Hunter L. Prillaman argued the cause, for intervenors American Water Works Ass'n and Ass'n of Metropolitan Water Agencies. With him on the brief were Kenneth A. Rubin, Robert J. Saner, II, and Rebecca Burke.

Jane Luxton argued the cause, for intervenor Lead Industries Ass'n, Inc. On the brief were Edwin H. Seeger and Kurt E. Blase.

Before GINSBURG and RANDOLPH, Circuit Judges, and SHADUR, Senior District Judge.*

Opinion for the Court filed by Circuit Judge GINSBURG.

The American Water Works Association and the Natural Resources Defense Council separately petition for review of the Environmental Protection Agency's final rule under the Safe Drinking Water Act promulgating a national primary drinking water regulation for lead. The NRDC challenges the EPA's decisions (1) establishing a treatment technique instead of a maximum contaminant level (MCL) for lead; (2) setting an extended compliance schedule; and (3) declining to regulate transient noncommunity water systems. The AWWA challenges (4) the EPA's inclusion of water lines owned by others in the definition of distribution facilities under the "control" of a public water system, and thus subject to the lead line replacement regulations. The Association argues that (a) the agency failed to provide notice and an opportunity to comment on its broad definition of control; (b) the definition is impermissibly vague; and (c) the EPA's interpretation unreasonably expands the agency's jurisdiction under the statute.

We grant in part and deny in part the NRDC's petition, as follows: the EPA is not required by the Act to set an MCL for lead at the tap, and the compliance schedule is not contrary to the statute, but the agency's explanation for its decision not to regulate transient noncommunity water systems is inadequate. We grant the AWWA's petition because the EPA failed to provide adequate notice that it might adopt a broad definition of control. Accordingly, we remand this matter to the EPA for a proper explanation of its noncommunity water systems policy and to provide an opportunity for public comment upon the definition of "control" it purported to adopt as part of the rule under review.

The Safe Drinking Water Act requires the EPA to promulgate drinking water regulations designed to prevent contamination of public water systems. 42 U.S.C. Sec. 300g-1(b). A national primary drinking water regulation (NPDWR) is one that specifies for a contaminant with an adverse effect upon human health either an MCL or a treatment technique, and establishes the procedures and criteria necessary to ensure a supply of drinking water that complies with that MCL or treatment technique. 42 U.S.C. Sec. 300f(1). An NPDWR is an enforceable standard applicable to all public water systems nationwide. In most of the NPDWRs promulgated to date the EPA has set an MCL for the particular contaminant being regulated. The EPA has the authority, however, to specify a treatment technique in lieu of an MCL if the Administrator finds that it is not "economically or technologically feasible" to determine the level of the particular contaminant in a public water system. 42 U.S.C. Sec. 300f(1)(C)(ii).

It is particularly difficult to determine the level of lead in a public water system. Less than one percent of all public water systems draw source water containing any lead. Notice of Proposed Rulemaking: Drinking Water Regulations; Maximum Contaminant Level Goals and National Primary Drinking Water Regulations for Lead and Copper, 53 Fed.Reg. 31,516, 31,526-27 (1988). Instead, most lead enters a public water system through corrosion of service lines and plumbing materials containing lead, such as brass faucets and lead solder connecting copper pipes, that are privately owned and thus beyond the EPA's regulatory reach under the Act. System-wide measurement is made still more difficult because the degree to which plumbing materials leach lead varies greatly with such factors as the age of the material, the temperature of the water, the presence of other chemicals in the water, and the length of time the water is in contact with the leaded material. Final Rule: Maximum Contaminant Level Goals and National Primary Drinking Water Regulations for Lead and Copper, 56 Fed.Reg. 26,460, 26,463-66, 26,473-76 (1991). Indeed, lead levels in samples drawn consecutively from a single source can vary significantly. Id. at 26,473-76. Measurement difficulties aside, treatment is made problematic because chemicals added to the drinking water supply in order to reduce the corrosion of pipes can increase the levels of other contaminants subject to MCLs. Id. at 26,486-87.

