WISDOM, Presiding Judge:

Introduction

The countless years of unfettered pollution of our natural resources by commercial industry is now well documented. Fortunately, as the nation's awareness of the magnitude of the problem grew, so did the nation's resolve to act. Several years ago, we as a nation, through our elected representatives, took a strong stand to end the contamination of our environment and to remedy the damage that had been done. That monumental undertaking, in turn, spawned a new body of law as the courts grappled with the allocation of responsibility and financial liability that the new laws assign. The matter presently before us is borne of that history.

Today we render an important decision on who shall bear the liability for the release of environmental contaminants over a period spanning decades in some of this country's busiest railyards. We recognize at the outset that this is a case with wide impact on the parties before us and, potentially, on many unknown to us now.

We have entertained several motions and cross-motions for summary judgment on the claims and cross-claims raised by the several parties. All focus on a deceptively simple question: Will this Court allow the government to proceed against Penn Central and the other named railroads for the costs of cleanup resulting from the contamination of the railyards? We hold that nothing in the statutes, agreements, or pertinent documents at issue here insulates the railroads from potential liability or bars the government from pursuing its environmental liability actions in the district courts.

A second question follows, which we address in the second half of this opinion: Will the railroads to whom the properties were conveyed in 1976 be insulated from liability for contamination that occurred before that date? We hold that the railroad defendants cannot be liable for recovery costs for contamination before the date of conveyance. We defer, however, to the district courts entertaining the cost recovery actions as to whether the divisibility of the harm allows for the allocation of liability other than the imposition of joint and several liability.

History of the Case

This case examines the intersection of two statutory schemes: the Regional Rail Reorganization Act of 1973 (the "Rail Act") and the Comprehensive Environmental Response, Compensation and Liability Act of 1980 ("CERCLA"), as amended by the Superfund Amendments and Reauthorization Act of 1986. Our task is to determine whether the Rail Act, considered in the light of the Final System Plan ("FSP") and this Special Court's conveyance orders, bars the United States from pursuing actions brought under CERCLA.

This case concerns who bears the responsibility and financial liability under CERCLA for the decades of unregulated pollution of certain railyards. We focus our attention on two yards: the Paoli yard in Paoli, Pennsylvania, and the Elkhart yard in Elkhart, Indiana.

For much of this century, the electric transformers in locomotives contained and often leaked polychlorinated biphenyls ("PCB's"). Today we understand that PCBs are deadly. They have been linked to cancer, suppression of the immune system, liver damage, birth defects, and impairment of reproductive functions.

The operation, service, repair, and storage of electric railroad cars resulted in widespread release of PCBs onto the land of the railyards and into the water table underlying the railyards, including the two yards at issue. The contamination polluted adjacent properties as well. For example, PCBs found in Valley Creek, located one-half mile north of the Paoli railyard, measure upwards of 5 parts per billion. It may seem small mathematically, but it is sufficiently high to prompt the Pennsylvania Fish Commission to ban consumption of fish taken from the creek. The levels in the sediment samples taken from the creek was significantly higher. The Elkhart yard reflects similarly dangerous levels of PCBs. In response, for approximately two years, the EPA conducted a removal action to excavate, seal, and cover the contaminated property.

The complexities of this case stem in large part from the fact that the Paoli and Elkhart yards have been owned and operated by different entities at different times — each of whom has a unique legal posture with the United States. Penn Central and its predecessors owned and operated the Paoli and Elkhart yards for 61 years, beginning in 1915. On April 1, 1976, by Congressional mandate, Penn Central transferred the properties to the Consolidated Rail Corporation ("Conrail"), a newly-created private railroad corporation. Conrail, in turn, transferred the Paoli yard to the National Railroad Passenger Corporation ("Amtrak"), which owns the yard currently. That ownership notwithstanding, Conrail continued to operate the commuter rail services at Paoli until the beginning of 1983, at which point the Southeastern Pennsylvania Transportation Authority ("SEPTA") took over. SEPTA operates the yard to this day. Conrail has remained the sole owner and operator of the Elkhart yard since the conveyances were effectuated in 1976.

Thus far, the government has filed two CERCLA actions involving property conveyed under the Rail Act, although the promise of more in the future is nearly certain. In 1986, the government filed suit in the Eastern District of Pennsylvania for CERCLA cleanup costs relating to the Paoli yard. In 1990, the government filed suit in the Northern District of Indiana to recover cleanup costs relating to the Elkhart yard.

Penn Central turned to this Court. The catalyst for that decision was the holding of the Court of Appeals for the Third Circuit that the bankruptcy reorganization proceedings of Penn Central's predecessor, Penn Central Transportation Company ("PCTC"), did not bar the government from pursuing CERCLA actions against Penn Central. The court held that Penn Central, as the reorganized company, continues to bear liability for PCTC's pre-reorganization activities. In so doing, the court reversed the reorganization court's decision that the government's CERCLA actions had been discharged as part of PCTC's reorganization. After the Supreme Court denied certiorari, the reorganization court, in accordance with the remand order of the Court of Appeals, authorized the government to proceed with the CERCLA cases.

Penn Central in turn came to this Court in the hope that we would prevent the government's CERCLA actions from going forward. In particular, Penn Central asks that we issue a declaratory judgment and a permanent injunction barring the present and any future CERCLA actions against it for its pre-conveyance activities at the Paoli and Elkhart yards.

Conrail, SEPTA, and Amtrak (collectively "the railroad defendants") seek to insulate themselves from any liability under CERCLA for contamination that occurred while the properties were under Penn Central's watch (until April 1, 1976). Accordingly, the railroad defendants filed counterclaims against Penn Central and cross-claims against the United States. They seek contribution from Penn Central and a declaration from this Court that they are insulated from liability for pre-conveyance activities.

The government initially argued that the Special Court is not vested with proper subject matter jurisdiction to hear this matter and so moved under Fed.R.Civ.P. 12(b)(1) to dismiss Penn Central's complaint and the railroad defendants' cross-claims. Without deciding the substantive issues raised in the complaint, counterclaims, or cross-claims, we held that the Special Court has exclusive jurisdiction to hear this case. Accordingly, the Court denied the government's motion to dismiss. That decision allowed the case to go forward to its current phase: the various motions for summary judgment.

Our analysis requires that we revisit the Rail Act, the FSP, the Settlement Agreement between Penn Central and the government in the Valuation Case, and the conveyance documents issued pursuant to the conveyance order. This analysis comprises the central functions of this Court. As we see our duties, we conclude that this Court cannot grant to Penn Central the relief it seeks; rather, the CERCLA actions must go forward.

Legal Background

As Congress contrived the special jurisdiction accorded this Court, the legal backdrop against which the plaintiff's claims are measured is unique. That framework has controlled our previous decisions as it guides the decision we render today. Accordingly, at this point, we go over only the basics.

A rail transportation "crisis" in this country occurred when the major railroads of the Northeast and Midwest, filed for bankruptcy. PCTC's problems to which we alluded played a central role in defining this crisis. Congress responded by creating a single system of rail transportation operated by a private, for-profit corporation. Specifically, the Rail Act created a government corporation charged with preparation of a Final System Plan ("FSP") for restructuring the railroads. The FSP designated the rail properties which would be the subject of a transfer from the railroads in bankruptcy reorganization to a private corporation, Conrail.

