Printer Friendly

Toxic apportionment: a causation and risk contribution model.

I. Introduction and Summary

A. The Basic Idea

In a typical toxic tort case,(1) the plaintiff or the plaintiffs property suffers some harm as a result of exposure to one or more sources of a toxic substance.(2) For example, if the plaintiff has been exposed to asbestos-containing materials(3) and has a history of cigarette smoking, and after years of exposure to both develops a cancer of the lungs, the question arises in tort law as to which of the exposures caused the cancer. However, under the current state of medical knowledge, the question of which toxic substance, if either, actually caused the cancer cannot be answered definitively.(4) If the exposed person sues asbestos manufacturers for products liability, he will have to attempt to prove that the exposure to asbestos-containing materials was either the but-for cause(5) of or a substantial causal factor in the cancer.(6) The trial of such a case will no doubt produce conflicting expert testimony, with the plaintiffs experts stressing the well-established causal relationship between certain cancers and asbestos exposure by inhalation,(7) and the defendant's experts stressing with equally well-documented data the causal relationship between smoking and lung cancer.(8) If the duration and extent of the exposures are significant enough that epidemiological data would support a plausible argument that each is a cause of the plaintiffs cancer, then the data do not warrant concluding that only the defendant's product or only the plaintiff's smoking should be regarded as the cause, thereby requiring either the defendant or the plaintiff to bear the entire loss.(9)

One method of avoiding an all-or-nothing outcome in such a case is to apply principles of apportionment: the plaintiffs cancer would be apportioned among the parties on the basis of the comparative cancer risks each created through its conduct. Nevertheless, this risk contribution model would not predicate liability based solely on risk creation, as the plaintiff must sustain actual harm and those risks created also must be "a' cause of that actual harm. Apportionment using a risk contribution model also avoids the rubric of "fault" that results when courts apportion liability under comparative negligence or fault-based schemes,(10) and it yields fairer and more efficient results than those achieved under some of the alternative causation models employed by courts over the last few decades.(11) For example, some courts hold that so long as a plaintiff demonstrates that asbestos exposure is a substantial factor in bringing about the cancer, the defendant is liable for the entire loss, whether or not there may be other causal factors.(12) In our smoking and asbestos exposure example, courts until recently were reluctant to find a reasonable basis on which to predicate an apportionment of the harm, and, as a result, our plaintiff would recover all of his damages despite the obvious relevance of a history of smoking.(13)

If the plaintiffs asbestos exposure and his smoking are both shown to be causal factors in the plaintiffs lung cancer, then the loss is necessarily capable of apportionment on the basis of the relative risks demonstrated for each kind of toxic exposure. From the perspective of individualized justice, the model produces a fairer result than nonapportionment. Because notions of individualized justice place emphasis on individual accountability, if a party's own conduct creates a kind of risk known to result in the harm complained of, then that party should bear a portion of the loss. Moreover, by apportioning the harm among the respective risk contributors, each pays no more than the portion of the harm fairly attributable to it. Thus, so long as individual responsibility is a valued goal of tort law, courts should apportion losses to reflect the contribution of each responsible party based on the risks each created.

Requiring the defendant to bear the entire loss is inefficient because it eliminates an incentive for persons to provide for their own health and safety. At the other extreme, having the plaintiff bear the entire loss is also inefficient because defendants then lack the incentive to avoid creating toxic-related risks. Apportionment of harm should yield the optimal measure of deterrence: if the defendants are liable only for a portion of the total harm, consistent with the risks they created, it will avoid both the underdeterrence and overdeterrence that result from placing the entire loss on either the plaintiffs or defendants. Moreover, the plaintiffs have an incentive to engage in safer lifestyles, an incentive that is missing if the defendants bear the entire loss.

In addition to requiring an analysis of the relative contribution of risk, this model demands proof of causation. In our smoking and asbestos exposure example, each party's proofs must be sufficient to make out a submissible issue on causation - that the toxic exposure was "a" cause of the harm, applying the tests for causation that have evolved in toxics cases. This Article develops the issue of precisely what kinds of proof would be sufficient for this purpose in part VI, below.(14)

The risk contribution model is not based on proportional recovery for risk alone, as has been advocated by some(15) and criticized by others.(16) Instead, it combines some of the benefits sought by those advocating proportional recovery for risk with the elements of causation and actual harm traditional to the toxic tort setting. The model represents a synthesis of ex ante risk and ex post harm. Moreover, it does not treat toxic exposure as an independent harm, as Professor Wright advocates.(17) Instead, this model requires proof of actual harm above and beyond exposure itself. Rather than looking to economic theory as the basis of apportiorment, it looks to science as the underlying basis for apportionment. By examining toxicology and epidemiology, along with whatever particularistic evidence may be available, as the basis for determining disease causes in humans, the model achieves both fairness and deterrence goals and will simplify trials in some cases. In the case involving the plaintiff exposed to asbestos who also smokes, the experts themselves, as seen below, examine the scientific literature to answer the question of what caused the plaintiffs lung cancer and of how much of a risk each source of toxic exposure created toward that resultant harm. Tort law should also he willing to base an apportionment of that harm on the same considerations that medical science applies to these questions.

In part II, the Article considers the existing black-letter rules and comments governing apportionment set forth in the Restatement (Second) of Torts, identifies the strengths and weaknesses of those rules, and assesses their compatibility with the model suggested here. Part III develops the risk contribution and causation model by focusing on one recent decision involving a plaintiffs asbestos exposure and smoking history. Part IV considers multiple-defendant, real property cases in which each defendant contributes to the risk of the resultant harm and develops how apportionment functions in that setting. Part V expands on the multiple-defendant apportionment issues by discussing the risk factors developed in litigation under the Comprehensive Environmental Response, Compensation, and Liability Act.(18) Thus, both personal injury and real property torts are considered. Part VI examines the sciences of toxicology and epidemiology, the two major sources of data relevant to risk assessment in toxic tort cases. Part VII is devoted to examining and identifying who are the risk contributors and explores in greater detail the causation threshold that must be crossed as a prerequisite to application of toxic apportionment. Parts VIII and IX consider some of the impediments to and rationales for the adoption of a risk contribution and causation model of toxic apportionment.

