The plaintiffs in this case, manufacturers of cigarettes and smokeless tobacco products,1 mounted a constitutional challenge to the novel ingredient-reporting requirements of Mass. Gen. L. ch. 94, § 307B (Section 307B). The district court granted the plaintiffs' motion for a preliminary injunction restraining two state officials (collectively, the Commonwealth) from enforcing these requirements. In this venue, the Commonwealth invites us to vacate or modify the injunction. We decline the invitation.

Regulation is not a stranger to the tobacco industry. The Federal Cigarette Labeling and Advertising Act, 15 U.S.C. § 1335a (1994) (the Labeling Act), mandates that "[e]ach person who manufactures, packages, or imports cigarettes shall annually provide the Secretary [of Health and Human Services] with a list of the ingredients added to tobacco in the manufacture of cigarettes," but this list need not "identify the company which uses the ingredients or the brand of cigarettes which contain the ingredients," and those required to furnish lists may designate proxies to do so on their behalf. Cigarette manufacturers typically comply with the Labeling Act's strictures through an internuncio; they submit information to a law firm which acts as a clearinghouse for the industry. The law firm then furnishes an annual list of all ingredients used by any of the companies to the Secretary. The law firm maintains the secrecy of the ingredients used in a particular brand from both the government and the brand's competitors.2 In short, though the Labeling Act obligates the Secretary to report to Congress health risks from tobacco products discerned directly or indirectly through the lists, it assures confidentiality for trade secrets.

Existing state law is not much more intrusive. Apart from Massachusetts, only Minnesota and Texas have required any reporting of tobacco ingredients. The Minnesota statute, Minn.Stat. § 461.17 (Supp.1997), compels tobacco manufacturers to report the use of any of several targeted additives in their products. The Texas law, Tex. Health & Safety Code Ann., §§ 161.251-255 (West Supp.1998), bears certain similarities to Section 307B, but provides protection for information submitted that "would be excepted from public disclosure as a trade secret under state or federal law." Id. § 161.254(c).

Massachusetts has gone further. When Section 307B was enacted as a means of regulating the tobacco industry, proponents billed it as an innovative regulatory effort which, incidentally, would protect public health. See Press Release Distributed by the Commonwealth upon Signing of Section 307B, August 2, 1996 (quoting then-Governor William F. Weld's description of Section 307B as "a common sense, pro-consumer bill that will give people all the information they need to make educated decisions about what they put in their bodies"). The statute significantly expands the reach of existing positive law. Its ingredient-reporting provisions are novel both because they demand brand-by-brand reporting of additives and because they permit public disclosure of this ingredient information.

Specifically, Section 307B stipulates that each manufacturer of tobacco products must report annually to the Massachusetts Department of Public Health (DPH) "[t]he identity of any added constituent other than tobacco, water or reconstituted tobacco sheet made wholly from tobacco, to be listed in descending order according to weight, measure, or numerical count" for each brand sold within the state. Any such information that DPH reasonably concludes "could reduce risks to public health, shall be public records," as long as the attorney general advises DPH that such disclosure would not work an unconstitutional taking.3 The historical archives clearly indicate the legislature's intent. For instance, in a letter urging colleagues to support the bill that eventually became Section 307B, a proponent explained that brand-specific reporting and disclosure are necessary because "[i]f you smoke Merits you want to know what is in Merits, not what may be in every brand of cigarettes on the market." Letter from Senator Warren E. Tolman to Colleagues 2 (June 14, 1996).

Because consumers choose brands based on flavor, taste, and aroma, and tend to remain loyal to those brands, small fortunes are spent creating the flavor formulas for tobacco products. The information needed to copy these formulas is, in turn, worth many millions of dollars. See, e.g., Kurt Badenausen, Blind Faith, Financial World, July 8, 1996, at 50-65 (describing Philip Morris's Marlboro brand as worth over $44,000,000,000 and rating it the most valuable of 364 brand names surveyed). It is no secret that tobacco companies, like other manufacturers of brand name products, employ elaborate procedures to safeguard their ingredient information. For example, suppliers sign confidentiality agreements and furnish their wares in coded packaging, devoid of proprietary names, to keep ingredient information under wraps. Even in house, copies of flavor formulas are retained under lock and key, and ingredient information is made available only on a "need to know" basis.

