MIKVA, Circuit Judge:

The process of loading and unloading vessels docked at marine terminals produces significant quantities of air pollution in harbor areas. Similarly, pollutants emitted as a ship approaches and leaves a marine terminal contribute substantially to the poor air quality of many harbors. Recognizing these facts, the Environmental Protection Agency (EPA) initially promulgated rules under which both "dockside" and "to-and-fro" vessel emissions were to be taken into account in regulating the construction and operation of marine terminals. See 45 Fed.Reg. 52,676, 52,736, 52,746 (1980) (vessel emission requirements). In 1981, however, EPA began to chart a new course, twice "staying" the vessel emission requirements and then, on July 25, 1982, revoking them. See 46 Fed.Reg. 36,695 (1981) (first stay); id. at 61,612 (second stay); 47 Fed.Reg. 27,544 (1982) (revocation). This about face did not result from EPA's reconsideration of whether ascribing vessel emissions to marine terminals was a wise means of implementing the agency's mandate under the Clean Air Act, 42 U.S.C. §§ 7401- 7642 (Supp. V 1981) (the Act); reversal followed instead from EPA's new conclusion that the agency had never had the authority in the first place to impose responsibility on marine terminals for emissions traceable to marine vessels. The central question in this case is whether, in revoking the vessel emission requirements, EPA properly construed its mandate under the Clean Air Act.

We hold that EPA was correct to interpret the key statutory term "mobile sources" to include marine vessels but that the agency acted far too precipitously in proceeding from that interpretation to the further conclusion that it "therefore" had no authority to attribute to marine terminal owners any emissions arising from stationary dockside activities involving both vessels and terminals. Absent an attempt by the agency to identify the various emissions and the way in which they are physically discharged into the atmosphere, we have no way of determining whether the statute authorizes some of those emissions to be attributed to the terminal and, if so, which ones. Accordingly, we reverse the agency's categorical exclusion of "the activities of any vessel" from the relevant definitions of a "stationary source" of pollution and remand so that the agency can undertake the analysis required by the statute. Because, however, it is entirely implausible that a vessel's "to-and-fro" emissions could be attributed to a marine terminal owner under any approach that the statute would tolerate, we affirm that portion of EPA's 1982 repeal which excluded "to-and-fro" vessel emissions from the definition of "secondary emissions."

I. BACKGROUND

In the Clean Air Act of 1970 Congress established a comprehensive state and federal scheme to control air pollution in the United States. See Pub.L. No. 91-604, 84 Stat. 1676 (1970). The central elements in this comprehensive scheme were the Act's new source review provisions, which required all major new "stationary" sources of pollution as well as all existing "stationary" sources that were being significantly modified to obtain a permit before construction. At the same time, a series of stringent requirements were established for the issuance of such permits. See 42 U.S.C. §§ 7475, 7502-7503 (Supp. V 1981). Seven years later, Congress returned to the Act, and, in an effort to adjust the balance between federal and state regulation in certain areas, carved out an exception from the federal controls on "Stationary" sources for facilities such as parking lots that traditionally had been exclusively within the regulatory domain of the states. See Pub.L. No. 95-95, 91 Stat. 685, 695-96 (1977). The 1977 amendments state that, in determining whether a new or modified source qualifies as "major" or "significant" and hence is subject to the new source review provisions, EPA cannot include, nor require states to regulate, emissions from "mobile" sources located at the stationary source. The effect of this bar is to prohibit EPA from promulgating, or from requiring as a condition for approval of a state implementation plan (SIP), any program for the facility-by-facility review of facilities, such as parking lots, which are not themselves significant sources of pollution but which attract substantial numbers of "mobile" pollution-emitting sources.

Such facilities are known as "indirect sources of pollution." From the outset, the Act's new source review provisions have applied to "stationary sources." Originally, the term stationary source "generally [had] been interpreted to mean facilities that affect or may affect air quality primarily because of their own air pollutant emissions." 38 Fed.Reg. 9,599 (1973). However, by 1973 EPA believed that, in addition to the transportation control measures then being considered — such as motor vehicle inspection and retrofitting used automobiles with emission control devices — preconstruction review of indirect sources would be a necessary further step "to insure the maintenance of the national ambient air quality standards, particularly for mobile source-related air pollutants, beyond 1975." Id.

In deciding to add such indirect source review to the list of transportation control measures states had to consider in their attempts to meet ambient air quality standard deadlines, EPA was no doubt prompted to some extent by our decision in NRDC v. EPA, 475 F.2d 968 (D.C. Cir. 1973) ( NRDC I). In that case, we found that EPA had no authority to extend the statutory deadlines for either state submission of the transportation control plan portions of their SIPs or for achievement of primary air quality standards for certain pollutants. Id. at 970. Rather, EPA was required to disapprove all "plans which do not provide for measures necessary to insure the maintenance of the primary standard after May 31, 1975. . . ." Id. at 972. See J. KRIER R. STEWART, ENVIRONMENTAL LAW AND POLICY 441-59 (1978) (chronology of EPA's early experience with transportation control plans and indirect source review).

