dissenting.
My colleagues’ opinion today reaches three important and dispositive conclusions:
That a parent corporation cannot, as a matter of law, be held directly liable under 42 U.S.C. § 9607(a)(2) as an “operator” of a facility owned by its subsidiary corporation, but may face only vicarious liability under state-law corporation-veil-piercing principles;
That piercing the corporate veil under Michigan law requires circumstances showing that the corporate form was used to “subvert justice”; and That the district court erred in finding that the defendants failed to prove their entitlement to the so-called third-party defense under 42 U.S.C. § 9607(b)(3).
*587In my judgment, the court is mistaken on all three grounds, and I therefore respectfully dissent.
I.
Direct Liability of a CPC Under Section 107(a)(2)
A.
Parent Corporations as “Operators”
CERCLA identifies two distinct categories of individuals or entities that may be held directly liable, as responsible parties, for the costs of cleanup of an environmentally contaminated facility:
(1) the owner and operator of a vessel or a facility, [or]
(2) any person who at the time of disposal of any hazardous substance owned or operated any facility at which such hazardous substances were disposed of[.]
42 U.S.C. § 9607(a) (emphasis added). The parties have stipulated that the site is a “facility” as defined by CERCLA, and that the site contains “hazardous substances.”
The terms “owner” and “operator,” as used in section 107(a)(2), are defined in the statute as “any person owning or operating such facility,” 42 U.S.C. § 9601(20)(A)(ii), a definition that is, at best, circular and unhelpful. At least one court has observed that “[t]he circularity strongly implies ... that the statutory terms have their ordinary meanings rather than unusual or technical meanings.” Edward Hines Lumber Co. v. Vulcan Materials Co., 861 F.2d 155, 156 (7th Cir.1988). Further, it is noteworthy that the definition of “owner” explicitly excludes one whose ownership interest is merely that of a stockholder and who does not participate in management of the facility, 42 U.S.C. § 9601(20)(A)(iii), suggesting that one who does participate in management is accordingly susceptible to liability, see United States v. Kayser-Roth Corp., 910 F.2d 24, 26 n. 6 (1st Cir.1990). Finally, “person” is defined expansively as “an individual, firm, corporation, association, partnership, consortium, joint venture, commercial entity, United States Government, State, municipality, commission, political subdivision of a State, or any interstate body.” 42 U.S.C. § 9601(21). The breadth of this definition plainly leaves room for a parent corporation. See Kayser-Roth, 910 F.2d at 25 n. 5.
The first question presented by this appeal is whether a parent corporation may be considered an “operator” under section 107(a)(2) when its subsidiary is the “owner.” The district court determined that a parent corporation may face potential liability as an operator of a contaminating facility because the plain language of section 107(a)(2) indicates that Congress intended to impose liability on any entity actually operating a facility, regardless of the nature of the entity’s ostensible interest in the facility. It is a conclusion with which I agree; indeed, it is a conclusion that the vast majority of circuits— eight out of nine considering the question— have reached as well. Certain Underwriters at Lloyd’s, London v. St. Joe Minerals Corp., 90 F.3d 671, 673-74 (2d Cir.1996); Schiavone v. Pearce, 79 F.3d 248, 253-55 (2d Cir.1996); FMC Corp. v. United States Dep’t of Commerce, 29 F.3d 833, 842 (3d Cir.1994) (en banc); Lansford-Coaldale Joint Water Authority v. Tonolli Corp., 4 F.3d 1209, 1221-22 (3d Cir.1993); Jacksonville Elec. Auth. v. Bernuth Corp., 996 F.2d 1107, 1110 (11th Cir.1993); John S. Boyd Co. v. Boston Gas Co., 992 F.2d 401, 408 (1st Cir.1993); see Kaiser Aluminum and Chem. Corp. v. Catellus Dev. Corp., 976 F.2d 1338, 1341-42 (9th Cir.1992); Kayser-Roth Corp., 910 F.2d at 27; see also United States v. TIC Investment Corp., 68 F.3d 1082, 1091-92 (8th Cir.1995), cert. denied, — U.S.-, 117 S.Ct. 50, 136 L.Ed.2d 14 (1996); Nurad, Inc. v. William E. Hooper & Sons Co., 966 F.2d 837, 842 (4th Cir.1992); cf. Sidney S. Arst Co. v. Pipefitters Welfare Educ. Fund, 25 F.3d 417, 420 (7th Cir.1994). But see Joslyn Mfg. Co. v. T.L. James & Co., 893 F.2d 80 (5th Cir.1990).