By 1976, we began the long process of determining the proper valuation of the properties to be transferred in accordance with the FSP. Over the course of the succeeding four years, the parties litigated numerous issues in their, and our, efforts to arrive at a constitutional minimum value due to Penn Central as compensation for the properties conveyed. That litigation came to be known as the Valuation Case. Ultimately, the parties executed a Settlement Agreement which resolved all outstanding issues regarding the valuation of the rail properties. The value of the settlement, including interest, totaled approximately $2.1 billion.

These transactions were consummated with an eye toward the future. Congress believed that the future profitability of the rail lines depended upon the degree to which they could be made more efficient. Accordingly, Conrail, as the recipient of the properties, was charged with rehabilitating, improving, and modernizing the rail properties. In exchange, Congress sought to give Conrail and the other post-conveyance railroads a clean slate, free from the financial burdens of their predecessor. This financial tabula rasa reflected the "fresh start policy" embodied in the Rail Act.

The Rail Act mentions environmental concerns only peripherally. Congress indicated that it believed travel by rail more commensurate with conservation of natural resources than other forms of transport. While that belief places a high value on the integrity of the environment, nothing in the Rail Act or FSP addressed the assignment of liability for the clean up of environmental contamination.

While the negotiations for the Settlement Agreement were underway, Congress debated how best to respond to the mounting environmental crisis and escalating public health risk coming from years of uncontrolled disposal of hazardous substances, particularly in our nation's industrial corridors. Congress enacted CERCLA in December of 1980. CERCLA primarily is designed to facilitate the clean up of environmental contamination. To achieve its purposes, CERCLA established the "Superfund" — a multi-billion dollar fund designated for cleanup costs. In addition, CERCLA has teeth: it places in the EPA the authority to use the Superfund money to move quickly on contaminated sites without awaiting a judicial decision.

As currently amended, CERCLA authorizes three means of responding to a contaminated site: (1) the EPA can conduct its own cleanup using Superfund money; (2) the EPA can order the responsible parties to carry out the cleanup; and (3) third parties may cleanup the site and recover their costs from potentially responsible parties or make a claim against the Superfund for reimbursement. When the EPA takes a direct response, it may engage in a "removal" action, whereby it moves quickly to stem the immediate threat to the public health, or it may take "remedial" action, whereby it orders a long term approach to the problems of the site. Remedial actions, unlike removal actions, may take place only at sites designated on the National Priorities List. These sites have been singled out as those that pose the greatest threat to the public health. The sites which form the basis of the present case are so designated.

CERCLA affects any person or entity that played a part in the ownership or disposal of hazardous substances. The list of potentially responsible parties who may be subject to both injunctive relief and the costs of cleanup includes: (1) the current owner or operator of a facility; (2) the owner or operator of a facility at the time the site became contaminated; (3) any party who arranged for the disposal or treatment of the hazardous substances owned or possessed by that party; or (4) any party who accepted the hazardous substances for transport.

CERCLA's mechanism for allocating liability is the subject of the dispute between the government and the Railroad Defendants. CERCLA imposes strict liability; moreover, unless the parties show that the harm is divisible, they are jointly and severally liable. There are but three narrow defenses to CERCLA's strict liability which must be proved by a preponderance of the evidence: an act of God, an act of War, or an act or omission of a non-employee/agent third party, providing the defendant took due care. While courts construe CERCLA liability broadly, they construe these defenses narrowly.

To clear up a side matter, we read the Settlement Agreement and CERCLA chronologies as follows. CERCLA was approved by the U.S. Senate on November 24, 1980, and by the U.S. House on December 3, 1980; the President signed it into law on December 11, 1980. The Settlement Agreement was submitted to this Court on November 19, 1980, and approved on December 5, 1980. Penn Central nonetheless seeks to draw this Court's attention to the date that the Settlement Agreement was "consummated" — January 15, 1981. This date marks only the formal closing of the Settlement Agreement and this Court's dismissal of the Valuation Case claims. We attach no legal significance to it, at least as it relates to this case. The parties had terminated their negotiations and this Court rendered its approval as of December 5, 1980. Penn Central must accept that CERCLA did not become law until six days later.

I. PENN CENTRAL AGAINST THE GOVERNMENT

The question, as framed by Penn Central, is "whether the obligation to bring the conveyed properties into environmental compliance was necessarily resolved by the disposition of the Valuation Case". Penn Central's position is that the Settlement Agreement represented an arduous process and lengthy, intense negotiations of a settlement between sophisticated parties. This Court, after receiving considerable evidence on the condition and worth of the properties, entered a judgment of dismissal in the subsequent settlement, this Court specifically found that Penn Central had received the constitutional minimum value ("CMV") for the conveyed properties. That final judgment, so Penn Central contends, acts as a bar to the current CERCLA actions.

Moreover, Penn Central argues, the releases that the two parties executed as part of the Settlement Agreement bar the government from asserting these cleanup recovery actions. As part of the agreement, the parties released each other from all claims relating to the valuation of the properties. Penn Central argues, then, that these CERCLA claims, of which the government knew long before the Settlement Agreement was consummated, were extinguished and released when the parties entered into and this Court approved the Settlement Agreement. As to both arguments, Penn Central asserts that the cost of environmental contamination was factored into the settlement when this Court determined the market value of the properties. To allow the government to proceed would in effect charge Penn Central twice for the contamination. Not only would it allow the government to re-ignite an extinguished and released claim, it would deprive Penn Central of the just compensation to which it is entitled in violation of the Fifth Amendment.

The government frames the issue as whether the Rail Act, the orders of this court, the conveyances documents, or the Settlement Agreement provides Penn Central with a defense to liability under CERCLA, when all preceded enactment of the statute. The government sees the intersection of the two statutory schemes as the determinative analysis: one with the purpose of continuing freight rail service in the Northeast (the Rail Act) and the other with the design to protect the public health by providing for the clean up of environmentally contaminated sites (CERCLA). The government contends, on the basis of the respective chronologies of the statutes, that the CERCLA claims were not (and could not have been) specifically addressed by the Settlement Agreement. Furthermore, the government argues that the CERCLA claims have nothing whatsoever to do with the valuation of the properties and in no way affect the constitutional minimum value that Penn Central received from the Valuation Case. Simply put, so the argument goes, these claims never dawned on the negotiating parties.

In reviewing a grant of summary judgment, this Court applies the familiar standard codified at Rule 56(c) of the Federal Rules of Civil Procedure. A grant of summary judgment is appropriate where there is "no genuine issue as to any material fact" and "the moving party is entitled to judgment as a matter of law." Summary judgment should be granted where the moving party presents evidence which negates any essential element of the opposing party's claim or where any essential element is without factual support.

The party moving for summary judgment has the initial burden of demonstrating the absence of any genuine issue of material fact. When the moving party's motion is supported, however, the non-movant cannot rest on mere assertions and must do more than show some metaphysical doubt as to the material facts. No disputed issues of fact exist in the present matter. We are able to grant judgment as a matter of law.

A. Penn Central's Fifth Amendment Argument

Penn Central alleges the government's CERCLA actions will deprive it of its right to just compensation for the transferred properties by retroactively diminishing the amount it received under the terms of the Settlement Agreement. The Fifth Amendment proscribes the taking of private property for public use unless the owner of that property receives just compensation. The Rail Act similarly binds the government to give the railroads the constitutional minimum value for the conveyed properties and no more. In most cases, the taker must pay the market value of the property in question unless the calculated market value would produce an unjust result or otherwise fail to accord just compensation. Fair market value is measured by what a willing purchaser would pay to a willing seller to buy the property. A willing purchaser would take all factors that affect the value of the property into consideration. One of the factors to be considered is the physical condition of the properties. The physical condition of property may include the environmental condition and any related cleanup costs.