Whatever the impetus, EPA went on to disapprove state implementation plans that did not provide for, inter alia, indirect source review, 40 C.F.R. § 52.22(a) (1974), and then established a mandatory federal indirect source review program for those states whose SIPs had been disapproved, id. at § 52.22(b). EPA's statutory authority to impose indirect source review was confirmed by South Terminal Corp. v. EPA, 504 F.2d 646, 667-71 (1st Cir. 1974).

Congress, however, was not so accepting of EPA's actions. Congress reacted negatively and immediately to EPA's attempt to regulate indirect sources of pollution, attaching a number of riders to appropriation bills in an effort to preclude EPA administration of programs designed to tax or limit indirect emissions from parking facilities. See, e.g., Pub.L. No. 93-245, 87 Stat. 1071 (1974). Because some states had already voluntarily adopted indirect source review programs in order to comply with the EPA regulations, Congress extended similar relief retroactively and amended the Act to make clear that states could not be required, though they were permitted, to regulate indirect sources of pollution. See 42 U.S.C. § 7410(a)(5)(A).

In the period between this court's 1973 decision in NRDC I and Congress' actions in 1977, however, neither Congress nor EPA focused any direct attention on the problem of emissions from vessels docked at marine terminals. The EPA regulation of indirect sources to which Congress reacted so immediately and decisively concerned only facilities that attracted automobiles. And the legislative history of that congressional response, up to and including the 1977 Amendments, similarly reveals no consideration of the vessel-emission problem.

The problem of the emission of substantial quantities of pollutants from the loading and unloading of marine vessels, including hydrocarbons, particulates, and sulfur oxides, was first addressed when EPA set out to define "stationary source." To implement the Act's mandate that new source review apply to all major new stationary sources of pollution and to all significant modifications of existing sources. EPA had to determine how to group pollution-emitting activities for purposes of determining the contours of a "stationary source" of pollution. After we rejected EPA's first effort to do so, see Alabama Power Co. v. Costle, 606 F.2d 1068, 1077-78 (D.C. Cir.), superseded in other aspects, 636 F.2d 323, 394-99 (D.C. Cir. 1979), and after several EPA proposed and modified definitions, see 45 Fed.Reg. 6,802-03 (1980); 44 Fed.Reg. 51,924, 51,952 (1979), EPA eventually defined a stationary source as "any building, structure, facility, or installation which emits or may emit any air pollutant subject to regulation under the Act." 45 Fed.Reg. 52,736 (1980). The agency then defined "building, structure, facility, or installation" as "all of the pollutant-emitting activities which belong to the same industrial grouping, are located on one or more contiguous or adjacent properties, and are under the control of the same person (or persons under common control)." Id. (emphasis added).

In the promulgation of this 1980 definition, EPA directly addressed the role that emissions from docked vessels were to play in regulating pollution from the activities carried on at marine terminals. EPA stated that it intended the final definition of stationary source "to encompass the activities of a marine terminal and only those dockside activities that would serve the purposes of the terminal directly and would be under the control of its owner or operator." Id. at 52,696. With regard to ships, the agency limited "dockside activities" to "those activities in which the ships would engage while docked at the terminal . . . [and which] would directly serve the purposes of the terminal, such as loading and unloading . . . [and] over which the owner or operator of the terminal would have control." Id. (emphasis added). To reach this result, EPA specifically had to reject the argument that ships at dockside were "mobile sources" within the meaning of the 1977 amendment to the Act that precluded indirect source review. Squarely confronting this question, the agency concluded that "an activity such as loading and unloading is certainly stationary, even if the ships that engage in it have mobility. Ships, moreover, are not `mobile sources' within the meaning of § 110(a)(5) of the Act, the provision restricting indirect source review." Id. The result of this process was that some of the emissions from the dockside activities of a vessel could be included in determining whether a new terminal would be a "major" source of pollution and whether a modification of an existing terminal would be "significant."

At the same time, EPA for more limited purposes also ascribed to marine terminals the emissions of vessels coming to and from the terminal. See 45 Fed.Reg. 52,737 (1980). These emissions were assigned to marine terminals under the rubric of "secondary emissions," defined by the agency as emissions that "occur as a result of the construction or operation of a major stationary source or major modification, but do not come from the major stationary source or major modification itself." Id. Secondary emissions are not used to determine whether a new or modified source would be a "major" emitter and hence subject to new source review, but once a source has been independently ruled major, secondary emissions become relevant for various other purposes. Secondary emissions, for example, must be included in the air quality impact analysis that an applicant who seeks to construct a major new stationary source or to significantly modify an existing source must perform; in this analysis, the applicant must demonstrate that all emission increases from the construction or modification — including secondary emissions — will not contribute to a violation of air quality standards. See 40 C.F.R. § 52.21(k) (1983).

