by the standards, must take its case to Congress, the only institution with the authority to remedy the problem.39

It may well be that underlying St. Joe's argument is its feeling that Congress could not or should not have intended this result, and that this court should supply relief by grafting a requirement of economic or technological feasibility onto the statute. The Supreme Court confronted a similar suggestion in the Tellico Dam case. TVA v. Hill, 437 U.S. 153, 98 S.Ct. 2279, 57 L.Ed.2d 117 (1978). There TVA argued that the Endangered Species Act should not be construed to prevent operation of the dam since it had already been completed at a cost of approximately $100 million, Congress had appropriated funds for the dam even after the Act was passed, and the species at risk the snail darter was relatively unimportant and ways might ultimately be found to save it. The Court rejected the invitation to "view the * * * Act 'reasonably,' and hence shape a remedy that 'accords with some modicum of common sense and the public weal.' " Id. at 194, 98 S.Ct. at 2301. The Court pointed out:

Our individual appraisal of the wisdom or unwisdom of a particular course consciously selected by the Congress is to be put aside in the process of interpreting a statute. Once the meaning of an enactment is discerned and its constitutionality determined, the judicial process comes to an end. We do not sit as a committee of review, nor are we vested with the power of veto. * * *

Id. at 194-195, 98 S.Ct. at 2302. Without suggesting that the Administrator's interpretation of the statute does not "accord( ) with some modicum of common sense and the public weal," the Supreme Court's admonition reinforces our decision to reject St. Joe's invitation to construe the statute as requiring the Administrator to consider economic and technological feasibility in promulgating air quality standards.

For its part, LIA maintains that its claim that the Administrator exceeded the bounds of his statutory authority does not depend on the supposition that he is required, or even permitted, to consider economic and technological feasibility in setting air quality standards. LIA contends that, instead, its argument is based on the fact that Congress itself was concerned about the question of the economic feasibility of compliance with air quality standards, a concern which was reflected in the statute it enacted. According to LIA, Congress was mindful of the possibility that air quality standards which are too stringent could cause severe economic dislocation. For this reason it only granted the Administrator authority to adopt air quality standards which are "designed to protect the public from adverse health effects that are clearly harmful (.)"40 LIA finds support for its interpretation of congressional intent in various portions of the legislative history of the Act. For example, it notes that the Senate Report on the 1970 legislation states that EPA "would be required to set a national minimum standard of air quality," S.Rep.No.91-1196, supra, at 10 (emphasis added), and that Senator Muskie pointed out during the floor debates that "air quality standards which will protect the public must be set as minimum standards * * *." 1 Legis.Hist., supra, at 125 (emphasis added).41 LIA then argues that the Administrator based the lead air quality standards on protecting children from "subclinical" effects of lead exposure which have not been shown to be harmful to health, that in so doing the Administrator ignored the clear limitation that Congress imposed on his standard-setting powers, and that the Administrator's action will in fact cause the very result that Congress was so concerned about avoiding.

LIA's argument appears to touch on two issues. The first concerns the type of health effects on which the Administrator may base air quality standards, i.e., the point at which the Administrator's regulatory authority may be exercised. This issue, as LIA suggests, does concern the limits that the Act, and its legislative history, may place on the Administrator's authority. The second issue appears to be more in the nature of an evidentiary question: whether or not the evidence in the record substantiates the Administrator's claim that the health effects on which the standards were based do in fact satisfy the requirements of the Act. Although these two issues are closely related, they are conceptually distinct, and they are best examined separately.

Section 109(b) does not specify precisely what Congress had in mind when it directed the Administrator to prescribe air quality standards that are "requisite to protect the public health." The legislative history of the Act does, however, provide some guidance. The Senate Report explains that the goal of the air quality standards must be to ensure that the public is protected from "adverse health effects." S.Rep.No.91-1196, supra, at 10. And the report is particularly careful to note that especially sensitive persons such as asthmatics and emphysematics are included within the group that must be protected. It is on the interpretation of the phrase "adverse health effects" that the disagreement between LIA and EPA about the limits of the Administrator's statutory authority appears to be based. LIA argues that the legislative history of the Act indicates that Congress only intended to protect the public against effects which are known to be clearly harmful to health,42 maintaining that this limitation on the Administrator's statutory authority is necessary to ensure that the standards are not set at a level which is more stringent than Congress contemplated. The Administrator, on the other hand, agrees that primary air quality standards must be based on protecting the public from "adverse health effects," but argues that the meaning LIA assigns to that phrase is too limited. In particular, the Administrator contends that LIA's interpretation is inconsistent with the precautionary nature of the statute, and will frustrate Congress' intent in requiring promulgation of air quality standards.

