DECISION AND ORDER

WILLIAM C. GRIESBACH, District Judge.

On July 5, 2011, this Court denied the government's motion for a preliminary injunction against Defendants NCR and Appleton Papers Inc. In doing so, I found that although the government had set forth grounds for relief against NCR, it had not done so against Appleton Papers because it was unlikely that Appleton Papers had successor liability under CERCLA. Appleton Papers ("API") has now moved for summary judgment on the issue of liability. For the reasons given below, I will deny the motion.

I. Successor Liability when the Seller Survives the Transaction

The issue of successor liability has already received significant treatment in this Court's denial of preliminary relief to the government. In sum, I have concluded that although API may have agreed to indemnify NCR as a part of its purchase of some of NCR's assets (more on this below), that agreement did not constitute a "successorship" to liability because (among other reasons) NCR continued to remain in business. Without an actual "succession," I concluded there could be no successor liability. In addition, I noted that CERCLA § 107(e)(1) precluded parties' efforts to shift liability to other entities. That statutory bar on the transfer of liability supported my conclusion that there was no succession here. Finally, I found that the equitable purposes of the successor liability doctrine were directed toward preventing the kind of fraud or injustice that would result if a liable entity were allowed to shirk its liability through a liability-shifting transaction. Because that did not occur here (given that NCR continued to remain liable), I found the successorship doctrine inapplicable. Having already effectively ruled in favor of API, I will focus my attention on the government's arguments that my preliminary conclusion was incorrect.

To recall, in the common law there are four established ways in which a purchaser of assets can be deemed a successor: "(1) Where there is an express or implied agreement of assumption; (2) where the transaction amounts to a consolidation or merger of the purchaser or seller corporation; (3) where the purchaser is merely a continuation of the seller; or (4) where the transaction is for the fraudulent purpose of escaping liability for the seller's obligation." (Moriarty v. Svec, 164 F.3d 323, 327 (7th Cir.1998) (quoting Vernon v. Schuster, 179 Ill.2d 338, 228 Ill.Dec. 195, 688 N.E.2d 1172, 1175 (1997))). Here, we are dealing only with the first of these, the express agreement of assumption. The principal focus of the government's present effort is its argument that NCR's continued existence as a viable company is irrelevant to the successorship analysis. It concedes that the seller's continued existence might preclude some ways of establishing successor liability, such as in the application of the "mere continuation" doctrine or a de facto merger. For example, if Company A sold some assets to Company B, it would be difficult to conclude that the transaction was actually a consolidation or merger of the two businesses if Company A remained a viable going concern. In that case, we would not view Company B as a successor.

But the government claims that the seller's continued existence is not relevant when successor liability is premised on an explicit agreement of liability assumption. If the buyer has signed an agreement explicitly stating that it is assuming the liabilities of the seller, then the concerns about continuity, fraud and de facto merger go by the wayside because the parties have saved us the trouble by negotiating successor liability as a matter of contract, a contract to which the public is a third-party beneficiary.

This argument faces at least two hurdles. First, as has been noted already, "the purpose of corporate successor liability is to prevent corporations from evading their liabilities through changes of ownership." United States v. Mexico Feed & Seed Co., 980 F.2d 478, 487 (8th Cir.1992). Here, because NCR remains viable, there has been no effort to "evade" liability. No one has disputed that NCR is a solvent corporation that has the ability to pay any judgment here, and neither has anyone suggested that the purpose of the asset sale to API was to shirk liability. Thus, the case, on its face, does not cry out for application of the successorship doctrine.

A second, and possibly related, problem is that there is almost no precedent for finding successor liability when the seller has remained a viable entity. For example, in United States v. Iron Mountain Mines, Inc., which the United States cites, the court found successor liability based on two assumption agreements. 987 F.Supp. 1233 (E.D.Cal.1997). But there, the predecessor companies had been dissolved. In fact, the court concluded that the requirements for a de facto merger were likely met, although it did not reach that issue given the existence of the express assumption agreement. Id. at 1242 n. 19. Except for an unpublished case decided more than two decades ago, it does not appear that successor liability has been found under the circumstances we have here. See United States v. Chrysler Corp., 1990 WL 127160, *4-7 (D.Del.1990) (finding assumption of liability through agreement even though selling company remained in-tact).