B. The Context

It is useful to see where the risk contribution model fits within the bigger picture of tort law. Much of the litigation involving apportionment has arisen in the context of joint torts, in which courts hold multiple actors liable for a loss caused to the plaintiff. Initially courts limited joint torts to cases of vicarious liability for concerted action, holding each defendant in a common scheme jointly and severally responsible for the entire harm.(19) No effort has ever been undertaken in these concert of action cases to allocate or apportion the harm among the respective actors based on the degree of culpability each one bore or on the causal contribution of each to the harm. Gradually courts expanded the concept of joint torts to incorporate situations where each of two or more persons acted independently but their actions concurred or combined to produce a single injury.(20) In these cases, the courts often speak of "single, indivisible" injuries, thereby implying that there exists no basis for determining how much of the harm can be attributed to each of the actors.(21) For example, where two negligent drivers collide, killing a third person, the courts have consistently regarded the harm as indivisible and have held both drivers jointly and severally liable for death damages.(22) In contrast, where two actors each inflict a separate injury on the plaintiff, albeit concurrently, there has been a willingness to regard the harm as two separate injuries and hold each actor responsible only for the harm it has caused.(23)

Difficulties arise when there is a reasonable basis for making some division among various causes even though the injury is singular. Once that division is made, the prerequisite for joint and several liability - single, indivisible injury - no longer exists, but proportionate liability is measured only by the harm each actor caused. Debate centers around whether such a division is purely arbitrary, or whether it is predicated on some rational and appropriate basis.(24)

Those who are advocates of joint liability are likely to focus on the point that each actor was "a" cause of the entire harm (however "cause" may be defined) and therefore that each ought to be chargeable with the entire loss, not simply some position of it.(25) The pivotal role of causation in the analysis is clear. Those courts that permit apportionment, say in certain nuisance cases, focus on the comparative degree, quantity, or magnitude of the interference with the plaintiff's rights that each defendant occasions. In such cases k may be possible to conclude that each defendant was not a cause of the entire harm but rather contributed to it in some measurable way. Problems of proof are also central. Can a party, usually the defendant, offer sufficient proof that apportionment is feasible and demonstrate that something akin to a separate invasion of the plaintiff's rights may be attributable to it? As in any case in which there may be a failure of proof, who bears the burden of demonstrating the basis of apportionment is critical. If the plaintiff proves that the defendant is a (partial) cause of the harm to which another has also contributed, some courts would hold the defendant liable for all of the harms unless it can carry its burden of showing some reasonable basis for apportionment.(26)

However difficult these issues are to resolve in the sporadic accident cases that have given rise to their development, toxic tort cases ushered in a whole new array of problems. While death may be a single, indivisible injury by everyone's definition, what if death can be attributed to two causes - asbestos exposure and smoking - and those two causes can be compared in terms of the extent to which each contributed to the death? Should apportionment be permissible in such cases? What about a case in which millions of dollars are expended to clean up contamination of real property brought about by multiple hazardous substances deposited by several firms over many years? The risk contribution model this Article develops argues that apportionment is appropriate in both the death case and the cleanup case so long as evidence is offered that permits the factfinder to assess the relative contributions each actor or source made to the harm.

If tortious acts are viewed as a spectrum, at one end is the case in which multiple sources produce a single, indivisible harm and it can be reasonably inferred that each source is a cause of the entire harm. At the other extreme are instances of pure risk, in which no harm results but one or more actors create significant risk of harm. Tort law has not entertained such pure risk cases but instead requires a plaintiff to suffer harm as a prerequisite to a cause of action. Many possibilities he between these extremes. For example, what if a defendant creates risks and a plaintiff suffers harm, and while no firm evidence of a causal nexus exists between the risk and the harm, the nature of the harm is consistent with the risks that the defendant created? Courts have entertained liability in some cases of this type, even though they involve some speculation or conjecture as to causation.(27) An alternative approach might be to redefine harm as "risk exposure per se" or "risk exposure that possibly leads to subsequent injury," an approach that finds little judicial support in the toxics field. Furthermore, at least in mass exposure litigation, it might also be possible to assess liability to correspond to the proportional increase in harm attributable to the risks the defendant created, in situations when it is not possible to determine the causal connection in individual cases, but the connection can roughly be determined for groups of individuals exposed.(28) The risk contribution and causation model does not go as far as these possible alternatives because it necessitates proof of actual harm, applying traditional definitions in terms of personal injuries, property damage, or economic loss; and requires proof of causation as has evolved over the last forty years. What is new, is that where multiple sources of causation-related risks exist that produce harm, liability is apportioned on the basis of the relative magnitude of those risks.

Let us now look briefly at some further extensions of the model that are developed in detail in later sections.

C. Extending the Idea to Multiple-Defendant Cases

In multiple-defendant cases in which a pond, stream, or the air is polluted, and the plaintiff suffers personal injury or property damage as a result, this Article advocates heavy reliance on apportionment based on the magnitude of risk created by each defendant's actions, rather than treating the harm as indivisible and therefore applying joint and several liability. Cases such as Landers v. East Texas Salt Water Disposal Co.(29) and Michie v. Great Lakes Steel Division(30) illustrate the now traditional view of indivisible harm in joint and several liability. In such cases, fairness and efficiency concerns argue persuasively for apportionment of the harm based on the risks created by each defendant. While the Michie(31) and Landers(32) opinions (both of which are described in part IV) assumed that no basis for apportionment existed, the parties in those cases could today supply data for apportionment if that were the law's preferred resolution. Over the last forty years, courts seem to have applied an unstated presumption against apportionment of damage in tort cases, including those involving toxic exposures.(33) The courts, with some support from the Restatement (Second) of Torts, place a heavy burden on parties seeking apportionment and demand more in the nature of distinct or obviously divisible harms.(34) But rarely is the harm itself - a disease, illness, death, flooded property, or contaminated ground water - capable of any meaningful division.(35) What is capable of division on a very reasonable basis is the contribution that each party makes to the forces that produce the harm; the emphasis should be on the inputs, not solely the end result. Unlike the resultant harms, the sources of and the risks associated with inputs can usually be measured in some reasonable fashion. As will be developed below in part II,(36) even the language of Restatement [sections] 433A appears to recognize this point, but the comments to that section fail to provide consistent application.(37)

