The government asks us to set aside a preliminary injunction stopping it from auctioning rights to drill for oil on 488 tracts near Georges Bank, a fishing area in the North Atlantic off the New England Coast. The district court, 560 F. Supp. 561, acting at the request of the Commonwealth of Massachusetts and the Conservation Law Foundation, issued the injunction (after a brief hearing) on March 28, 1983--sixteen days after the Commonwealth and the Conservation Law Foundation asked for a preliminary injunction and one day before the scheduled oil lease sale was to take place. The court based its order on four separate grounds, namely that the proposed sale was likely to violate four separate statutes: the National Environmental Policy Act, 42 U.S.C. Secs. 4321 et seq., the Endangered Species Act, 16 U.S.C. Secs. 1531 et seq., the Coastal Zone Management Act, 16 U.S.C. Secs. 1451 et seq., and the Outer Continental Shelf Lands Act, 43 U.S.C. Secs. 1331 et seq., 1801 et seq. Appellant challenging the issuance of a preliminary injunction must carry the heavy burden of proving that the district court abused its discretion. National Tank Truck Carriers, Inc. v. Burke, 608 F.2d 819, 823 (1st Cir.1979). On this appeal, we conclude that the district court was correct at least as to its finding of a likely violation of NEPA. Because the likelihood of a violation of NEPA, in the circumstances of this case, is sufficient to support a preliminary injunction, we need not consider the other alleged statutory violations. We affirm the district court's order.

* NEPA requires that the Department of the Interior, before auctioning its oil leases, prepare a statement that describes "the environmental impact of the proposed action, ... any adverse environmental effects which cannot be avoided ..., [and] alternatives to the proposed action ...." 42 U.S.C. Sec. 4332(C). Further regulations, binding throughout the Executive Branch, Andrus v. Sierra Club, 442 U.S. 347, 358, 99 S. Ct. 2335, 2341, 60 L. Ed. 2d 943 (1979), require the Department to prepare a supplement to this Environmental Impact Statement (EIS) if there are "significant new circumstances or information relevant to environmental concerns and bearing on the proposed action or its impacts." 40 C.F.R. Sec. 1502.9(c). In this case, after the Department prepared its Final Environmental Impact Statement (FEIS), it radically revised its estimates of oil likely to be found on the tracts it intended to lease. It lowered its mean estimate from 1.73 billion barrels to 55.7 million barrels--a 97% reduction. The NEPA issue is whether the fact that the Department now believes there is only 1/31 as much oil to be found off Georges Bank is a "significant new circumstance" sufficient to require an EIS supplement.

We judge the lawfulness of the Department's decision not to supplement the EIS by asking whether that decision was reasonable under the circumstances. California v. Watt, 683 F.2d 1253, 1267 (9th Cir.1982) ("[a]n agency's decision not to supplement an EIS will be upheld if it was reasonable"), cert. granted on other grounds, --- U.S. ----, 103 S. Ct. 2083, 77 L. Ed. 2d 295 (1983); cf., e.g., Sierra Club v. Froehlke, 534 F.2d 1289, 1299 (8th Cir.1976) ("courts must employ a rule of reason in the examination for adequacy of these statements, lest the litigation have no end"); Scientists' Institute for Public Information, Inc. v. Atomic Energy Commission, 481 F.2d 1079, 1092 (D.C.Cir.1973) (rule of reason applies); Village of False Pass v. Watt, 565 F. Supp. 1123 (D.Alaska, 1983) (same). Having read the approximately five-hundred-page FEIS and the various decision documents placed before the Secretary, we conclude that the Department of Interior's decision not to supplement was not reasonable. Without a supplement, the FEIS did not describe the likely environmental harms well enough to allow the Secretary to make an informed decision. See, e.g., Vermont Yankee Nuclear Power Corp. v. Natural Resources Defense Council, 435 U.S. 519, 558, 98 S. Ct. 1197, 1219, 55 L. Ed. 2d 460 (1978); Alaska v. Andrus, 580 F.2d 465, 474-75 (D.C.Cir.1978); County of Suffolk v. Secretary of Interior, 562 F.2d 1368, 1375 (2d Cir.1977); Silva v. Lynn, 482 F.2d 1282, 1284-85 (1st Cir.1973); Calvert Cliffs' Coordinating Committee, Inc. v. Atomic Energy Commission, 449 F.2d 1109, 1114 (D.C.Cir.1971).