Recognizing the peculiar difficulty of establishing an MCL for lead in public water systems, the EPA proposed regulations that distinguish between control of lead in source water and control of lead due to corrosion. First, the EPA proposed an MCL for lead in source water, to be measured at the point where the water enters the distribution system. Second, the EPA proposed to require a treatment technique--an "optimal corrosion control treatment" supplemented with a program of public education--to be tailored specifically by each public water system in such a way as to minimize lead contamination in drinking water without increasing the level of any other contaminant to the point where it violates the NPDWR for that substance. 53 Fed.Reg. at 31,537-38.

The EPA solicited comments on this two-part monitoring and treatment proposal and on several alternatives that it was not then proposing. One such alternative was to require each public water system to replace the lead service lines it owns or controls that, after treatment to reduce corrosion, still contribute a significant amount of lead to tap water. Id. at 31,546, 31,547-48. Under this approach the EPA would erect a "rebuttable presumption" that the public water system "owns or controls and therefore can replace, the lead components up to the wall of the building served." 53 Fed.Reg. at 31,548.

In the final rule the EPA abandoned its two-part monitoring and treatment proposal in favor of a rule under which all large water systems must institute corrosion control treatment, while smaller systems must do so only if representative sampling indicates that lead in the water exceeds a designated "action level." 56 Fed.Reg. at 26,550 (to be codified at 40 C.F.R. Sec. 141.81(d) & (e)). The agency also adopted a schedule that allows a public water system five or more years to comply with the regulation, depending upon the number of persons it serves, see id. at 26,480 (codified at 40 C.F.R. Sec. 141.81(a)); see also id. at 26,479-80, 26,494-97. The EPA required larger systems to come into compliance sooner than smaller systems because they are generally more sophisticated technically and have a greater impact upon the purity of drinking water; also the states, which are responsible for implementing the regulation, would benefit from experience gained with larger systems before reviewing treatment plans for smaller systems. The EPA exempted from the rule all transient noncommunity public water systems, such as those in restaurants, gas stations, and motels. 40 C.F.R. Sec. 141.80(a)(1).

Unlike the proposed rule, the final rule requires each public water system to replace each year at least 7% of the lead service lines it controls that when tested exceed a designated action level. 40 C.F.R. Sec. 141.84(b) & (d). A public water system is said to "control" a service line if it has

authority to set standards for construction, repair, or maintenance of the line, authority to replace, repair, or maintain the service line, or ownership of the service line.

40 C.F.R. Sec. 141.84(e). The rule establishes a presumption that the public water system controls every service line up to the wall of the building it serves; the system can rebut the presumption only by demonstrating that its control is limited by state statute, local ordinance, public service contract, or other legal authority. 40 C.F.R. Sec. 141.84(e). A public water system that controls only part of a service line must replace the portion under its control and must offer to replace the remaining portion, although not necessarily at the system's expense. 40 C.F.R. Sec. 141.84(d).

Together the petitioners raise four challenges to the EPA's final rule, which we review under the familiar framework of Chevron U.S.A., Inc. v. Natural Resources Defense Council, 467 U.S. 837, 104 S. Ct. 2778, 81 L. Ed. 2d 694 (1984). Where the "Congress has directly spoken to the precise question at issue," we must give effect to its "unambiguously expressed" intent; but where the Congress has been silent or its statement is ambiguous, we will defer to the EPA's interpretation if it is reasonable in view of the text, the structure, and the underlying purpose of the statute. Id. at 842-43, 104 S.Ct. at 2781-82; see also American Mining Congress v. United States EPA, 824 F.2d 1177, 1184 (D.C.Cir.1987).

The NRDC first contends that, because it is economically and technologically feasible to ascertain the level of lead in water, the Safe Drinking Water Act requires that the EPA set an MCL for lead. See 42 U.S.C. Sec. 300f(1)(C); 42 U.S.C. Sec. 300g-1(b)(7). Further, because the tap is the delivery point to the user of a public water system, the NRDC concludes that the MCL must be set at the tap.