Under this model, even the famous twin fires case, a staple of introductory torts courses, was possibly decided wrongly. In Anderson v. Minneapolis, St. Paul & Sault Ste. Marie Railway Co.,(38) two fires, one attributable to the defendant-railroad's negligence and the other of unknown origin, combined to destroy the plaintiffs property.(39) The court rejected application of the but-for test of causation because, since the innocent fire was a sufficient cause, the plaintiff would have sustained the loss regardless of the defendant's fire.(40) After concluding that leaving the plaintiff without a remedy would be unfair, the court approved the lower court's use of a substantial factor test of causation and held the defendant liable for the entire loss.(41) Under the model advanced here, however, the loss would be apportioned because the court could compare the two fires by size, strength, area, or other criteria.(42) The central idea is to allocate the resistant harm by the degree of risk created by each source. One justification for apportionment on these facts is that some (or all) of the plaintiff's property would have been destroyed by fire even if the railroad had not been negligent. Allocating the loss among the two sources of fire is fairer than having one source bear the entire loss. Moreover, it should not matter whether the other source of fire is unknown (as in the Anderson case), another tortfeasor, or the plaintiff herself. The fairest and most efficient resolution of Anderson would have been for each party to bear some of the loss. Were it not for the defendant's fire, the plaintiff would have borne the entire loss since the unknown fire would have destroyed the property. Having the defendant absorb a portion of the loss is fairer to the plaintiff than applying the but-for test and fairer to the defendant than refusing to apportion and requiring it to bear the entire loss. Further, the defendant's tortious conduct requires some deterrence, so apportioning part of the harm to the defendant is necessary. But it is both unfair and inefficient for the defendant to absorb all of the loss, and the plaintiff, who would otherwise have absorbed the entire loss, cannot complain about her reduced recovery. The causation prong of the risk contribution model is also fulfilled in Anderson; each source could satisfy tort law causation rules since each was a sufficient cause of the harm.(43)

In Landers, too, some of the harm would have occurred even if either defendant were not negligent.(44) Each contributed something to the pollution in the pond, large quantities of salt water or salt water and oil, that killed the fish in the pond. The Texas Supreme Court opted for indivisibility and the imposition of joint and several liability because of what it perceived as the unfairness of having the plaintiff bear the loss and the difficulty of requiring the plaintiff to apportion the harm among the two defendants. Today, however, the parties could offer evidence on the chemicals each deposited into the water, their respective volume, toxity, duration, and other factors bearing on the quantum of risk each created by its negligence or tortious conduct. Interestingly, even the comments to the Restatement (Second) of Torts section 433A suggest that apportionment is preferred on precisely these facts, based on the volume of pollution each contributed to the water.(45)

As another example, assume that a plaintiff with a smoking history suffering from cancer sues several asbestos firms, which is usually what occurs in asbestos litigation today. In this case, it is necessary to apportion the harm, first between smoking and asbestos exposure (plaintiff-defendant apportionment), and second, among the asbestos manufacturers or distributors responsible for the plaintiffs exposure (multiple-defendant apportionment). As to the first factual inquiry, the risk contribution model functions precisely as described above and as explored below in greater detail in part III: the risks created by smoking that contributed to the cancer are compared to the aggregate risks created by exposure to products containing asbestos. The necessary comparison is based on the relative risks established by epidemiological and other evidence, as developed in part VI. Thus, if a twenty-year smoking history produces a risk of contracting lung cancer that is ten times the general risk of developing cancer, and twenty years of asbestos exposure yields a risk of contracting lung cancer that is also ten times the risk in the general population, then a comparison of those two relative risks would suggest an apportionment of harm of fifty percent to the plaintiff and fifty percent to the asbestos firms in the aggregate. As to the apportionment among the defendants, more particularized evidence of the exposure to each defendant's product is essential: the duration of exposure to each product, the relative toxicity of each, the conditions and intensity of that exposure, and possibly other factors. By developing evidence on these factors and others, a factfinder has a reasonable basis for apportioning one-half of the harm among these defendants.

Similarly, if a plaintiff who smoked sued both tobacco companies and asbestos firms, the court could still allocate the harm on the basis of the risks each of the firms created. As to the smoking-related portion of the harm, if the plaintiff smoked brand X for five years, brand Z (produced by a different firm) for ten years, and brand Q (produced by a third firm) for five years, the number of years of smoking each brand, assuming daily doses remained relatively constant, serves as a rational means for apportioning the harm among the three firms.

D. Related Concerns

Toxic tort cases require a doctrinal bias in favor of apportionment rather than indivisibility whenever either the plaintiff or other defendants create risks material to that harm. Even though the burden may rest on the defendants to establish the basis for apportionment, the courts should exercise a preference for divisibility of the harm and the limitation of liability based on that division. This model of apportionment should not be dependent on the various theories of liability the plaintiffs assert. Some recent literature suggests that the efficiency of apportionment rules varies according to whether negligence or strict liability is asserted.(46) However, toxic tort cases come in a variety of theories - negligence, nuisance, trespass, strict liability for abnormally dangerous activities, and strict products liability - but the divisibility of the harm is the same regardless of the theory. In part VIII, this Article develops in greater detail rationales for adoption of this risk contribution apportionment model that demonstrate that both fairness and deterrence can be achieved regardless of the underlying theory of liability.