Despite these hurdles, I am satisfied that API may be deemed a successor to the liability of NCR even though NCR itself remains liable to the government. First, all of the CERCLA cases addressing successor liability recognize that a company may become liable as a successor by expressly agreeing to become liable. Moriarty v. Svec, 164 F.3d at 327. These cases do not condition liability-by-agreement on the non-existence of the selling corporation. Perhaps the problem is that assumption of CERCLA liability through an agreement is not actually a "succession," because in common parlance a succession implies that the successor has assumed liability in lieu of the transferor. Instead of talking about succession, it might be clearer to state that a party may assume direct CERCLA liability by agreement even though it may not "succeed" to it in the traditional understanding of the term. In any event, CERCLA case law is clear that parties may assume liability through agreement (though they may not transfer it away), and none of the cases requires that an assumption is only valid if the seller ceases to exist. Accordingly, the fact that NCR continues to be liable should not be an obstacle to finding API liable.

A second consideration motivating my denial of preliminary relief was the fact that CERCLA explicitly prevents parties from shifting liability to other entities. 42 U.S.C. § 9607(e)(1). "Although the law of corporate succession contemplates that corporate parties may allocate liabilities in an asset sale, CERCLA § 107(e)(1) nullifies any attempted transfer of CERCLA liability." A.C. Reorganization v. Dupont, 1997 WL 381962, *7 (E.D.Wis.1997). If CERCLA invalidates such attempts to transfer, API argued, then how could the direct liability of NCR be assumed by API (even if that were the parties' intent)? Section 107(e)(1)'s prohibition on liability transfer appeared to bolster my conclusion that API was not a successor.

The government now persuasively argues that although § 107(e)(1) would preclude a party from eliminating liability through a liability transfer agreement, it does not preclude parties from creating additional liability, in effect, on the part of the buyer or anyone else. The court in A.C. Reorganization cited Harley — Davidson, Inc. v. Minstar, Inc. for the principle that CERCLA precludes efforts to divest liability. 41 F.3d 341, 342 (7th Cir.1994). But that is not the same as saying that CERCLA prohibits a non-liable party from entering an agreement to take on direct liability in addition to that of the already-liable party: the only condition CERCLA imposes is that the directly liable party must remain liable. It might be argued that such an agreement would have no purpose: if the seller cannot escape direct liability, then what is the point of "transferring" liability to the buyer if the seller still remains on the hook for that liability? It is true that the primary value in such an arrangement might manifest itself in the indemnification provision that makes the buyer compensate the seller for any liability, rather than the assumption of direct liability itself. But by making the buyer itself directly liable, the seller has obtained something of value: an additional defendant to help share the burden of defense. Where there was one defendant there are now two. Although this might not be the overarching purpose of such an arrangement, neither is it a trifle.1 The point is that not only is an assumption agreement allowed by CERCLA (subject to the conditions noted above), such an agreement also makes commercial sense. Thus, CERCLA's bar on the transfer of liability does not preclude a finding that API could be liable in addition to NCR.

With those considerations favoring the imposition of successor liability, the only remaining problem is that the purpose underlying the successor liability doctrine does not appear to apply here. To recall, the cases tell us that the successor liability doctrine is intended to prevent corporations from dying "paper deaths" only to reemerge having eliminated their environmental liability. United States v. Mexico Feed and Seed Co., Inc., 980 F.2d 478, 487 (8th Cir.1992). Obviously such a concern is absent here. But just as the continued existence of the seller does not matter in an assumption-by-agreement case, the concerns underlying the successor liability doctrine fall by the wayside as well. As noted above, the problem may be one of nomenclature: the "succession" doctrine applies when a buyer has actually succeeded (exclusively) to the liabilities of the seller, who has either been dissolved or merged in some fashion. In these circumstances, courts rightly view the doctrine as a means of preventing the injustice that would occur if a liable entity were allowed to simply paper away that liability through private agreements. But when a buyer expressly agrees to assume the environmental liabilities of the seller, we need not worry about such concerns because shirking environmental liability — something CERCLA prohibits — was never the intent of the agreement in the first place. In a case like this, it is not so much that a "doctrine" of successor liability is being applied as it is simply a matter of enforcing an explicit contract that created additional liability. Viewed in that light, the fact that the concerns underlying the succession doctrine are absent in an assumption-by-agreement case is not all that surprising.