Both aspects of the vessel emissions requirements — the dockside component and the to-and-fro component — became effective on August 7, 1980. Two months later, GATX Terminals Corp. (GATX), intervenors in the present case, simultaneously filed a petition for reconsideration before EPA and a petition for review in this court, see Docket No. 80-2212 consolidated with Chemical Manufacturers Ass'n v. EPA, No. 79-1112 and consolidated cases, the latter of which is currently undergoing settlement negotiations. GATX's petitions for reconsideration lay dormant before EPA until July 15, 1981, when EPA announced that it had, without notice or comment, temporarily stayed the vessel emissions requirements as of July 7, 1981 for ninety days (temporary stay). 46 Fed.Reg. 36,695 (1981). At that time, EPA announced that it had decided to reconsider its treatment of vessel emissions and solicited comments on whether the requirements should be stayed pending completion of that process and, if so, upon what conditions. Id.

The temporary stay ended on October 7, 1981, bringing the vessel emissions requirements back into effect. This state of affairs lasted only until December 17, 1981, however, when EPA announced that it had again stayed the requirements, this time from December 7 until whenever it reached a final decision on the GATX reconsideration petitions. 46 Fed.Reg. 61,612 (1981) (extended stay). Proposing to rescind the vessel emissions requirements altogether, EPA also announced a notice of proposed rulemaking at this time. EPA suggested, as GATX had argued, that recission was appropriate because the requirements had been promulgated without adequate notice and because they were in violation of the 1977 ban on indirect source review.

On June 25, 1982, EPA revoked the vessel emissions requirements in their entirety. 47 Fed.Reg. 27,554 (1982). In doing so, the agency found it unnecessary to decide a host of questions which call for the application of its expertise, including whether owners or operators of a marine terminal control the vessels docked there, for the agency accepted both the inadequate notice and the statutory interpretation arguments put forward by GATX. Acceptance of the latter constituted rejection of EPA's earlier interpretation of the Act, under which ships at dockside were not considered to be "mobile sources" within the meaning of the 1977 congressional prohibition on indirect source regulation. Instead, the agency maintained, as it now does on review, that vessels are "mobile sources" even when being unloaded at dockside, and hence that none of the emissions traceable in any way to the activities of marine vessels can be made the responsibility of a marine terminal. Petitioners, the Natural Resources Defense Council, Inc., the State of California, and the Illinois Environmental Protection Agency, challenge EPA's about-face.

II. DISCUSSION

EPA's decision to stay and then revoke the vessel emission requirements rests on two premises: first, that the requirements are barred by the 1977 ban on indirect regulation of mobile sources, and second, that the requirements were not promulgated in accordance with the Act's procedural requisites. We find the first premise too sweeping to be sustained in full and the second premise too flawed to accept at all. Consequently, we invalidate certain portions of EPA's 1982 revocation. Our holding on these issues makes it unnecessary to decide whether, and in what circumstances, the Act authorizes EPA to stay existing regulations without notice and comment.

A. Whether the Clean Air Act Grants EPA Power to Attribute to Marine Terminals Any of the Emissions Produced by Marine Vessels 1. Standard of Review

The parties sharply contest the standard of review we are to apply to determine whether EPA's abnegation of all power to reach vessel emissions is "not in accordance with law." 42 U.S.C. § 7607(d)(9)(A) (Supp. V 1981). One reason for this dispute is that the case law under the Administrative Procedure Act, 5 U.S.C. § 706(2)(A) (1982), has not crystallized around a single doctrinal formulation which captures the extent to which courts should defer to agency interpretations of law. Instead, two "opposing platitudes" exert countervailing "gravitational pulls" on the law. Black Citizens for a Fair Media v. FCC, 719 F.2d 407, 423 (D.C. Cir. 1983) (Wright, J., dissenting). At one pole stands the maxim that courts should defer to "reasonable" agency interpretive positions, see Udall v. Tallman, 380 U.S. 1, 16, 85 S.Ct. 792, 801, 13 L.Ed.2d 616 (1965), a maxim increasingly prevalent in recent decisions. See, e.g., Public Service Comm'n v. Mid-Louisiana Gas Co., ___ U.S. ___, 103 S.Ct. 3024, 3035, 77 L.Ed.2d 668 (1983). Pulling in the other direction is the principle that courts remain the final arbiters of statutory meaning, see FTC v. Colgate-Palmolive Co., 380 U.S. 374, 385, 85 S.Ct. 1035, 1042, 13 L.Ed.2d 904 (1965); that principle, too, is embossed with recent approval. See, e.g., FEC v. Democratic Senatorial Campaign Comm., 454 U.S. 27, 32, 102 S.Ct. 38, 42, 70 L.Ed.2d 23 (1981).

A second reason for the vigor with which the parties contest the standard of review is that the agency interpretation under attack constitutes a complete reversal of the position the agency took only two years ago. Here, too, the case law contains conflicting strands. There can be little doubt that a "`settled course of behavior embodies the agency's informed judgment that, by pursuing that course, it will carry out the policies committed to it by Congress. There is, then, at least a presumption that those policies will be carried out best if the settled rule is adhered to.'" Motor Vehicle Manufacturers Ass'n v. State Farm Mutual Automobile Insurance Co., ___ U.S. ___, 103 S.Ct. 2856, 2866, 77 L.Ed.2d 443 (1983) ( quoting Atchison, T. S.F.R.R. v. Wichita Bd. of Trade, 412 U.S. 800, 807-08, 93 S.Ct. 2367, 2374-75, 37 L.Ed.2d 350 (1973)). At the same time, a change in administrative construction which is "thought necessary to match the statute's construction to the original congressional intent" remains entitled, under some circumstances, to judicial deference. See National Muffler Dealers Association v. United States, 440 U.S. 472, 485, 99 S.Ct. 1304, 1311, 59 L.Ed.2d 519 (1979).