The Administrator begins by pointing out that the Act's stated goal is "to protect and enhance the quality of the Nation's air resources so as to promote the public health and welfare and the productive capacity of its population (.)" Section 101(b)(1), 42 U.S.C. § 7401(b)(1). This goal was reaffirmed in the 1977 Amendments. For example, the House Report accompanying the Amendments states that one of its purposes is "(t)o emphasize the preventive or precautionary nature of the act, i.e., to assure that regulatory action can effectively prevent harm before it occurs; to emphasize the predominant value of protection of public health(.)" H.R.Rep.No.95-294, 95th Cong., 1st Sess. 49 (1977). The Administrator notes that protecting the public from harmful effects requires decisions about exactly what these harms are, a task Congress left to his judgment. He notes that the task of making these decisions is complicated by the absence of any clear thresholds above which there are adverse effects and below which there are none. Rather, as scientific knowledge expands and analytical techniques are improved, new information is uncovered which indicates that pollution levels that were once considered harmless are not in fact harmless. Congress, the Administrator argues, was conscious of this problem, and left these decisions to his judgment partly for this reason.43 In such situations the perspective that is brought to bear on the problem plays a crucial role in determining what decisions are made. Because it realized this, Congress, the Administrator maintains, directed him to err on the side of caution in making these judgments. First, Congress made it abundantly clear that considerations of economic or technological feasibility are to be subordinated to the goal of protecting the public health by prohibiting any consideration of such factors. Second, it specified that the air quality standards must also protect individuals who are particularly sensitive to the effects of pollution. Third, it required that the standards be set at a level at which there is "an absence of adverse effect" on these sensitive individuals. Finally, it specifically directed the Administrator to allow an adequate margin of safety in setting primary air quality standards in order to provide some protection against effects that research has not yet uncovered. The Administrator contends that these indicia of congressional intent, the precautionary nature of the statutory mandate to protect the public health, the broad discretion Congress gave him to decide what effects to protect against, and the uncertainty that must be part of any attempt to determine the health effects of air pollution, are all extremely difficult to reconcile with LIA's suggestion that he can only set standards which are designed to protect against effects which are known to be clearly harmful to health.

We agree that LIA's interpretation of the statute is at odds with Congress' directives to the Administrator. As a preliminary matter, though it denies this, LIA does at times seem to be arguing, along with St. Joe, that the Administrator should have considered economic and technological feasibility in setting the standards,44 a claim that must be rejected for reasons we have already stated.45 Be that as it may, it is not immediately clear why LIA expects this court to impose limits on the Administrator's authority which, so far as we can tell, Congress did not. The Senate Report explains that the Administrator is to set standards which ensure that there is "an absence of adverse effects." The Administrator maintains that the lead standards are designed to do just that, a claim we will examine in due course. But LIA would require a further showing that the effects on which the standards were based are clearly harmful or clearly adverse. We cannot, however, find the source of this further restriction that LIA would impose on the Administrator's authority.46 It may be that it reflects LIA's view that the Administrator must show that there is a "medical consensus that (the effects on which the standards were based) are harmful * * *."47 If so, LIA is seriously mistaken. This court has previously noted that some uncertainty about the health effects of air pollution is inevitable.48 And we pointed out that "(a)waiting certainty will often allow for only reactive, not preventive regulat(ory action)." Ethyl Corp. v. EPA, supra, 541 F.2d at 25. Congress apparently shares this view; it specifically directed the Administrator to allow an adequate margin of safety to protect against effects which have not yet been uncovered by research and effects whose medical significance is a matter of disagreement.49 This court has previously acknowledged the role of the margin of safety requirement. In Environmental Defense Fund v. EPA, 598 F.2d 62, 81 (D.C.Cir.1978), we pointed out that "(i)f administrative responsibility to protect against unknown dangers presents a difficult task, indeed, a veritable paradox calling as it does for knowledge of that which is unknown then, the term 'margin of safety' is Congress's directive that means be found to carry out the task and to reconcile the paradox." Moreover, it is significant that Congress has recently acknowledged that more often than not the "margins of safety" that are incorporated into air quality standards turn out to be very modest or nonexistent, as new information reveals adverse health effects at pollution levels once thought to be harmless. See H.R.Rep.No.95-294, supra, at 103-117. Congress' directive to the Administrator to allow an "adequate margin of safety" alone plainly refutes any suggestion that the Administrator is only authorized to set primary air quality standards which are designed to protect against health effects that are known to be clearly harmful.