In sum, I do not believe the hurdles identified above suffice to preclude enforcement of an explicit agreement to assume CERCLA liability. Nothing in federal common law requires that the seller cease to exist before another entity may assume its CERCLA liability. And the concerns underlying the other "successor" liability doctrines, such as de facto merger or mere continuation, are not relevant when two companies have explicitly agreed that the buyer will become liable along with the seller. Finally, nothing within CERCLA itself invalidates an attempt to create additional CERCLA liability, so long as the agreement does not purport to transfer that liability. Thus, the next question is whether the agreement signed by NCR and API actually does operate to create direct liability on the part of API.

II. The 1978 Agreement

In 1978 NCR sold its Appleton Papers Division to API's predecessor, a company called Lentheric, Inc., hereinafter referred to simply as API. As part of the asset purchase agreement memorializing that transaction, API agreed to assume several liabilities and to indemnify NCR as follows:

(Dkt. 139, Ex. 3 at 19-21.)

Two of these assumption of liability clauses refer to Schedule A of the asset purchase agreement. As relevant here, Schedule A contains the following clause:

(Dkt. # 195, Ex. 3 at 4.)

The United States argues that each of these clauses shows that API's predecessor explicitly assumed liability for the cost of environmental cleanup at issue in this action. The question is whether the language in the asset purchase agreement is broad enough to encompass the liability at issue here even though neither CERCLA nor the extent of the PCB problem had been in the minds of the parties at the time the contract was signed.

"A party may indemnify another party for liability arising out of a law not in existence at the time of contracting." Kerr-McGee Chemical Corp. v. Lefton Iron & Metal Co., 14 F.3d 321, 327 (7th Cir.1994). But when that happens the parties cannot be said to have a meeting of the minds as to the specific liability at issue. Instead, if the parties have expressed a meeting of the minds, their agreement goes to the division of liability when some unforeseen liability emerges; in other words, the parties may contract to shift the risk of the unknown. In Kerr-McGee, for example, the court found a pre-CERCLA agreement broad enough to encompass CERCLA liability when the indemnitor had agreed to pay for "the maintenance of any action, claim or order concerning pollution or nuisance." 14 F.3d 321, 327 (7th Cir.1994). Given the breadth of this language ("any action ... concerning pollution") the court had no trouble concluding that the indemnitor had indeed agreed to pay for the CERCLA liability in question even though such liability was not specifically envisioned by the parties when the contract was signed.

Similarly, in Olin Corp. v. Consolidated Aluminum Corp., the Second Circuit noted that indemnification agreements are interpreted strictly under New York law (which applies here at the agreement of the parties). Even so, and despite the fact that CERCLA had not been enacted at the time of the agreement in question, the court found in the parties' agreement an intent to indemnify for CERCLA liability:

5 F.3d 10, 15 (2d Cir.1993).

Both of these cases reflect the scenario in which although the parties did not — could not — know about CERCLA liability, they made a business decision to shift the risk of the unknown (both "known unknowns" and "unknown unknowns," of which CERCLA was likely the latter) from party A to party B.

Although two of the assumption clauses in this case are similar, § 1.4.4 appears to be broader than § 1.4.5 because it includes "any state of facts" or "matters," whereas the latter section merely applies to governmental investigations or claims. Either way, the question is whether the CERCLA liability at issue here arises out of any "matters," etc., or governmental actions disclosed in Schedule A. (Schedule M is not relevant here.) As noted earlier, the relevant portion of Schedule A reads as follows:

(Dkt. # 195, Ex. 3 at 4.)