Finally, this model rejects the availability of contribution as an answer to concerns regarding joint and several liability. Indeed, the fact that a party could successfully obtain contribution from a codefendant usually demonstrates that some basis for apportionment is feasible. Contribution assumes that a determination has been made that multiple defendants are joint tortfeasors, but because apportionment would obviate the need for finding that liability is joint, there would be no occasion to resort to contribution. A tortfeasor who is held liable for only its portion of the harm would neither seek contribution nor be liable for it. Such liability is thus analogous to the party who settles under a release for only a portion of the plaintiffs claim.(47)

II. General Rules of the Restatement (Second) of Torts

A. The Current Framework

The Restatement (Second) of Torts(48) in section 433A sets forth the guiding principles that apply in determining the propriety of apportionment. It reads:

[Sections] 433A. Apportionment of Harm to Causes (1) Damages for harm are to be apportioned among two or more causes where

(a) there are distinct harms, or

(b) there is a reasonable basis for determining the contribution of each cause to a single harm.

(2) Damages for any other harm cannot be apportioned among two or more causes.(49)

The most striking features of section 433A are that subsection (1)(b) recognizes that "single" harms are capable of being apportioned and that such apportionment is to be predicated on the "contribution" of each cause. This language appears consistent with the risk contribution model advanced here. In most toxic tort cases, the harm will be singular, such as death from cancer, the cancer itself, or contaminated real property or groundwater. However, the absence of distinct harms does not preclude apportionment if rational means are available for measuring the contribution of each source or, party to the harm. Consequently, the black-letter statement of section 433A is supportive of a risk contribution model of toxic apportionment.

The lengthy comments to section 433A contain material both supportive of and inconsistent with the risk contribution model. First, the comments describe a broad range of coverage; apportionment principles are said to apply to tortious and nontortious conduct; those joined as defendants and those not joined; plaintiffs as well as defendants; pre-existing conditions; and forces of nature.(50) In other words, section 433A is to be applied to all forces that can be regarded as contributing causes of the harm. Second, section 433A applies to distinct harms, as set forth in subsection (1)(a), and divisible harms, as set forth in subsection (1)(b). Regarding distinct harms, the Restatement offers the example of two defendants shooting a plaintiff simultaneously, one in the leg and one in the arm. This example affords a "logical, reasonable and practical" basis for apportionment of the harm, even as to pain and suffering and medical expenses.(51) While there may be some difficulty as to such expenses, nevertheless it is "possible to make a rough estimate which will fairly apportion such" damages.(52)

A second example, denominated "successive injuries," posits that if two defendants pollute a stream over successive time periods, apportionment is appropriate because "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."(53) While this example may not be as "clear" as the Restatement suggests, it at least acknowledges that nuisance-type cases are not per se excluded from the application of apportionment principles.(54)

Third, and most important to this inquiry, the comments elaborate on divisible harms that are not as "clearly marked out as severable into distinct parts," but that are nevertheless amenable to "division upon a reasonable and rational basis."(55) Another toxic-tort example demonstrates the point:

Such apportionment is commonly made in cases of private nuisance, where the pollution of a stream, or flooding, or smoke or dust or noise, from different sources, has interfered with the plaintiff's use or enjoyment of his land. Thus where two or more factories independently pollute a stream, the interference with the plaintiff's use of the water may be treated as divisible in terms of degree, and may be apportioned among the owners of the factories, on the basis of evidence of the respective quantities of pollution discharged into the stream.(56)

Thus, the Restatement regards the quantity of pollution (the volume of hazardous waste, to take an even more contemporary example) as the reasonable basis upon which a court and jury can undertake the apportionment. As discussed below,(57) this paradigm is the very stuff of which cases under the Comprehensive Environmental Response, Compensation, and Liability Act (CERCLA)(58) are made. In such cases, the Environmental Protection Agency (EPA), courts, arbitrators, or parties attempt to allocate the costs of cleanup among many potentially responsible parties or endeavor to avoid joint and several liability to the CERCLA plaintiff (usually the government) by establishing the divisibility of the harm.(59)

Fourth, after describing the availability of apportionment in fairly broad terms, the Restatement injects in comment h an override for "exceptional cases" when "injustice to the plaintiff may result from an application of the rule."(60) The comments further describe "injustice" as equating to a tortfeasor's insolvency, which would preclude the plaintiffs full recovery if each defendant were responsible only for its proportionate share of the total liability. In other words, the Restatement comment authorizes joint and several liability to trump apportionment whenever the "innocent plaintiff" must otherwise absorb the share of the loss attributable to an insolvent tortfeasor.(61)

Finally, in comment i, the Restatement declares that "[c]ertain kinds of harm, by their very nature, are normally incapable of any logical, reasonable or practical division."(62) Citing examples such as death, destruction of a house by fire, and the sinking of a barge, it observes that "by far the greater number of personal injuries, and of harms to tangible property, are thus normally single and indivisible."(63) It then gives the example of the twin fires case in which either cause (one fire innocent, the other tortious) would have been sufficient to bring about the destruction of a building.(64)

In section 433B, the Restatement addresses burdens of proof, declaring in subsection (1) that, except as noted below, the plaintiff bears the burden of proving that the defendant's tortious conduct "has caused the harm to the plaintiff."(65) In subsection (2), however, is the rule most relevant for this inquiry:

(2) Where the tortious conduct of two or more actors has combined to bring about harm to the plaintiff, and one or more of the actors seeks to limit his liability on the ground that the harm is capable of apportionment among them, the burden of proof as to the apportionment is upon each such actor.(66)

Thus, when the tortious conduct of two defendants "combines" to bring about the harm, the burden shifts to any defendant seeking to limit its liability - that is, a defendant who seeks to shift a portion of the loss to the plaintiff or seeks to avoid joint liability with other defendants for the entire loss must establish the basis for apportionment. As a result, in cases where the loss is capable of apportionment (those involving distinct or divisible harms), the defendant, not the plaintiff, bears the burden of proving the basis for apportionment so long as the plaintiff has proven that each defendant's tortious conduct is a cause of the harm. The language of section 433B(2) and comment c suggests that this rule applies only when there are multiple tortious actors. It is unclear whether it applies to the single plaintiff-single defendant case. However, because comment d refers to the "entirely innocent plaintiff," the burden-shifting principle apparently is intended to apply to multiple-defendant cases only. The burden-shifting rule applies to the "typical example" of the pollution of a stream by a number of factories, in which the harm is capable of apportionment.(67)