In this case, we are concerned with whether the negotiating teams to the Settlement Agreement ever considered the environmental condition of the properties. Penn Central asserts that the foreseeability of CERCLA's passage and the existing environmental statutes would have influenced any buyers and sellers. Hence, the environmental condition of the property had to have been considered during the valuation process, in order to arrive at a just compensation figure. If the government is allowed to assert a CERCLA claim based on that contamination, Penn Central would not have received just compensation as mandated by the Fifth Amendment.

The parties construe the CERCLA claims differently. The government sees these CERCLA claims as the cost of responding to environmental contamination in order to protect the public health and the environment bears no relation to the underlying economic value of the property, and indeed in some cases may dwarf that value. Penn Central, in contrast, views the potential cost of cleanup as a direct substraction from the value of the properties. The government emphasizes the distinct nature of its CERCLA claims by noting that the government also seeks reimbursement for cleaning up neighboring properties that were in no way related to the Valuation Case.

Penn Central has misconstrued the nature of a CERCLA claim; we thus side with the government. Penn Central's liability is derived from its role as the polluter, not merely as the owner or even operator of the properties. In that sense, its potential responsibility for the cleanup costs comes from its responsibility for creating the contamination. This distinction informs our entire discussion. It is the foundation for our conclusion that, while a CERCLA claim relates in a sense to the physical condition of the properties, it concerns a fundamentally different subject — the protection of human health and the environment — from the type of property assessment undertaken in the Valuation Case. We do not dispute that the environmental condition of the railyards is an aspect of their value and that these CERCLA claims affect that value. That admission, however, does not force the outcome that Penn Central urges.

Penn Central views the money received under the Settlement Agreement as residing in a sealed vault, immune from any later attempts to reach it by the United States. Penn Central asserts that anything relating in any way to the condition of the properties has been litigated, negotiated, and paid for. Penn Central cannot be forced to "further improve" the properties by remedying the environmental contamination. The imposition of CERCLA liability, we are told, would have precisely that effect.

We disagree. Although CERCLA relates to the physical condition of the land, it does not concern the valuation of the railyards, as the parties understood that process. Without any evidence whatsoever that either negotiating team considered the environmental condition of the properties in the course of those negotiations, it is impossible to conclude that the amount that Penn Central received reflected the consideration of these environmental handicaps. Accordingly, we will not hold up the Valuation Case as a shield to potential CERCLA liability.

CERCLA is unique. It radically changed the horizon of environmental law by giving the government enforcement tools far beyond its previous capacity. For Penn Central it is but an unfortunate fact that the government also happened to be the party with whom it was negotiating. That coincidence does not change the outcome.

In so holding, we reject any implicit challenges to the retroactive imposition of CERCLA in this case. To the degree that Penn Central argues that the Settlement Agreement is final and immune from the reach of subsequent litigation, we cite the now well-established rule that CERCLA's retroactivity is constitutional.

The absence of evidence that CERCLA claims were considered makes sense, given that the CERCLA claims which constitute the legal vehicle for recovering cleanup costs did not arise until after the Settlement Agreement was concluded. The value of the property was calculated in the light of known factors at that time. For this reason, any effect that these CERCLA actions or the cleanup efforts would have now — fourteen years later — on the value of the properties is irrelevant to this discussion. Penn Central ceased to maintain an interest in these properties as of the time of conveyance.

If the environmental contamination of the properties had been considered as part of the evaluation of the pre-conveyance condition of the properties, the compensation calculated would have reflected that consideration. In the Valuation Case, that amount would have been subtracted from the value of the properties.

Yet, Penn Central offers only suggestions that the government was aware of the gravity of the environmental consequences of the PCB release. Penn Central charges that "the government's arguments in the Valuation Case were marked by their relentless emphasis on the deplorable condition of the rail properties". The condition that worried the government, however, concerned the physical impediments to present and future railroad efficiency, not the level of invisible contaminants in the soil and water table. This distinction is borne out by the documents which Penn Central cites.

No impact study, report, assessment, or inspection document has been offered to show that either side contemplated these concerns in 1980. Penn Central offers evidence that the government — through its various agencies — was aware of the environmental condition of the Paoli yard. The affidavits offered in support of the government's position, however, demonstrate that the negotiating parties never considered the environmental condition as part of the Valuation Case. This is a crucial distinction. Both sides were represented by talented counsel. The settlement of the Valuation Case was, as both parties acknowledge, an arduous negotiation process. If these concerns were considered, the affidavits would have so indicated.

Penn Central asserts that the parties need not have specifically quantified environmental cleanup costs for them to be subsumed into the valuation process. That is true; many aspects of the yards' value were not specifically quantified. Still, Penn Central fails to acknowledge the obvious: the environmental claims at issue measure into the tens of millions of dollars. The parties would have contested the value of these significant costs, had they been considered. Although any reconstruction of events is a precarious endeavor, it is safe to say that the representatives in the matter at hand would not have failed to record evidence concerning a dispute of that import.

Penn Central resorts to two other arguments as a means of underscoring its Fifth Amendment discussion. First, Penn Central raises the doctrine of res judicata as a variation of its assertion that, because the physical condition (including potential environmental contamination) of the properties already was litigated in the Valuation Case and Settlement Agreement, the government is barred from pursuing its CERCLA claims.

Res judicata bars the relitigation of all claims that were asserted or could have been asserted in a prior action between the parties or those with whom they shared privity. To preclude a subsequent claim on the basis of res judicata, there must be: (1) a final judgment on the merits in a prior suit resolving, (2) claims by the same parties or their privies, and (3) a subsequent suit based on the same cause of action. The Valuation Case certainly was a final judgment on the merits. Moreover, the parties to that action are essentially the same as those in the pending or threatened CERCLA actions.

The harder question is whether the government had a full and fair opportunity to litigate its CERCLA claims (or other claims for environmental contamination) in the Valuation Case. Penn Central's argument here is just another twist on their Fifth Amendment argument; i.e., the valuation of the Paoli and Elkhart yards in the Valuation Case and Settlement Agreement necessarily took the physical and environmental condition of the properties into consideration. For that reason, Penn Central asserts, the government may not do so now.

The linchpin to this argument once again is whether the valuation of the properties included an evaluation of their environmental condition. Res judicata is not the proper vehicle for deciding the central issue of this case. This doctrine offers no dimension not covered in the Fifth Amendment analysis above. The claim now being asserted (CERCLA) is distinct from any claim that could have been raised in the Valuation Case (e.g., a pre-CERCLA tort claim for environmental damage). There is no such thing as a CERCLA-type claim.

The second argument that Penn Central advances in conjunction with its Fifth Amendment contention is based on the doctrine of compensable unconstitutional erosion ("CUE"). Unconstitutional erosion arises when a business that is losing money is forced by the government to continue in operation for more than a reasonable period despite its unprofitability. The "erosion" usually consists of financial erosion — e.g., expenses which arise as a result of the period of forced operation; or physical erosion — e.g., the deterioration of the physical properties during the forced operation. Penn Central argues that, even if this Court does not agree that the CERCLA claims would reduce the compensation that it received to below the constitutional minimum value, the CERCLA claims would constitute CUE.

The viability of this argument turns on whether this Court classifies these threatened cleanup recovery actions as operating costs. The logic goes that these environmental cleanup costs would have been a deferred maintenance expense (Penn Central was deferring other operating costs at the time) during the period that it was forced to continue operations against its will. This cleanup obligation matured prior to April 1, 1976; the expense therefore constituted an unconstitutional erosion of its estate.