My colleagues offer several reasons for parting company with this impressive wealth of authority. First, they argue that courts have relied on the so-called remedial purpose of CERCLA to justify their conclusion that CERCLA should be given an expansive meaning, including a meaning that admits of liability for parent corporations as operators. Certainly, the shibboleth “remedial purpose” *588is a weak basis for legal analysis and I do not find it necessary to resort to it. But what the majority sees as the logical second step to a rejection of the remedial-purpose litany, namely, its “adhere[nce] to the tenet that liability attaches only to those parties who are culpable in the sense that they, by some realistic measure, helped to create the harmful conditions,” is, to put it mildly, a non sequitur. (Maj. op. at 578.) Implicit in the majority’s chosen “tenet” is rejection of the universally accepted principle that liability under CERCLA is both strict, as well as joint and several. See, e.g., Kayser-Roth, 910 F.2d at 26; FMC Corp., 29 F.3d at 835. Whether CERCLA’s purpose is “remedial” or not, the sounder proposition, I think, is that liability attaches only to those entities whom Congress has singled out for liability. For the reasons set forth below, the statutory language leaves no room to doubt that Congress has singled out operators for liability, irrespective of their corporate form.
The majority is also critical of the district court’s language that liability should be predicated on a “new, middle ground,” and contends that this choice of language makes it unclear “whether the [district court’s] basis for finding parental liability as an operator ... is the actual operation of the subsidiary’s business or ... the exertion of power or influence through active participation in the subsidiary’s business.” (Maj. op. at 579.) It may be that trial courts’ inventive “new, middle ground” expression is more conferring than clarifying, but my colleagues’ conclusion that the district court’s holding “threaten[s] the efficacy of time-honored liability protections afforded by the corporate form,” simply does not follow. And my colleagues’ ultimate conclusion, that a parent can only be found liable “when the requirements necessary to pierce the corporate veil are met,” contradicts the plain language of section 107(a)(2). Moreover, it is a conclusion that begs the question this case presents, which, under subsection (a)(2), is not whether a parent corporation may be held vicariously liable for abuse of its subsidiary’s corporate form— clearly it may — but whether Congress has created direct liability if the facts show that the parent corporation was the actor actually operating a contaminating facility. Stated differently, the issue is whether Congress has excused a parent corporation that is in fact operating a contaminating facility from direct liability, simply because it is doing so in the name of a corporate subsidiary. The majority’s belief that such an excuse may be found in the statute does not, for me, withstand close scrutiny.
The structure of the statute, on its face, requires a recognition that “ ‘owner’ liability and ‘operator’ liability denote two separate concepts and hence require two separate standards for determining whether they apply.” Lansford-Coaldale, 4 F.3d at 1220. “ ‘CERCLA’s language ... indicates an intent to hold a corporation liable for the environmental violations of its subsidiaries and sister corporations, if it is otherwise determined to have operated the facility in question.’ ” Schiavone, 79 F.3d at 255 (quoting Lansford-Coaldale, 4 F.3d at 1221 n. 11). As the Kayser-Roth court observed,
Congress, by including a liability category in addition to owner (“operators”) connected by the conjunction “or,” implied that a person who is an operator of a facility is not protected from liability by the legal structure of ownership. Given this grammatical construction and the broad definition of “person,” corporate status, while relevant to determine ownership, cannot shield a person from operator liability.
Kayser-Roth, 910 F.2d at 26; see Schiavone, 79 F.3d at 254. In short, direct CERCLA liability under subsection (a)(2) may be predicated on one’s status as an owner or on one’s status as an operator.
Alternatively, of course, a parent corporation may also be derivatively liable if circumstances warrant piercing the corporate veil in order to treat that parent corporation as an owner, when its subsidiary is in fact the owner. The majority opinion conflates and confounds the two types of liability, which are analytically distinct, and erroneously concludes that the latter is the exclusive basis for liability. See Sidney S. Arst Co., 25 F.3d at 420; cf. Schiavone, 79 F.3d at 253. As a result, the majority’s analysis founders on a perceived tension between the standard of a derivative liability arising out of common-law *589principles of corporate law and the standard of direct liability arising out of an application of the statutory language of CERCLA. The tension is chimerical. The statute explicitly provides for distinct, direct operator liability, and “the basis for such operator liability is wholly independent of any liability on the part of’ the subsidiary, Schiavone, 79 F.3d at 254; it is a direct liability that “stem[s] directly from [the parent’s] control over the plant,” that is, its own actions as an operator, id. The mere fact that derivative owner liability requires a veil-piercing analysis simply does not speak to the appropriate analysis of direct operator liability.