Section 433B makes clear that it is not necessary that all tortfeasors be joined as defendants to trigger the burden-shifting rule. Comment d offers a fairness rationale for burden-shifting, as it is unfair to permit a tortfeasor who caused some harm to escape liability on the fortuity that others also contributed to the harm, when the nature of the harm necessitates the introduction of apportionment-related evidence.(68) The comment concludes: "[a]s between the proved tortfeasor who has clearly caused some harm, and the entirely innocent plaintiff, any hardship due to lack of evidence as to the extent of the harm caused should fall upon the

In comment e to section 433B, the Restatement addresses the case of many actors, each of whom contributes a de minimis amount to the total harm, in which application of the burden-shifting rule would cause "disproportionate hardship to defendants."(70) Comment e gives the example of one hundred factories each causing a small but incalculable amount of pollution to a stream in which application of burden-shifting would render such a small contributor potentially liable for the entire harm. However, it notes that "such cases have not arisen, possibly because in such cases some evidence limiting the liability always has been in fact available."(71) With the ushering in of CERCLA litigation, however, such cases are widespread, and courts are struggling with rationales and rules for apportioning liability.(72)

B. Some Problems with the Restatement

This brief survey of the Restatement's principles for apportioning harm reveals several problems. First, while the back-letter rules appear to permit apportionment in a significant number of cases, some of the comments suggest otherwise. This tension is created because at some points, the Restatement emphasizes the harm, and at others, the contributing causes. For example, the comments provide that pollution of streams by multiple actors can be apportioned on a volumetric basis by measuring the quantity of contaminants from each source, thus focusing on the sources of the harm.(73) In contrast, the polluted water - the actual harm - is not amenable to division, although the forces that brought it about can be reasonably measured and compared. Therefore, the black-letter rule that refers to the "contribution" of each cause and comment d's polluted stream example apply the same basic principle. Later, however, comment i declares that damage to real property is indivisible, which is precisely what occurs in many nuisance, trespass, and abnormally dangerous activity cases. This inconsistency arises because the Restatement comments at some points consider only the resistant harm itself, as when they state that injury to real property is indivisible, and at other points are to examine the inputs responsible for the harm.

In addition, comment i states that death is the quitessential indivisible harm - and indeed it is - but in deaths attributable to toxic causes, as when a plaintiff dies from lung cancer brought about by the combined effects of smoking and asbestos exposure, each of the contributing causes can be compared and the harm apportioned on that basis.(74) Comment i also offers the example of property destroyed by twin fires as being indivisible, and therefore not subject to apportionment. But here again, if the sources rather than the charred debris are examined, then division on a reasonable and practical basis is quite feasible. Indeed, why is injury to real property from two sources of pollution regarded as capable of apportionment, but injury to real property from two fires not so capable? If injury to real property were truly indivisible, then the cause of injury makes no difference. The difference in outcomes is best explained by the fact that the Restatement authors are at some points talking about causes and sources, and at other points about resultant harms.

A second problem with the Restatement is the comment that authorizes an override of apportionment whenever one of the tortfeasors is insolvent.(75) This is an unprincipled basis for ignoring the black-letter rule. In every tort case, the plaintiff faces the risk that the defendant may be insolvent, uninsured, underinsured, absent from the jurisdiction, or otherwise not amenable to suit. In multiple-defendant cases, it is highly improbable that all the defendants will be financially incapacitated, but the plaintiff nevertheless runs the risk that one among them may be. In a single-defendant case, an insolvent defendant translates into no recovery for the plaintiff. In multiple-defendant cases, however, plaintiffs win always have much better odds of satisfying at least portions of the judgment. But there is no reason, if apportionment is otherwise called for, to dispense with it because one of the multiple tortfeasors is unable to satisfy its portion of the judgment. If that defendant is the only tortfeasor, the plaintiff recovers nothing. Because of multiple tortfeasors, the plaintiff will recover for all of the harm apportioned to those with sufficient resources. In addition, of course, it is in a plaintiffs interest to attempt to demonstrate that such solvent defendants should have most of the harm assigned to them.

It is one thing to conclude that joint and several liability should apply when a harm and its causes are indivisible(76) and cannot be apportioned on any basis, but it is quite another to declare that even though apportionment is appropriate, it should be barred because, one of the defendants is impecunious. Where the harm can reasonably be apportioned but is not - for so-called reasons of "injustice" - the defendant is subjected to an amount of liability that bears no necessary relation to the degree of responsibility" for the plaintiff's injury.(77)

Moreover, imposing joint and several liability may create risks of either overdeterrence or underdeterrence when a tortfeasor's compensation to the plaintiff is disproportionate to its contribution to the harm. As Professor Kenneth Abraham notes, because each defendant bears the risk of the other's insolvencies, it may have to pay both its own non-causal share and a portion of the shares of co-defendants."(78) This excessive liability will not result in optimum safety expenditures: a defendant who knows that, even if it invests in safety, it will have to pay for the failure of others to invest in equal levels of precaution will have an incentive to invest less, not more, for such purposes. Professor Richard Epstein also makes this point, in criticizing the effect of joint and several liability in CERCLA cases.(79)

At the other extreme, a "deep-pocket" defendant who, because it anticipates liability for insolvent or immune parties, does invest up to the level of its ultimate, disproportionate liability will be overinvesting in safety when that investment exceeds the losses which its tortious conduct has actually caused. This does not mean that joint and several liability should be rejected in all cases, but only that where apportionment is appropriate under the risk contribution model, it ought not be jettisoned solely because of the insolvency of one tortfeasor or any similar reason that may deny the plaintiff one hundred percent of her damages. Professors Mario Rizzo and Frank Arnold state the point convincingly:

Until now, we have implicitly assumed that all causally relevant defendants are present in the suit and that each is capable of satisfying the apportioned judgment. If, on the other hand, a tortfeasor either is not joined in the action or is judgment-proof, on whom should the burden of his causal contribution fall?