Furthermore, we agree with the Administrator that requiring EPA to wait until it can conclusively demonstrate that a particular effect is adverse to health before it acts is inconsistent with both the Act's precautionary and preventive orientation and the nature of the Administrator's statutory responsibilities. Congress provided that the Administrator is to use his judgment in setting air quality standards precisely to permit him to act in the face of uncertainty.50 And as we read the statutory provisions and the legislative history, Congress directed the Administrator to err on the side of caution in making the necessary decisions. We see no reason why this court should put a gloss on Congress' scheme by requiring the Administrator to show that there is a medical consensus that the effects on which the lead standards were based are "clearly harmful to health." All that is required by the statutory scheme is evidence in the record which substantiates his conclusions about the health effects on which the standards were based.51 Accordingly, we reject LIA's claim that the Administrator exceeded his statutory authority and turn to LIA's challenge to the evidentiary basis for the Administrator's decisions.

LIA does not question a number of the steps in the Administrator's analysis. It does not disagree with his selection of children between the ages of one and five years as the target population, or the decision to set a standard that would keep 99.5 percent of the children below the maximum safe individual blood lead level. In addition, LIA does not challenge the Administrator's suggestion that the standards should be based on an assumption that non-air sources contribute 12 ug Pb/dl to blood lead levels. LIA does, however, challenge other key elements in the Administrator's analysis.

LIA attacks the Administrator's determination that 30 ug Pb/dl should be considered the maximum safe individual blood lead level for children, maintaining that there is no evidence in the record indicating that children suffer any health effects that can be considered adverse at this blood lead level. As previously noted,52 the Administrator's selection was based on his finding that EP elevation at 30 ug Pb/dl is the first adverse health effect of lead exposure, and his determination that a maximum safe individual blood lead level of 30 ug Pb/dl will allow an adequate margin of safety in protecting children against the more serious effects of lead exposure anemia, symptoms of which appear at blood lead levels of 40 ug Pb/dl and central nervous system deficits which begin to occur at blood lead levels of 50 ug Pb/dl.

LIA challenges each of these findings. First, it contends that nothing in the record supports the suggestion that EP elevation at 30 ug Pb/dl is harmful to health, arguing that EP elevation is a mere "subclinical effect" a biological response to lead exposure which is without health significance, and noting that a number of its experts brought this matter to EPA's attention in their comments on the proposed standards.53 In LIA's view, the Administrator did not explain precisely how impairment of heme synthesis at blood lead levels of 30 ug Pb/dl adversely affects the health of children.54 Second, LIA challenges the Administrator's determination that a maximum safe individual blood lead level of 30 ug Pb/dl is justified by the need to allow an adequate margin of safety in protecting children against anemia and central nervous system deficits. It maintains that the evidence in the record does not support the Administrator's conclusion that the blood lead threshold for the symptoms of anemia in children is 40 ug Pb/dl. LIA claims that this error was brought to the Administrator's attention by comments on the proposed standard, but that he failed to respond to these comments, thereby violating the statutory provision requiring him to respond to "significant comments, criticisms, and new data submitted * * * during the comment period."55 Third, LIA contends that the preamble to the final regulations does not state the basis for the Administrator's finding that central nervous system deficits occur in children at blood lead levels of 50 ug Pb/dl, thereby precluding this court from being able to test the soundness of this determination. Finally, LIA argues that even if it were to concede that EPA's conclusions about the blood lead thresholds for anemia and central nervous system deficits are correct, there is still no explanation of why the Administrator concluded that a maximum individual safe blood level of 30 ug Pb/dl rather than 35 ug Pb/dl, for example is necessary to provide an adequate margin of safety against these effects.