At bottom the NRDC and the EPA disagree over the meaning of the word "feasible" as it applies to ascertaining the level of lead in drinking water. The NRDC argues that the Congress clearly expressed its intent that "feasible" be understood to mean "physically capable of being done at reasonable cost"; accordingly it argues that the EPA's rule is contrary to the plain meaning of the statute. See Chevron, 467 U.S. at 842-43, 104 S. Ct. at 2781-82. For its part, the EPA does not dispute that it is "feasible" to monitor lead under the definition advanced by the NRDC; instead the agency interprets "feasible" to mean "capable of being accomplished in a manner consistent with the Act." The agency argues that if public water systems were required to comply with an MCL for lead, they would have to undertake aggressive corrosion control techniques that might reduce the amount of lead leached from customers' plumbing but would also increase the levels of other contaminants. The EPA argues that because the Congress apparently did not anticipate a situation in which monitoring for one contaminant, although possible, is not conducive to overall water quality, it impliedly delegated to the agency the discretion to specify a treatment technique instead of an MCL.

We agree with the EPA that the meaning of "feasible" is not as plain as the NRDC suggests. Although we generally assume that the Congress intends the words it uses to have their ordinary meaning, see Securities Industry Ass'n v. Board of Governors of Federal Reserve System, 468 U.S. 137, 149, 104 S. Ct. 2979, 2985, 82 L. Ed. 2d 107 (1984), case law is replete with examples of statutes the ordinary meaning of which is not necessarily what the Congress intended. See, e.g., Young v. Community Nutrition Institute, 476 U.S. 974, 980, 106 S. Ct. 2360, 2364, 90 L. Ed. 2d 959 (1986) (EPA's interpretation of unclear statute held rational though not the "more natural interpretation"); American Mining Congress, 824 F.2d at 1185-86. Indeed, where a literal reading of a statutory term would lead to absurd results, the term simply "has no plain meaning ... and is the proper subject of construction by the EPA and the courts." Chemical Manufacturers Association v. Natural Resources Defense Council, Inc., 470 U.S. 116, 126, 105 S. Ct. 1102, 1108, 84 L. Ed. 2d 90 (1985). If the meaning of "feasible" suggested by the NRDC is indeed its plain meaning, then this is such a case; for it could lead to a result squarely at odds with the purpose of the Safe Drinking Water Act.

The Congress clearly contemplated that an MCL would be a standard by which both the quality of the drinking water and the public water system's efforts to reduce the contaminant could be measured. See 42 U.S.C. Sec. 300g-1(b)(5). Because lead generally enters drinking water from corrosion in pipes owned by customers of the water system, an MCL for lead would be neither; ascertaining the level of lead in water at the meter (i.e. where it enters the customer's premises) would measure the public water system's success in controlling the contaminant but not the quality of the public's drinking water (because lead may still leach into the water from the customer's plumbing), while ascertaining the level of lead in water at the tap would accurately reflect water quality but effectively hold the public water system responsible for lead leached from plumbing owned by its customers.

We must defer to the EPA's interpretation of "feasible" if it is reasonable, Chevron, 467 U.S. at 842-43, 104 S. Ct. at 2781-82, and we think that it is. A single national standard (i.e., an MCL) for lead is not suitable for every public water system because the condition of plumbing materials, which are the major source of lead in drinking water, varies across systems and the systems generally do not have control over the sources of lead in their water. In this circumstance the EPA suggests that requiring public water systems to design and implement custom corrosion control plans for lead will result in optimal treatment of drinking water overall, i.e. treatment that deals adequately with lead without causing public water systems to violate drinking water regulations for other contaminants. 56 Fed.Reg. 26,487.

Viewing the Act as a whole, we cannot say that the statute demonstrates a clear congressional intent to require that the EPA set an MCL for a contaminant merely because it can be measured at a reasonable cost. In light of the purpose of the Act to promote safe drinking water generally, we conclude that the EPA's interpretation of the term "feasible" so as to require a treatment technique instead of an MCL for lead is reasonable.

We turn next to the NRDC's contention that the compliance schedule promulgated by the EPA is contrary to the statutory injunction that NPDWRs "shall take effect 18 months after the date of their promulgation." 42 U.S.C. Sec. 300g-1(b)(10). According to the petitioner, in "ordinary English" the phrase "shall take effect" means "shall be fully implemented and enforced against public water systems." Therefore, we are told, the meaning of the statute is plain and the EPA's interpretation to the contrary should be struck down, Chevron, 467 U.S. at 842-43, 104 S. Ct. at 2781-82, with the result that the rule must be fully implemented within eighteen months of when it was promulgated.

