The Comprehensive Environmental Response, Compensation, and Liability Act of 19801 ("CERCLA" or the "Act") directed the President to compile a list identifying top priorities among the nation's known hazardous waste sites.2 The President, in turn, delegated this responsibility to the Environmental Protection Agency ("EPA" or the "Agency").3 The petitioners4 take exception to the EPA's selection of five specific sites for the list.5 Upon careful review of the EPA's decisions, we conclude that in each instance the listing comports with both the Act and the regulations promulgated thereunder,6 and is not arbitrary.7

Eagle-Picher Industries assails the listing of the Picher Field, a region spanning the Oklahoma-Kansas border that was mined for lead and zinc ores from 1904 to 1958. The Picher Field owes its inclusion on the National Priorities List to contamination of both surface water--Tar Creek--and groundwater--the Roubidoux and Boone aquifers underlying the mining field.8 The mines penetrate the Boone aquifer, and groundwater that now fills abandoned mines interacts with minerals in the mine walls, affording presumably the principal mechanism of contamination.9 Periodically, this polluted mine water discharges through mine shafts, bore holes, and airshafts to the surface, debasing Tar Creek.10 Additionally, because the Boone and Roubidoux aquifers are linked by bore holes and other connections, polluted mine water migrates from the Boone into the Roubidoux, which supplies water to the local residents.11 Rainwater runoff from piles of tailings left scattered about the area by the mining operations further contaminates the surface water.12

The EPA included the Eagle-Picher mining area--denominated simply Tar Creek, Oklahoma--on an interim priorities list13 and later on the proposed National Priorities List.14 The EPA divided the mining area into two sites because it spanned two Agency regions,15 and listed these sites as Tar Creek (Ottawa County), Oklahoma, and Cherokee County, Kansas, on the National Priorities List as ultimately adopted.16 In the interim between issuing the proposed and final lists, the EPA lowered the Hazard Ranking System scores given to these two locations, but both sites remained on the list.17

Eagle-Picher advances three objections to the listing of the Tar Creek site. First, it contends that the EPA violated its own "aquifer of concern" rule for calculating the population served by the contaminated groundwater. Second, it asserts that the reason that the Agency assigned maximum scores for releases to groundwater and surface water at the Tar Creek location was that the Agency failed to consider the most up-to-date information. Third, Eagle-Picher claims that the EPA did not respond meaningfully to its comments. Addressing these challenges seriatim, we find each lacking in merit.

The EPA has incorporated an "aquifer of concern" principle into its regulations,18 requiring a scorer to consider the same aquifer when scoring for the observed release or route characteristics as when calculating target characteristics such as population served. Eagle-Picher argues that in rating Tar Creek for an observed release to groundwater,19 the EPA ignored this aquifer-of-concern concept. It asserts that the Agency measured the level of contamination in the Boone, from which drinking water is no longer drawn, but used the Roubidoux, which supplies the area with drinking water, for calculating the target population.20 Although it measured groundwater contaminants in samples of water primarily from the Boone aquifer, not the Roubidoux, the Agency concluded that the pollution of the Boone itself constituted a release to the Roubidoux, and therefore scored for an observed release to the aquifer of concern, because of the documented existence of bore holes and the possibility of other links between the Boone and the Roubidoux.21

Since the purpose of the Hazard Ranking System is to estimate the potential hazard from a site,22 we think it clear that the aquifer-of-concern principle does not preclude the combination of hydrologically connected aquifers for scoring purposes.23 If the site-contaminated aquifer spreads hazardous substances to a second aquifer from which people draw water, the site presents a threat to that population. Therefore, as the EPA explained in the Support Document for the National Priorities List,24 given the hydrological connections between the two aquifers,25 the Agency reasonably treated them as a unit for purposes of the Hazard Ranking System.