From these discordant themes petitioners emerge with the conclusion that EPA's interpretation warrants no deference at all, while the government concludes that "considerable" deference is owing. We do not find it necessary in this case to resolve this dispute nor to attempt to harmonize the various precedents. We would affirm the agency's conclusion that vessels are "mobile sources" for purposes of the ban on indirect source review even if we were to adopt the least deferential standard of review. See Section 2 at pages 768-771, infra. Similarly, we would sustain the agency's revocation of the secondary emission component of the vessel emission requirement under that same standard. And our invalidation of EPA's treatment of dockside emissions, see Section 3(a) at pages 771-772, infra, is not based upon our assessment of the accuracy of the result reached by the agency, but rather upon the agency's complete failure to consider the criteria that should inform that result; as a consequence, whatever deference might be owing to the agency's conclusions under other circumstances, on this issue none at all is warranted.

2. Whether Ships are "Mobile Sources" of Pollution for Purposes of the Clean Air Act

We must begin, as always, with the language of the statute. See International Brotherhood of Teamsters v. Daniel, 439 U.S. 551, 558, 99 S.Ct. 790, 795, 58 L.Ed.2d 808 (1979). The statutory guidance afforded us here to determine whether vessels docked at marine terminals or coming to and from such terminals are "mobile" sources of pollution is contained in the following sentence from the 1977 ban on EPA regulation of indirect pollution sources: "the term `indirect source' means a facility, building, structure, installation, real property, road, or highway which attracts, or may attract, mobile sources of pollution." 42 U.S.C. § 7410(a)(5)(C) (Supp. V 1981).

Nowhere in the statute is the term "mobile sources" defined. In ordinary usage, however, that term would seem to apply to any source that emits pollution and is capable of moving — which would, of course, include marine vessels. That fact is of some significance in our attempt to determine whether Congress intended marine vessels to be treated as "mobile sources" of pollution; "[a]fter all, legislation when not expressed in technical terms is addressed to the common run of men and is therefore to be understood according to the sense of the thing, as the ordinary man has a right to rely on ordinary words addressed to him." Addison v. Holly Hill Co., 322 U.S. 607, 618, 64 S.Ct. 1215, 1221, 88 L.Ed. 1488 (1944) (Frankfurter, J.).

We do not, however, treat that fact as dispositive, for inquiry into statutory meaning cannot cease at the bounds of a statute's "plain" words. As Justice Frankfurter also realized, "the notion that because the words of a statute are plain, its meaning is also plain, is merely pernicious oversimplification." United States v. Monia, 317 U.S. 424, 431, 63 S.Ct. 409, 412, 87 L.Ed. 376 (1943) (Frankfurter, J., dissenting) (quoted with approval in FBI v. Abramson, 456 U.S. 615, 625 n. 7, 102 S.Ct. 2054, 2061 n. 7, 72 L.Ed.2d 376 (1982)). Only after examination of the context in which statutory words are set — the statute's purpose, structure, and history — is it possible to determine reliably whether Congress used a word as a technical term of art or instead intended that word to bear merely its "plain" meaning. See, e.g., R. DICKERSON, THE INTERPRETATION AND APPLICATION OF STATUTES 230 (1975) ("[T]o exclude consideration of context would be to ignore one of the most basic principles of communication."). Absent consideration of context, the plain meaning rule becomes merely a rorschach onto which judges cast their own conflicting visions of what statutory words ought to mean.

We turn first, then, to the legislative history and administrative background of the 1977 ban on indirect source review. In this case, however, those contextual signposts turn out to be of little aid in directing us to a resolution of the issue before us. All that can be gleaned from that history and background is that, in 1977, Congress simply did not consider the specific problem of marine terminals and vessels.

The EPA regulation of indirect sources, which began in 1973 after this court's decision in NRDC I, 475 F.2d 968, was expressly directed at "major highways and airports, large regional shopping centers, major municipal sports complexes or stadiums, major parking facilities, and large amusement and recreational facilities." 38 Fed.Reg. 15,837 (1973). Though the agency stated that this list of indirect sources was not necessarily exhaustive, at no time was there any mention in the agency's proposal of marine terminals or of vessel emissions. In contrast, the agency's proposals for indirect sources such as parking lots, highways, and airports were elaborately detailed: pre-construction review, for example, was required for any new parking facility in an urban area, or "associated parking" in such an area, that had a capacity of 1,000 or more cars, while the similar figure for non-urban areas was 2,000 cars; review was also required for new highway projects with an anticipated daily traffic volume of 20,000 or more vehicles within ten years. 36 Fed.Reg. 960 (1973).