Our reading of the FEIS shows us that it was written with the 1.73 billion barrel mean in mind, and suggests that an FEIS written with a 56 million barrel mean expectation would be a very different document. We can best demonstrate the inadequacy of the FEIS--showing why, and how, expecting 31 times as much oil significantly affected the description of environmental consequences--by choosing as an example a key decision that the Secretary had to make: the decision whether or not to accept the Conservation Law Foundation's preferred choice, "Alternative Five." That alternative would have removed 207 blocks from the sale and left 333 for leasing. Did the FEIS adequately inform the Secretary about the adverse environmental consequences of not adopting this alternative?

The Secretary had before him a Secretarial Issue Document (SID) that framed the relevant issues for his choice. The SID was based upon the new (97% reduced) oil recovery estimates. It stated that by rejecting Alternative Five--by leasing 540 blocks instead of 333--the Secretary would create only about $35 million in additional oil benefits, instead of the hundreds of millions of dollars previously thought possible. Had the Secretary simply balanced this $35 million benefit against the serious environmental harms described in the FEIS, he might well have chosen Alternative Five, and taken 207 tracts out of the sale.

The SID, however, goes on to tell him not to balance the "FEIS described" harms against this new benefit. It correctly tells him (in two sentences) that the new resource recovery estimates imply that the likely environmental harms have changed too. It states:

Since the revised resource estimates for this option are much lower than those assumed for the basis of the FEIS, a corresponding smaller number of platforms, wells, and pipelines needed to explore for and develop these resources should be assumed. This would significantly lower the expected impacts.

Although the SID thereby tells the Secretary that the likely environmental harm is less than described in the FEIS, it does not tell him by how much the previous FEIS estimate of likely environmental harm should be reduced, (nor does it purport to do so).

The Secretary, alerted by the SID's reference to a "corresponding smaller" amount of drilling and "significantly lower" impacts, might have looked to another document before him--an Environmental Assessment (EA)--for further explanation. But the EA contains only a very limited amount of further information. The EA states, as to Alternative Five:

The revised oil and gas resource estimates for Georges Bank mean fewer production platforms but ... approximately the same exploration and delineation structures.... [R]eductions would occur in estimated volumes of drill muds and cuttings, chronic and acute discharges, air emissions, number of spills and number of pipelines needed. Reduced activity results in a reduction of estimated impacts as described in the FEIS (see Table II.D.1).... While estimated impacts on all resources are reduced, impact severity for fisheries and biological resources are only slightly lower.

The reference to a table is presumably a reference to Table II.B.10 of the FEIS, for we could not find a Table II.D.1. Table II.B.10 compares the environmental impacts of the various alternatives in fifteen different categories, ranking each alternative in each category on a scale of "0" ("none") to "3" ("major"). The FEIS states that leasing 540 blocks, instead of Alternative Five's 333 blocks, would increase the severity of likely environmental harms from "minor" to "moderate" or "moderate" to "major" in seven of the fifteen environmental categories. The EA provides a revised version of this table in which all harms diminish across the board, and the relative harms of the full sale compared with Alternative Five increase from "minor" to "moderate" in four categories. In other words, the EA tells the Secretary more about harms than the SID, but not very much more about the specifics.

The Secretary might also have looked to the FEIS itself for further information on the environmental consequences of the reduced oil estimates, for the FEIS contains a one-page "Addendum" that mentions the changed oil recovery estimates. The Addendum, however, simply states that there will probably be "fewer oil spills" and "fewer wells drilled;" it goes no further. These words might be sufficient if it were obvious that the likely environmental harm would decline proportionately with the expected oil recovery, but the FEIS, EA, and SID make clear that the anticipated decline will not be proportional.