Causal apportionment demands that the burden rest on the plaintiff. If A and B jointly cause harm to C, and A's contribution is 70 per cent while B's is 30 per cent, C is entitled to collect only 70 per cent of the damages from A. C's inability to gain full redress from B is irrelevant to the question of A's responsibility to C, because the limit of a tortfeasor's responsibility to his victim is defined by his causal contribution.(80)

Thus, the Restatement, ostensibly motivated by a sense of corrective justice, requires defendants to pay more than their fair share to avoid an outcome of a plaintiff recovering nothing, and that is precisely the rationale offered for joint and several liability in Michie v. Great Lakes Steel Division(81) and Landers v. East Texas Salt Water Disposal Co.,(82) discussed below in part IV. If courts permitted apportionment to stand when it can be established, however, there would be virtually no prospect of plaintiffs going empty-handed from the courthouse. The chance that all tortfeasors would be insolvent is remote. Moreover, the Restatement takes a giant leap when it declares that even if apportionment is reasonable and practical, "injustice" to the plaintiff may require holding solvent defendants jointly and severally liable to minimize the risk that a plaintiff would not recover one hundred percent of her damages because one of the defendants is insolvent. It is one thing to justify joint liability to avoid a zero recovery when apportionment is deemed not possible and quite another to justify joint liability when the plaintiff will recover some damages and apportionment is reasonable and practical. Courts that rushed to adopt joint and several liability in cases of single, indivisible injury later seemingly forgot the underlying rationale for doing so and extended the principle to situations in which it had no application whatsoever (apportionable harms) and where the plaintiffs were not left without remedy. The situation in these multiple-defendant toxic tort cases is very analogous to market-share liability, in which at least some courts crafted their own version of market-share to permit proportionate liability only.(83) Those courts, such as the California Supreme Court in Brown v. Abbott Laboratories, Inc.,(84) are concerned that if the underlying theory is based on the probability of actual causation, absent or insolvent tortfeasors must diminish a plaintiffs recovery, lest remaining defendants end up paying more in judgments than what their tortious conduct actually caused. Thus, motivated by a concern for fairness to defendants, some courts refuse to apply joint and several liability in market share cases because it inevitably produces disproportionate liability.

A third problem with the current Restatement is the burden-shifting rule, set forth in section 433B. The plaintiff and defendants will have a legitimate conflict over how to apportion harm based on the risks each party created that resulted in the harm. But if apportionment is the rule and not the exception in toxic tort litigation, it will be in each party's interest to develop the most convincing factual record to support its proposed apportionment. In Dafler v. Raymark Industries, Inc.,(85) discussed in part III, it was the plaintiff, not the defendant, who offered the testimony that served as the basis for apportionment between smoking and asbestos exposure. Indeed, if toxic apportionment were the predominant rule, many cases that now go to trial based on the all-or-nothing extremes would probably be settled because of the ability to reach agreement on the comparative risks each created. That evidence, which would be toxicological and epidemiological, is more amenable to statistical comparisons and is therefore more likely to create reasonable foundations for settlement. Cases like Dafler would be less likely to go to trial on the issues of causation and damage; each party's experts Would have the available scientific studies, and any divergences in their conclusions could be the subject of mediation.(86) While fault or the basis of liability might still require a jury's determination, a major piece of controversy could be disposed of more economically.

III. The Plaintiff-Defendant Apportionment Paradigm

A. The Dafler Decision

One recent case that vividly illustrates the risk contribution model is Dafler v. Raymark Industries, Inc.,(87) in which the jury, relying on expert testimony regarding the risks each party created by its conduct, found that the plaintiff contributed seventy percent to his lung cancer and the defendant thirty percent and apportioned the damages on that basis.(88) The plaintiff, a shipyard worker, was exposed to asbestos fibers from 1939 to 1945 as a result of working in proximity to pipefitters who used asbestos-containing materials to cover pipes in the engine and boiler rooms. The record also showed that the plaintiff smoked a pack of cigarettes a day for forty-five years, until he was diagnosed with asbestosis in 1984. The plaintiff's two experts testified that the background or baseline risk for developing lung cancer for persons who do not smoke and are not exposed to asbestos was eleven cases out of one hundred thousand persons.(89) They testified further that the epidemiological data disclosed that people with occupational exposure to asbestos comparable to the plaintiffs had a relative risk of lung cancer of between 5:1 and 7:1, or a five- to seven-fold increase over the baseline risk, while the relative risk associated with cigarette smoking for that period of time was between 10: 1 and 12: 1, a ten- to twelve-fold increase over the baseline risk.(90) Both experts disavowed any apportionment of the cancer itself; neither could say how much of the cancer each risk factor caused. However, both testified that it was the synergistic effect of both risk factors that caused the cancer. The relationship between the two factors was not additive, but multiplicative, such that the relative risks of 5:1 for asbestos and 10:1 for smoking yielded a 50:1 relative risk for those persons exposed to both carcinogens.(91) The defendant's expert, in contrast, discounted any role for asbestos exposure, opining that smoking was the sole cause of plaintiff's lung cancer.(92)

The trial judge found that this evidence was sufficient to submit the issue of apportionment to the jury. The jury, obviously accepting the opinions of the plaintiff s experts, apportioned the damages almost exactly as the relative risks would suggest: roughly two-thirds to the plaintiffs smoking history and one-third to the defendant. The plaintiff argued on appeal that the trial judge erred in submitting the question of apportionment to the jury because the record demonstrated that the harm itself was indivisible and no reasonable basis was offered to enable the jury to undertake the apportionment.(93) The New Jersey Superior Court rejected that argument:

We conclude that there was ample basis in the record of this trial to submit the issue of apportionment to the jury. The extant legal precedent supports rational efforts to apportion responsibility in such circumstances rather than require one party to absorb the entire burden. The jury obviously accepted the epidemiological testimony based on relative risk factors .... The result was rational and fair We can ask no more. This is fairer than requiring the defendant to shoulder the entire causative burden where its contribution in fact was not likely even close to 100%. Or fairer, for certain, than no recovery at all for the plaintiff ....(94)