When Congress reacted vigorously to EPA's actions, it was thus a specific regulatory-lightning rod — one which did not include marine terminals or vessels — that drew the congressional response. See, e.g., H.R. Rep. No. 95-294, 95th Cong., 1st Sess. 222 (1977) ("[T]he Committee is especially cognizant of the potentially sweeping consequences and potentially socially and economically disruptive impacts which may result from efforts to reduce automobile pollution through mandated restrictions in parking supplies and restrictions on new parking facilities."). As a result, it is not surprising that throughout the debate on indirect source review — stimulated by EPA's attempt to require airports, highways, and parking lots to obtain federally-controlled permits before construction or significant modification — no mention was made of marine terminals or vessels. EPA did not attempt to regulate marine terminals; Congress did not respond to EPA's inaction.

Petitioners appeal to another, more tangible, part of the 1977 legislative history — a provision in the House bill defining the term "mobile source-related air pollutant" as any air pollutant "subject to regulation under [Title II] of this Act." H.R. 6161, 95th Cong., 1st Sess. § 201(e) (1977), reprinted in 4 Legislative History of the Clean Air Act Amendments of 1977, at 2107 (1978) [hereinafter cited as "Leg.Hist."]. The only sources of pollution explicitly subject to regulation under Title II are motor vehicles and airplanes. Arguably, as petitioners insist and respondents concede, see Brief for Respondent at 21, this definition therefore identified not only the nature of the pollutants in question but also their sources; that is, mobile-sources related pollutants were those produced only by motor vehicles or aircraft. Since marine vessels are not among the sources of "mobile source-related pollutants" covered by Title II, petitioner's syllogism concludes that the House did not consider vessels to be "mobile sources" of pollution.

Even if this interpretation of the House bill is correct, a point upon which we have some doubt, it does not follow that the term "mobile sources" as it appears in the enacted legislation remained confined to automobiles and airplanes. The Conference Committee, while accepting the essence of the House bill, deleted without explanation the term "mobile source-related pollutant" and substituted, again without explanation, the term "mobile source." Compare H.R. Rep. No. 95-294, 95th Cong., 1st Sess. 437 (1977), 4 Leg.Hist. at 2904, with H.R. Rep. No. 95-564, 95th Cong., 1st Sess. 93 (1977), U.S. Code Cong. Admin.News 1977, 1077, 3 Leg.Hist. at 473. Whether that substitution preserved the underlying concept of the House bill with regard to which sources were "mobile," or whether the change instead expanded from the House bill the class of sources that were "mobile," is a choice we cannot make in the absence of some basis for understanding the reasons animating that change. Because the Conference Report does not even allude to this change, and because we can discern no clear explanation for it in the history, background, or logic of the indirect source review ban, we refuse to read any dispositive significance into the initial House definition of "mobile source-related pollutant." Thus, we can draw no conclusion about the etiology of the term "mobile sources."

As a result, all we can glean from the 1977 Act's legislative history is that Congress simply had no specific intent as to whether vessel emissions were to be treated like auto emissions. That Congress has not specifically considered a controversial issue does not, however, mean that courts are rendered impotent to determine whether one resolution or another of that issue best comports with the statutory scheme as a whole. "[S]ilence . . . is no reason for limiting the reach of federal law . . . the inevitable incompleteness presented by all legislation means that interstitial federal law-making is a basic responsibility of the federal courts." United States v. Little Lake Misere Land Co., 412 U.S. 580, 593, 93 S.Ct. 2389, 2397, 37 L.Ed.2d 187 (1973). The limitations on Congress' ability to foresee or consider particular problems that may in the future arise under a statutory scheme often require that legislation be drafted by reference to general categories rather than to specific classes of activities within those categories.

That requirement, in turn, demands that courts seek out the broader purposes — the overriding statutory goals — constitutive of the general categorical term in which Congress has embodied its will. Only by engaging in such purposive interpretation can we accord fidelity to Congress' objectives. See, e.g., Moragne v. States Marine Lines, Inc., 398 U.S. 375, 392, 90 S.Ct. 1772, 1783, 26 L.Ed.2d 339 (1970) ("[A] statute may reflect nothing more than the dimensions of the particular problem that came to the attention of the legislature, inviting the conclusion that the legislative policy is equally applicable to other situations in which the mischief is identical."); see generally Fuller, Positivism and Fidelity to Law, 71 HARV L.REV. 630, 661-69 (1958).

In this case, the salient congressional objection to indirect source review was the inequity and technological irrationality of regulating indirectly various sources of pollution:

Considerations of equity and common sense demand that primary emphasis be placed on the prevention of pollution by reducing the emissions from each and every new automobile tail pipe. Efforts based on indirect control of the use of automobiles through restrictions on parking lots, shopping centers, and other indirect sources, rather than full and prompt controls for new autos, trucks, buses, and motorcycles are inherently inequitable.