As previously alluded to, many courts have analyzed the issue of direct operator liability of a parent corporation. Most have followed the approach of Kayser-Roth, in which the court declared that “[t]o be an operator requires more than merely complete ownership and the concomitant general authority or ability to control that comes with ownership. At a minimum it requires active involvement in the activities of the subsidiary.” 910 F.2d at 27. “This standard requires an investigation into the relationship between the parent and subsidiary, in order to reveal the requisite level of corporate involvement.” John S. Boyd, 992 F.2d at 408. Thus, the degree of control required is “more ... than simple ownership and the general authority or control that comes with it.” Certain Underwriters, 90 F.3d at 674. In Lansford-Coaldale, 4 F.3d 1209, the court embraced the KayserRoth standard, emphasizing that “operator liability may be established even without evidence that a [parent] corporation controlled the environmental decisions of an affiliated corporation as long as there exist other factors which sufficiently demonstrate pervasive control.” Id. at 1222 n. 13. Similarly, in Jacksonville Electric, 996 F.2d 1107, the court agreed that the test for direct parent corporation liability as an operator was whether the parent “ ‘exercises actual and pervasive control of the subsidiary to the extent of actually involving itself in the daily operations of the subsidiary. Actual involvement in decisions regarding the disposal of hazardous substances is a sufficient, but not a necessary, condition to the imposition of operator liability.’” Id. at 1110 (citation omitted). As is clear from the language of these cases, and as the Kayser-Roth court emphasized, it would “obviously not [be] the usual ease that the parent of a wholly owned subsidiary is an operator of the subsidiary.” Kayser-Roth, 910 F.2d at 27; see John S. Boyd, 992 F.2d at 408.
Among the many courts that have recognized that a parent corporation may be directly liable as an operator under section 107(a)(2), there has been a difference of opinion “as to whether operator liability should be predicated on the actual control one corporation has over the other, or whether the corporation’s capacity or authority to control is sufficient.” Lansford-Coaldale, 4 F.3d at 1220. The minority view has been adopted by the Fourth and Ninth Circuits, which would hold a parent corporation directly liable as an operator not because the parent corporation actually controlled the subsidiary, but rather because it had the authority to do so. See, e.g., Nurad, 966 F.2d at 842; Kaiser Aluminum, 976 F.2d at 1341. Those courts have justified their test as “one which properly declines to absolve from CERCLA liability a party who possessed the authority to abate the damage caused by the disposal of hazardous substances but who declined to actually exercise that authority by undertaking efforts at a cleanup.” Nurad, 966 F.2d at 842.
This view, I think, goes too far, because it assigns liability to the parent corporations beyond that intended by Congress. But in holding, as the majority of circuits have, that a parent corporation may be liable when it actually operates the facility in question, a court simply faithfully interprets plain statutory language.
Only the Fifth Circuit has adopted the limited view of operator liability described in the majority opinion here, in which the corporate form of the alleged operator is dispositive. The Joslyn court framed its inquiry as being whether to “impose direct liability on parent corporation for the violations of their wholly owned subsidiaries.” Joslyn, 893 F.2d at 81. This somewhat disingenuous framing of the question ignores the direct link contemplated by the majority *590of other courts that have discussed operator liability, and suggests that operator liability for parent corporations, like owner liability, would simply be another type of derivative liability, albeit of a less demanding nature than corporate-veil-piercing. The Joslyn court reasoned that because “CERCLA does not [explicitly] define ‘owners’ or ‘operators’ as including the parent company of offending wholly-owned subsidiaries,” id. at 82, there can be no operator liability for parents in the absence of veil-piercing. This analysis completely misses the point, which is that since a parent corporation is certainly within the statutory definition of “person,” the relevant question is simply whether the particular corporation was, in fact, an operator of the facility in question. Thus, the Joslyn court’s invitation to Congress to “extend liability to parent corporations” if it wishes to do so, id. at 83, is simply meaningless. It is obvious that CERCLA does extend liability to all corporations, including parent corporations, so long as they satisfy the statutory prerequisite that they be operators. I note, too, that another panel of the Fifth Circuit, in a case decided after Joslyn, observed in dicta that individual shareholders or officers could be held directly liable as an operator, when “they themselves actually participate in the wrongful conduct prohibited by the Act,” and recognized that “this personal liability is distinct from the derivative liability that results from ‘piercing the corporate veil.’” Riverside Mkt. Dev. Corp. v. International Bldg. Prods., Inc., 931 F.2d 327, 330 (5th Cir.1991). The Riverside holding indicates, at a minimum, a reluctance within the Fifth Circuit to broadly apply Joslyn.