We decline to adopt this suspect reasoning. The CERCLA recovery costs at stake cover a period of pollution stretching a minimum of three decades. The period of Penn Central's forced operation represents a tiny fraction of that period. Hence, these CERCLA claims can hardly be considered operating expenses incurred during the period of forced operation. Moreover, CERCLA was not enacted until 1980; to consider the remedies outlined in its provisions as a cost of operations several years earlier strains logic. This is particularly so where, as here, the bulk of the contamination took place during the 1940's, 50's, and 60's.

As such, we do not construe these CERCLA claims as forcing Penn Central to pay twice for these burdens. We thus hold that the retroactive enforcement of obligations under CERCLA does not drop the compensation that Penn Central received to below the constitutional minimum value mandated by the Fifth Amendment.

B. Penn Central's Release Argument

As a starting point, the government argues that it could not have released its CERCLA claims because CERCLA did not exist at the time that the Settlement Agreement was executed and approved. While some district courts have held that a release which predates CERCLA may nonetheless release CERCLA claims, those cases involve broader release agreements than those here at issue. Whether the government released future CERCLA claims in the releases at issue, then, raises a question of contract construction.

In evaluating Penn Central's release argument, we recite some of the principles applicable to the interpretation of the Settlement Agreement which inform our analysis. This Court's function is, in part, to interpret the Settlement Agreement and related conveyance documents. In so doing, we are guided by general federal contract law. Moreover, general federal contract law governs the interpretation of contracts to which the government is a party. The development of federal contract law relies on and borrows from the best rules and decisions in the development of the general common law of contract interpretation.

A Settlement Agreement, while a slightly different animal, is examined within the familiar strictures of ordinary contract interpretation. Where the language of a contract provision is clear and unambiguous, that language controls. It is thus error to depart from the plain meaning of a settlement agreement. A release is unambiguous unless it is reasonably susceptible to more than one interpretation. Whether the contract is clear and unambiguous is itself a question of law. Courts should not hunt for ambiguity in a plain provision. At all times, we keep in mind that a court's interpretation of an agreement ultimately should reflect the intentions of the parties.

The unique setting out of which the Settlement Agreement came is not lost in all of these hornbook rules. The terms of the release must be construed in the light of the circumstances of its formation and the context in which it arose. The valuation proceedings were extraordinary and prolonged in every sense; they arose in a very special factual and legal context. The upshot is that we must construe the terms of the releases so as to cover all that the negotiating teams considered as part of them.

We are, however, limited to the terms of those releases. The contract before us contains a merger clause which bars the admission of parol evidence. Section 8.04 of the Settlement Agreement provides: "This Agreement contains the entire agreement between the parties with respect to the subject matter of this Agreement." Where a contract contains such an integration clause, parol evidence cannot be used to supplement, vary, or contradict the terms of the provision in question.

The releases in question are those the government granted as part of the Settlement Agreement and in exchange for releases granted by the Penn Central transferors. The government's release, contained in section 4.04(b) of the agreement, provides:

[E]ach Government Party hereby releases all of its claims, to the extent specified in Exhibit 5, against each Penn Central Transferor.

Exhibit 5, referred to in that clause, provides that the following claims are released:

(a) any claim that the terms of transfer or conveyance of Rail Property . . . under Section 303 of the Rail Act were more fair and equitable than required as a constitutional minimum;

(b) any claim for recovery of VOB [value of other benefits] in excess of the amount set forth in Table 1;

(c) any claim for the recovery of any amount pursuant to Section 215(b)(2) of the Rail Act;

(d) any claim for adjustment of the value of transferred or conveyed Rail Properties in the Penn Central System on account of value attributable to maintenance or improvement provided pursuant to an agreement under Section 213 or Section 215 of the Rail Act; and

(e) any claim . . . for the recovery of any amount paid to any Penn Central Transferror pursuant to Section 213 or 215 of the Rail Act[.]

Our task is to delineate the breadth of the releases and determine whether the terms are susceptible to a construction that encompasses the release of the pending CERCLA claims. We conclude that they are not.

Examination of the releases strikes the reader not for the breadth so typical in commercial contracts but, rather, for their specificity in their enumeration. The releases demonstrate that the parties had some specific concerns and addressed them in five narrowly focused categories. We follow the rule that, "if the agreement appears to be limited to specific disputes or particular types of liability, CERCLA liability will be excluded unless the agreement contains a clear, unambiguous reference to such liability." No such reference exists here.

Paragraph (a), quoting directly from the Rail Act, releases a claim that the terms of the conveyances did not satisfy the constitutional minimum value. Paragraph (b) releases any claim for the value of any other benefits (VOB) other than those enumerated in an accompanying table. This release also is guided by a specific provision of the Rail Act. Last, paragraphs (c) through (e) release any claims resulting from the interim program for keeping the railroads in operation during the Rail Act planning process.

In sum, the releases are unambiguous; they are concerned exclusively with the valuation of the properties transferred and with ensuring that interim rail service would be unaffected by the transfers. This plain construction is underscored by the purpose of the Rail Act and, in turn, the Settlement Agreement: to create an economically viable rail network. The narrow scope of the releases reflects that purpose by removing only the potential hurdles that could have waylaid that endeavor. They are no more ambitious than that.

Penn Central nonetheless argues that the CERCLA claims fit within one or more of the specifically enumerated categories. Release (a) bars any claim that Penn Central received more than the CMV. Penn Central would have the Court construe the present CERCLA claims as a diminishment of the value of the properties and, thus, a reduction of the compensation Penn Central received. Penn Central argues that the environmental condition of the properties was considered in the valuation process and when the government drew up its releases. Hence, by this logic, any claims regarding the environmental condition of the railyards would be, in effect, claims that Penn Central received more than the CMV for its properties.

All of Penn Central's arguments are different roads leading to the same destination: this argument turns on whether the parties considered the contamination. The government charges that "Penn Central is simply reading its theory of this case into the language of the Settlement Agreement, rather than reading the language itself." We agree. Once again, we reject Penn Central's attempt to fit its theory that it has already paid once for the CERCLA claims or that the valuation of the properties necessarily subsumed assessment of the environmental condition of the properties.

If it cannot win outright, Penn Central hopes to create, at the least, a genuine issue of material fact on the question of knowledge. The garnered evidence demonstrates at best that the government knew that the properties were contaminated. We need not postpone our decision for it is plain that, whatever knowledge the government may have had of the environmental condition of the properties, it was not sufficient to read into the releases a bar to future CERCLA claims. Without CERCLA in existence, neither the government nor Penn Central could have factored potential monetary liability for cleanup costs of the present magnitude into the negotiations.

Although those negotiations covered all factors relating to the physical condition of the properties, these CERCLA claims transcend that relationship. It is true that the environmental condition of the properties is a subset of the physical condition of them. But, as we have already explained, it is more. Environmental condition as we understand it today encompasses the property's worth in the light of all possible uses to which it could be put and any potential hazards to human beings. The Settlement Agreement, on the other hand, was crafted in an effort to preserve a viable rail system. There is simply no reason to believe that the latter considered the former.

Claims for the costs of cleanup derived from statutory causes of action that did not exist could not have been contemplated by the negotiating parties. While it may appear unfair to make Penn Central retroactively vulnerable to CERCLA liability, that is precisely the scheme that Congress approved. In sum, although the government may have understood the likelihood that the environmental condition of the properties was serious, it does not follow that the parties considered these costs as part of the Settlement Agreement.