The tobacco companies claim that the operation of Section 307B threatens to destroy these enormously valuable trade secrets. The industry submits aggregate lists of all ingredients included in tobacco products sold in the United States in compliance with federal law. However, at the current state of technology, these lists cannot feasibly be used to copy a tobacco product's taste or aroma. Divulging brand-specific lists of ingredients in descending order of volume, as required by Section 307B, is quite a different story; the plaintiffs aver--and the Commonwealth, for purposes of this proceeding, does not contradict--that such lists, when and as disclosed, will allow pirates to "reverse engineer" products possessing flavors and aromas indistinguishable from popular brands, with substantially reduced research and development costs. The threat of this increased ease of entry into, and competition within, the tobacco industry fuels the plaintiffs' challenge to Section 307B.

The cigarette and smokeless tobacco companies brought separate suits attacking Section 307B. Their complaints claimed that the statute was preempted by federal law and that it ran afoul of various constitutional impediments, including the Takings Clause, the Commerce Clause, and the Due Process Clause. The district court consolidated the cases. Early on, it resolved the preemption question in favor of the Commonwealth, and we affirmed that determination. See Philip Morris, Inc. v. Harshbarger, 122 F.3d 58, 87 (1st Cir.1997).

The plaintiffs had greater success when they moved for a preliminary injunction to prevent the enforcement of Section 307B's ingredient-reporting requirements. Finding that the plaintiffs were likely to succeed on the merits of their takings claim and that they faced irreparable harm in the absence of interim relief, the district court restrained the enforcement of the ingredient-reporting provisions pendente lite. This interlocutory appeal ensued. We have jurisdiction under 28 U.S.C. § 1292(a)(1).

In considering a request for a preliminary injunction, a trial court must weigh several factors: (1) the likelihood of success on the merits, (2) the potential for irreparable harm to the movant, (3) the balance of the movant's hardship if relief is denied versus the nonmovant's hardship if relief is granted, and (4) the effect of the decision on the public interest. See Ross-Simons of Warwick, Inc. v. Baccarat, Inc., 102 F.3d 12, 15 (1st Cir.1996); Narragansett Indian Tribe v. Guilbert, 934 F.2d 4, 5 (1st Cir.1991). Likelihood of success is the touchstone of the preliminary injunction inquiry. See Ross-Simons, 102 F.3d at 16; Weaver v. Henderson, 984 F.2d 11, 12 (1st Cir.1993). Mindful of this reality, the Commonwealth confines its challenge here to this element.

We review the trial court's grant or denial of a preliminary injunction only for abuse of discretion or mistake of law. See EEOC v. Astra USA, Inc., 94 F.3d 738, 743 (1st Cir.1996). This standard requires "a party who appeals from the issuance of a preliminary injunction [to] bear[ ] the considerable burden of demonstrating that the trial court mishandled the fourpart framework." Ross-Simons, 102 F.3d at 16. Our analysis proceeds accordingly.

The Takings Clause of the Fifth Amendment is incorporated in, and applies to the states by virtue of, the Fourteenth Amendment. See Chicago, Burlington & Quincy R.R. Co. v. Chicago, 166 U.S. 226, 239, 17 S. Ct. 581, 41 L. Ed. 979 (1897); Culebras Enters. Corp. v. Rivera Rios, 813 F.2d 506, 515 (1st Cir.1987). Case law under the Takings Clause has developed along two parallel lines, one addressing physical invasions (sometimes called per se takings) and the other addressing regulatory takings. See Lucas v. South Carolina Coastal Council, 505 U.S. 1003, 1015, 112 S. Ct. 2886, 120 L. Ed. 2d 798 (1992). Here, the plaintiffs' principal claim is that Section 307B works a regulatory taking.4 The thrust of their argument is that the Commonwealth's action in requiring disclosure and permitting the subsequent publication of brand-specific ingredient information is not everyday regulation, the inconveniences of which individuals in a civilized society must bear, but, rather, goes so far that it impermissibly takes their property for public use without just compensation, in violation of the Takings Clause. See Pennsylvania Coal Co. v. Mahon, 260 U.S. 393, 415, 43 S. Ct. 158, 67 L. Ed. 322 (1922).