The EPA, on the other hand, contends that the plain meaning of the statute is that the agency cannot impose the requirements of an NPDWR upon public water systems any earlier than eighteen months after promulgation. According to the agency, the Congress included the 18-month provision in the Act not in order to force the agency to adopt a hasty implementation schedule but in order to "constrain[ ] the Agency's authority under the Administrative Procedure Act, 5 U.S.C. Sec. 553(d), to make rules effective 30 days after their publication in the Federal Register." 56 Fed.Reg. at 26,494-95.

We start, as usual, with the terms of the statute. The spare mandate--that NPDWRs "shall take effect eighteen months after the date of their promulgation"--is in our view considerably less clear on its face than the NRDC suggests. As we only recently noted, depending upon the context, "take effect" can mean either "take legal effect" (as the EPA here suggests), or "produce results" (as the NRDC suggests). See Boehner v. Anderson, 30 F.3d 156, 161-62 (D.C.Cir.1994); Natural Resources Defense Council v. Browner, 22 F.3d 1125, 1137-40 (D.C.Cir.1994). Because we must examine the effective date provision in its statutory context in order to determine which meaning the Congress intended, we cannot say that either the NRDC's or the EPA's reading is the uniquely "plain meaning" of the provision.

Turning to that context, we see first that the effective date provision refers only to drinking water regulations, not to their implementation and enforcement. The Act defines an NPDWR as a regulation that (1) applies to public water systems; (2) identifies a contaminant that could adversely affect human health; (3) specifies either an MCL or a treatment technique to control the contaminant; and (4) contains criteria and procedures, such as operating and maintenance standards, to assure a reasonably safe drinking water supply. 42 U.S.C. Sec. 300f(1). An NPDWR does not, however, set an implementation schedule or enforcement procedures for the MCL or treatment technique it specifies. On the contrary, the Act requires the States, which have primary enforcement responsibility, to promulgate regulations to implement and enforce the NPDWRs. See 42 U.S.C. Secs. 300g-2, 300g-3. Clearly, therefore, the Congress contemplated that the date upon which a drinking water regulation takes effect under 42 U.S.C. Sec. 300g-1(b)(10) would not necessarily be the date upon which the regulation will be implemented or enforced. See, e.g., Boehner v. Anderson, 30 F.3d 156 (D.C.Cir.1994) (law may have delayed effect).

Second, the 18-month provision applies equally to all drinking water regulations, whether they promulgate a treatment technique or an MCL. Although it may be reasonable to assume that an MCL can be implemented and enforced 18 months after promulgation, we doubt that the Congress expected that each state could in 18 months approve the various treatment plans submitted by all the public water systems in the state; promulgate and implement enforcement regulations; grant exemptions and variances for public water systems that cannot comply with the NPDWR; and establish a reporting mechanism, which must be designed on a site-by-site basis in order to minimize exposure to the contaminant in drinking water without adversely affecting the public water system's compliance with other MCLs or treatment techniques. See 42 U.S.C. Secs. 300g-2, 300g-4, 300g-5, 40 C.F.R. Secs. 142.4-142.19; see also 56 Fed.Reg. at 26,487 (describing factors to take into account in designing treatment technique), 26,494-95 (describing actions by public water systems and States necessary to implement treatment technique). Compressing these activities into 18 months could compromise effective treatment for lead; it is surely not unreasonable, therefore, for the EPA to interpret the 18-month provision so as to prefer the Act's overall goal of safe drinking water over a hasty implementation and enforcement schedule. Accordingly, we conclude that the EPA's interpretation of the 18-month provision is reasonable.

The NRDC next challenges the EPA's decision not to apply the NPDWR for lead to transient, non-community water systems (defined as those that serve fewer than 25 individuals regularly for six months per year). See 56 Fed.Reg. at 26,478. The NRDC claims that this limitation is in direct conflict with the requirement of the Act that NPDWRs "shall apply to each public water system in each State." 42 U.S.C. Sec. 300g (with certain exceptions not relevant here). The NRDC asserts that transient, non-community water systems--such as those at restaurants, motels, and parks--are "public water systems" within the meaning of the Act, which defines a public water system as one that provides "piped water for human consumption" regularly to at least twenty-five individuals or to at least fifteen service connections. 42 U.S.C. Sec. 300f(4). As a corollary, the NRDC contends that any exemption from an NPDWR must be made pursuant to the variance or exemption provisions of the Act. See 42 U.S.C. Secs. 300g-4, 300g-5.

