For more than four years, the CERCLA world has watched to see if certain parties at the Lower Fox River and Green Bay Superfund Site could avoid CERCLA's  joint and several liability through the assertion of the  divisibility defense. (Fox River Litigation). In May 2015, after numerous district court opinions, a bench trial, and two trips to the Seventh Circuit Court of Appeals, the Eastern District of Wisconsin finally agreed with the NCR Corporation that its liability for a portion of the Site known as operable unit 4 (OU4) was divisible and should be limited to 28%. By proving its divisibility case, NCR hoped to exit the joint and several liability regime under which it was ordered to bear most of the clean-up costs.

That victory, however, was short lived. On October 19, 2015, the court reversed its holding in response to multiple motions for reconsideration and held instead that the expert opinions underlying NCR's divisibility arguments were unreliable, and therefore could not serve as the basis for making a finding of divisibility. 

This is unlikely to be the end of this saga, but, even at this juncture, the tortuous history of the divisibility defense in the Fox River Litigation serves as an excellent case study of the divisibility defense in the wake of the US Supreme Court's seminal decision in Burlington Northern & Santa Fe Ry. Co v. United States, 556 U.S. 599 (2009), in which the Court unanimously upheld the district court's finding that a PRP's liability at a CERCLA site was divisible. Much has been written regarding howBurlington Northern has not been the sea change that many hoped for, and others feared, regarding CERCLA joint and several liability. 

Many of the reasons for this appear in the Fox River Litigation, including: (1) courts continue to develop what a PRP must establish to prevail on a CERCLA divisibility defense; (2) a PRP bears the burden of proof and failure to meet this burden results in CERCLA's default position-joint and several liability; and (3) the divisibility defense requires rigorous fact and expert discovery. 

I.   CERCLA Divisibility

The notion that liability under CERCLA can be divisible is almost as old as CERCLA itself. As the Supreme Court recognized in Burlington Northern, as early as 1983, courts found that "Congress intended the scope of [CERCLA] liability to 'be determined from traditional and evolving principles of common law.'"1 As such, "the universal starting point for divisibility of harm analyses in CERCLA cases in § 433A of the Restatement (Second) of Torts."2 

Section 433A contains two steps. First, "whether the harm at issue is theoretically 'capable of apportionment,'" which is a "question of law" to be made by the court.3 "Second, if the harm is capable of apportionment, the fact-finder must determine how actually to apportion the damages, which is a question of fact."4 For both of these steps, "the burden remains on the party seeking apportionment . . . 'to prove[] that a reasonable basis for apportionment exists.'"5

Although simply stated, the divisibility defense has proven difficult to establish in the CERCLA context. The Fox River Litigation courts have resolved several important questions regarding the divisibility defense, creating a somewhat clearer path for future litigants seeking to present a divisibility defense, but uncertainties remain. 

A.  The "Harm" to Be Apportioned Is the Contamination Caused By Each PRP

In the early stages of the Fox River Litigation, the district court wrestled with how to define the "harm" addressed in its divisibility analysis and considered: (1) "harm as measured by remediation costs imposed by the pollution;" (2) "harm as measured by danger to the public;" and (3) "harm as measured by the amount of pollution in the sediment."6 

In 2012, the Seventh Circuit Court of Appeals held that the "harm" to be apportioned is the "'contamination traceable to each defendant.'"7 The Court recognized that for some sites (e.g., landfill cases with known waste-in totals) this would be easier than other sites where the harm is not solely to the volume of contaminants. 

Nonetheless, this decision simplified matters considerably. For example, if the relevant "harm" was the resulting remediation costs, as the United States had advocated,8 the analysis became prohibitively complicated and even a complex model that divided OU4 into over forty polygons and attempted to trace the PCBs within each polygon to a specific PRP proved insufficient.9 Similar complexities would face PRPs attempting to demonstrate harm in terms of danger to the public, especially for sites like those at issue in the Fox River Litigation where the route of human exposure was not direct, but indirect (i.e., consumption of fish exposed to PCBs). Such complexities, however, dropped away under the Seventh Circuit's approach.  

B.  The Harm Must Be Correlated with the Amount of Contamination In Order to Be Divisible

The Seventh Circuit found that the harm was capable of apportionment only once NCR demonstrated that the harm-approximated by the remediation costs10-was positively correlated with the level of PCB concentrations; in other words, the greater the PCB concentration, the greater the remediation costs. 