The second paragraph just quoted indicates not only that APD had received notices from governmental authorities in the past, but that it continued to do so at the time of the asset purchase. These notices not only had resulted in fines and corrective action but "may result" in fines and corrective action in the future. That is a "matter" disclosed in Schedule A. When API received notices from the EPA that it was a PRP for the Fox River PCB problem, that was a notice from a federal authority claiming violation of environmental laws. Although the PCB problem was not the subject of any notices APD had received as of 1978, the clause is broad enough to indicate that the division generally received such notices and would continue to do so in the future. Such notices "may result" (as they did here) in corrective action. By disclosing the division's proclivity for receiving corrective notices from the government, the seller was alerting the buyer that this was an issue for which it was accepting responsibility, both for problems already disclosed and for those that had not yet arisen. It would therefore not be difficult to conclude that liability for the PCB problem arose out of a matter disclosed on Schedule A.

Although I may be able to reach that conclusion from the asset purchase agreement alone, the parties recognize that the purchase agreement may no longer be read in a vacuum. In 1995 NCR sued API in the Southern District of New York to resolve liability for the PCB cleanup. The parties ultimately reached a two-part settlement of that issue. First, they agreed to split the first $75 million of any liability between them. Second, they agreed to submit the question to binding arbitration with respect to any amounts in excess of $75 million.

The arbitration panel concluded that API was liable for 60% and NCR 40% of any expenses in excess of $75 million. The panel found that the contractual language "is not sufficiently clear and unambiguous with respect to the issue of responsibility for the environmental costs at issue to permit an award based solely on the contract language." (Dkt. # 208, Ex. 1 at 4.) Based on testimony and documents produced to the panel, however, the panel decided to impose a larger share of liability upon API.

Although the panel did not make its decision solely on the basis of the contract, the parties had agreed that the arbitrators would settle the question of liability for cleanup expenses once and for all. By agreeing to have the matter resolved by arbitration, the arbitration, in effect, altered the terms of the original purchase agreement. The entirety of the agreement is thus the product of the arbitration, which imposed liability upon API. In fact, given the agreed structure and mission of the arbitration, which guaranteed that API would be found liable for some part of the expenses, API had essentially already conceded liability just by virtue of entering into the arbitration.

API protests that in entering into the settlement agreement that led to arbitration, API and NCR had explicitly agreed that neither party was admitting liability of any kind. (Dkt. # 124, Ex. 1.) Although that may be true, that was simply an agreement between those private parties that neither of them were conceding liability. That does not mean, however, that one of the signatories could not assume the liability that another party was actually found to have, regardless of the lack of any concessions or admissions. If NCR were ultimately found to be liable under CERCLA, as it has been, API could have agreed to assume that liability. In other words, the fact that neither one was conceding it was liable in 1978 has no impact on whether or not one party assumed the liability that the other actually was later found to have (irrespective of any concessions). For these reasons, I am satisfied that the purchase agreement, as interpreted and applied by the arbitration panel, could impose liability upon API. Accordingly, API's motion for summary judgment will be denied.

III. Conclusion

For the reasons given above, I conclude that the government is correct that the continued existence and liability of NCR does not preclude a finding that API assumed CERCLA liability. Moreover, I further conclude that the terms of the 1978 assumption agreement, as applied in the parties' arbitration, are broad enough that they could encompass that liability. Accordingly, API's motion for summary judgment is DENIED.

DECISION GRANTING MOTION FOR RECONSIDERATION

On July 5, 2011, this Court denied the government's motion for a preliminary injunction against Defendants NCR and Appleton Papers Inc. ("API") on the basis that the government was unlikely to prove Appleton Papers was a liable party. In a December 19, 2011 decision, I denied API's motion for summary judgment on that point and instead accepted the government's argument that it appeared API had in fact agreed by contract to assume CERCLA liability when it purchased the Appleton Papers Division from NCR. API soon moved to reconsider, citing several points of error. For the reasons given below, I agree with API to the extent that the purchase agreement in question was not drafted broadly enough to encompass API's direct liability for the CERCLA liability at issue in this case. The motion for reconsideration will therefore be granted in part.