To evaluate the propriety of a preliminary injunction on a regulatory takings claim, an inquiring court must sort through a takings analysis in addition to the multi-factored preliminary injunction determination. This takings analysis should include consideration of "the character of the governmental action, its economic impact, and its interference with reasonable, investment-backed expectations." PruneYard Shopping Ctr. v. Robins, 447 U.S. 74, 83, 100 S. Ct. 2035, 64 L. Ed. 2d 741 (1980). Although the articulation of these factors makes the takings inquiry seem much like any other multi-pronged test, the Supreme Court has stated in no uncertain terms that a regulatory takings analysis should not be governed by a "set formula," but must be determined by an "essentially ad hoc, factual inquir[y]." Penn Cent. Transp. Co. v. New York City, 438 U.S. 104, 124, 98 S. Ct. 2646, 57 L. Ed. 2d 631 (1978). Thus, the three elements enumerated in PruneYard operate primarily as lenses through which a court can view and process the facts of a given case rather than as a checklist of items that can be ticked off as fulfilled or unfulfilled.

In the case at hand, the lower court determined that the plaintiffs enjoyed a likelihood of success on their regulatory takings claim. The Commonwealth disputes this determination on two main grounds. First, it faults the lower court's conclusion that the plaintiffs' reasonable, investment-backed expectations of nondisclosure of ingredient information sufficed to legitimate the finding of a taking. Second, it challenges the court's characterization of the governmental action, asseverating that the application of Section 307B's ingredient-reporting provisions to the plaintiffs lacks legal compulsion sufficient to create an actionable taking. After a brief detour, we will consider these contentions sequentially.

Before proceeding to address the Commonwealth's claims, we think it is useful to clarify what the Commonwealth does not claim in this proceeding. For one thing, it does not now dispute that information provided to DPH under Section 307B will be disclosed to the public. For another thing, it concedes for purposes of these appeals that such information will include valuable trade secrets, susceptible to destruction if exposed. Finally, because the statute was enacted as a regulatory measure, it is perforce grounded in the state's police power over matters of public health. Although the Commonwealth suggests with scant elaboration that the police power alone offers a sufficient justification for the statute, the parties primarily have briefed and argued the issue of whether the Takings Clause may invalidate the statute. We have therefore focused our likelihood of success analysis on this Takings Clause issue. At this stage of the proceedings, the Commonwealth has not developed any independent "police power" rationale to justify its position and, accordingly, we have before us insufficient "police power" rationale to reach a decision on that issue.

In debating whether the plaintiffs possess the requisite expectations to support a takings claim, both sides embrace the Supreme Court's decision in Ruckelshaus v. Monsanto Co., 467 U.S. 986, 104 S. Ct. 2862, 81 L. Ed. 2d 815 (1984). Monsanto is a complex case based on intricate facts and it ultimately propounds several holdings. Despite the palpable difficulty of doing so, we believe it is necessary to explicate the factual scenario that confronted the Monsanto Court in order to assess the conflicting claims asserted here.

Monsanto involved sequential amendments to the Federal Insecticide, Fungicide, and Rodenticide Act (FIFRA), 7 U.S.C. § 136 et seq., the first set of amendments occurring in 1972 and the second set in 1978. See Monsanto, 467 U.S. at 990-97, 104 S. Ct. 2862. The timing of these amendments created three distinct FIFRA regimes. From 1947 (when Congress first enacted FIFRA) until the effective date of the 1972 amendments, FIFRA operated primarily as a licensing and labeling statute; its terms required all pesticides sold in interstate or foreign commerce for use within the United States to be registered with the Secretary of Agriculture and appropriately labeled. See id. at 991, 104 S. Ct. 2862. In addition, FIFRA empowered the Secretary to require applicants for registration to produce testing data (including pesticide formulas) and to substantiate claims asserted on product labels. See id. The statute forbade the Secretary to disclose formula information, but made no mention of what protection, if any, attended the submission of other testing data. See id.