Our review of the record persuades us that there is adequate support for each of the Administrator's conclusions about the health effects of lead exposure and, consequently, that LIA's challenges to the evidentiary support for these findings must be rejected. Under the statutory scheme enacted by Congress, the Criteria Document prepared with respect to each pollutant is to provide the scientific basis for promulgation of air quality standards for the pollutant. We have already noted that the Lead Criteria Document was the product of a process that allowed the rigorous scientific and public review that are essential to the preparation of a document "accurately reflect(ing) the latest scientific knowledge useful in indicating the kind and extent of all identifiable effects (of lead exposure) on (the) public health * * *."56 In our view, the Criteria Document provides ample support for the Administrator's findings.

The Criteria Document concluded that EP elevation, which begins in children at blood levels of 15-20 ug Pb/dl, is one of the more significant effects of low level lead exposure because it indicates that lead has already begun to affect basic biological functions in the body. We have already examined in some detail the findings that led the Criteria Document to the conclusion that EP elevation is an indication of a physiological impairment which affects human health.57 Briefly stated, it concluded that EP elevation indicates an impairment of the functioning of the mitochondria58; that EP elevation indicates that lead has begun to affect the process of heme synthesis; that EP elevation may indicate a reduction in any reserve capacity there might be in the heme synthesis system; and that lead's impairment of heme synthesis in the blood suggests that lead may also interfere with production of heme proteins in other organ systems. Relying on the Criteria Document's discussion, as well as other evidence in the record, the Administrator made a judgment that for purposes of setting air quality standards for lead, EP elevation at 30 ug Pb/dl must be considered the first adverse effect on the health of children, and he determined that the maximum safe individual blood lead level should be no higher than 30 ug Pb/dl.

The Administrator's judgment echoes the consensus of a group of clinicians who, in 1975, participated in preparation of a statement issued by the Center For Disease Control and endorsed by the American Academy of Pediatrics. These experts agreed that EP elevation "should be used as an indicator of a significant and worrisome body burden of lead." CD 11-14, JA 1236. Moreover, the Center For Disease Control uses EP elevation at 30 ug Pb/dl as the cutoff point for detection of lead poisoning in children in its screening programs,59 a factor that influenced the Administrator's decision. 43 Fed.Reg. 46253, JA 2955. The Administrator's reliance on this was, in our view, entirely appropriate. While the Center has indicated that children with a blood lead concentration of 30 ug Pb/dl have a lead problem,60 it is also significant that the Center's screening program has much the same preventive and precautionary approach that Congress directed the Administrator to apply to the task of setting air quality standards designed to protect the public health.61 Further support for the Administrator's decisions is provided by the testimony of various medical and other scientific experts who participated in the lead standards rulemaking. These experts endorsed the Administrator's (and Criteria Document's) conclusions about the effects of low level lead exposures, and agreed with his assessment of the health significance of these effects.62 The Administrator's decision is, of course, precisely the sort of issue that Congress specifically left to his judgment, and where there is evidence in the record which supports these judgments, this court is not at liberty to substitute its judgment for the Administrator's. In this instance the Administrator has acted properly under the terms of the statute. He has explained his factual findings and policy judgments, and there is an adequate basis in the record for these decisions. No more is required of him.

LIA's challenge to the Administrator's findings concerning the health significance of EP elevation also stresses that this phenomenon is only a "subclinical" effect. But the clinical/subclinical distinction has little to do with the question whether a particular effect is properly viewed as adverse to health. Rather, the distinction pertains to the means through which the particular effect may be detected: observation or physical examination in the case of clinical effects, and laboratory tests in the case of subclinical effects.63 Thus describing a particular effect as a "subclinical" effect in no way implies that it is improper to consider it adverse to health.64 While EP elevation may not be readily identifiable as a sign of disease, the Administrator properly concluded that it indicates a lead-related interference with basic biological functions. Expert medical testimony in the record confirms that the modern trend in preventive medicine is to detect health problems in their "subclinical" stages, and thereupon to take corrective action.65 Moreover, as we have already noted, the Center For Disease Control uses the same "subclinical" effect as the key indicator of the need for medical intervention in its lead poisoning screening program. The accepted use of this "subclinical" effect to determine the need for medical observation or intervention properly influenced the Administrator's decision.66 Thus the fact that the effects the Administrator relied on in setting the lead standards are "subclinical" does not detract from their significance for human health, or make them an improper basis for setting air quality standards.