2. Agency Consideration of the Task Force Report

Eagle-Picher further argues that the EPA failed to consider all relevant factors in scoring Tar Creek for groundwater release, because allegedly it based its decision to list Tar Creek exclusively on the preliminary Hittman Report, rather than the more extensive Task Force Report, thus ignoring hard data in the latter in favor of hypothesis and speculation in the former.26 The Agency did not, however, disregard the Task Force Report; instead, it noted in the National Priorities List Support Document that it had considered that study and found it consistent with its earlier conclusions predicated upon the Hittman Report.27 Because the Hittman Report adequately supported the EPA's determination that the site merited listing, the Agency did not perceive any need to perform a complete reevaluation based on the confirming documentation supplied by the Task Force Report.28

This decision was entirely reasonable, for the Agency quite rationally did not find any material contradiction between the two reports. Although data obtained in the course of the Task Force study arguably conflicted with the Hittman Report's speculations on the permeability of the formations overlying the Roubidoux, the Task Force Report acknowledged that other evidence suggested that in fact "water from mines in the Boone Formation has reached the Roubidoux Formation."29 The Report further observed that "[t]he most likely route by which the acid mine water could reach the Roubidoux Formation from the Boone Formation is by direct access through active and abandoned deep wells," and stated that despite the low permeability of intervening layers, "downward migration from the mine workings through the Chattanooga Shale, Cotter Dolomite and Jefferson City Formations is possible."30 Thus, although it offered a different explanation for the leakage than did the Hittman Report, the Task Force Report supported the EPA's ultimate conclusion that contaminated water from the Boone reaches the Roubidoux. Accordingly, we find no impropriety in the Agency's decision not to rescore Tar Creek as a result of the Task Force Report.

In support of its contention that the listing of the Tar Creek site was arbitrary and capricious, Eagle-Picher also argues that the EPA inadequately responded to significant problems that it raised in four of its comments to the Agency. To the extent that these comments presented material issues, we find that they received ample consideration and explanation by the Agency.

Eagle-Picher first asserts that the EPA did not sufficiently respond to a comment advocating rescoring of Tar Creek on the basis of the Task Force Report. As we have already pointed out, however, the Task Force and Hittman Reports are substantially consistent in their relevant conclusions, and thus the Agency rationally denied the reevaluation request.

Second, Eagle-Picher points to the alleged specific inconsistency between the two reports on the issue of permeability of the geologic formation separating the Boone and the Roubidoux.31 However, the quotations that the petitioner selects to demonstrate this contradiction must be read in the context of the Task Force's appraisal that its data are "not conclusive about the overall permeability of the formations which overlie the Roubidoux Formation."32 Furthermore, although Eagle-Picher maintained in its comments that the Task Force Report's core analysis showed that "sealing" by secondary cementation of cracks and fissures has occurred,33 we have searched the Task Force Report's data in vain for data buttressing Eagle-Picher's claim,34 and we agree with the EPA's conclusion that Eagle-Picher did not supply evidence to support the alleged contradiction.35

Eagle-Picher also claims that the EPA has inadequately answered a third comment: that, because the Agency based its assessment of an observed release to the Boone aquifer on data representing contamination of water in the mines, it improperly assumed that the contaminated mine water actually reaches the Boone aquifer. In reply to this comment, the Agency stated that because data indicated hydrological connections between the abandoned, water-filled mines and the Boone aquifer, any contamination of mine water naturally spread to the aquifer.36 We find the EPA's response to be adequate and reasonable.

Eagle-Picher's fourth challenge to the EPA's responses to its comments pertains to the scoring of an observed release to the Roubidoux aquifer, assertedly based on an observed release into the Boone aquifer. Although the EPA relied not only on the release into the Boone aquifer but also on the existence of bore holes connecting the Boone with the Roubidoux,37 Eagle-Picher commented that "no empirical data [were] presented to demonstrate that mine water is, in fact, migrating to the Roubidoux formation through 'boreholes.' "38 The EPA responded that reasonable inferences were appropriate and necessary, and that documentation of the Boone's contamination and the existence of numerous boreholes connecting it with the Roubidoux strongly supports the inference that the Roubidoux receives polluted water from the Boone.39 As we have already observed, the Agency pointed out that Eagle-Picher had identified no data indicating the extent, if any, to which these connections had sealed. We find the Agency's response to be appropriate and adequate.

Thus, the EPA adequately responded to all significant challenges presented by Eagle-Picher's comments. These responses evidence none of the arbitrariness Eagle-Picher would have us find; instead, they demonstrate a careful and reasoned consideration of the problems presented.