In fact, after reading the FEIS, the SID, and the EA, the Secretary would have no clear idea how to visualize the environmental harms under the new mean estimates. The entire FEIS is couched in terms of the old mean; the new documents merely say that the harm is "not as great." But the crucial question for a rational decisionmaker is by how much the likely environmental harm will be reduced. And here the documents offer contradictory signals. On the one hand, they suggest that estimates of some environmental harms do decline proportionately as the estimates of oil recovery decline. They state, for example, that because of the new oil estimate, the projected number of major oil spills falls from 10 in 30 years to 0.3 in 30 years. On the other hand, the documents strongly suggest that the reduction in many other harms is not proportional. A comparison of Tables II.B.10 in the FEIS and in the EA, for example, reveals that for eight categories of environmental harm the new oil recovery estimate reduces likely environmental harm by only half a step (from, say, "2" to "1-2"), for five categories it reduces harm by a full step, and for two categories it leaves the estimate of harm unchanged. Although resource recovery estimates have changed from "lots" to "very little," the chart suggests that harms have not diminished correspondingly. Similarly, analysis of the FEIS suggests that the likelihood of fewer oil spills does not necessarily reduce environmental risks proportionately. The harm that an oil spill causes depends in part on where the oil is spilled; the FEIS makes clear, for example, that some oil spills are more likely to harm coastal areas than others. Yet, nothing in the FEIS, the SID, or the EA suggests that the "fewer wells" now contemplated will be dispersed in a geographical pattern similar to that thought likely when the FEIS was written.

In sum, the SID, the EA, and the FEIS addendum state that the likely environmental harms associated with rejecting Alternative Five have changed. The SID makes clear that it is important to know the probable magnitude of this change. The documents indicate that this magnitude is considerable; they point out that there is no obvious way in which a decision maker could readily calculate this magnitude (by, for example, applying a rule of proportionality); and, they suggest that considerably more analysis can fairly easily be done. Under these circumstances, we do not see how the FEIS can be said to have given the Secretary a reasonably adequate picture of the likely environmental harms associated with his choice. And, the Department of Interior's decision not to supplement the FEIS therefore seems unreasonable.

We have considered several arguments to the contrary. First, the government claims that the change in oil recovery estimates is not important because the estimates were not very reliable in the first place; even for purposes of the FEIS the estimated recovery was expressed in terms of a range, which ran from a low expectation of 17 million barrels to a high expectation of 6.35 billion. And, the new mean of 55.7 million barrels falls within this range.

This argument is not convincing, however, for the FEIS primarily focuses on the mean estimate of 1.73 billion barrels. The FEIS was written and its alternatives were developed with this "mean" in mind. Of the several hundred pages in the FEIS and the numerous charts and tables, only six pages that we could find focus specifically upon the environmental impact under the low, or the high, range estimates.

Second, the government cites California v. Watt, 683 F.2d 1253 (9th Cir.1982), cert. granted on other grounds, --- U.S. ----, 103 S. Ct. 2083, 77 L. Ed. 2d 295 (1983), to support its claim that the changed circumstances do not warrant an EIS supplement. In California v. Watt the Ninth Circuit held that a new oil recovery estimate did not require a supplement. But there the resource estimate doubled, while there was no reason to believe that environmental harms could increase more than proportionately. Here, the oil estimate has changed by a multiple of 31, while (as discussed above) there is every reason to think that the likely change in environmental harm is less than proportional.

Watt applied a standard of "reasonableness" to the agency's decision not to supplement, and the court considered

the environmental significance of the new information, the probable accuracy of the information, the degree of care with which the agency considered the information and evaluated its impact, and the degree to which the agency supported its decision not to supplement with a statement of explanation or additional data.

683 F.2d at 1267-68. Here these factors cut in favor of requiring a supplement. As previously explained, a supplement would describe the magnitude of the change in likely environmental harms. To what extent will there be less drilling or fewer platforms? Will some tracts be affected more severely than others? Where are oil spills now likely to take place? The new oil recovery estimates are based on drilling (in previously-leased nearby areas) that produced three dry holes, and more recent drilling apparently has produced five more dry holes. It seems likely that the drilling produced new information about the area, yet none of the documents discuss this. At the same time, the EA's only discussion of the decision not to supplement the FEIS consists of the sentence: "Based on the analysis done for this assessment, it has been determined that a supplemental EIS is not necessary."

Third, it was argued that the EA itself (or the SID and EA together) constitutes an adequate supplement. We have previously explained why we think that the EA did not contain enough information or analysis to inform the Secretary of the magnitude of the change in environmental harm. But, even if we are wrong, the EA and SID are legally insufficient as FEIS supplements, for they were not made public until the beginning of this litigation. As we have previously held, unless a document has been publicly circulated and available for public comment, it does not satisfy NEPA's EIS requirements. Grazing Fields Farm v. Goldschmidt, 626 F.2d 1068, 1073-74 (1st Cir.1980); see I-291 Why? Association v. Burns, 517 F.2d 1077, 1081 (2d Cir.1975); Appalachian Mountain Club v. Brinegar, 394 F. Supp. 105, 122 (D.Mass.1975); 42 U.S.C. Sec. 4332(c); see also Alaska v. Andrus, 580 F.2d 465, 474 (D.C.Cir.1978) (emphasizing the importance of public comment); 40 C.F.R. Secs. 1502.1, 1503.1, 1506.6, 1508.9.