In sum, I conclude, as the district court did, that a parent corporation may be held directly liable as an operator of a contaminating facility under section 107(a)(2) if the facts of the case show that its domination and control of the subsidiary corporation ostensibly operating the facility is so pervasive that the parent is the operator in fact.
B.
CPC’s Liability Under Section 107(a)(2)
In addition to generally repudiating the possibility of direct operator liability for parent corporations, which has been adopted in a majority of circuits considering the question, my colleagues have a factual dispute with the district court. That is, given my conclusion that CERCLA plainly contemplates the possibility of direct operator liability for parent corporations, the analysis necessarily turns to whether such liability is appropriately imposed here: whether CPC, as a factual matter, operated the facility. My colleagues acknowledge that, arguendo, “where the parent directly operates the facility itself, either independently of its subsidiary, or as an actual co-operator alongside the subsidiary,” liability would be appropriate, but they assert that this theory “is not the one relied upon by the district court and, in any event, is not supported by the facts in the record before us.” (Maj. op. at 581.) This assertion simply belies the record and the district court’s opinion, and further, it fails to recognize that our review on this point is limited to review for clear error. Lansford-Coaldale, 4 F.3d at 1219; John S. Boyd, 992 F.2d at 408; Kayser-Roth, 910 F.2d at 27.
The district court found that CPC, Ott II’s parent corporation, so totally and completely controlled Ott II that CPC was the actual operator of the contaminating facility in Dalton Township, Michigan. There is an abundance of evidence in the record to support this factual finding, and in all events, not even my colleagues claim it is clearly erroneous. It follows, therefore, as the district court found, that CPC is directly hable under section 107(a)(2) as the operator of the site. I agree.
Section 107(a)(2) provides that a parent corporation is hable if it is the operator-in-fact of the facility as indicated by the extent of its domination and control of the subsidiary, the ostensible operator. The district court pointed out that some of the factors relevant to deciding whether the parent is the operator-in-fact include considerations such as the parent corporation’s involvement in the subsidiary’s board of directors and daily operations, and the parent corporation’s control over the subsidiary’s pohcy making in areas such as personnel, finance, and waste *591disposal. Also relevant are the facts leading up to the subsidiary’s origin and the reasons for its existence, and the parent’s level of financial monitoring and its cooperation or consolidation with the subsidiary’s accounting, legal, and research functions.
The specific facts with regard to CPC, as found by the district court, are these:
From CPC’s acquisition of Ott II in October 1965 through April 1966, all four directors on Ott II’s board were CPC officers. Over the next three and one-half years, at least three of the eight board members were officers of CPC, and for the following two and one-half years, until CPC sold Ott II, CPC officials comprised the majority of the then eleven-director board. At all times during Ott II’s existence, the chairman of its board was a high-level CPC executive, appointed by CPC’s president. In addition, the managers of Ott II who exerted active control over the subsidiary’s day-to-day activities also were officers of CPC. Arnold Ott, who had been chief executive officer of Ott I, continued as Ott II’s chief executive officer until 1969, during which time he also was CPC’s vice president for scientific research and president of CPC’s development company, a subsidiary with oversight responsibility for several CPC subsidiaries including Ott II. In addition, James Eiszner, who had been Ott I’s vice president of marketing, served as Ott II’s president from 1967 to 1970. During Eiszner’s tenure with Ott II, he also served as vice president of CPC’s development company, and eventually became CPC’s chief executive officer. Moreover, Eiszner, in particular, was criticized during his tenure as an Ott II official for paying too much attention to his CPC responsibilities and not enough attention to Ott II. Beverly Warner served as Ott II’s chief executive officer from 1970 until it was sold in 1972, at the same time serving as president of CPC’s development company.