For its part, the EPA claims (on brief) that it decided not to impose the NPDWR for lead upon transient, non-community systems because lead poses a significant health risk only with chronic exposure. This is consistent with long-standing agency practice. See 40 Fed.Reg. 59,566 (1975) (stating that MCLs are "based on the potential health effects of long-term exposure"); 52 Fed.Reg. 25,694-95 (1987) (excluding transient systems from certain regulations because they pose no "long-term health risk"); see also 40 C.F.R. Secs. 141.61(a) & (c), 141.62(a) (with exception of nitrate and nitrite, MCLs applied only to community and non-transient, non-community systems). For factual and policy support of its decision, the EPA cites its previous documentation that lead in drinking water poses adverse health effects only with chronic exposure, see, e.g., Environmental Defense Fund, Inc. v. Costle, 578 F.2d 337, 348-49 (D.C.Cir.1978); the administrative burden of regulating the large number of transient, non-community systems; the absence of any appreciable public health benefit from doing so; and the inability of the generally unsophisticated owners of such systems to comprehend and comply with complex drinking water regulations. See Comment Response Document for the August 18, 1988 Proposed Lead and Copper Rule, Response to Comment 1061-46B.

Be that as it may, the EPA concedes that in adopting the final rule it failed adequately to explain its basis for excluding transient, non-community systems from the NPDWR for lead. Nonetheless, the agency suggests that the rule not be vacated but instead that the court should remand the record in order for the agency to try again.

That the agency has a long-standing policy of excluding transient systems from NPDWRs for non-acute contaminants is clear; that it inadvertently failed to document its decision to treat lead pursuant to this policy is also clear. See Response to Comment 1061-46B: (although "lead can pose acute toxic effects, such effects only occur at extremely high exposure levels (give an indication of how high must be) which are not found in drinking water (explain what we mean by this. Do we mean that the concentrations found in drinking water in the past have not been high enough, or is it more that the levels would have to be so high that we would never expect to see them in water)"). Because the agency's error is apparently a technical one, and we think it more likely than not that the agency can justify its exemption decision when it gets down to trying, vacatur would be unnecessarily disruptive to the exempted industries. See Allied-Signal, Inc. v. U.S. Nuclear Regulatory Comm'n, 988 F.2d 146, 150-51 (D.C.Cir.1993). Rather than vacate the exemption, therefore, we remand this matter to the EPA for a more detailed justification of its decision to exclude transient, non-community water systems from compliance with the drinking water regulation for lead.1D. Definition of Control

Finally, we consider the AWWA's challenge to the agency's criteria for determining whether a water system service line is under the "control" of the system operator, and thus subject to the lead service line replacement regulations. Although public water systems generally own only that part of the service line that underlies public property, 56 Fed.Reg. at 26,503, the EPA established in its final rule a rebuttable presumption that the public water system "controls" the water service line up to the wall of the building unless the system (1) does not own the line; and neither (2) has the authority to replace, repair, or maintain the service line, nor (3) has the authority to set standards for construction, maintenance, or repair of the line. 56 Fed.Reg. at 26,504, 26,553 (codified at 40 C.F.R. Sec. 141.84(e)).

The AWWA contends first that the definition of "control" in the final rule was not prefigured in the proposed rule and therefore that the public did not have notice of and opportunity to comment upon it, as required by the Administrative Procedure Act, 5 U.S.C. Sec. 553(b), (c). For its part, the EPA points out that it stated in its notice of proposed rulemaking that the agency was considering adopting a lead service line replacement requirement under which it would establish a presumption, rebuttable only by invoking a statute, ordinance, or contract to the contrary, that a public water system owns or controls the line up to the wall of the building it serves. 53 Fed.Reg. at 31,548. The agency's definition of control in the final rule, it argues, merely states "in slightly more precise terms the alternative discussed in the [notice of proposed rulemaking]."