Finally, the government suggested at oral argument that the Secretary's decision reflects the fact that a supplement could not have changed his mind. One cannot, however, argue that a NEPA statement is not legally necessary because the agency already knows what the statement will contain or because the agency plans to make its decision without regard to what the FEIS shows. In some instances, the statement may change a mind that previously thought itself unchangeable; in other instances the statement will simply allow the public to judge more fully the merits of the decision that was made. We note that the Secretary did delete 52 tracts from the proposed sale; 41 were in areas which most commenters urged should not be leased, and the rest were deleted at the request of the Navy. In any event, the statute requires an EIS according to its terms.

In sum, we conclude that the agency's decision not to supplement the FEIS is likely to be found to be unreasonable, and, consequently, we find adequate support in the record for the district court's finding of a probable NEPA violation.

The government argues that, regardless of any NEPA violation, the district court erred in issuing a preliminary injunction because the plaintiffs will suffer no "irreparable harm." The government correctly states that the propriety of a preliminary injunction depends upon consideration of the plaintiffs' likelihood of success on the merits, of "irreparable harm," of an appropriate "balance" of the harms to the plaintiffs and defendants, and of the effect upon the public interest. Planned Parenthood League of Massachusetts v. Bellotti, 641 F.2d 1006, 1009 (1st Cir.1981); see Weinberger v. Romero-Barcelo, 456 U.S. 305, 311-13, 102 S. Ct. 1798, 1802-03, 72 L. Ed. 2d 91 (1982). The government is wrong, however, in arguing that there is no "irreparable injury" here.

The government points out that the lease sale does not necessarily entitle the lease buyers to drill for oil. There are several further steps that must be taken, and further governmental permission must be obtained before oil exploration can begin. See Outer Continental Shelf Lands Act (OCSLA), 43 U.S.C. Sec. 1331 et seq.; North Slope Borough v. Andrus, 642 F.2d 589, 593 (D.C.Cir.1980). This fact, in the government's view, shows that the lease sale alone cannot hurt the environment. The government concludes that the district court should have allowed the sale to proceed while the court made a more thorough determination of its lawfulness. If the court were to find the lease sale unlawful, it could always set it aside after the event. See, e.g., Conservation Law Foundation v. Andrus, 623 F.2d 712, 720 (1st Cir.1979) (dictum); County of Suffolk v. Secretary of Interior, 562 F.2d 1368, 1374 (2d Cir.1977); see also New York v. Kleppe, 429 U.S. 1307, 1312-13, 97 S. Ct. 4, 6-7, 50 L. Ed. 2d 38 (Marshall, Circuit Justice, 1976).

The government's argument, however, ignores an important feature of NEPA. NEPA is not designed to prevent all possible harm to the environment; it foresees that decisionmakers may choose to inflict such harm, for perfectly good reasons. Rather, NEPA is designed to influence the decisionmaking process; its aim is to make government officials notice environmental considerations and take them into account. Thus, when a decision to which NEPA obligations attach is made without the informed environmental consideration that NEPA requires, the harm that NEPA intends to prevent has been suffered. Alaska v. Andrus, 580 F.2d 465, 485 (D.C.Cir.1978); Jones v. District of Columbia Redevelopment Land Agency, 499 F.2d 502, 512-13 (D.C.Cir.1974). NEPA in this sense differs from substantive environmental statutes, such as the Federal Water Pollution Control Act, 33 U.S.C. Secs. 1251 et seq. The Federal Water Pollution Control Act focuses upon the "integrity of the Nation's Waters, not the permit process," Weinberger v. Romero-Barcelo, 456 U.S. at 314, 102 S.Ct. at 1804. NEPA does the converse. Moreover, to set aside the agency's action at a later date will not necessarily undo the harm. The agency as well as private parties may well have become committed to the previously chosen course of action, and new information--a new EIS--may bring about a new decision, but it is that much less likely to bring about a different one. It is far easier to influence an initial choice than to change a mind already made up. See Alaska v. Andrus, 580 F.2d at 485; Jones v. District of Columbia Redevelopment Land Agency, 499 F.2d at 512-13 ("So long as the status quo is maintained, so long as the environmental impact statement is not merely a justification for a fait accompli, there is a possibility that the statement will lead the agency to change its plans....").