At the preliminary injunction hearing, NCR "did not put forth any evidence to refute the government's contention that NCR's contributions of PCB would, alone, require approximately the same remedial measures."11 Consequently, both the district court and the Seventh Circuit Court of Appeals were left with the impression that at the preliminary injunction phase the harm was binary-"once the PCBs rise above a threshold level, their presence is harmful" and "the need for cleanup triggered by the presence of a harmful level of PCBs in the River is not linearly correlated to the amount of PCBs . . . discharged."12 Accordingly, the district court and the Seventh Circuit agreed that the harm was not divisible.13

At trial, however, NCR showed that remediation costs were "positively correlated with the concentrations of PCBs near the surface."14 Specifically, NCR demonstrated that "higher PCB concentrations near the surface are more likely to require dredging," which is the most expensive remedial option, "while lower concentration at greater depths are more likely to be eligible for capping or even sand covering," which are less expensive options.15 This was critical, as it allowed the Seventh Circuit to hold: (1) "we think the harm would be theoretically capable of apportionment if NCR could show the extent to which it contributed to PCB concentrations in OU4;" and (2) "if NCR cleared that hurdle, we think a reasonable basis for apportionment could be found in the remediation costs necessitated by each party."16

The Fox River Litigation counsels that, if a PRP is pursuing a contaminate amount-based divisibility argument, it must demonstrate not only the extent of contamination and its contribution to it, but also the connection between the contamination and the remediation costs.

C.  Divisibility Is Possible for River Remedies and Other Sites With Complex Conceptual Site Models

Initially, the district court rejected the notion that the harm in the Fox River Litigation was even theoretically capable of apportionment because "the riverbed is a complex organism with numerous factors working independently on it," including changes in river flow direction, historical dredging, and boat traffic.17 The court found that these factors constituted independent factors that rendered "the harm an indivisible one" and, more generally, "that harm is unlikely to be found divisible" when such independent factors exist.18 The Seventh Circuit effectively overruled these findings when it subsequently concluded that "the harm would be theoretically capable of apportionment if NCR could show the extent to which it contributed to PCB concentrations in OU4."19 

The district court's seemingly categorical bar to the use of the divisibility defense in complex CERCLA sites would have narrowed the defense considerably and introduced a new hurdle for PRPs to overcome on their route to a divisibility finding (i.e., is the site as complex as a riverbed). Indeed, as the Fox River Litigation demonstrates, it is often the more complex sites that require PRPs to resort to the divisibility defense. Although a PRP still must establish that the harm is capable of apportionment, the mere fact that the site is complex does not appear to be a persuasive basis for rejecting the divisibility defense.20

D.  A Successful Divisibility Argument Requires the PRP to Address All of the Contamination At the Site

The district court's October 2015 reversal demonstrates that, even when a PRP establishes a path to divisibility that courts endorse, a divisibility defense can fail because the PRP did not adequately address, in the court's opinion, all the harms at the site. Specifically, the October 2015 decision found that NCR failed to satisfy its burden on both parts of the divisibility analysis because one of NCR's experts, among other things: (1) overstated the PCB discharges of another entity; and (2) did not account for all of the sources of PCBs into OU4.21 The Fox River Litigation court is not the first to have denied a divisibility argument because a PRP did not adequately apportion all of the harms at the site.22 

E.  A Party's Burden In Proving a Divisibility Defense Remains Steep

A successful divisibility defense continues to be "the exception . . . not the rule"23 in large part because the burden  is on the party asserting the defense to prove: (1) whether the harm is capable of apportionment; and (2) if so, whether a reasonable basis for apportionment exists. In practice, however, this burden has numerous, very complex components, including: (1) the requirement for a party to create a site model that explains fate and transport;24 (2) that model must account for all the contaminates being addressed in the remediation, and not just the subset attributable to the PRP;25 and (3) the PRP must show the relative contributions of all PRPs.26

This burden is compounded by the nature of the divisibility defense, which seeks to attribute specific numerical allocations of liability for complex sites and in an adversarial setting. Compare, for example, how the same arguments are handled differently in the context of the divisibility defense and the context of a CERCLA equitable allocation. In an allocation, if Party A argues its liability is 25% and Party B argues it is 75%, the court could conclude Party A's liability was somewhere between 25% and 75%. 

In the divisibility context, however, does the court have the same ability? The October 2015 decision in the Fox River Litigation suggests perhaps not, and that the PRP instead must convince the court that its liability is very close to precisely as it claims-e.g., 25%-or else the court can deny the divisibility defense on that basis alone. What if the court, after weighing the facts, believes that Party A's liability is really 50%, 30%, or 27.5%,? Is that a basis to deny the divisibility defense? Some courts have refused to take such a hardline approach,27 but the October 2015 decision in the Fox River Litigation arguably suggests an unwillingness to adjust a PRP's calculations regarding each PRP's relative contributions and, instead, to simply deny the divisibility defense.

II.  Conclusion

The district court's October 2015 decision, reversing its prior holding that NCR established that the harm is divisible, likely is not the last decision in the Fox River Litigation regarding the divisibility defense. Although this case has developed the details of the CERCLA divisibility defense in several respects, making the path to a divisibility finding somewhat clearer for future litigants, several questions remain.