Indeed, as a response to the Court's request for such direct evidence at oral argument, counsel for Penn Central underscored the paucity of its evidence by referring only to documents suggesting the government's awareness of the problem at the yards. No evidence was offered showing that the negotiation of the Settlement Agreement included potential CERCLA claims or other environmental claims that could have grossly affected the amount of the Agreement.

Our discussion regarding Penn Central's CMV claim combined with our interpretation of release "a" forecloses once and for all Penn Central's valuation argument. We instead agree with the government's reading that these claims represent not the valuation of the properties but, rather, the "cost of responding to environmental contamination in order to protect the public health and the environment".

Release "b" releases claims relating to the value of other benefits (VOB). Penn Central argues that the government's release of any claim that seeks the VOB bars the United States from bringing a CERCLA claim. The government argued in the Settlement Agreement that the value of other benefits includes "executory contracts and inchoate obligations". Penn Central attempts to paint the costs of environmental cleanup as an "inchoate obligation" that it would have faced in the absence of the Rail Act. This is an attempt to force a square peg into a round hole. It is the least convincing of Penn Central's arguments.

Releases "c" through "e" related only to the interim operation of the railroads. Hence, they ceased to have effect at the same time that the conveyances were finalized.

Compare to these narrow, specific releases the type of broad language that has been found to release a future CERCLA claim. In Olin Corp. v. Consolidated Aluminum Corp. , the Court held that a broad release could encompass a future CERCLA claim. The release provision in that case bears little resemblance to the provision at issue here. It stated that Concalco, the defendant in the suit,

hereby releases and settles all claims of any nature which Conalco now has or hereafter could have against Olin . . . whether or not previously asserted, under or arising out of the [Agreement] . . ., or the transactions contemplated thereby.

No such all-encompassing language exists in the releases executed pursuant to the Settlement Agreement.

In conclusion, we hold that the releases do not encompass the CERCLA claims at issue. When the terms of the releases are placed alongside the familiar canon that the expression of one thing is the exclusion of another, it is plain that the costs of environmental cleanup were neither contemplated nor included. Hence, the Settlement Agreement acts as no bar to those actions. These releases are plain and specific; they can be read without resort to extrinsic evidence.

Next, Penn Central looks to the claims that the government preserved. In addition to releasing five valuation-related claims, the government specifically preserved three types of claims. Penn Central argues that the United States could have preserved a CERCLA claim and, by failing to do so, released its right to later bring one. This argument plainly lacks merit chiefly because, as we stated above, the government's release was confined to specific categories. A party typically needs to preserve a claim only when that claim would otherwise have been released; i.e., in the case of a broad, general release provision.

The government reserved tax claims, claims on loans for pre-conveyance obligations, and claims at issue in Matter of Penn Central Transp. Co. Penn Central urges that we construe this list of preserved claims as exhaustive; i.e., that any claims (such as claims for cleanup costs) not included in that list were released. In this way, Penn Central seeks to run with the foxes and chase with the hounds: that same logic, if applied to the releases, would have undercut all of Penn Central's arguments.

Although Penn Central provides the appropriate starting point, namely, that "[w]here a contract contains specific exclusions, it is to be assumed that no others were intended," that general rule does not end our analysis. Here, it is unclear what the government should have preserved. The government did not yet have CERCLA as an enforcement tool. Hence, even if the United States knew about the contamination of the railyards, it did not yet have this statutory mechanism for recovering cleanup costs. Accordingly, it could not have preserved future CERCLA claims.

As a final word, we note that Penn Central's position deserves some sympathy. The Settlement Agreement was supposed to end the interaction between Penn Central and the government once and for all. Furthermore, Penn Central owned and operated the yards at a time when our collective knowledge of the safety and health threat posed by environmental hazards was woefully inadequate. We all are paying for that mistake. CERCLA is but one mechanism for remedying these decades of abuse. Sympathetic or not, however, Penn Central cannot escape the fact that Congress passed a statute which launched similar retroactive actions everywhere. In that sense, it is important to distinguish between the valuation of the railyards (which the Settlement Agreement resolved) and the retroactive imposition of liability through a subsequent legislative enactment (which, we hold, the Settlement Agreement did not).

When Penn Central owned and operated these yards for profit with little regard for (and, ultimately, at the expense of) their environmental condition, Penn Central became primarily responsible for that pollution. Nothing in the legal framework within which this Court functions works to shield Penn Central from being legally available to answer for it. Nor has Penn Central demonstrated that further discovery will shed light on any dispute of material fact. We conclude that no such dispute exists. Accordingly, we hold that Penn Central has not asserted valid defenses that prevent the government from going forward with its CERCLA actions in the district court.

We grant the government's motion for summary judgment and deny Penn Central's motion for summary judgment on the foregoing issues.

II. PENN CENTRAL AND CONRAIL

In this section, we turn to the claims between Penn Central and Conrail. Two questions demand our focus: First, whether Penn Central conveyed the properties "as is, where is" and, in so doing, terminated its liability for cleanup costs; and, second, whether Conrail's obligation to rehabilitate the conveyed properties under § 302(c) of the Rail Act includes the duty to pay for the environmental cleanup. The parties filed cross-motions for summary judgment.

A. As Is, Where Is

The Bill and Sale Assignment ("BSA") executed pursuant to the conveyances states that the "Grantee takes the property as is and where is." Penn Central contends that this "as is" provision transferred Penn Central's liability to Conrail, the transferee and, consequently, exonerated Penn Central from CERCLA liability to the government. The linchpin to Penn Central's argument is that the "as is" provision in the BSA pertains to real property — and specifically, the Paoli and Elkhart railyards — conveyed from Penn Central to Conrail. Because that assertion is fundamentally flawed, we conclude that the "as is" provision in the BSA has no bearing on the issues at hand.

Penn Central's argument is based upon a mistaken reading of the BSA itself. Penn Central attaches great significance to meaningless permutations of the word "property". For example, Penn Central explains that whenever non-real estate properties were discussed, the term "properties" or "railroad properties" was used. In contrast, the argument goes, when the term "property" was used — without the plural — it impliedly referred to real estate. The contention does not impress us. Nothing in the use of these terms implies that some refer to real estate while others do not.

A proper construction of the BSA reveals that the "as is" provision applies only to personal property and administrative assets, but not to real property such as the railyards conveyed. The BSA was separate from the conveyance documents which detailed the transfer of real property. The BSA's classification scheme and contents lists make this clear.

The scope of what the BSA was to encompass is delineated in schedules B through J, attached to the BSA itself. The BSA contains specifically enumerated contents limited to personal property and administrative assets. This property was transferred "as is" to foreclose the possibility of a future dispute regarding the value of a comparably insignificant part of the total transaction.

The only mention of real property — Schedule A — directs the reader to the other documents. Schedule A merely identifies the personal and administrative property subject to the BSA. In some cases, it was necessary to refer to the real property to which the administrative assets were appurtenant as means of identifying them. In other words, real property is mentioned only as a guide to get to the personal and administrative assets covered by the BSA — no real property was affected by the BSA's "as is" clause.

The parties vigorously dispute whether an "as is" clause is capable of limiting CERCLA liability. It seems that a strong line of cases holds that an "as is" clause cannot work to limit a conveyor's CERCLA liability. Other cases have approved of private contracts which seek to allocate CERCLA liability. In the light of our construction of the BSA, we must side-step this issue entirely.

