District Court Reverses Course and Rejects Divisibility Cap on Superfund Liability

October 23, 2015

A Wisconsin district court has reversed its prior ruling that NCR Corporation’s liability for contamination in the Fox River was divisible and therefore limited, holding that the evidence used to establish the divisibility defense was unreliable.

Five months after ruling that NCR Corporation’s (NCR’s) liability for contamination in the Fox River is limited to NCR’s share of contamination contributed to the river, a Wisconsin district court has reversed itself, holding the evidence that NCR used to establish its divisibility defense was unreliable.[1] The court’s decision comes after the United States and multiple other parties asked the court to reconsider its earlier finding that NCR had established that its contribution to the harm was divisible and, therefore, that NCR was not jointly and severally liable for costs to clean up the Fox River. In support of their position, the parties seeking reconsideration argued that the evidence offered to establish divisibility was speculative and therefore, the court erred in relying on it. The court agreed, holding—in its third opinion on the issue—that NCR failed to meet its burden of establishing both that the harm was theoretically capable of divisibility and that there was a reasonable basis for apportionment. Notably, however, the court left undisturbed its prior conclusion that evidence of volumetric contribution may suffice to establish the divisibility defense.

Liability Under Superfund

Liability under the Comprehensive Environmental Response, Compensation, and Liability Act (CERCLA or Superfund)[2] is, by default, joint and several, meaning that any contributor of pollution to a site can be held liable for the entire cost to clean up the site, notwithstanding other parties’ contribution to the pollution. The US Supreme Court confirmed in Burlington Northern & Santa Fe Railway Co. v. U.S., 556 U.S. 599 (2009), however, that where a potentially responsible party (PRP) carries its burden of showing that the harm is capable of being divided and damages reasonably may be apportioned among the parties, the PRP is only liable for its own contribution to the harm. Proving divisibility of harm and a reasonable basis for apportionment therefore defeats joint and several liability. Proving divisibility of harm, however, has proved difficult.

The Fox River Decisions

The case before the district court involved liability for polychlorinated biphenyl (PCB) contamination in portions of the Fox River and Green Bay, Wisconsin. Remediation of the river and bay (the “site”) is ongoing and constitutes one of the most expensive Superfund cleanups to date. The United States sued NCR, seeking to compel it to clean up the site. NCR presented a divisibility defense—that a reasonable basis for apportionment existed, and, therefore, NCR could not be held jointly and severally liable for the entire site cleanup.

After a 2012 trial, the district court held NCR had failed to prove that the harm at issue was divisible for purposes of determining liability under CERCLA, making NCR jointly and severally liable for the entire cleanup. However, on appeal, the US Court of Appeals for the Seventh Circuit took issue with the district court’s “binary” test for determining whether the harm was divisible, reversed the district court’s determination of joint and several liability, and directed the district court in assessing divisibility to focus on NCR’s contribution to the PCB contamination that caused harm to human health and the environment rather than on the propensity of NCR’s releases to require remedial action.

On remand, the district court noted that, prior to the Seventh Circuit’s decision, the prevailing practice in assessing whether harm was capable of being divided was to focus on whether the PRP’s releases triggered the need for a remedy. After the Seventh Circuit’s opinion, the court concluded that “the harm is more properly defined as a release’s toxicity or danger to human health and the environment” as opposed to the remedy or costs triggered by NCR’s releases. The court established a new framework where “the harm is the contamination,” and the second step in the analysis, the apportionment of “damages,” is to be assessed in terms of “each party’s causation of remedial response costs.” Using this framework, the court reviewed the expert testimony at trial and found NCR had demonstrated that the evidence was sufficient to establish an upper limit of how much of the contamination in the relevant section of the site could be attributed to NCR.

In reaching its conclusion, the court relied on an apportionment model developed by NCR’s expert, who, in turn, relied on contamination estimates developed by another expert, Dr. Wolfe. In its opinion finding that the harm was divisible, the court acknowledged Dr. Wolfe’s estimates were subject to multiple uncertainties, including “assumptions about releases, dredging and the like,” but dismissed those uncertainties as issues that “go with the territory” of applying the divisibility defense to a river with multiple PRPs and an imperfect historical record.

Divisibility Revisited (Again)

Upon reconsideration, the court reversed its holding that NCR had established the divisibility defense. The court determined that NCR’s reliance on Dr. Wolfe’s testimony was misplaced because Dr. Wolfe’s estimates contradicted the court’s own findings of fact concerning the contribution of PCBs to the river from another source, which called into question the estimates presented by NCR (and used by the court) of NCR’s contribution of contamination to the river. As the court explained, NCR’s “use of Wolfe’s flawed estimates [could not] be expected to produce a reliable answer to the question of how much NCR contributed to PCB concentrations in [the river].” Therefore, NCR had failed to carry its burden of establishing its contribution to the harm (i.e., that the harm was theoretically divisible).

The court also concluded that NCR failed to establish a reasonable basis for apportionment because the apportionment model developed by NCR was based on incomplete and flawed information. In particular, the court noted that NCR’s expert was missing two key estimates from Dr. Wolfe, and therefore, it was “a mystery” as to how NCR’s expert developed his apportionment of costs to NCR. Moreover, as noted above, the court considered the estimates of Dr. Wolfe that NCR did rely on to be flawed.


In its earlier opinion finding that NCR had established the divisibility defense, the district court adopted a “simplified” approach to divisibility, which focused on a PRP’s volumetric contribution. On reconsideration, the court did not deviate from its simplified volumetric analysis. The court did, however, scrutinize closely the evidence relied on to establish divisibility. In short, the court’s opinion suggests the challenges that parties face in proving divisibility may persist, even under a volumetric approach to divisibility.


If you have any questions or would like more information on the issues discussed in this LawFlash, please contact any of the following Morgan Lewis lawyers:

Denise Scofield

Los Angeles
James J. Dragna
Rick R. Rothman

New York
Judith A. Walkoff

Jeffrey N. Hurwitz
Glen R. Stuart

Stephanie R. Feingold
John McGahren

San Francisco
Greg A. Christianson

Washington, DC
Duke K. McCall, III
Michael W. Steinberg

[1]. United States v. NCR Corp., No. 10-C-910 (E.D. Wis. Oct. 19, 2015).

[2]. 28 U.S.C. § 9601 et seq.