In reaching this conclusion, the court relied on section 433A of the Restatement (Second) of Torts and other authorities.(95) Although the plaintiff's experts provided the evidentiary basis for undertaking the apportionment, the burden of proof was on the defendant to justify the apportionment once the plaintiff offered proof that the defendant's products were "a" cause of the harm. If the burden was on the defendant, why was it that the plaintiff's experts provided the ammunition for apportionnment? Perhaps defense counsel thought the evidence might be so overwhelming as to the role of smoking causing the plaintiffs cancer to permit a jury to assign one hundred percent of the responsibility to the plaintiff, although that strategy has obvious risks. Conversely, perhaps plaintiff's counsel, too, was fearful of a verdict placing all the responsibility on the plaintiff and elicited the relative-risk testimony to increase the chances of at least a partial recovery. Nevertheless, prior to Dafler the weight of authority was against apportionment on these facts, and plaintiffs counsel may have felt that even the lesser relative risk for asbestos was more than adequate to render it "a" cause or contributing factor, thereby entitling him to full recovery; further, he may have believed that the trial court or the jury would not regard the relative risks alone as a sufficient basis for undertaking the apportionment.

The New Jersey court in Dafler got it right. The defendant, who was responsible for exposing the plaintiff to a significant risk of cancer, should have borne part of the loss, and the plaintiff, who also was responsible for exposing himself to an even greater risk of cancer, should have borne part of the loss. Imposing the entire loss on either party would work an injustice, resulting in the defendant bearing more of a loss than its risks comparatively created and in the plaintiff being freed of any responsibility for the risks he created. Conversely, placing the entire liability on the defendant would create overdeterrence by holding it liable for losses exceeding the risks it created; simultaneously, relieving the plaintiff of any liability would create disincentives for persons to engage in safer lifestyles. Apportionment on these facts produces optimum levels of deterrence for both parties and yields an outcome that is fair to all concerned.

Dafter also illustrates, although implicitly, another important principle: apportionment is not the same as causation.(96) From a purely medical and scientific standpoint, the plaintiffs smoking history is riot relevant to the question of whether asbestos exposure was "a" cause of his lung cancer. The following table shows the data for a group of insulation workers. The study data demonstrate that the relative risks are nearly identical, about 5.0, for both smokers and nonsmokers who are exposed to asbestos.
                               Table I
            Annual Lung Cancer Death Rate (per 100,000)(97)




Smoking         Asbestos            US            Relative
 Status        Insulators        Population         Risk


 Smoker           601.6             122.6            4.9
Non-Smoker         58.4              11.3            5.2


The implications of these data for litigation are significant: they suggest that two individuals who were exposed to comparable levels of asbestos would stand on equal footing as to their ability to demonstrate that asbestos exposure caused his or her lung cancer, even though one was a chronic smoker and the other a nonsmoker. Asbestos exposure multiplies the incidence of background lung cancer in all cases, whether or not the background includes smoking. A relative risk of about 5.0 translates into an etiological probability of 8096, meaning that exposure to asbestos causes 80% of the lung cancers among smokers and 8096 of the lung cancers within the nonsmoking group.(98) Philip Enterline, a noted epidemiologist, concludes, "I can think of no asbestos lung cancer cases where knowledge of cigarette smoking habits would help" in proving causation.(99) Thus, based on these data, the defendant's expert in Dafler was possibly wrong to opine that asbestos was not "a" cause.

The data in Table I also reveal another relevant phenomenon: smokers have a relative risk of about 10.0 compared to nonsmokers,(100) making it about 90% likely that smoking causes an individual's lung cancer, even though for the same person the probability of asbestos exposure being the cause is 80%.(101) The existence of a significant absolute interaction between asbestos and smoking in the production of lung cancer explains this apparent paradox. Therefore, in the great majority of cases, asbestos and smoking act together to produce lung cancer.(102)

Therefore, while the relative risk data on smoking relied on by the Dafler court bear on apportionment, such data may not, from a probabilistic standpoint, bear on whether asbestos exposure was "a" cause-in-fact. Both smoking and asbestos were sufficient causes of the plaintiffs lung cancer, based on the probability that each would have independently caused the plaintiffs lung cancer. In a sense, such lung cancer is akin to the twin fires cases: each fire was a sufficient cause and would have, standing alone, burned the plaintiffs barn. But if courts could attribute to each fire some dimension of magnitude (that is, a relative risk) they could rationally apportion the harm among these multiple causes.

Despite the persuasiveness of the result in Dafler, few courts have actually apportioned the harm in a similar fashion.(103) Indeed, in most of the reported cases involving asbestos exposure and smoking, neither party sought apportionment. Instead, plaintiff's and defense counsel usually offer conflicting testimony on the causation-in-fact issue only.(104) One would anticipate that the kinds of testimony offered in Dafler will now become more commonplace in future asbestos litigation, although such data have been available for over thirty years!

Finally, the kind of evidence offered in Dafler suggests another possible means for undertaking apportionment more favorable to the plaintiff. While the plaintiffs smoking increased his risk, standing alone, by a ratio of 10:1, he should not have to bear the burden of the synergistic effect with asbestos. The defendant's asbestos did not merely increase the plaintiff's risks by 5:1, but instead increased it, because of the multiplicative effect, from 10:1 to 50:1. Therefore, instead of comparing relative risks of 10.0 to 5.0, as the jury and court did in Daft", perhaps the relevant comparisons are 10.0 to 50.0, so that the defendant is responsible for five-sixths or eighty-three percent of the harm, and the plaintiff one-sixth or eighteen percent of the harm. The defendant, of course, would argue the converse, maintaining that the relevant comparisons are 5.0 to 50.0, so that the defendant is responsible for only one-eleventh or nine percent of the harm, and the plaintiff is responsible for ten-elevenths or ninety-one percent of the harm. The plaintiff's response to this argument is that because the defendant's acts were tortious and the plaintiff's were not, it is inappropriate that he bear the brunt of the synergistic effect. Although beyond the scope of this Article, the relative merits of these arguments deserve further consideration.