Eagle-Picher also challenges the listing of a second location in the Picher Field: the Cherokee County, Kansas site just across the Oklahoma-Kansas border from the Tar Creek site. Eagle-Picher asserts that in scoring the Kansas portion of the mining area the EPA improperly relied upon information collected at the Oklahoma site. Eagle-Picher adds that, as a result, the Kansas listing suffers from all of the infirmities that allegedly invalidate the Oklahoma listing. Since we reject Eagle-Picher's challenge to the listing of Tar Creek, we likewise reject the same challenges insofar as they would apply to Cherokee County. However, Eagle-Picher also argues that the inclusion of Cherokee County on the list was arbitrary because the site-specific evidence that supplemented the Oklahoma data was irrelevant and inadequately analyzed.

We attach less significance to the EPA's initial reliance on the Oklahoma data than does the petitioner. Moreover, we disagree with Eagle-Picher's assessment of the subsequently obtained site-specific information as irrelevant. And finally, although we share Eagle-Picher's concern that the EPA could have offered a clearer explanation of how it evaluated the new data, we find that a simple examination of the evidence makes the Agency's rationale sufficiently clear to withstand the petitioner's challenge.40

Eagle-Picher's argument that the Cherokee County, Kansas listing was based on data improperly borrowed from the Oklahoma site exalts form over substance. The Agency indeed used Oklahoma scores to support its proposal to include the Kansas portion of the mining fields.41 However, Eagle-Picher ignores the fact that the Kansas and the Oklahoma sites are actually one large mining area that was divided into two parts solely for administrative convenience. The fact that identical mining activities have been pursued on both sides of the state line42 further attests to the artificiality of the boundary. Since the two sites are in practical effect one, the EPA acted neither arbitrarily nor irrationally in initially using data collected in Oklahoma to support inferences about contamination in Kansas.

In any event, the EPA did not rely solely on Oklahoma data. Rather, it both developed data indicating that, for the most part, the Oklahoma conclusions accurately described the Kansas site, and gathered independent information on characteristics unique to Kansas. Eagle-Picher contends, however, that the EPA's inclusion in the administrative record of site-specific material for the Cherokee County site constitutes merely "a smoke screen for its continued reliance on Oklahoma data as the basis for listing the Cherokee County, Kansas site."43 Eagle-Picher suggests that the EPA should have removed Cherokee County from the National Priorities List until new data could be gathered, but that the Agency instead "cast around for some type of information, no matter how scanty, that the Agency could use to disguise the fact that its listing of Cherokee County was based on data borrowed from Tar Creek."44 Nothing in the record supports this allegation. On the contrary, the EPA agreed that site-specific evidence should be obtained, and revised the Cherokee County score downward from 66.74 to 58.15 as a result of its site reevaluation.45

We need not reach the question whether the EPA could have relied entirely on Oklahoma data to support the listing of the sister site in Kansas, for plainly it has not done so. We merely hold that the more limited use it did make of the Oklahoma information in assessing the Kansas listing was proper.

Eagle-Picher challenges the data used by the EPA to test its hypothesis that the kinds of releases occurring on the Oklahoma side of the state line also occur on the Kansas side.46 The Agency relied on three sets of data and included each of them, reduced to computer printout form, in the administrative record. One set of data, from the STORET system, provided information on surface water quality for several points along Tar Creek.47 Additional information compiled by the United States Geological Survey analyzed surface water and groundwater.48 The third collection of material supplied information on populations served by wells in Cherokee County.49

The EPA found that this site-specific material supported the finding of an observed release to both surface water and groundwater at the Kansas site.50 Eagle-Picher challenges this finding with respect to each route. We can find no legal infirmity in the Agency's assessment.

a. Surface Water

Eagle-Picher launches its attack on the Kansas surface-water data by contending that the EPA has not explained how it used the data in its analysis. It appears obvious, however, that the STORET data and the data from the United States Geological Survey provided background levels of contaminants against which the degree of contamination in Kansas might be tested. Eagle-Picher asserts, however, that these data are insufficient to demonstrate an observed release to surface water.