H.R. Rep. No. 95-294, 95th Cong., 1st Sess. 221 (1977), U.S. Code Cong. Admin.News 1977, at 1300. Two principal reasons underlay this objection. First, the House believed "that the primary emphasis of any air pollution control program must be on the reduction of air pollution — including that generated by automobiles — at the source." Id. at 221, U.S. Code Cong. Admin.News 1977, at 1300. Building technological controls into an indirect source, rather than into the polluting source itself, was considered inefficient. Second, it was considered inequitable to impose upon owners of the indirect source facilities which EPA sought to regulate — parking lots, shopping centers, airports — the cost of cleaning up someone else's emissions.

Both of these purposes are equally applicable to the relationship between ships and harbors. It would be just as unfair, and would make as little technological sense, to require marine terminal owners to clean up vessel emissions as Congress thought it was in 1977 to require parking lot owners to bear the regulatory brunt of automobile emissions. As a result, we conclude that vessels are "mobile sources" of pollution for purposes of the 1977 ban on indirect source review. On this point, we accept EPA's 1982 interpretation of the Act.

3. The Treatment of Mobile Sources Under the Act a). Dockside Emissions

From the fact that vessels are mobile sources of pollution, EPA leapt to the conclusion that all emissions in harbor areas which were traceable in some way to the activities of such vessels could not be regulated. The fact that vessels are "mobile sources" of pollution, however, should have been merely the starting point for EPA's inquiry — not, as the agency made it, the end point. To maintain, as EPA now does, that vessel emissions can never be considered part of a terminal's pollution — even if the emissions are of a type produced only at dockside, where the vessel is associated with or connected to terminal facilities, and if the means to control the emissions involve the application of the most advanced land-based technology — is to sweep too broadly in an area where precision is the requisite skill. For it is assuredly not the case that the ban on indirect source review was intended to go so far as to prohibit the attribution to a stationary source of all emissions which happen to emanate from or even merely physically contact a mobile source. Indeed, the statute itself provides that "[d]irect emissions sources or facilities at, within, or associated with, any indirect source shall not be deemed indirect sources for the purpose of this [ban on indirect source review]." 42 U.S.C. § 7410(a)(5)(C) (Supp. V 1981). The agency should therefore have examined the nature of specific interactions between a mobile source and an attendant stationary source to determine which categories of emissions can (as a matter of statutory authority) and should (as a matter of policy) be attributed to the stationary source. Accordingly, we vacate EPA's blanket repeal of the dockside component to the vessel emission requirements.

On remand, the agency must engage in the following analysis to bring its rules into compliance with the statutory mandate. First, the agency must apply its "control and proximity" regulations to define which emissions from dockside activities of marine vessels are "stationary source" emissions of the marine terminal. Second, the agency must develop attribution rules to determine which, if any, of the emissions defined above cannot be assigned to the terminal because to do so would be to violate the ban on indirect source review. In ordering this second step, we do not suggest that the agency must conclude that some dockside vessel emissions which are under the control of the terminal owner and which directly serve his purposes, see 45 Fed.Reg. 52,696 (1980), cannot be assigned to the terminal owner; it may be open to the agency to conclude that its control and proximity regulations already adequately divide responsibility for dockside vessel emissions between the ship and the terminal. We therefore express no opinion on whether the first and second steps must yield distinct conclusions, or if they do, what legal test ought to govern at the second stage.

In rejecting the agency's blunderbuss approach, we consider it highly relevant that the more finely calibrated approach we order EPA to apply on remand is precisely the approach EPA has taken in recent regulations that establish the new source performance standards for bulk gasoline terminals. 48 Fed.Reg. 37, 578 (Aug. 18, 1983). Those regulations include in the emissions of such terminals the vapors displaced from a tank truck when liquid gasoline is pumped from the terminal into the tank. There is no question that such trucks, whether moving or stationary, are mobile sources under the 1977 ban, nor that, were these same vapors emitted at a parking lot, EPA could not ascribe the emissions to the parking lot. Nonetheless, by requiring terminal owners to restrict loadings to trucks equipped with vapor-recovery systems, these regulations recognize that emissions from mobile sources can, in some contexts, be attributed to the stationary source. EPA's supporting rationale for the bulk gasoline terminal regulations nicely captures our reasons for remanding in this case: "Since escape of these vapors is caused by stationary activities [of a mobile source] at a stationary facility, they are `stationary source' emissions subject to regulation under Section 111 — even though the tank trucks from which they escape during that activity have the capability to move." Id. at 37,586-87. Cf. Brief for Petitioner Illinois Environmental Protection Agency 31 ("If both vessels and trucks are mobile sources, and if both engage in loading and unloading activities at each respective's terminal, it is irrational to claim that the emissions from one amount to direct sources emissions controllable under Section 110(a)(5)(C) and the emissions from the other do not."). We of course express no opinion on the validity of the bulk gasoline terminal regulations, nor even suggest that some vessel emissions are necessarily attributable to the marine terminals — we cannot know, on the basis of the record before us, enough about the nature of those emissions to impose our original conception upon the agency. We reverse and remand for EPA's inexplicable failure to consider the point. What EPA must do on remand is apply the same general approach employed in the bulk gasoline terminal regulations to vessel emissions to determine which, if any, of those emissions can and should be assigned to marine terminals.