Not only did CPC participate in Ott II’s environmental matters via the CPC’s environmental affairs director, G.R.D. Williams, but it was instrumental in setting Ott II’s environmental policies. For example, because Williams did not believe Ott II needed a biological waste treatment facility, Ott II officers abandoned presenting plans for such a facility at a meeting with the state of Michigan. Moreover, Williams repeatedly controlled the interaction between Ott II officials and state and federal regulators, and instructed Ott II’s officers “to consult with CPC before responding to regulatory questionnaires or other inquiries.” CPC Int’l, Inc. v. Aerojet-General Corp., 777 F.Supp. 549, 561 (W.D.Mich.1991). Finally, as the district court found, CPC’s involvement in Ott II’s financial affairs involved more than mere review and oversight. For example, CPC made loans to Ott II in excess of $5 million, while assuming many of Ott II’s existing loans. In addition, CPC commingled its funds with Ott II’s funds. CPC also limited the amount of capital expenditures that Ott II could approve without further approval by CPC’s board of directors. As to personnel matters, CPC repeatedly participated in Ott II’s labor negotiations with local unions.
Based on these and many other findings, the district court concluded that “[t]he evidence shows a level of participation and control by CPC that exceeds the bounds of an interested investor and enters the realm of an active operator.” Id. at 575. These findings are amply supported in the record. As I have said, my colleagues do not claim they are clearly erroneous. Accordingly, the district court’s conclusion that CPC was directly liable under section 107(a)(2) as an operator should be affirmed.
II.
Aerojet’s Liability Under Section 107(a)(1)
The district court found Aerojet hable as a present “owner” of a contaminated facility under section 107(a)(1). It did so after finding that the facts justified piercing the corporate veil that thinly shielded Cordova/Michigan from Aerojet. My colleagues hold that the district court erred because it misapplied Michigan law: “[The district court’s findings do] not suggest that the company acted to subvert justice or with fraudulent intent or otherwise sought to distort the legitimate *592purposes of the corporate form.” (Maj. op. at 582.)
While there is no question that fraudulent intent or a subversion of justice justifies piercing the corporate veil, there is ample authority under Michigan law for finding parent corporation liability through veil piercing for a less egregious reason, such as unjustified use of the corporate form. Indeed, the Michigan Supreme Court has only recently restated the applicable rule:
Michigan law presumes that, absent some abuse of corporate form, parent and subsidiary corporations are separate and distinct entities. This presumption, often referred to as a “corporate veil,” may be pierced only where an otherwise separate corporate existence has been used to “subvert justice or cause a result that [is] contrary to some other clearly overriding public policy.” More specifically, Michigan courts have generally required that a subsidiary must “become ‘a mere instrumentality’ of the parent” before its separate corporate existence will be disregarded.
Seasword v. Hilti, Inc., 449 Mich. 542, 537 N.W.2d 221, 224 (1995) (emphasis added) (citations omitted); see Wells v. Firestone Tire and Rubber Co., 421 Mich. 641, 364 N.W.2d 670, 675 (1984); Herman v. Mobile Homes Corp., 317 Mich. 233, 26 N.W.2d 757, 762-63 (1947); Potter v. Michigan Bell Tel. Co., 246 Mich. 198, 224 N.W. 438, 439 (1929). But the most instructive language of Seasword, for purposes of this case, is in the court’s illustrative listing of “relevant factors ... showing that a subsidiary is a ‘mere instrumentality’ of its parent”:
[T]he parent and subsidiary shared principal offices, or had interlocking boards of directors or frequent interchanges of employees, that the subsidiary is the parent’s exclusive distributing arm, or the parent’s revenues are entirely derived from sales by the subsidiary.
Seasword, 537 N.W.2d at 224 n. 10.