We also reject LIA's claim that the evidence in the record does not support the Administrator's determination that the blood lead threshold for symptoms of anemia in children is 40 ug Pb/dl. According to the Criteria Document, the onset of anemia is marked by a decline in the level of hemoglobin per unit of blood, CD 13-4, JA 1334. And the Criteria Document states that "(r)eports on children indicate that statistically significant decreases in hemoglobin levels begin to appear at a blood lead level of 40 ug (Pb)/dl or somewhat below." Id. At another point the Criteria Document explains, "In children, a threshold blood lead level for production of * * * symptoms of anemia is approximately 40 ug Pb/dl * * *." CD 1-6, JA 1110. See id. 11-13, JA 1235; JA 2583-2585 (Needleman and Piomelli). The Criteria Document's conclusions were reached after a review of various studies that have examined the subject, and we cannot, in light of these findings, say that the Administrator's decision about the threshold blood lead level for the symptoms of anemia in children does not have an adequate basis in the record.67

Finally, our examination of the record also reveals ample support for the Administrator's determination that lead-induced central nervous system deficits begin to occur in children at blood lead levels of 50 ug Pb/dl. The central nervous system damage about which the Administrator was concerned was not the severe brain damage that can occur at relatively high levels of lead exposure 80-100 ug Pb/dl.68 Rather, his focus was on more subtle and largely irreversible neurological and behavioral impairment that has been detected in children at lower blood lead levels, 43 Fed.Reg. 46253, JA 2955.69 The Criteria Document candidly admitted that "(t)he literature on this subject is somewhat limited and controversial," but it was nevertheless able to conclude that "certain statements (can) be made about the possible hazard of low to moderate lead exposure levels." CD 11-18, JA 1240. The conclusion it reached, after a detailed review of various studies that have examined the subject, was that:

(The) evidence tends to confirm that some type of neural damage does exist in asymptomatic children, and not necessarily only at very high level of blood lead. The body of studies on low- or moderate-level lead effects on neurobehavioral functions * * * present overall a rather impressive array of data pointing to that conclusion. Several well-controlled studies have found effects that are clearly statistically significant, whereas others have found nonsignificant but borderline effects. Even some studies reporting generally nonsignificant findings at times contain data confirming statistically significant effects, which the authors attribute to various extraneous factors. * * *

CD 11-26, JA 1248. The Criteria Document reported that the blood lead levels associated with these neurobehavioral deficits are 50-60 ug Pb/dl. Id. These conclusions were endorsed by several of the experts who participated in the lead standards rulemaking proceedings, including one of LIA's experts.70 Some of these experts even suggested that these effects may occur at blood lead levels lower than the levels indicated by the Criteria Document.71 Contrary to LIA's suggestion, the evidence in the Criteria Document and the testimony of the experts provides an adequate basis for this court to undertake a review of the Administrator's findings concerning these effects. Accordingly, we reject LIA's challenge to the Administrator's conclusion that central nervous system deficits begin to occur in children at blood lead levels of 50 ug Pb/dl.

Our conclusion that there is ample support for the Administrator's determination that EP elevation at 30 ug Pb/dl is the first adverse health effect that children experience as a result of lead exposure is, of course, sufficient to sustain his selection of 30 ug Pb/dl as the maximum safe individual blood lead level. Given this, we cannot say that his further determination that a maximum safe individual blood lead level of 30 ug Pb/dl would in addition provide protection against the more serious adverse health effects of lead exposure was irrational.