The Hazard Ranking System manual provides that the EPA may score a site for an observed release to surface water only if the Agency has "quantitative evidence that the facility is releasing contaminants into surface water."51 As an example of such quantitative evidence, the manual cites "the measurement of levels of contaminants from a facility in surface water, either at the facility or downhill from it, that represents a significant ... increase over background levels."52

The Kansas data show that the mining facility measurably contaminates surface water, rendering appropriate the EPA's scoring for an observed release. Levels of contaminants measured downstream in the mining field near Treece, Kansas, substantially exceeded, over a period of time, amounts found upstream in Tar Creek at Cravensville, Kansas.53 Clearly, as Eagle-Picher itself surmises in its brief, the EPA used the upstream samples to arrive at background levels, and interpreted increased contamination downstream as demonstrating pollution resulting from the creek's passage through the mining area.54

The Tar Creek data indicated that the samples of water taken upstream of the mining facility contained less lead, cadmium and zinc than did the specimens collected downstream.55 This constitutes an observed release into surface water under the Hazard Ranking System, and supports the EPA's conclusions.

b. Groundwater

In scoring the Cherokee County site for an observed release into the Roubidoux aquifer, the EPA relied upon documented contamination of the Boone aquifer in Kansas.56 6] The EPA also obtained data showing that the conditions creating hydrological connections in Oklahoma57 exist in Kansas as well.58 We have already concluded that the geological data evidencing connections between the Roubidoux and Boone aquifers support the score for an observed release into the Roubidoux. We therefore find no fault in the methodology the EPA utilized to rate the Cherokee County site for an observed release into the Roubidoux aquifer.

The EPA obtained data to supplement the Oklahoma data that supported its initial proposed listing of Cherokee County, and its analysis of this new data confirmed its belief that the pertinent conditions existing on the Oklahoma side of the state boundary likewise exist on the Kansas side. The EPA did not explain its reliance on this new information with the same degree of thoroughness it provided for its use of the Oklahoma data, but the Kansas data and the Agency's explanation nonetheless lead us to the conclusion that its determination was rational and adequately supported. We accordingly do not disturb its decision to include the Cherokee County, Kansas site on the National Priorities List.59

Whitewood Creek flows through the Black Hills area of South Dakota, the site of gold mining and milling for over 100 years.60 As a result of these operations, large amounts of tailings have been deposited in Whitewood Creek and washed downstream.61 Concern for the environmental impact of these wastes on ground and surface water resulted in a 1982 study agreement among petitioner Homestake Mining, the EPA, and the State of South Dakota,62 and led South Dakota to designate Whitewood Creek as its highest priority for inclusion on the National Priorities List.63

The two grounds for the EPA's listing of the Whitewood Creek site were South Dakota's top-priority designation and a final hazard ranking score of 63.76.64 Homestake contests both grounds, and also alleges procedural improprieties. We find no error either substantively or procedurally in the EPA's analysis and accordingly reject Homestake's challenge.65

We have previously considered and rejected Homestake's challenges to the substance of the Hazard Ranking System.66 Accordingly, we now address only its arguments questioning the system's application to the Whitewood Creek site.

1. Scoring of Observed Release to Groundwater

Homestake first contends that the EPA erred in scoring the site for an observed release to groundwater by using an inadequate sampling technique and improperly analyzing the sampling result. Homestake admits that four of the wells tested in the studies upon which the EPA relied showed arsenic consistently in excess of the drinking water standard. In its comments, however, Homestake hypothesized that the "extremely high degrees of variation [among the amounts detected at the various sampling locations] probably are related more to sample techniques than to accurate findings."67 Homestake criticized the sampling methodology and offered possible explanations for the amounts detected at various points.68 However, the EPA's response--that it had reviewed the data and had substantiated the presence of arsenic above background levels in six wells69--demonstrates that the Agency acted reasonably in resolving these factual issues.

The data Homestake cites as contradicting the studies relied upon by the EPA include an earlier report prepared by the Agency in 1973.70 The 1973 report merely states, however, that arsenic levels in the Whitewood Creek wells approached the recommended limit for domestic and stock water.71 Thus, the 1973 report is not inconsistent with the later studies on which the EPA relied.72

Homestake's comments also cited well samplings by another investigator, Jim Rouse, who detected no arsenic,73 and a report submitted by PedCo74 in connection with the Solid Waste Disposal Act.75 Homestake asserts that the PedCo report shows only low arsenic levels in settled solids in Homestake's active tailings pond, fresh tailings, and underground mine-waste rock, and suggests that, if the tailings deposited in Whitewood Creek come from similar ores, their arsenic levels must likewise be insignificant.76 As already noted, however, the EPA relied on reports directly documenting arsenic contamination of wells. The Agency clearly did not act arbitrarily in refusing to lay aside direct evidence of well pollution in favor of the mere hypothesis that tailings piles in Whitewood Creek were composed of ores similar to those known to contain only low levels of arsenic.