In 1970, Congress shifted the responsibility for administering FIFRA from the Department of Agriculture to the Environmental Protection Agency (EPA). See Reorganization Plan No. 3 of 1970, 3 C.F.R. 1074 (1966-1970), reprinted in 5 U.S.C. app. at 1552 (1994). Shortly thereafter, the 1972 amendments metamorphosed FIFRA into a comprehensive regulatory scheme for pesticides. This scheme, in effect from 1972 to 1978, required disclosure to the EPA and the public of environmental, health, and safety data--but it provided specific protections for that data so as to avoid the revelation of trade secrets. See Monsanto, 467 U.S. at 992, 104 S. Ct. 2862. During this period, FIFRA allowed applicants to designate submitted information as "trade secrets or commercial or financial information," and simultaneously prohibited the EPA from publishing such information. Id. (quoting applicable statutory provision). If the EPA and an applicant disagreed as to the status of submitted information and the EPA proposed to make such information public, FIFRA authorized the applicant to bring a declaratory judgment action in a federal district court prior to publication. See id. The 1972 amendments also allowed the EPA to use information provided by one applicant in its consideration of another applicant's request for registration of a similar chemical, provided that the latter agreed to compensate the former. See id. It must be noted, however, that this information-sharing requirement only applied to data designated as "trade secrets or commercial or financial information" if the initial applicant consented to such use. Id.

When Congress amended FIFRA again in 1978, it altered the safeguards against disclosure with respect not only to data thereafter submitted, but also with respect to data that had been supplied during earlier periods. See id. at 994-95, 104 S. Ct. 2862. Under the 1978 iteration of the statutory scheme, applicants who submitted health, safety, or environmental information to EPA for pesticides registered after September 30, 1978, received a ten-year period of exclusive use for any such data that related to new active ingredients. See id. at 994, 104 S. Ct. 2862. Any other data that had been tendered after December 31, 1969 were to be made available for citation and consideration in support of other applications for fifteen years after the original submission date, provided that the later applicant agreed to compensate the original submitter. See id. The 1978 amendments also allowed all health, safety, and environmental data to be divulged upon request, notwithstanding the prohibition on disclosing trade secrets, but did not authorize revelation of manufacturing or quality control processes without a determination by the EPA that such disclosure was "necessary to protect against an unreasonable risk of injury to health or the environment." Id. at 995-96, 104 S. Ct. 2862 (quoting applicable statutory provision).

The Monsanto plaintiff, a pesticide manufacturer, argued that use or disclosure of the trade secrets that it had submitted to the federal sovereign during any of the three periods would constitute a regulatory taking. The Court decided as a threshold matter that the data constituted "property" under state law and thus enjoyed protection under the Takings Clause. See id. at 1003-04, 104 S. Ct. 2862. The Court then addressed each of the three statutory intervals. Apropos of the 1972-78 period, the Court held that uncompensated (or undercompensated) use or disclosure of trade secret data submitted during that time frame would constitute a taking. See id. at 1010-14, 104 S. Ct. 2862. In contrast, the Court ruled that there could be no taking for either the pre-1972 or the post-1978 periods because the pesticide manufacturer had no reasonable, investment-backed expectation of governmental nondisclosure during those periods. See id. at 1006-07, 1009-10, 104 S. Ct. 2862. Speaking of this last period, the Court explained: "[A]s long as [a pesticide manufacturer] is aware of the conditions under which the data are submitted, and the conditions are rationally related to a legitimate government interest, a voluntary submission of data by an applicant in exchange for the economic advantages of registration can hardly be called a taking." Id. at 1007, 104 S. Ct. 2862.

The Commonwealth uses the statement we have just quoted to support its claim that, after the effective date of Section 307B, divulgement of submitted ingredient lists cannot work a taking because the statute's enactment vitiates any reasonable investment-backed expectation of nondisclosure on the tobacco companies' part. We think that it is unfair to read Monsanto for this proposition because the part of the Court's trifurcated holding to which the Commonwealth clings depended on the existence of a voluntary exchange. See Nollan v. California Coastal Comm'n, 483 U.S. 825, 833-34 n. 2, 107 S. Ct. 3141, 97 L. Ed. 2d 677 (1987) (drawing this distinction). Under the post-1978 FIFRA scheme, submitters of environmental, health, and safety data received significant benefits in return for the disclosure of their data, including rights of exclusive use for a term of years and rights to compensation from later applicants who wished to utilize submitted data. Since this exchange afforded tangible compensation to pesticide manufacturers, the post-1978 version of FIFRA did not work an uncompensated taking (and, hence, did not work an unconstitutional taking). Section 307B effects no comparable bargain.