In vacating EPA's rescission of the dockside component of the 1980 vessel emission requirements, we are not unmindful that our opinion today casts a cloud over the very regulations it implicitly reinstates. Although the validity of the 1980 regulations is, not properly before the court, we note that there, too, the agency failed to engage in the careful examination of the specific nature of dockside vessel emissions which we mandate in this opinion. We thus cannot know whether EPA would have reached the same conclusion in 1980 had it not relied in part in 1980 on the erroneous premise that vessels are not mobile sources. We therefore anticipate that the agency will expeditiously apply itself to the completion of the task left unfinished by both the 1980 and 1982 vessel emission rules. As things now stand, the agency cannot approve any SIP provision regarding marine terminal review on the basis of the 1982 dockside emission rule. To apply our decision today, the agency before approving such a SIP must examine, either through new rulemaking or adjudication, attribution factors similar to those used in the bulk gasoline terminal regulations to determine which dockside vessel emissions to attribute to the terminal.

b). To-and-Fro Emissions

We affirm that portion of the 1982 repeal which rescinds the "to-and-fro" component of the vessel emission requirements. The original regulations treated emissions produced by a ship moving to or from a marine terminal as "secondary emissions" of the terminal itself. We need not decide whether the Act itself bars EPA from doing so, for since the agency's development of the secondary emission concept, the agency itself has refused to include in secondary emissions those emissions produced by a mobile source as it approached and left a stationary facility. See 41 Fed.Reg. 55,528 n. 3 (1976) (exempting auto emissions from secondary emissions of "parking-type facilities"). Nor do the recently promulgated bulk gasoline terminal regulations attempt to use the secondary emission concept to make terminal owners responsible for the emissions of tank trucks as the latter come to or leave such terminals. It would be arbitrary for EPA to treat car and truck emissions in this way while treating "to-and-fro" vessel emissions as secondary emissions of a marine terminal. The only explanation EPA offered in 1980 for such a distinction is that vessels are not "mobile sources" of pollution. We have already held to the contrary. As a result, the 1980 regulations arbitrarily treated vessel emissions as secondary emissions of marine terminals when similar treatment was not accorded auto emissions in parking lots. Rescission of the secondary emissions component of those regulations was therefore valid.

B. Promulgation of the Vessel Emissions Requirements

The second premise on which EPA initially based its stay and revocation of the vessel emission requirements was that the notice it gave when the requirements were promulgated in 1980 was deficient. See 47 Fed.Reg. 27,554, 27,558-560 (1982). Although the agency's current position on this issue is not completely clear, it appears from the briefs that, at least on appeal, the agency now justifies its revocation primarily on the basis of the statutory interpretation arguments discussed above. Intervenor GATX, however, continues to press deficiency of the notice as an independent basis for sustaining EPA's revocation of the vessel emission requirements. GATX argues that EPA failed to give advance notice that it was going to take into account the emissions from a docked vessel when determining whether a new or modified marine terminal is a major source of pollution. We must therefore address this claim.

We begin by noting that, under the Clean Air Act, the standard of review and remedial discretion of a court for alleged procedural errors by the agency is rather limited. A court can vacate a rule for procedural irregularity only if: 1) the alleged procedural error was arbitrary and capricious, 2) an objection was timely raised, and 3) "the [alleged] 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." 42 U.S.C. § 7607(d)(8) (Supp. V 1981). Notice is sufficient, and thus the error is not arbitrary and capricious, if the final rule is a "logical outgrowth" of the proposed rule, Sierra Club v. Costle, 657 F.2d 298, 352 (D.C. Cir. 1981), and if "interested persons were sufficiently alerted to likely alternatives to have known what was at stake," South Terminal Corp. v. EPA, 504 F.2d 646, 659 (1st Cir. 1974). In this case, it appears that GATX was "sufficiently alerted" and that the final rule was a "logical outgrowth" of the proposed rule, but in any event, it is clear that EPA's decision to issue the vessel emissions requirements would not have been changed even had the alleged error not been made.

In August 1979, EPA stated to this court that emissions from vessels going to, staying at, and coming from a marine terminal would be classified as secondary emissions and that secondary emissions would not be counted in determining whether a terminal was a major source of pollution. Just one month later on September 5, 1979, however, EPA promulgated a notice of proposed rule-making which included only the emissions of vessels going to and coming from a terminal in the definition of secondary emissions. 44 Fed.Reg. 51,949 (1979). EPA's failure to include the emissions from vessels staying at a terminal left open the possibility that the agency might include those emissions in the emissions of the terminal. Moreover, that same proposal also announced that EPA intended to define "stationary source" in terms of the pollutant-emitting activities on contiguous or adjacent properties "owned or operated" by the same person. Id. at 51,948, 51,952. Although at this time the definition of stationary source did not specifically define those activities under "control" of a single person to constitute a single source, GATX understood it to do so impliedly and commented accordingly. See PSD Rulemaking Docket, No. A-79-35, Doc. No. III-B-118 at 4-5 (written comments of GATX).