The coincidence between these hypothetical factors and the reality of Aerojet’s relationship with its subsidiaries is, to put it mildly, striking. The district court found that Aerojet was the 100% shareholder of Cordova/California; that Cordova/California, in turn, was the sole shareholder of Cordova/Michigan; and that the boards of directors of Cordova/California and Cordova/Michigan were titular only, not even convening for meetings. At least twenty Aerojet officers simultaneously held the same or nearly identical positions in Cordova/Califomia and Cordova/Michigan. Aerojet so completely controlled the finances of all companies that neither Cordova/California nor Cordova/Michigan were permitted to maintain separate bank accounts. In addition, there was evidence that Aerojet used Cordova/Michigan by transferring to Cordova/California millions in worthless debt owed to Aerojet by Cordova/Michigan, effectively canceling debt owed by Aerojet to Cordova/California. These findings are supported by the record and are not clearly erroneous. The evidence established that Cordova/Michigan operated as a “mere instrumentality” of Aerojet.
Even more compelling are the findings of the district court regarding Cordova/Michigan’s corporate purpose. When Aerojet began negotiations with MDNR for the Dalton Township property, Aerojet negotiated side-by-side with its then-unincorporated division, Cordova. After Aerojet entered two short-term stipulations with MDNR, and merely eleven days before the sale was concluded, Aerojet incorporated Cordova as a wholly-owned subsidiary. Although Aerojet had initially drafted the stipulation and consent order with MDNR, it was “Cordova Chemical Company” that actually signed the agreement. Then, in November 1978, with the remodeling of the facility complete and manufacturing about to begin, Cordova/California incorporated Cordova/Michigan, transferring to it Cordova/Califomia’s ownership of the facility. Despite the separate corporate form of Cordova/Michigan, throughout operations, Aerojet actively participated in negotiations with prospective buyers for the possible sale of the facility. Once Cordova/Michigan ceased operations at the site, it was Aerojet that took responsibility for leasing portions of the site to third parties. It is clear that Aerojet took pains to insulate itself from environmental liability for the situation they *593knew existed at the site. Aerojet admits as much in its brief:
By using well-capitalized, non-fraudulent, separate corporate subsidiaries, such as Cordova/California and then Cordova/Michigan, Aerojet could justify an attempt to reclaim and make the waste Site productive without risking all of its corporate assets. A rule of law imposing enormous environmental liability on parent corporations whose subsidiary neither perpetrated a fraud nor contributed to actual contamination would result in contaminated waste sites being permanently abandoned as unproductive, orphan properties, because no rational corporate officer could support a decision to rehabilitate a contaminated site if such liability were unavoidable.
Thus, Aerojet admits that Cordova/Michigan was established solely as a facade, to avoid any legal obligation to pay for further environmental cleanup at the site. Under Michigan law, its admission is sufficient to justify piercing the corporate veil. See Potter, 224 N.W. at 440. As Aerojet points out, it is possible that a refusal to allow a prospective purchaser of a contaminated site to avoid liability will result in a scarcity of willing buyers. Certainly, both EPA and MDNR have a substantial interest in locating conscientious purchasers, who are willing to reclaim environmentally corrupt facilities. However, there is no evidence that the Michigan courts would view this interest as an exception to the state’s veil-piercing standard, especially in light of the competing interest in imposing environmental cleanup costs on private industry rather than on taxpayers. Congress certainly was not deterred by this argument, given its balancing of interests in favor of imposing liability on new owners. Moreover, Aerojet and the Cordovas are not blameless, as they would have the court believe. The district court found that the entities actively contributed to the contamination and then failed to take remedial action, despite knowledge that contamination was continuing to migrate.
Accordingly, the district court’s conclusion that, by piercing the corporate veil, Aerojet may be held hable as an owner under 42 U.S.C. § 9607(a)(1) should be affirmed.
I further agree with the district court that Aerojet was directly hable as an operator under section 107(a)(2):
In light of the same facts that were probative in concluding Aerojet is hable under section 107(a)(1), the court concludes that Aerojet operated the site through active participation and pervasive control over the businesses of both Cordova/California and Cordova/Michigan.
As with CPC’s involvement with Ott II, Aerojet’s participation and control over the board, management and decision-making at Cordova/California and Cordova/Michigan shows that the parent operated the facility. Aerojet’s conduct toward its subsidiaries extended well beyond the activities that are merely indicative of a parent’s general oversight of a wholly owned subsidiary.
Accordingly, the court concludes the [sic] Aerojet is directly hable as an operator under section 107(a)(2).
CPC Int’l, Inc., 777 F.Supp. at 580.
III.