In its reply brief, Homestake for the first time suggests that the EPA could not properly have documented an observed release because no means existed to determine contemporaneous background levels in the area, where more than a century of mining activity has tainted all of the groundwater in the vicinity.77 Homestake has not demonstrated that this issue was raised during the rulemaking, and thus it is barred from raising the issue in this petition for review.78 Even if the argument were properly before us, however, we would reject it on the merits. Assuming that the Agency cannot ascertain the naturally occurring level of arsenic in the area's groundwater, this obstacle certainly does not preclude it from listing the site.79 The EPA's selection of the safe drinking standard as the "background level" was an entirely rational alternative, and one that is consistent with the purposes of the Hazard Ranking System.

Homestake next attacks the EPA's scoring of the groundwater target factors for the site--the nature of the groundwater use, the distance to the nearest well, and the population served.80 The site received a maximum score for groundwater use because the groundwater is used for irrigation and constitutes the sole available source of drinking water.81 The Agency calculated that the nearest well was 200 feet away, and that the groundwater served 2500 people within a three-mile radius.82

As to groundwater use, Homestake asserts that there is no evidence that the local population uses area water for drinking. Once again, Homestake has improperly raised an issue for review without demonstrating that the issue was presented to the Agency in the first place, and for that reason we reject this challenge without addressing the merits. Homestake also contends that the EPA overestimated the population served. However, Homestake's challenge to the score given to population count in essence contests the Hazard Ranking System's preference for using formulas even where actual figures are available, a preference which we have already upheld.83

The EPA gave Whitewood Creek the maximum score for proximity of the waste to the nearest well because it documented a domestic well 200 feet from the site. Homestake argues, however, that "[t]he data indicate ... that uses of the water take place more than 2,000 feet from the tailings."84 Because Homestake has not identified this "data," because it has failed to show that this facet of the groundwater score was challenged during the rulemaking, and because the record supports the EPA's finding,85 we will not upset the Agency's conclusion.

3. Scoring for Observed Release to Surface Water

Homestake contends that the EPA rated Whitewood Creek for an observed release to surface water without properly documenting such a discharge. The petitioner argues that the site does not meet the Hazard Ranking System's observed-release requirement of "measurement of levels of contaminants from a facility in surface water ... that represents a significant ... increase over background levels."86 The EPA reasoned that discharge of the arsenic-laden tailings into the waters of Whitewood Creek in itself constitutes an observed release.87 We find that this construction of the Hazard Ranking System is a reasonable interpretation of that regulation, and we therefore hold that the Agency properly subjected Whitewood Creek to a score for an observed release to surface water.

The EPA scored surface water proximity to a sensitive environment at the highest level, indicating that a 1000- by 600-foot marsh area formed part of the site.88 Homestake asserts, however, that the administrative record does not support the conclusion that this "sensitive environment" is within 100 feet of the hazardous substance,89 a prerequisite to receiving the maximum score in this category under the Hazard Ranking System.90 Both the August, 1982 Hazard Ranking System worksheets91 and the final Hazard Ranking System worksheets92 describe this marsh area. Although the final worksheets refer to a map93 that forms no part of the administrative record, we need not be detained by this defect in documentation, for even if we substract the points scored for proximity to a wetland, this would not reduce the site's score enough to remove it from the National Priorities List, and thus any error was not prejudicial.94

B. Consistency with Administrative Procedure Act

Homestake argues that its comments on the proposed National Priorities List did not receive consideration adequate to meet the requirements of the Administrative Procedure Act.95 Homestake first claims that the manner in which the EPA dealt with its comments violated the rule that "[t]he one who decides must hear,"96 because the EPA official who scored the site allegedly did not consider the comments. Additionally, Homestake contends that the EPA must not have considered the comments contemporaneously with its rulemaking decision, because it did not release the Support Document responding to the comments until two months after promulgation of the Final Rule.