To recall, the result reached in my December 19, 2011 Decision and Order was based on two discrete conclusions. First, I concluded that NCR's continued existence did not per se preclude an agreement to create liability to a third party such as the government. Although it was not a traditional "successorship" liability situation (because NCR did not go out of business), I found that there was nothing within CERCLA that would preclude parties, as a matter of contract, from effectively creating additional liability under CERCLA (although presumably that would be a rare scenario). The second aspect of my decision required interpretation of the purchase agreement, between NCR and API's predecessor, for NCR's Appleton Papers Division. I concluded that the language and matters disclosed in Schedule A to that agreement were broad enough to encompass the CERCLA liability at issue in this case. Because I now conclude otherwise, I do not address the first prong of that decision, which is now moot.

I. The 1978 Agreement

In 1978 NCR sold its Appleton Papers Division to API's predecessor, which for ease of understanding will be referred to simply as API. A key clause in that agreement provided that API agreed to "assume, pay, perform, defend and discharge... all of Seller's obligations and liabilities of any kind, character or description relating to the period subsequent to the Closing Date which arise out of or in respect of any state of facts, matter, event or disclosure set forth on an attachment to the agreement that was designated as Schedule A." (Section 1.4.4; Dkt. # 139, Ex. 3 at 19.)

Schedule A contains the following clause:

(Dkt. # 195, Ex. 3 at 4.)

In my previous decision, I concluded that because NCR had disclosed that the Appleton Papers Division "receives" (present tense) notice of various environmental violations, which "may result" (in the future) in corrective action, the buyer was accepting liability for the Division's "proclivity" for environmental violations. In other words, the buyer was on the hook not just for specific past violations but for any future environmental issues as well.

A. The present liability does not arise from any "violation" of law or any "compliance" issue

I am now convinced that there are at least two problems with the approach taken at the summary judgment stage. First, the clauses that trigger liability require notice of the "violation" of environmental laws and standards, as well as past and potential fines for those violations. API notes that no one has argued that the PCB pollution at issue in this case was the product of any legal or regulatory violations. CERCLA did not yet exist in 1978, of course, and the PCBs were released into the environment primarily in the 1960s before they were regulated. In fact, as noted in the parallel contribution action, No. 08-C-16, in more recent years PCBs were released in smaller quantities at least partly with governmental acquiescence due to the difficulty in separating them from recyclable paper.

In response, the government now attempts to bolster my earlier ruling by citing, for the first time, the federal Refuse Act, as well as Wisconsin law barring the unauthorized disposition of refuse into waterways. 33 U.S.C. § 407. It also cites the Clean Water Act, 33 U.S.C. § 1311(a), which regulates discharge of pollutants. Because the Appleton Papers Division may have been operating in violation of those laws, the government argues, that should trigger the liability clause in Schedule A.

API notes that these are new arguments and, as such, are waived. Even if the argument is not waived, however, the violations of various environmental laws now alleged would not be enough to create CERCLA liability under the terms of the 1978 agreement. To recall, Schedule A discloses that the Division may have been operating in violation of various environmental laws and regulations, and thus the buyer would be assuming any liability arising out of those violations. Here, the massive CERCLA liability at issue, which is based on the discharge of PCBs, does not "arise out of" any Clean Water Act, Refuse Act or other state or federal statutory or regulatory violations. First, for the period in question the release of PCBs was not known to be environmentally toxic, and so their release would not have given rise to any statutory or regulatory violations.1 Second, violations of laws like the Clean Water Act are not a precondition to CERCLA liability, which is strict. Put another way, the liability that CERCLA creates does not depend on any "violations" of, or compliance with, then-existing environmental laws. Thus, even if the Division had been operating in violation of, say, the Refuse Act, that does not mean that CERCLA liability "arises out of" those violations. CERCLA liability is its own creature.