As our analysis should make clear, the "as is" clause in the BSA not only fails to transfer Penn Central's liability to the successor railroads, it similarly fails to impede the government's ability to go directly after Penn Central. In addition to the reasoning laid out, we note the general rule of law that private parties cannot contract to avoid liability to the government. Although these circumstances make this conveyance unique — Penn Central made this conveyance at Congress's direction and to a government-created entity — the "as is" clause is still a provision created by two private parties. Penn Central cannot avoid liability to the government on the basis of that clause.

B. Section 302(c) of the Rail Act

We have examined closely the Rail Act and the legislative history leading to its enactment. Molded by the context in which the clause was drawn, we conclude that Congress intended Conrail to "rehabilitate, improve, and modernize" the rail system to make it more efficient. The railyards were in a state of disrepair that had rendered the yards dangerous and inefficient. Congress wanted to rectify the situation quickly and placed that burden on its new entity, Conrail.

This clause is the expression of that intent. In construing it, this Court must look beyond mere words on a page. Congress had a specific objective and addressed it using broad terminology. The facial breadth of the provision notwithstanding, we cannot in good conscience stretch this provision so far as to interpret a congressional mandate to relieve others from future statutory liability, let alone liability based on environmental contamination.

Similarly, nowhere is there any evidence that, if faced with the question, Congress would have chosen to shoulder Conrail with the entire financial burden of remedying the decades of Penn Central's environmental abuse. In fact, the fresh start policy embodied in the Rail Act counsels for the exact opposite conclusion.

Conrail's motion for summary judgment is granted on these issues.

III. THE UNITED STATES AGAINST THE RAILROAD DEFENDANTS

As a starting point, we outline the basic lines of contention. Like the preceding parts of the opinion, the question presented is facially quite simple. The Railroad Defendants assert that the Rail Act, the FSP, the conveyance deeds for the Paoli and Elkhart Yards, and this Court's Conveyance Order provide that the railroads were to take the railyards free from any obligation or liability for Penn Central's pre-conveyance activity. In support of that construction, the railroads stress that the government mandated the process which led to the conveyances; accordingly, the government should not now be allowed to come forward and impose this CERCLA liability on them. Much of this debate focusses on whether the environmental damage is capable of divisibility so that the railroads would be liable only for their period of ownership.

The government's position is, at times, circular and amorphous. On one hand, the government disputes the railroads's characterization of the Rail Act and the conveyance documents. The government repeatedly asserts that CERCLA overpowers those provisions. On the other hand, however, the government states that it seeks to impose no liability on the railroads for pre-conveyance activity. It concedes that it presses no theory of successor liability. That position merely frames the issue differently for, pursuant to CERCLA, the government will seek to impose joint and severable liability on the Railroad Defendants and Penn Central. If successful, the railroads would be on the hook for all liability — whether pre- or post-conveyance.

A. The Rail Act and the Conveyance Documents

The Fresh Start Policy Embodied in the Rail Act.

Congress viewed the successor railroads as the phoenix rising from Penn Central's ashes. The Rail Act and related conveyances are the mechanisms that Congress chose to effectuate its goals. Chief among Congress's priorities was to give the railroads a fresh start; that is, to "wipe the slate clean". Relying on this "fresh start policy" as the glue that holds the Rail Act together, the railroads contend that the imposition of CERCLA liability for pre-conveyance contamination would run contrary to this congressional mandate. They seek our declaration to that effect.

The government concedes that Congress sought to give Conrail and the subsequent railroads a fresh start. Where it parts company, however, is the degree to which faithful adherence to that policy stands as a hurdle to these CERCLA actions. To the government, the fresh start policy embodied in the Rail Act is superseded by the later-enacted explicit and detailed CERCLA statute.

Both parties turn to the Bankruptcy Code as an analogue — it gives debtors a fresh start — though each camp thinks the Code militates in its favor. The crucial distinction between the Bankruptcy Code and the Rail Act is that the Code actually codified a fresh start scheme, whereas under the Rail Act it is just a pervasive policy. In this sense, the fresh start policy in the Rail Act is less compelling than its sister statute.

The Code notwithstanding, we conclude that the fresh start policy, taken alone, is insufficient to ward off potential CERCLA liability. For one, the fresh start that Congress sought to bestow upon Conrail had nothing to do with considerations such as environmental burdens. Congress sought to give the railroads a clean slate so that they could operate freely without some of the impediments that had threatened their predecessors' survival and might threaten theirs. These impediments were rooted in Penn Central's financial crisis and the dilapidation of the railyards that resulted.

Taken in this context, it becomes clear that CERCLA or other environmental damage liability simply was of no moment to Congress at that time. "[T]here is just no indication that Congress regarded the possibility of successor tort liability to be an obstacle to Conrail's continued operations. . . ." The railroads cite Congress's intent to wipe the slate clean as a talismanic, statutory mandate barring any potential CERCLA liability. On the contrary, we hold that CERCLA, a massive statutory enactment, must trump whatever policy permeates the Rail Act.

The two can be viewed in harmony rather than in conflict. Congress intended to wipe clean the slate of fiscal burdens that plagued Conrail's predecessors. This congressional objective the Rail Act achieved. The fresh start policy is in no way impeded by allowing the government to go forward with its CERCLA claims at this juncture, for these claims were not the fiscal burden of either Conrail or its predecessors. We note also that, as the same Congress that passed CERCLA amended the Rail Act, Congress could have removed the threat of a CERCLA action from Conrail's future, had it so wished.

The "Free and Clear" Provision.

Section 303(b)(2) of the Rail Act provides that the conveyances shall be made "free and clear of any liens and encumbrances." The railroads argue that this "free and clear" provision is not a simple indemnity but, rather, a broad statutory codification of the fresh start policy. It is broad enough, they argue, to mean that the railroads took the railyards free and clear of these CERCLA claims. That is not the case.

In contrast to the purpose the railroads ascribe to it, the "free and clear" provision was designed to verify the conveyance of marketable title vis-a-vis the properties. No evidence exists that this language was additionally intended to insulate the defendants from the type of statutory liability at issue here or any other obligations. Accordingly, the "free and clear" provision is meaning less in the face of the government's CERCLA enforcement action.

This construction is borne out by a textual analysis as well. Neither the term "lien" nor "encumbrance", taken in its ordinary legal meaning, could encompass a statutory claim for the recovery of cleanup costs. At least one court has rejected the notion that a potential CERCLA claim constitutes a de facto lien. We see no reason to break from that line.

This is the first point at which we become privy to the government's alternative case theories. The government disputes, point for point, the railroads' arguments as to the provisions and policies at issue. Within each argument, however, the government argues that these disputes are irrelevant because the government seeks to impose only post-conveyance liability. At great lengths and in exasperated tones, the government asserts time and time again that it is not pressing any theory of successor liability by which the Railroad Defendants assumed the liability of Penn Central.

For reasons which we detail below, this is the classic wolf in sheep's clothing. CERCLA's liability scheme would expose the railroads to liability for pre- and post-conveyance contamination if they are found jointly and severally liable. Hence, the government can afford to lose all of these battles, knowing that if joint and several liability is imposed, it still will win the war. We defer further discussion of this problem to our analysis of liability allocation below.

The "Other Charge" Provision.

A clause in this Court's Conveyance Order allocates liability between Conrail and Penn Central for payment of "any tax, assessment, license fee, or other charge imposed by a governmental authority on or with respect to any such property or any use thereof or thereon." The railroads attempt to stretch the phrase "other charge" to include CERCLA claims. That specious interpretation would stretch this phrase well beyond its intended limits.