We reject both contentions. First, the EPA's Administrator, not the site-scorer, ultimately decided whether to include particular sites on the National Priorities List.97 Regardless of input received from Agency staff, the decision to list Whitewood Creek was for the Administrator, who "bears full legal and personal accountability."98

Homestake's second argument--that the Administrator could not have considered Homestake's comments when he approved the list because the Support Document was not then publicly available, is likewise wanting in merit. The record shows that when the Administrator rendered the final decision, he had before him both the National Priorities List, which included hazard ranking scores broken down by pathway together with an explanatory preamble, and the Support Document that summarized and responded to comments, and which included five pages devoted exclusively to Whitewood Creek.99 That the Support Document was not available to the public in its final form does not mean that it was not available to the Administrator.

Homestake further argues that the EPA's failure to change the Whitewood Creek site scores in response to Homestake's comments indicates that the Agency did not consider those comments prior to promulgation of the list. However, the EPA's decision not to alter Whitewood Creek's score resulted from its determination that Homestake's comments did not support any change. That the Administrator declined to subscribe to the comments does not mean that he failed to consider them.

Thus, the record does not support the petitioner's allegations of impropriety in decisionmaking, nor does Homestake's proffer of evidence on the information available to the scorer or the date of public issuance of the Support Document raise any such issue. Given our obligation to presume procedural regularity,100 we would require a far stronger showing than Homestake has made before we would reverse the Agency on these grounds.

Homestake Mining Company has also operated a uranium mill near Milan, New Mexico, since 1958.101 Two on-site tailings ponds, one of which remains active, contain the waste from the 3500 tons of uranium ore that the mill processes daily at full capacity. When the mill operates at that level, 150 gallons of waste per minute seep from the active tailings pond into groundwater, contaminating it with selenium and uranium. Seepage into groundwater also occurs at the inactive pond.102

Homestake argues that its injection-collection system, installed to reduce seepage of contaminated water from tailings ponds to groundwater supplying area wells, prevents additional waste from reaching the wells, and therefore should reduce the groundwater score assigned to the site. This court has already rejected the argument that the EPA must consider the effects of remedial measures in scoring a site under the Hazard Ranking System.103 Homestake insists, however, that its injection-collection system interposes a barrier between the tailings pond and the wells--that the system constitutes a "discontinuity in the aquifer ... between the hazardous substance and all wells,"104 and therefore removes the wells from the scope of the Hazard Ranking System.105

Homestake contends, in essence, that the EPA must depart from its rule against considering remedial measures in site-scoring whenever those measures create conditions that reduce ratings under the Hazard Ranking System. We cannot accept this position. Most remedial measures are likely to have score-reducing results comparable to the discontinuity assertedly created by the injection-collection system at Milan. Even if that measure produces an effect expressly addressed by the Hazard Ranking System manual, the Agency's rule that remedial measures must be disregarded for purposes of Hazard Ranking System scoring is reasonable and must stand.

United Nuclear Corporation challenges the EPA's listing of its Churchrock, New Mexico operation, where it mines uranium-bearing ores, crushes the rock to extract uranium, and impounds the waste rock in tailings ponds at the site. The EPA scored the Churchrock site for observed releases to groundwater, to surface water, and to air.106 United Nuclear protests the ratings of all three of these pathways. We find that the EPA's application of the Hazard Ranking System to Churchrock was proper as to each pathway.

United Nuclear challenges the EPA's calculation, required by the Hazard Ranking System, of the number of wells within three miles of the release, and the population served by those wells.107 The EPA based its final score for the Churchrock site on the presence within that radius of seven wells serving 114 people.108 Data on the number of wells and the population served varied widely, from three wells and sixty-eight people to sixteen wells and 100 to 500 people, depending on the season.109 The EPA based its calculations on a comparison of data and maps submitted by United Nuclear with similar materials furnished by the Bureau of Indian Affairs and the Public Health Service,110 and, indeed, revised its original figures downward.111 We find the Agency's conclusions rational, and leave them undisturbed.