An agency fulfills the notice requirement of the APA if it "provide[s] sufficient factual detail and rationale for the rule to permit interested parties to comment meaningfully." Florida Power & Light Co. v. United States, 846 F.2d 765, 771 (D.C.Cir.1988). In most cases, if the agency then alters its course in response to the comments it receives, little purpose would be served by a second round of comment. The test we have developed for deciding whether a second round of comment is required in a particular case is whether the final rule promulgated by the agency is a "logical outgrowth" of the proposed rule. See, e.g., Chemical Waste Management v. EPA, 976 F.2d 2, 28 (D.C.Cir.1992); Small Refiner Lead Phase-Down Task Force v. EPA, 705 F.2d 506, 546-47 (D.C.Cir.1983). We apply that standard functionally by asking whether "the purposes of notice and comment have been adequately served," Fertilizer Institute v. EPA, 935 F.2d 1303, 1311 (D.C.Cir.1991), that is, whether a new round of notice and comment would provide the first opportunity for interested parties to offer comments that could persuade the agency to modify its rule.

The EPA referred often in the notice of proposed rulemaking to the scope of control exerted by a public water system but it never once treated "control" as a term of art or otherwise in need of definition. On the contrary, in discussing a public water system's control over lead service lines, the agency allowed that private "ownership and/or control" of a service line is a limitation upon its authority without so much as a suggestion that private ownership might not preclude a public water system from having "control" over a service line. Consider:

A further complication of the pipe replacement issue is that ownership and/or control of the service line often is split between the public water system and the property owner....

. . . . .

EPA believes that, in general, its authority to require replacement of service lines and other connections ends where the water supplier ownership or control of the lines end. The EPA has conducted a limited study of some large cities, which indicates that these water suppliers generally limit maintenance on water pipes to those portions they own. However, several cities have authority to enter private property to perform work on water lines under special circumstances; in such cases the property owner is generally billed for the work performed.

. . . . .

EPA also solicits information on the extent of the authority of public water systems over lead service lines and connections under State law and local ordinances.

53 Fed.Reg. at 31,535-36. If it does not necessarily imply the contrary, this passage certainly gives no affirmative indication that the EPA was contemplating that a public water system might be said to control a service line merely because it has the authority (but not the obligation) to set construction standards or to repair.

There is a further reason to believe that the AWWA, which commented extensively upon the lead service line replacement program, was not asleep at the switch when it failed to comment upon the scope of a public water system's control over service lines. Seven months before the EPA published its notice of proposed rulemaking, the Georgia Supreme Court had interpreted the definition of a "public water system" under that state's Safe Drinking Water Act--which is identical to the definition in the federal statute--as confining the regulatory authority to portions of the service line not underlying private property. See Bass v. Ledbetter, 257 Ga. 738, 363 S.E.2d 760, 761 (Ga.1988). Yet the EPA said nothing in the notice of proposed rulemaking about the Georgia case, which was then the only authority to have considered the definition of a public water system. In these circumstances it was quite reasonable for the regulated industry not to perceive any doubt on the EPA's part that its control over a service line ends at the private property line.

Because we find that the interested parties could not reasonably have "anticipated the final rulemaking from the draft [rule]," Anne Arundel County v. EPA, 963 F.2d 412, 418 (D.C.Cir.1992) (quoting Natural Resources Defense Council, Inc. v. EPA, 863 F.2d 1420, 1429 (9th Cir.1988)), we conclude that the EPA failed to provide adequate notice that it would adopt a novel definition of control. Accordingly, we vacate the rule insofar as it deems privately owned lead service lines to be within the "control" of a public water system for the purpose of obligating the system to replace them.

The AWWA also contends that the EPA's expansive definition of "control" would extend the agency's reach beyond its lawful grasp under the Safe Drinking Water Act, and that the definition is impermissibly vague in that it seems to require water suppliers to enter private property but "does not indicate if it is intended to create a right of entry." Because we vacate the rule for lack of public notice, we need not reach these substantive issues. See, e.g., National Family Planning v. Sullivan, 979 F.2d 227, 240-41 (D.C.Cir.1992); Solite Corp. v. EPA, 952 F.2d 473, 499-500 & n. 13 (D.C.Cir.1991).

For the foregoing reasons, we deny the petition of the NRDC insofar as it challenges both the EPA's decision to require a treatment technique in lieu of an MCL for lead and the compliance schedule for public water systems. We remand the matter to the EPA for a better explanation of its transient, non-community water systems policy, and such further action as it may wish to take in light of our holding that the agency failed to provide adequate notice of how it would define "control."

So ordered.