Moreover, Schedule A appears to premise liability on receipt of a "notice" of a violation. Schedule A tells the buyer that the Division has received notices of noncompliance with laws and regulations and might receive similar notices in the future. This creates liability for those "claims" and anything "arising" from them. The clause underscores the fact that the liability being assumed is not open-ended environmental liability but is instead linked to specific claims and notices of environmental violations. Needless to say, the Appleton Papers Division had never received a "notice" that it was "violating" CERCLA. Had the parties wanted to draft a broader clause, it would have been much easier to simply say so in the text of the agreement itself. Instead, by drafting a schedule, the parties were clearly limiting liability to the particular circumstances disclosed therein.

The government also cites other sections of the purchase agreement (§§ 1.4.3 and 1.4.9) that create liability arising out of "compliance" with applicable environmental laws. It argues that the present CERCLA enforcement action is itself an action to assure "compliance" with CERCLA's requirement that liable parties clean up pollution sites. That is, the government appears to argue that this lawsuit is itself a trigger for liability because it is being brought to ensure compliance with CERCLA.

The claim that the government, simply by bringing a lawsuit, has the power to trigger the very liability it is seeking to enforce is a wholly circular argument. That is, the government's argument begs the question of API's liability: this is only an action to enforce "compliance" with CERCLA to the extent API is actually liable under CERCLA, and of course that is one of the centerpiece questions being adjudicated in this lawsuit. It makes little sense to argue that the very act of suing someone under CERCLA makes a defendant "non-compliant" with CERCLA. Accordingly, I cannot find that liability exists on the basis of any compliance issues.

In sum, even if the Appleton Papers Division were operating in violation of certain environmental laws, I am unable to conclude that CERCLA liability "arises out of such laws. The various environmental laws now cited by the government have their own provisions for liability and their own remedies, none of which are integral to a CERCLA action. The government has not explained how, for example, violating the Clean Water Act could make a party liable to pay for the billion-dollar cleanup of a large river. Because liability under CERCLA is distinct from these other provisions, it does not arise out of (or even relate to) those alleged violations. And, as API points out, had the parties wanted to include all environmental liability, it would have been simple enough to do so.

B. At a Minimum, the Contractual Language is Silent, Which Means No Liability

Ultimately, perhaps the most important point is that the 1978 agreement is silent about CERCLA liability and lacks a broad "catch-all" environmental liability clause. The question of API's liability has a long history. In 1995 NCR sued API in the Southern District of New York to resolve liability for the PCB cleanup. In a brief ruling, the district court concluded that it was unable to determine from the 1978 asset purchase agreement whether API had, in fact, assumed liability for PCB cleanup expenses. (Dkt. # 208, Ex. 2 at 8.) It found that the contract was negotiated before CERCLA came into existence and that none of the clauses in the asset purchase agreement was conclusive as to liability. The parties agreed to arbitrate the matter, and an arbitration panel also concluded that the agreement was unclear. The arbitration panel concluded that API pay 60% and NCR 40% of any expenses in excess of $75 million. The panel found, like the district judge, that the contractual language "is not sufficiently clear and unambiguous with respect to the issue of responsibility for the environmental costs at issue to permit an award based solely on the contract language." (Dkt. # 208, Ex. 1 at 4.) Thus, three judicial bodies (including this one) have now concluded that there is no clear language indicating that API's successor agreed to assume liability to the government for any CERCLA claims. At most, as the arbitrators found, API agreed to indemnify NCR for a portion of such liability.