We start with an oft-cited canon of interpretation, ejusdem generis: where a list of specific and particular items is followed by general terms, those general terms are restricted in scope to the topic or class of the specific items. Claims for environmental cleanup costs are well outside the temporal and topical parameters suggested by the specifically enumerated tariffs in this clause. Rather than being a periodic tax, assessment, or charge, a CERCLA claim is an equitable action seeking reimbursement of costs. Even the powerful congressional objective that the transferee railroads start with a clean slate cannot stretch the language here to the lengths urged by the defendants.

The "Assumes No Obligation" Provision.

In the deeds for the Paoli and Elkhart Yards, Conrail agreed to assume and perform "all obligations . . . that arise or accrue . . . under all licenses, easements . . . and operating, trackage right and joint facility agreements." The deeds further provide that Conrail would assume no obligation or liability for any pre-conveyance activity; where the obligation at issue occurred both before and after the conveyance date, Conrail would be responsible for only the liability allocable to its post-conveyance ownership. The Railroad Defendants argue that this provision was designed to shield them from liability for any pre-conveyance activity, including Penn Central's contamination of the yards.

The government urges that we give this provision a restrictive construction in the light of its immediate context. The government believes that this clause meant that Conrail assumed no obligations derived from "licenses, easements, and leases" — the terms contained in the immediately adjacent provision. According to the government, the juxtaposition of the two provisions renders them interdependent. Hence, the "assumes no obligation" language has no bearing on the environmental contamination of the railyards.

While the deed language is admittedly not clear-cut when viewed as a whole, we think that the railroads have properly construed the operation of these two provisions. The language in the excerpt is broad, using phrases such as "no obligation or liability" and "any event, act or failure to act".

This is where the fresh start policy embodied in the Rail Act tips the scales. These deeds were not drawn in a vacuum; they were drawn with the intent to give the railroads a clean slate as of the date that they took the yards. That policy cannot insulate them from their responsibility for contaminating the yards, but it certainly indicates that they should not be responsible for the contamination that came before them.

We thus conclude that the excerpt means what it says: Conrail, and the other railroads, "assumes only that portion of the obligation or liability which is reasonably allocable to the part or the period after [April 1, 1976]". It is on this basis that we expressly limit the Railroad Defendants' liability exposure to the post-conveyance period only; we thus exonerate them from any obligations or liability that accrued prior to the time that they took the properties. Pre-conveyance contamination is Penn Central's responsibility.

The FELA Cases.

The railroads rely heavily upon two Federal Employers' Liability Act ("FELA") cases. In both cases, the courts held that the predecessor railroads remained liable for employee exposure to asbestos while the successor railroads did not. While superficially similar to the matter at hand, an important distinction renders these cases inapposite.

The FELA claims existed and were known at the time that the predecessor conveyed its assets to the successor railroad, whereas in the present matter, at the time of the conveyances CERCLA was still several years away. In the FELA cases, the parties would have been well-versed in FELA claims. Hence, the valuation and subsequent releases from future liability in the conveyancing process would have taken tort claims brought under FELA into account. In our case, Congress could not have considered whether the "fresh start" they envisioned would shield the successor railroads from liability under CERCLA because CERCLA did not yet exist. The "Iron Ore" Case.

The government relies heavily on one case in particular, In re Lower Lake Erie Iron Ore Antitrust Litig. ("Iron Ore"). In that case, the district court held that Congress did not intend to shield Conrail from the consequences of its participation in a criminal antitrust conspiracy that also involved predecessor Penn Central employees. The government asserts that the case stands for the proposition that, in spite of the fresh start policy benefitting Conrail, Congress did not intend to immunize the transferee railroads from the consequences of Penn Central's pre-conveyance activities.

Iron Ore is distinguishable from the present matter on two grounds: First, Iron Ore involved a criminal statute (as opposed to CERCLA's strict liability scheme) and, second, Iron Ore involved a conspiracy that stretched from the predecessor (Penn Central) to the successor railroads. This latter distinction is dispositive. In Iron Ore, the court held that Conrail's culpability was based not on some theory of successor liability but, rather, on the affirmative conspiratorial acts of Conrail's own management. The fact that a conveyance took place in the interim did nothing to alter the equation.

With that, it is plain that Iron Ore has no bearing on the instant matter. Here, the Railroad Defendants have done nothing to incur pre-conveyance liability. If they are made to pay for Penn Central's pre-conveyance activities, it will be based solely on their status as successor railroads. No continuity of culpability exists.

Conclusion.

We hold that the conveyance documents, read in the light of the Rail Act's fresh start policy, limit the liability of the Railroad Defendants to the time of the conveyances forward. Penn Central is responsible for its activities; it is not Conrail's burden. Although this decision is rooted in the language of the deeds construed with the Rail Act, we also accept the government's assertion — really, concession — that it seeks to impose only post-conveyance liability on the railroads. It is clear that no theory of successor liability applies here.

In the next section, however, we examine whether that holding is meaningless in the face of the threat of joint and several liability. From the railroads' perspective, if joint and several liability is imposed, our statement that they are free from pre-conveyance liability will be likened to the old saying, "That and a quarter will get you a cup of coffee". Regardless, we grant the Railroad Defendants summary judgment on the foregoing issues.

B. CERCLA's Liability Framework

In this section we address whether, pursuant to CERCLA, we can allocate the environmental damage at issue so that the railroads will be potentially liable only for post-conveyance contamination. After canvassing CERCLA's provisions and the applicable case law, we conclude that this question cannot be resolved without a further evidentiary hearing. It is our view that the district courts hearing the CERCLA claims are the appropriate fora to receive that evidence.

Two significant determinations must be made in the allocation of liability under CERCLA. First, the court must decide if the parties are jointly and severally liable. That is strictly for the district courts. Second, the court must determine how the allocation of damages among the parties is affected by other factors. Although this is also the province of the district courts, we expect that our holding relieving the railroads of liability for pre-conveyance activities will strongly counsel the courts when making this decision.

The Railroad Defendants do not allege that they did not harm the properties or that they owned or operated only a percentage of the sites. Rather, they assert that April 1, 1976, is the first date for which they became potentially liable. This date of conveyance, they argue, is a bright-line cut off of their potential liability by law. As such, no further evidentiary showing should be required. As support for that construction, the railroads again look to the Rail Act's fresh start policy.

In the light of our conclusion that the Rail Act and conveyance documents shield the railroads from pre-conveyance liability, they now seek a declaration that the environmental harm that resulted is divisible according to period of ownership and, thus, joint and several liability is not appropriate.

The Various Approaches to Apportioning Liability.

The central issue is whether the respective periods of ownership and operation (beginning April 1, 1976 for the defendants) are an accurate reflection of the degree to which each party is liable for the environmental harm to the railyards. If not, joint and several liability must be imposed.

Different courts have taken various approaches to this task. A starting point is the statute itself, and the legislative history that led to its enactment. It is clear that Congress deleted a proposed provision that would have mandated joint and several liability under CERCLA. The result of that deletion has been that, while joint and several liability is commonly imposed under CERCLA, courts do not view it as automatic. If the harm can be fairly divided, joint and several liability is unnecessary.

A recent Fifth Circuit case, Matter of Bell Petroleum Services, Inc., explained that the nature of the harm is the determinative factor in this analysis. The question is "whether there is a reasonable and just method for determining the amount of harm that was caused by each defendant". Our inquiry, then, is whether the nature of the harm to the railyards is susceptible to divisibility as to the various owners and operators.