To be sure, the Administrator's conclusions were not unchallenged; both LIA and the Administrator are able to point to an impressive array of experts supporting each of their respective positions.72 However, disagreement among the experts is inevitable when the issues involved are at the "very frontiers of scientific knowledge," and such disagreement does not preclude us from finding that the Administrator's decisions are adequately supported by the evidence in the record. It may be that LIA expects this court to conclude that LIA's experts are right, and the experts whose testimony supports the Administrator are wrong.73 If so, LIA has seriously misconceived our role as a reviewing court. It is not our function to resolve disagreement among the experts or to judge the merits of competing expert views. AFL-CIO v. Marshall, 617 F.2d 636, 651 & n.66 (D.C.Cir.1979); cf. Hercules Inc. v. EPA, 598 F.2d 91, 115 (D.C.Cir.1978) ("(c)hoice among scientific test data is precisely the type of judgment that must be made by EPA, not this court"). Our task is the limited one of ascertaining that the choices made by the Administrator were reasonable and supported by the record. Ethyl Corp. v. EPA, supra, 541 F.2d at 35-36. That the evidence in the record may also support other conclusions, even those that are inconsistent with the Administrator's,74 does not prevent us from concluding that his decisions were rational and supported by the record. AFL-CIO v. Marshall, supra, 617 F.2d at 651 n.66; Environmental Defense Fund, Inc. v. EPA, 510 F.2d 1292, 1298 (D.C.Cir.1975); accord, Bayside Enterprises, Inc. v. NLRB, 425 U.S. 298, 302, 97 S.Ct. 576, 579, 50 L.Ed.2d 494 (1976). Cf. Universal Camera Corp. v. NLRB, 340 U.S. 474, 488, 71 S.Ct. 456, 465, 95 L.Ed. 456 (1951) ("a court may (not) displace the (agency's) choice between two fairly conflicting views, even though the court would justifiably have made a different choice had the matter been before it de novo ").

Having determined that we must uphold the Administrator's decisions concerning the health effects that are the basis for the lead standards, we turn to petitioners' other challenges to the Administrator's analysis.

Both LIA and St. Joe argue that the Administrator erred by including multiple allowances for margins of safety in his calculation of the lead standards. Petitioners note that the statute directs the Administrator to allow an "adequate margin of safety" in setting primary air quality standards, and they maintain that as a matter of statutory construction the Administrator may not interpret "margin" of safety to mean "margins" of safety. In petitioners' view, the Administrator in fact did just this insofar as he made allowances for margins of safety at several points in his analysis. They argue that margin of safety allowances were reflected in the choice of the maximum safe individual blood lead level for children, in the decision to place 99.5 percent of the target population group below that blood lead level, in the selection of an air lead/blood lead ratio at 1:2, and in the Administrator's estimate of the contribution to blood lead levels that should be attributed to non-air sources.75 The net result of these multiple allowances for margins of safety, petitioners contend, was a standard far more stringent than is necessary to protect the public health. St. Joe suggests that EPA should have adopted an approach which required decisions on:

1) The maximum level of lead in air which is protective of health; i.e., a threshold beyond which the public health is not protected; and

2) An adequate margin of safety by which the level which is protective of health must be reduced. (76

EPA responds by maintaining that allowances for a margin of safety were made only at two points in its analysis: in the selection of a maximum safe individual blood lead level of 30 ug Pb/dl and in the decision to set a standard designed to keep 99.5 percent of the target population below that blood lead level. It argues that the statutory requirement of a margin of safety does not mandate adoption of the method suggested by St. Joe. Rather, EPA suggests, it indicates the precautionary orientation the Administrator is to bring to bear on the task of setting air quality standards. How conservative he must be in making particular judgments must, the Agency maintains, depend on such factors as the amount of uncertainty involved, the size of the population affected, and the severity of the effect. EPA argues that petitioners' claims about multiple allowances for margins of safety indicate that they have failed to recognize the difference between providing for a margin of safety and making a scientific judgment in the face of conflicting evidence.77