The contract's silence on the point is enough to support a finding that API did not agree to assume direct CERCLA liability. In Olin Corp. v. Consolidated Aluminum Corp., the Second Circuit noted that indemnification agreements are interpreted strictly under New York law (which applies here at the agreement of the parties). 5 F.3d 10, 15 (2d Cir.1993). If an arbitration panel and another district court could not even conclude that API had agreed to indemnify NCR for its CERCLA liability, it should go without saying that the notion that API had agreed to become liable to a third party is even more tenuous. In Olin, where the Second Circuit found an assumption of liability, the district court had noted that "One would be hard pressed to draft broader or more inclusive indemnification provisions than those entered into by Conalco and Olin." 807 F.Supp. 1133, 1142 (S.D.N.Y.1992). Here, the opposite is true. The parties drafted a number of indemnification and liability clauses, but each of those clauses contains limitations linking liability to Schedule A or compliance with applicable regulations and the like. They are not the narrowest of clauses, but neither are they the "extremely broad language" at issue in Olin. 5 F.3d at 15. Ultimately, API's overarching point remains salient: if the parties had wanted to make the buyer liable for all future unknown environmental liabilities, it would have been much easier to simply use the kind of broad language used in Olin and the other cases. That lawyers and courts have spilled so much ink on the question for more than seventeen years is itself suggestive of an intent not to create CERCLA liability.

C. Given NCR's continued viability, the negating clause precludes the creation of additional CERCLA liability to third parties

API has also cited what it describes as the purchase agreement's negating clause, which in its view bars the government from seeking to enforce the contract as a third party beneficiary. That clause provides that "Nothing in this Agreement, express or implied, is intended to confer upon any other person not a party to this Agreement any rights and remedies hereunder." (Dkt. # 139, Ex. 4 at § 10.10.) The government asserts that this is not a true negating clause and, in any event, it could not operate to bar the federal government from suing a party under CERCLA.

I agree with the government in part. Specifically, I conclude that a negating clause (or "no third parties" clause) cannot be dispositive of the issue of successor liability in cases in which the seller ceases its existence. Otherwise, a simple negating clause could leave both the seller and the buyer off the hook for CERCLA liability, even if the buyer would otherwise be deemed a successor. Clearly that would not be a satisfactory result, as private parties cannot simply "contract out" of CERCLA liability to the government.

But where, as here, the seller remains in existence, we are not dealing with successorship in an equitable sense, we are dealing with successorship in a contractual sense, which means we must explore the question of the parties' contractual intentions. In that context, a negating clause is dispositive. Ultimately, the clause underscores the point made above: if the parties had intended to create liability to the government, surely they would have done so more clearly. They would not have transferred very specific environmental liabilities referenced in Schedule A and then used a negating clause to make clear that they wanted no third parties to be able to benefit from the contract. Thus, even though a negating clause could not be determinative of the issue in a traditional successorship context, I conclude here that it precludes any reading of the Agreement that would make API directly liable to the government for CERCLA-type liability.

D. No estoppel applies

The government and some of the other Defendants in this action argue that the arbitration order, which was confirmed by the New York district court, means that API is estopped from arguing that it is not liable under CERCLA. Yet the arbitration panel, like the district court, was not persuaded that API had actually assumed CERCLA liability through the asset purchase agreement. Instead, the arbitration was the product of a settlement between NCR and API, a settlement made advisable for the principal reason that API's actual liability was not crystal clear (as the district court found). In dividing responsibility 60/40 between the parties, the panel was not concluding that API was directly liable under CERCLA or that it had become a successor to NCR's liability. Instead, the 60/40 division appears to have been the result of a number of quasi-equitable factors that pointed to requiring API to bear a larger share of responsibility.

More importantly, the arbitration and the award itself were an assessment of how much each party should pay, which is an entirely different question than whether API had assumed direct CERCLA liability. No one ever posed that question to the arbitrators, and in fact it is doubtful that private arbitration could ever resolve a question involving one party's liability to the federal government. Accordingly, it would be improper to view the arbitration award as having any kind of estoppel effect on API's ability to argue that it never agreed to become directly liable under CERCLA.2

II. Conclusion

For the reasons given above, I conclude that the terms of the 1978 assumption agreement are not broad enough to encompass the CERCLA liability at issue here. Accordingly, the motion for reconsideration is GRANTED in part, and API is entitled to summary judgment that it is not a liable party under CERCLA. All claims against API are DISMISSED.

SO ORDERED