Most courts have held that this is threshold determination. The Third Circuit, for example, decided that the question of joint and several liability should be decided at the liability stage, not during the contribution action which inevitably follows. The Second Circuit agreed, although that court concluded that the timing of the joint and several liability determination is best left to the individual court. We agree that the determination of divisibility should be made as early in the action as possible.

During the contribution stage, the court can examine other factors in determining how the liability and damages ought to be allocated. Those factors center on the degree of involvement of the parties in the generation, transportation, and disposal of the subject waste. As we previously stated, our discussion of pre- and post-conveyance liability will be relevant in this assessment.

The problem with the divisibility inquiry is that simple solutions such as dividing liability according to periods of ownership is inapplicable to a harm like the long-term effects of the seepage of chemicals. Often, the harm simply is incapable of divisibility because of the migratory or toxicity potential of the released chemicals. Those chemicals, mixed with others, may have caused synergistic reactions which render the simple volume of chemicals released an inaccurate measure of liability.

An admittedly simplistic example can be nonetheless illustrative. An operator may have dumped millions of gallons of chemicals over a period of decades with limited damage. A successor to the property may dump a few gallons of a relatively innocuous chemical that becomes highly toxic when it mixes with the previously dumped liquid. The Court must decide, usually several years later when it is impossible to tell exactly what happened, who is responsible for what. While joint and several liability is not always the equitable answer, it often is the most practicable and sometimes the only possible option.

Is the Harm in the Railyards Divisible?

In the case at hand, the railroads urge that the date of conveyance is precisely the sort of quantifiable demarcation the courts have in mind when they discuss whether the harm is susceptible to divisibility. They rely heavily on Comment c to section 433A of the Restatement (Second) of Torts which reads, with an uncanny factual similarity to the present case:

c. Successive Injuries. The harm inflicted may be conveniently severable in point of time. Thus if two defendants, independently operating the same plant, pollute a stream over successive periods, it is clear that each has caused a separate amount of harm, limited in time, and that neither has any responsibility for the harm caused by the other. (Emphasis added).

The question remains, however, whether there is a reasonable basis for determining the contribution of each cause to a single harm.

The answer can be reached only after the study of extensive evidentiary and testimonial material, usually in the form of expert testimony. For that reason, we must decline the invitation to resolve the divisibility question in this Court and, as such, we deny the Railroad Defendants' motions for summary judgment. Although the railroads see a bright line starting at April 1, 1976, time periods of operation are but one factor to consider. As stated, an examination must be made of the "relative toxicity, migratory potential and synergistic capacity of the hazardous waste at issue." A decision reached without the benefit of that showing would be premature.

The Bell case provides an effective contrast to the present matter. The court there ultimately held that the harm at issue was capable of divisibility and, thus, joint and several liability was inappropriate. Only one chemical had been released; there was no need to look to synergistic effect. The volume released by each defendant was, thus, an appropriate indicator of percentage of liability. Although the court admitted that its liability apportionment could not be proved with an absolute certainty, it nonetheless concluded that the facts of the case allowed for divisibility.

Unlike the Bell decision, the Fourth Circuit's decision in Monsanto is closer to the case at hand. The court there stated that "some evidence disclosing individual and interactive qualities of the substances deposited" is necessary before a determination of joint and several liability could be made. That is the problem we face. We cannot proceed to a determination of whether the harm is divisible without a great deal of information. Had we received the voluminous evidence and expert testimony heard in Bell, it is possible that we could reach that court's conclusion. We are, however, in a markedly different position.

Which Court Should Receive The Evidence?

Given the need to receive evidence on the divisibility of the harm, we must decide whether we should proceed to hear that evidence or defer to the district courts entertaining the CERCLA claims. Revisiting our respective tasks resolves the matter.

This Court is the final arbiter of the correct interpretation of the conveyance documents, Rail Act, and FSP. No CERCLA claims will ever be pressed here. We do not conduct conventional trials. Our rules and procedures are, in many respects, unique. Federal district courts, on the other hand, are well-versed in receiving complicated evidence and testimony. Given that they are the tribunals that will ultimately hear the CERCLA claims themselves, they are better situated to receive the evidence that will affect those determinations. In sum, we conclude that the district courts hearing the CERCLA claims are the proper fora to analyze evidence on the degree of contamination and the number of toxic and hazardous chemicals released. The decision whether to impose joint and several liability is theirs.

The Fresh Start Policy, Redux.

As a reiteration of their previous argument, the Railroad Defendants again seek shelter in the Rail Act's fresh start policy, arguing that CERCLA cannot trump the Rail Act. We do not agree.

The task, they say, is to reconcile the Rail Act with CERCLA and construe them in pari materia. The railroads rely heavily on the canon which favors construing two applicable statutes harmoniously. The only harmonious construction, they urge, is to divide the harm and bar the imposition of joint and several liability.

The railroads hold up three reasons supporting their contention: (1) the Rail Act's fresh start policy; (2) Congress's decision not to require joint and several liability in CERCLA; and (3) the fact that CERCLA seeks to avoid imposing liability on an involuntary transferee.

The government, on the other hand, asserts that no conflict between its sought enforcement of CERCLA and the fresh start policy of the Rail Act exists because the Rail Act is silent on the issue of recovery of cleanup costs. This argument is hollow. The fresh start policy embodied in the Rail Act cannot so easily be dismissed simply because its authors didn't mention the environment.

The government's better argument is that CERCLA takes precedence over this general fresh start policy because Congress specifically stated that CERCLA liability arises "[n]otwithstanding any other provision or rule of law." In a conflict, CERCLA prevails.

In either case, for the reasons we have outlined, the fresh start policy may work to limit the railroads' liability to post-conveyance contamination; it does not, however, affect our decision to leave for the district courts the decision whether to impose joint and several liability.

Other Equitable Considerations.

Aside from the fresh start policy, equities exist which favor the Railroad Defendants. We outline them here not because they affect the decision reached in this case, but because we expect that the subsequent district courts hearing these CERCLA actions will consider them, if only at the contribution stage.

One is the principle of involuntary ownership. Conrail, by virtue of its involuntary ownership and operation of the Paoli and Elkhart railyards, could become jointly and severally liable for decades of contamination long before it existed. Conrail had nothing to do with the railyards prior to the conveyances and, yet, it might end up paying for it. Similarly, SEPTA was forced to assume responsibility for "hopelessly unprofitable commuter rail services" and, like Conrail, was not given a choice concerning the transfers.

Worse, the railroads are in dire straits financially. In an era where they routinely teeter on the precipice of financial crisis, the government seeks the huge additional burden of the cost of environmental cleanup. For SEPTA, the $28 million at stake in this case could be the margin of its survival.

Conclusion.

Although we have decided that the Railroad Defendants should be shielded from liability for the release of any substances prior to April 1, 1976, a district court hearing the CERCLA claims is in the best position to determine how that shield will affect their liability for the costs of cleanup under CERCLA. We thus decline the invitation to decide the issues of apportionment or allocation on the grounds that an evidentiary showing of the divisibility of harm is necessary. That showing is more appropriately made in the district courts.

In those district court actions, we expect that the fresh start policy embodied in the Rail Act will be taken seriously. If the evidence demonstrates that the harm in this case is capable of divisibility, we urge the district courts to divide it accordingly. That disposition would acknowledge that Penn Central polluted these yards for decades and, more, that the Railroad Defendants should not bear the brunt of that tragedy.

It is so ordered.

GASCH and JUNE L. GREEN, JJ., concur.