United Nuclear next argues that the stream known as the Rio Puerco constitutes a discontinuity in the aquifer, and thus only those wells located on its Churchrock bank should be counted in scoring population. Because, in its comments, United Nuclear cited no authority for this proposition, the EPA sought confirmation from the United States Geological Survey that the Rio Puerco created a barrier to groundwater flow.112 The Geological Survey responded that streams may, but do not always, act as barriers.113 Because United Nuclear did not support its position and because the EPA's investigation revealed that a stream does not necessarily establish a barrier, the Agency acted reasonably in declining to take this alleged discontinuity into account in scoring the site.

United Nuclear further argues that the EPA did not meet its burden, imposed by the Hazard Ranking System manual, of showing a likelihood that contaminants will migrate across the discontinuity.114 However, because United Nuclear did not carry its initial burden of showing that the stream created a discontinuity, the EPA had no obligation to demonstrate migration. Thus, in calculating the targets, the Agency did not err when it included wells separated from the Churchrock site by the Rio Puerco.115

Lastly, United Nuclear charges that the EPA incorrectly scored for an observed release to groundwater, because water naturally cleanses itself of contaminants as it moves through geological formations to the wells from which the water is drawn. At the time it adopted the Hazard Ranking System, the EPA stated that collection of data to establish the likelihood of such cleansing effects would be "an expensive and extensive ... effort."116 We have already sustained the simplification wrought by the observed-release scoring process of the Hazard Ranking System.117 Because United Nuclear's challenge is not materially different from the arguments we rejected previously, we accordingly find it to be without merit.

The EPA scored the Churchrock site for an observed release to surface water because of a spill of at least 93 million gallons of tailings fluid in 1979.118 United Nuclear does not deny that the spill occurred, but argues that it has since cleaned up the spill and that its redesigned and reconstructed tailings systems have released no further waste to surface water. United Nuclear further contends that the Act requires that hazards be currently imminent and therefore prohibits scoring based on 1979 site conditions rather than current conditions.119

The EPA's scoring for an observed release to the surface water at Churchrock in no way violates the Act's mandate. The purpose of the Hazard Ranking System is to ascertain the likelihood of a release,120 and the EPA may reasonably conclude that a past discharge increases the probability of future pollution from the same facility. We have already upheld the EPA's determination that remedial measures taken at a site are not to be considered in scoring under the Hazard Ranking System.121 We therefore find no infirmity in the Churchrock surfacewater scoring.

United Nuclear also claims that the EPA improperly scored the Churchrock site for an observed release to air. The first ground assigned in support of this claim is that uranium is not toxic in the concentrations in which it is released at the site, a position we have already rejected.122

United Nuclear next contends that its facility actually discharges only a few hundred pounds of uranium into the air, and that the EPA improperly based its score for waste quantity upon the tailings piles at the site. United Nuclear maintains that releases of uranium into the air do not result from the tailings, but rather from the nearby milling operation, and that it is improper for the Hazard Ranking System to measure the quantity of hazardous material in air releases by relying on the quantity of such material calculated for releases to water. However, the Hazard Ranking System manual provides that, for both the air and water pathways, "[h]azardous waste quantity includes all hazardous substances at a facility."123 United Nuclear does not challenge this rule as irrational. Because we find the rule to be dispositive, we reject United Nuclear's argument.

Finally, United Nuclear objects to the EPA's failure to consider (1) the concentration of the pollutant released, and (2) the fact that most people living within a four-mile radius of United Nuclear's operations are upwind of releases. United Nuclear contends that the Agency's approach is an oversimplification that flouts the congressional mandate that the National Priorities List be "based upon relative risk or danger to public health or welfare or the environment ... taking into account the population at risk."124 We have already sustained the Hazard Ranking System against the arguments that "the basis for scoring the 'target' population threatened by a release is unreasonable because it is rooted in a formula for estimating the population ... and does not utilize actual population figures," and that "the basis for scoring waste characteristics is unreasonable because it fails to consider the relatively low concentration of harmful substances present in high-volume mining wastes."125 We see no reason to depart from our previous rejection of these arguments.

Petitioners have challenged the inclusion on the National Priorities List of five specific sites located throughout the United States. Having considered the propriety of each contested listing in light of the relevant statutes, the Hazard Ranking System, and the arguments of the parties, and with due regard for the rule of prejudicial error, we conclude that the EPA's listing of these sites was reasonable and lawful. We accordingly deny the petitions for review.

So ordered.