The Commonwealth demurs. The exchange, it says, consists of permitting the tobacco companies to continue doing business in Massachusetts in return for the companies' compliance with Section 307B. This construct will not wash. A Monsanto-type exchange requires that the government grant a benefit of real value to compensate a property owner for a taking. In constructing this balance, not all benefits bestowed by the sovereign will possess sufficient substance to ameliorate the taking--and the state's self-interested characterization of a right as a benefit cannot change the underlying calculus. Permitting a company to continue conducting business within a state, while a benefit of sorts, lacks sufficient substance to create a Monsanto-type exchange.

Nollan illustrates this point. There, a governmental entity required a landowner to dedicate a public easement across his beachfront property in order to obtain a building permit to improve the existing structure. See id. at 828, 107 S. Ct. 3141. To counter the landowner's assertion that the compelled easement comprised a taking, the dissent called the permit a benefit and claimed that its conferral triggered an exchange akin to that in Monsanto. See id. at 860 n. 10, 107 S. Ct. 3141 (Brennan, J., dissenting). The majority disagreed, stating that "the announcement that the application for (or granting of) the permit w[ould] entail the yielding of a property interest cannot be regarded as establishing the voluntary 'exchange' that we found to have occurred in Monsanto." Nollan, 483 U.S. at 834 n. 2, 107 S. Ct. 3141 (citations omitted). The Nollan Court explained that the ability to improve one's own property, though subject to some regulation, is incomparable to the type of government benefit proffered in exchange for use and disclosure of trade secret information in Monsanto. See id. Thus, Nollan teaches that the mere granting of permission to engage in routine activities, incident to existing property rights, does not afford compensation sufficient to support a Monsanto-type exchange.

Applying Nollan 's rationale here, it is pellucid that the Commonwealth's unilateral announcement that the privilege of continuing to do business in Massachusetts henceforth will entail the yielding of a tobacco company's trade secrets cannot, in itself, establish a benefit sufficient to support a voluntary exchange within the Monsanto paradigm. The ability to conduct (and, more especially, to continue to conduct) a lawful business in Massachusetts, though subject to some governmental requirements, simply is not analogous, either in kind or in degree, to the benefit that effected the exchange and extinguished the takings claim in Monsanto. In context, then, the Monsanto Court's discussion of FIFRA's post-1978 regime offers the Commonwealth cold comfort.

The Commonwealth finds somewhat sturdier support for its position in the Monsanto Court's resolution of the takings issue for the pre-1972 period. Even though the earliest versions of FIFRA included no conditions explicitly permitting public disclosure of submitted data, and the Trade Secrets Act, 18 U.S.C. § 1905, effectively barred disclosure of trade secrets by government agencies and employees, the Court held that when Monsanto submitted data during the pre-1972 period it "could not have had a 'reasonable investment-backed expectation' that EPA would maintain those data in strictest confidence and would use them exclusively for the purpose of considering the Monsanto application in connection with which the data were submitted." Monsanto, 467 U.S. at 1010, 104 S. Ct. 2862. Consequently, the Court found no unconstitutional taking for that period.5

The analogy between Monsanto 's pre-1972 period and Section 307B cannot be brushed aside lightly. Historically, Massachusetts has granted protection to trade secrets both statutorily, see Mass. Gen. Laws, ch. 93 §§ 42, 42A (1997), and under state common law, see, e.g., Peggy Lawton Kitchens, Inc. v. Hogan, 18 Mass.App.Ct. 937, 466 N.E.2d 138, 139-40 (Mass.App.Ct.1984). The Commonwealth makes a plausible argument that these protections together comprise a general, external source of protection comparable to the Trade Secrets Act. Under the Monsanto Court's reasoning regarding submissions during the pre-1972 period, this argument holds, the tobacco companies would have no founded expectation of nondisclosure in respect to information submitted pursuant to the strictures of Section 307B. In the last analysis, however, Monsanto itself neutralizes this argument.