The role that control was to play as a factor in determining which emissions should be attributed to a stationary source became more clear on January 30, 1980, when EPA published a supplement to the September 5 notice. See 45 Fed.Reg. 6,802 (1980). Issued in response to this court's order that EPA give notice as to whether the term "source" would "include the unloading of vessels at marine terminals," see Alabama Power Co. v. Costle, 636 F.2d at 397, the supplement explained how EPA would apply the September 5th proposed definition of stationary source:

Obviously, that definition, like the definition of the term in the existing regulations, would make proximity and control the key factors in deciding whether to combine a particular activity with other ones for applicability purposes. EPA intends to use those factors in a reasonable way. For example, it does not intend to treat a pipeline that stretches across many states as a single "stationary source." The agency solicits comment on the use of proximity and control as the sole factors in grouping pollutant-emitting activities. In particular, it asks for comment on whether additional factors, such as the functional relationship of one activity to another, should be used. The agency also asks for concrete and specific examples of cases where a literal application of the proposed definition would be unreasonable.

45 Fed.Reg. 6,803 (1980) (emphasis added). Even though this notice did not specifically mention vessel emissions, it is reasonably clear that GATX understood its implications. Eight days after this notice was promulgated, GATX's attorney wrote EPA's General Counsel, stating that "it seems appropriate at this time to open communication and conversations with you and with other relevant parties within EPA to consider the agency's position concerning the issue of vessel emission attribution to marine terminals." Addendum to Petitioner NRDC's Reply Brief at 1. Thus, it is clear that GATX understood the implications of this notice.

Moreover, even if GATX had not understood the implication of this notice, there still would not be a "substantial likelihood" that EPA's decision would have been changed had sufficient notice in fact been given. The issues with which GATX was concerned — whether vessels were controlled by the terminal and whether EPA had the statutory authority to regulate dockside vessel emissions — were specifically considered by EPA in 1980. See 45 Fed.Reg. 52,676, 52,696 (1980). After discussing the issues, EPA squarely rejected the arguments GATX makes today and claims it would have made then. Id. Thus, EPA cannot justify staying and then revoking the vessel emissions requirements on the basis of the argument that the 1980 notice was deficient.

C. Additional Challenges

Petitioners also challenge on substantive and procedural grounds both the temporary and the extended stays of the vessel emission requirements. These challenges raise a number of complex questions regarding the scope of EPA's power to issue stays under section 307(d)(7)(B) of the Clean Air Act, 42 U.S.C. § 7607(d)(7)(B) (Supp. V 1981), the interplay of that section with the requirements of the Administrative Procedure Act, and the extent to which agencies possess general equitable power to stay for extended periods validly promulgated regulations. Moreover, the fact that we today approve the substantive basis on which the secondary emission rules were stayed does not validate the stays themselves; the question remains whether EPA may stay a legally effective rule before the agency validly rescinds that rule, and if so, under what circumstances, through what procedures, and for how long. Similarly, that the 1982 dockside emission rules are substantively invalid does not necessarily mean that the accompanying stays were invalid ab initio.

We find it unnecessary to address these difficult issues today, however, for at this time no party has alleged sufficient injury from either of the stays to guarantee that an actual controversy exists with regard to them. As of now, we have not been presented with a concrete case of a marine terminal that, on the basis of the stays, ignored dockside emissions of marine vessels, or the emissions of vessels coming to and from the terminal, between July 7, 1981, the date of the first stay, and June 25, 1982, the date repeal became legally effective. Should such a case later be brought before the court, the parties, or others, remain free to challenge the validity of the stays.

Finally, when EPA first stayed the vessel emission requirements, EPA also stayed regulations requiring that any voluntarily adopted emissions limitations be federally enforceable before a new or modified source could rely on them as a means of meeting the standards set forth in the Act. The stay of these federally enforceability requirements is also challenged in this case. Because those requirements are back in effect today, and no injury has been alleged as a result of the suspension of those requirements, we find this issue moot.

CONCLUSION

That EPA's statutory authority is broader than that which the agency now asserts does not preclude EPA from revisiting the policy choices to be made in determining how best to carry out its mission under the Act. Whether vessel emissions at dockside should be controlled, and whether a failure to control them would be arbitrary on the merits, are questions not now before the court. We hold only that, as a matter of law, EPA is not barred from engaging in such a policy analysis. Accordingly, we vacate that portion of EPA's revocation which "excepts the activities of any vessel" from the emissions attributable to marine terminals, see, e.g., 40 C.F.R. § 51.24(b)(6) (1983), 47 Fed.Reg. 27,560 (1982). At the same time, we uphold that part of the agency's decision that refuses to include vessel emissions in the secondary emissions of a marine terminal. See, e.g., 40 C.F.R. § 51.24(b)(18) (1983), 47 Fed.Reg. 27,560 (1982). The case is remanded to the EPA for further proceedings consistent with this opinion.

It is so ordered.