Third-Party Defense Under Section 107(b)(3)
Finally, I do not agree that the district court should be required, upon remand, to “revisit its treatment of the [third-party] defense raised by Aerojet, Cordova/California, and Cordova/Michigan” under section 107(b)(3). (Maj. op. at 582-83.) To succeed under section 107(b)(3), the defendants are required to prove all four elements of a third-party defense, which are:
1. That they did not contribute to the contamination;
2. That they were not in a direct or indirect contractual relationship with any person who, in connection with the contractual relationship, caused the contamination;
3. That they exercised due care throughout their ownership or operation of the contaminating facility; and
4. That they protected against those acts and omissions of the polluting persons, *594and the consequences of those acts and omissions, that were foreseeable.
See 42 U.S.C. § 9607(b)(3); see also Kerr-McGee Chem. Corp. v. Lefton Iron & Metal Co., 14 F.3d 321, 325 (7th Cir.1994).
The district court found that the defendants failed to carry their burden on all four elements of the section (b)(3) defense. My colleagues, in desiring that the district court “revisit its treatment of the defense” raised by the defendants, address only the “contractual relationship” ground. Even if the district court erred in its treatment of that element of the defense, its findings that the defendants have not carried their burden of proving the remaining three elements of the section 107(b)(3) defense, which my colleagues do not question, is amply supported in the record.
The district court committed no clear error in finding that the defendants demonstrated neither the exercise of due care nor the use of appropriate precautions. For example, in a May 1980 environmental risk report covering Aerojet and several subsidiaries including Cordova/Michigan, the defendants repeatedly stated their intent to take a head-in-the-sand approach with respect to the contamination problem:
The most significant environmental problems associated with the facility are those relating to the residues of past industrial occupants of the site. Management has adopted the position that any injury to others arising out of contamination from these residues is the responsibility of the State of Michigan Department of Natural Resources and that, accordingly, Cordova should insulate itself from any knowledge of, or involvement in monitoring these wastes.
The report went on to acknowledge that waste drums remained buried, despite completion of MDNR’s removal efforts. Moreover, the report acknowledged that the stipulation entered into by Cordova and MDNR was never intended to resolve the contamination problem. In describing the responsibilities assumed by the parties under the stipulation, the report admitted that MDNR’s duties were limited to removing 8700 drums and a portion of the contaminated soil and sludge, and that the stipulation’s hold-harmless clause relieved the defendants only of liability arising out of these specified removal efforts. In addition, as the report acknowledged, the MDNR absolved the defendants of liability relating to procurement of an alternative community water source.
As the district court pointed out, and as the report confirms, the stipulation did not resolve responsibility for the remaining drums and contaminated soil and sludge, as well as responsibility for groundwater contamination. In this regard, the report specifically opined:
Between 65 and 100 monitoring wells for testing groundwater were either installed by Story or have been installed more recently under a State/Federal study of groundwater contamination at the site. The study is being carried out by several consulting organizations under contract to the state. Cordova management believes the studies show little or no contamination but has avoided any participation or liaison with the study teams____ Because of the possibility that Cordova’s potential liability for groundwater contamination may have survived the Consent Order, it would appear desirable for Cordova management to keep abreast of current monitoring results. In addition, although a high chloride content would show continuation of problems from the old Story wastes, a high sulphate concentration would indicate seepage problems arising out of Cordova’s current operations.
Fully aware that waste drums remained buried beneath the site and that the majority of the contaminated soil had not been removed, and cognizant of groundwater contamination to which they may have been contributing, the defendants believed the solution to these problems was to don blinders. Their willful blindness can hardly be characterized as the exercise of due care.
For the same reasons, it cannot be said that the defendants took adequate precautions to protect against the consequences of Ott II’s and Story’s omissions and acts. While the parties, in their briefs, debate whether reimplementation of the purge wells would have been an adequate precau*595tion, they overlook the big picture. The defendants, fully aware that contamination problems on their property were not being addressed, chose to take no precautions to protect against the foreseeable consequences of these problems — namely, further migration.
Accordingly, because the defendants have failed to prove at least two of the requisite elements of the third-party defense, the district court properly held that they were not entitled to invoke it. It is simply unnecessary to consider whether they sustained their burden regarding the remaining two elements, including whether any of the pollution was the act of a third party “in connection with” the contractual relationship with the defendants.
IV.
The judgment of the district court should be affirmed.