The Monsanto Court's holding that no uncompensated taking occurred during the pre-1972 and post-1978 periods is neither the be-all nor the end-all of its opinion. The Justices also held that Monsanto had reasonable, investment-backed expectations sufficient to support a regulatory takings claim for data submitted during the intermediate 1972-78 period. See Monsanto, 467 U.S. at 1011, 104 S. Ct. 2862. This undermines the Commonwealth's argument because the 1972-78 period presents the closest, most persuasive analogy to the situation created by Section 307B. The FIFRA scheme then in effect provided specific protections for trade secret information--and the Court determined that pesticide registrants might reasonably rely on these protections. See id. at 1010-11, 104 S. Ct. 2862. The statutory and common law protections for trade secret information in place in the Commonwealth create a very similar prophylaxis and thus form the basis for a reasonable expectation of continued confidentiality.

Because this matter is before us on appeal from the grant of a preliminary injunction, we need not rule definitively on the point. Likelihood-of-success determinations in such a context require only that courts formulate statements of probable outcomes. See Cohen v. Brown Univ., 991 F.2d 888, 902 (1st Cir.1993); Narragansett Indian Tribe, 934 F.2d at 6. While we cannot entirely dismiss the Commonwealth's argument, we are comfortable in concluding that it probably will bear no fruit.

This is especially so because other signposts point in a direction favoring the tobacco companies' position. Most notably, recent Supreme Court cases share a greater affinity with the Nollan Court's distinction of Monsanto--a distinction that did not explicitly differentiate among the case's three holdings--than with the Commonwealth's isthmian focus on Monsanto 's treatment of the pre-1972 period. See, e.g., Dolan v. City of Tigard, 512 U.S. 374, 391, 114 S. Ct. 2309, 129 L. Ed. 2d 304 (1994) (holding that burdens of municipal exactions required in exchange for building permits must achieve a "rough proportionality" with benefits received by the landowner to avoid municipal liability for a taking); Lucas, 505 U.S. at 1031-32, 112 S. Ct. 2886 (requiring South Carolina to prove that a landowner's intended residential use of his land would create a common law nuisance in order to avoid a finding of a taking without just compensation). These authorities show the Court's increasing concerns in this area and counsel persuasively that the Court will demand substantial, rather than nominal, compensation to legitimize governmental takings. In light of this guidance, we cannot accept the Commonwealth's claim that mere leave to continue one's business activities in a state will duly compensate a taking of valuable private property rights.

The Commonwealth's remaining theory posits that Section 307B cannot work a taking as a matter of law because it lacks "legal compulsion"--in other words, the law works no taking because it does not force the tobacco companies to sell their products in Massachusetts (and, thus, they can avoid any need to grapple with it merely by limiting their business activities to more hospitable climes). In pressing this theory, the Commonwealth relies chiefly upon Hinesburg Sand & Gravel Co. v. Chittenden Solid Waste Dist., 959 F. Supp. 652 (D.Vt.1997). In that case, a landowner sued a municipal authority to recover legal costs incurred in defending against the attempted condemnation of his property, alleging that there had been a taking. See id. at 656-57. The court ruled that no taking had occurred because the landowner was not legally compelled to spend funds defending his property. See id. at 657-58.

Hinesburg is small solace to the Commonwealth. The court's legal reasoning is suspect and, in all events, the case is plainly inapposite.6 Here, unlike in Hinesburg, a state statute forces a party to make a Hobson's choice: either submit ingredient lists containing valuable trade secrets without adequate safeguards or cease doing business in an important market. This is the essence of legal compulsion.