It is appropriate for the courts to recognize this type of injury in a NEPA case, for it reflects the very theory upon which NEPA is based--a theory aimed at presenting governmental decision-makers with relevant environmental data before they commit themselves to a course of action. This is not to say that a likely NEPA violation automatically calls for an injunction; the balance of harms may point the other way. Cf. Weinberger v. Romero-Barcelo, supra (a violation of the Federal Water Pollution Control Act does not automatically warrant an injunction). It is simply to say that a plaintiff seeking an injunction cannot be stopped at the threshold by pointing to additional steps between the governmental decision and environmental harm.

In the present case plaintiffs would suffer harm if they were denied an injunction, if the lease sale then took place, and if the court then held that a supplemental EIS was required. In that event, the successful oil companies would have committed time and effort to planning the development of the blocks they had leased, and the Department of the Interior and the relevant state agencies would have begun to make plans based upon the leased tracts. Each of these events represents a link in a chain of bureaucratic commitment that will become progressively harder to undo the longer it continues. Once large bureaucracies are committed to a course of action, it is difficult to change that course--even if new, or more thorough, NEPA statements are prepared and the agency is told to "redecide." It is this type of harm that plaintiffs seek to avoid, and it is the presence of this type of harm that courts have said can merit an injunction in an appropriate case. See California v. Watt, 520 F. Supp. 1359, 1371 (C.D.Cal.1981) ("leasing sets in motion the entire chain of events which culminates in oil and gas development"), cited with approval in California v. Watt, 683 F.2d 1253, 1260 (9th Cir.1982), cert. granted, --- U.S. ----, 103 S. Ct. 2083, 77 L. Ed. 2d 295 (1983); W. Rodgers, Environmental Law Sec. 7.7 at 767 (1977) ("[NEPA's] purpose is to require consideration of environmental factors before project momentum is irresistible, before options are closed, and before agency commitments are set in concrete.").

The more difficult question here involves balancing the harm caused plaintiffs without the injunction against the harm the injunction will cause defendants. Planned Parenthood League of Massachusetts v. Bellotti, 641 F.2d at 1009. We have already expressed our view that the agency action--the lease sale--will probably be found unlawful on NEPA grounds, and we have just described plaintiffs' harm. We believe that the countervailing factor--the harm the injunction will cause defendants--is fairly small. For one thing, defendants do not argue that the Georges Bank oil is needed immediately. Nor, given the apparent recent "oil glut," do we see how they could make such an argument. For another thing, the government does not claim that the oil could make a significant difference to the national defense, or to any economic decisions that it might make. Finally, the government has already--in the SID and EA--begun to gather the information necessary to supplement the FEIS. At oral argument, government counsel said she believed that a supplement could be prepared within 120 days. See also Greene County Planning Board v. Federal Power Commission, 455 F.2d 412, 422-23 (2d Cir.1972) (delay is a natural part of the NEPA process); Calvert Cliffs' Coordinating Committee, Inc. v. Atomic Energy Commission, 449 F.2d 1109, 1122 (D.C.Cir.1971) (the spectre of a power crisis "must not ... blackout ... environmental consideration[s]").

Thus, weighing the NEPA-type harm to plaintiffs against the fairly short delay, in light of the likelihood of unlawfulness, we conclude that the district court's decision as to the "irreparable harm" and its balancing of the relevant harms was reasonable and adequate to support on appeal the issuance of this preliminary injunction. The district court's weighing of the public interest and its conclusions thereon were likewise well within its sound discretion.

In concluding that the preliminary injunction is adequately supported by the likelihood of a NEPA violation, we have deliberately avoided passing upon the ESA, OCSLA and CZMA issues. It is unnecessary for us to do so now. They involve complex issues of law, and we doubt that the district court, in the short time available, had an adequate opportunity to consider them. If we are to decide them, it would be easier to do so with a more complete record before us. Presumably the district court will soon proceed to a full hearing on the merits. If the government issues a supplemental EIS before this litigation is concluded and seeks to have the preliminary injunction dissolved, the district court may wish to consider these other statutory issues again. If it does so, it should not regard its prior opinion concerning them as governing; but it should consider the issues anew.

The decision of the district court here under review is

Affirmed.