We agree with the Administrator that nothing in the statutory scheme or the legislative history requires him to adopt the margin of safety approach suggested by St. Joe.78 Adding the margin of safety at the end of the analysis is one approach, but it is not the only possible method. Indeed, the Administrator considered this approach but decided against it because of complications raised by the multiple sources of lead exposure.79 The choice between these possible approaches is a policy choice of the type that Congress specifically left to the Administrator's judgment. This court must allow him the discretion to determine which approach will best fulfill the goals of the Act. As we pointed our in Hercules Inc. v. EPA, supra, 598 F.2d at 108, "Decision between the alternatives is a quintessential policy judgment within the discretion of EPA. We cannot accept (the) notion that the administrator of the agency created to protect the environment lack(s) even the capability to exercise the discretion with which he was entrusted by Congress." (Emphasis in original.)80 Where, as here, the Administrator has provided an explanation of why he chose one method rather than another, and this explanation and his choice are not irrational, we must accept his decision. See Industrial Union Dep't, AFL-CIO v. Hodgson, supra, 499 F.2d at 475-476.

We also agree with the Administrator's suggestion that petitioners have ignored the distinction between scientific judgments based on the available evidence and allowances for margins of safety. In every instance in which the Administrator's judgment on a particular issue differed from petitioners' they attributed his decision to an allowance for a margin of safety. To be sure, there is no bright line that divides these two types of decisions, but they are nonetheless conceptually distinct. In any event, whatever the nature of the decision, the real test, as petitioners recognize,81 is whether the decision is reasonable when examined in light of the evidence in the record. We have already found that at least one of the decisions that the petitioners attribute to an allowance for a margin of safety the selection of the maximum safe individual blood lead level for children satisfies this test. Accordingly, we turn to petitioners' claims that the other steps in the Administrator's analysis cannot withstand critical scrutiny.

VII. ALLEGED ARBITRARY AND CAPRICIOUS DECISIONS

Petitioners contend that a number of the findings which constitute the very core of the Administrator's analysis violate one or more of the decisionmaking requirements of the Act. While arguing that each of these violations is sufficient ground for remand of the regulations to EPA, LIA maintains that "cumulatively they paint a picture of an agency that had prejudged the result from the very outset and was bent on adhering to its original proposal no matter what the evidence showed, the very converse of the fair and impartial rulemaking to which litigants * * * are entitled."82

LIA contends that the Administrator's choice of an air lead/blood lead ratio of 1:2 as the appropriate ratio for calculating the lead standards was arbitrary and capricious. LIA's claim is largely based on its disagreement with the Administrator's interpretation of the results of three studies that have examined the relationship between air lead exposure and blood lead levels. It argues that the Administrator's analysis of these studies is inconsistent and designed solely to support his decision to arrive at an air quality standard of 1.5 ug Pb/m 3.83 In addition, LIA contends that the Administrator erred by calculating the air lead/blood lead ratio on the basis of studies involving both adults and children, when the standards are designed to protect children. LIA maintains that the Administrator would have arrived at a ratio of 1:1.3, had he focused solely on the studies involving children.

We do not agree that the Administrator's selection of an air lead/blood lead ratio of 1:2 was arbitrary or capricious. The Criteria Document reported that air lead/blood lead ratios for the whole population, adults as well as children, range between 1:1 and 1:2, with children at the upper end of the range or even slightly above it. CD 12-38, JA 1325. And the range of ratios for children reported by the studies that were reviewed in the Criteria Document was 1:1.2 to 1:2.3, CD 12-25 (Table 12-28), JA 1312 (Table 12-28).84 The Administrator's choice of a ratio of 1:2 for purposes of calculating the lead standards is consistent with each of these findings. Moreover, the Administrator calculated that each of three particularly relevant and well-documented studies that were reviewed by the Criteria Document suggested an air lead/blood lead ratio close to 1:2. 43 Fed.Reg. 46250, 46254, JA 2954, 2956.85 Finally, the Administrator's choice of a ratio of 1:2 was endorsed by several experts who participated in the rulemaking proceedings.86 Indeed, the issue of the proper relationship between air lead exposure and blood lead levels was extensively discussed in the comments on the initial drafts of the Criteria Document, with several experts severely criticizing the suggestions in early drafts that the appropriate ratio is 1:1.87 Given all the evidence in the record which supports the Administrator's choice of a ratio of 1:2, we would be exceeding the scope of our reviewing function if we were to agree with LIA's suggestion that the Administrator's decision was either arbitrary or capricious.