The other authorities cited by the Commonwealth are no more convincing. See, e.g., Bowles v. Willingham, 321 U.S. 503, 517, 64 S. Ct. 641, 88 L. Ed. 892 (1944) (holding that wartime rent control did not work a taking and noting that "[t]here is no requirement that the apartments in question be used for purposes which bring them under the Act"); Meriden Trust & Safe Deposit Co. v. FDIC, 62 F.3d 449, 455 (2d Cir.1995) (concluding that FDIC cross-guarantee provisions did not unconstitutionally take private property because they presented financial institutions with a choice between insuring and not insuring); Garelick v. Sullivan, 987 F.2d 913, 916 (2d Cir.1993) (rejecting an anesthesiologist's claim that Medicare fee-for-service regulations constituted a taking and noting that doctors are "under no legal duty to provide services to the public and to submit to price regulations"). Underlying these cases, and others like them, is the reality that a governmental entity which creates a market's supply or sets its prices may be expected to alter property rights in the course of modifying its regulations. Thus, when an individual voluntarily participates in such a price-regulated program or market, the Takings Clause does not protect him from changes in his property rights due to changes in applicable regulations.

The situation created by Section 307B is entirely different. The plaintiffs historically have participated in a lawful, non-price-regulated market, in which state government hitherto has not been directly involved. They now face the potential loss of their valuable trade secrets merely to remain in business in Massachusetts. The Commonwealth cannot by some mysterious alchemy transform this situation into one akin to that which existed in the regulated-market cases. Were the law otherwise, any government entity could avoid the due operation of the Takings Clause by the simple expedient of stating its intentions in advance.

The Commonwealth derives its final support for its "legal compulsion" argument from a footnote to the Monsanto Court's discussion of why use and disclosure of data submitted after 1978 would not constitute a taking. In this note, the Justices explained that a pesticide manufacturer could choose to forgo registration in the United States and sell its pesticides solely in foreign markets. See Monsanto, 467 U.S. at 1007 n. 11, 104 S. Ct. 2862 (dictum). Using footnote 11 as a springboard, the Commonwealth maintains that the tobacco companies suffer no taking under Section 307B because they may refrain from selling their products in Massachusetts and thereby thwart disclosure.

This argument wrenches footnote 11 loose from its contextual moorings. The Supreme Court appended the footnote to its discussion of the voluntary exchange component of Monsanto 's post-1978 regime. Voluntary exchange is a far cry from the situation at hand, in which the only benefit offered by the government in return for releasing the tobacco companies' trade secrets is the right to continue doing business in the Commonwealth. As we already have explained, see supra at 677-78, permission to continue operating a lawful business is not the type of government benefit on which a Monsanto-type exchange validly may be predicated.

In sum, the fact that the tobacco companies may cease doing business in Massachusetts if they do not wish to submit ingredient information to the DPH is true as far as it goes, but, as a principle of constitutional law, it does not go very far.

At a last gasp, the Commonwealth insists that the lower court swept too broadly in fashioning the preliminary injunction and, therefore, abused its discretion. In the Commonwealth's view, the district court could have met the plaintiffs' legitimate needs by allowing the ingredient information to be furnished to DPH, as required by Section 307B, and enjoining only public disclosure of the data.

We agree that, in the exercise of the district court's discretion, a narrower order might have been appropriate. Still, there is a rub: the Commonwealth never tendered this suggestion in the district court. Having pursued the advantages of an all-or-nothing strategy in arguing against the injunction, the Commonwealth may not belatedly obtain the benefits of the more moderate approach that, in the light of its defeat, now looks more attractive.

There is no reason to tarry. As a general rule, a disappointed litigant cannot surface an objection to a preliminary injunction for the first time in an appellate venue. See United States v. Zenon, 711 F.2d 476, 478 (1st Cir.1983) (explaining that parties are required to "state their objections to the injunction to the district court, so that the district court can consider them and correct the injunction if necessary, without the need for appeal"). Having failed to comply with this basic rule, the Commonwealth has forfeited the opportunity to obtain consideration of whether the preliminary injunction, as framed, is overbroad.

We need go no further. The short of it is that neither the Commonwealth's "absence of reasonable, investment-backed expectations" argument nor its "legal compulsion" construct satisfies its weighty burden of demonstrating that the district court committed a clear error of law or an abuse of discretion. The Commonwealth's effort to fault the breadth of the district court's decree is similarly unavailing. Consequently, we are unable to conclude, at the preliminary injunction stage, that the district court erred in finding that the plaintiffs had demonstrated a likelihood of success on the merits.

Affirmed. Costs in favor of appellees.