B. Other Courts' Approaches

Illustrative of decisions declining to permit apportionment is Martin v. Owens-Corning Fiberglas Corp.,(105) in which the Pennsylvania Supreme Court reversed the U& and intermediate appeals courts for permitting a jury's apportionment to stand. In Martin, the plaintiff was an insulation worker exposed to asbestos for thirty-nine years and who smoked for thirty-seven years.(106) The plaintiffs experts testified that the interaction of emphysema (attributable to smoking) and asbestosis (attributable to occupational asbestos exposure) caused his lung cancer and, similar to the testimony in Dafler, that it was not possible to apportion the lung disease itself between the two causes.(107) Also as in Daft", the defendant's expert opined that the plaintiffs cigarette smoking solely caused his lung diseases. However, neither side offered any epidemiological evidence or relative risks data, as were offered in Dafler. The Pennsylvania Supreme Court, in a plurality opinion, concluded that the jury "was provided no guidance in determining the relative contributions of asbestos exposure and cigarette smoking to [the plaintiffs] disability' and that "the jury cannot be expected to draw conclusions which medical experts ... could not draw."(108) It further observed that "`rough approximation' is no substitute for justice."(109) Three justices dissented, one commenting that "rough approximations" are acceptable,(110) and another observing that he was "at a loss to imagine what additional testimony would satisfy the majority" and eschewing a rule that would require "experts to speak in terms of numerical percentages," which would "introduce[ ] a false precision" and "[m]athematical exactitude ... not found in the real world of medicine."(111) In Dafler, the experts did indeed offer evidence involving numerical percentages to enable the jury to make its apportionment.(112) Whether the Martin plurality would have been satisfied had the record contained the same expert testimony as the parties offered in Dafler is questionable. The Martin plurality relied heavily on the experts' statements that while asbestos exposure and smoking "both play[ed] a significant role" and both were "important in producing the effect," it was nevertheless not possible to apportion the harm itself.(113)

One source of confusion here is that the medical experts in Martin and Dafler employed the term apportion' in a narrower sense than that contemplated by section 433A of the Restatement and considerably narrower than the model suggested here. Both of these sources envisage an inquiry not into the fact of causation or the actual division of the harm itself, but rather the feasibility and practical convenience of splitting up the total harm into separate parts which may be attributed to each of the two or more causes."(114) The Pennsylvania Supreme Court erred by focusing only on the cancer itself and failing to recognize that the risk factors that contributed to the cancer are critical in fashioning apportionment.

The Pennsylvania Superior Court in Martin,(115) which the Pennsylvania Supreme Court reversed as described above, also relied on section 433A of the Restatement in concluding that apportionment was appropriate on the record of the case.(116) Indeed, the superior court found that the record persuasively demonstrated the reasonable basis demanded by the Restatement:

Here, in contrast, appellant's witnesses's [sic] testified that the smokers'

diseases

of emphysema and chronic bronchitis as well as asbestosis were significant

"specific parts" of appellant's disability. Moreover, the jury saw the x-rays

and heard Dr. Wiot's explanation of the "specific parts" - bullae, indicative

of

emphysema, in the upper portion of the lungs, and pleural plaques, indicative

of asbestosis, in the lower portions. The jury knew, from appellant's

testimony,

the length of time he had been an asbestos worker and a smoker. That

the testimony did not establish the exact proportion that each disease

contributed

to appellant's disability suggests, not that the damages should not have

been apportioned, but only that medical science has not yet been able to

calculate

the proportions as exact percentages. This inability does not diminish the

fact that the causes of the harm were ... distinct and capable of rough

approximation.(117)

The superior court in Martin, unlike the supreme court, examined the contributing causes or sources of the harm and apportioned damages based on a comparison of the two sources, rather than looking solely to the apportionability of the end result, the plaintiff's disability. In addition, the record in this case was unusual because expert testimony identified different portions of the plaintiff's lungs that were affected by smoking and asbestos, thus enabling apportionment of the harm itself. Nevertheless, subsequent federal and state cases have followed the Pennsylvania Supreme Court's opinion in Martin disallowing apportionment.(118)

A few decisions arrive at apportionment of liability but rely upon rationales distinct from that in Dafler. For example, the Michigan Supreme Court in Brisboy v. Fibreboard Corp.(119) relied on comparative fault as the basis for apportioning the plaintiff's damages between smoking and asbestos exposure.(120) The decedent smoked two packs of cigarettes a day for thirty years and was exposed to asbestos for twenty-six years.(121) The plaintiff, the decedent's representative, sued nine asbestos manufacturers and distributors, but settled with eight before or during trial.(122) The decedent was exposed to the remaining defendant's asbestos products for six to nine months.(123) The jury found and the court affirmed that such a limited exposure could be a "substantial factor" in producing the plaintiff's fatal lung cancer condition.(124) The jury also found that the decedent's smoking history, based on the record, justified assigning him forty-five percent of the fault, with the defendant being assigned forty-five percent.(125) However, the trial court overturned this finding and the Michigan Court of Appeals affirmed on the grounds that the decedent was unaware that his smoking heightened his risk of developing lung cancer and, therefore, his smoking could not be considered fault or negligence.(126) The Michigan Supreme Court reversed and reinstated the jury's finding, on the basis that the risk of developing lung cancer was within the scope of the risks assumed by a smoker, even though other factors, such as asbestos exposure, enhance that risk.(127) The court concluded that the record provided a "rational basis for the jury's apportionment of fault."(128)

C. Shortcomings of Comparative Fault

Under the risk contribution and causation model advanced in this Article, the approach in Brisboy v. Fibreboard Corp.(129) is to be faulted for several reasons. First, by compelling the jury to allocate the damages between only the remaining defendant and the plaintiff, over twenty-five years of contributing causes are ignored. At no point is the trial court or the jury afforded the opportunity to examine what risk contribution six to nine months of exposure to the defendant's asbestos made as compared to over twenty-five years of exposure to other manufacturers' asbestos products.(130) The problem is that the substantial factor test of causation obfuscates the question of apportionment: although the record may have justified an affirmative finding on this causation-in-fact inquiry, epidemiological evidence most assuredly would not have justified assigning the defendant forty-five percent of the harm based on the risks that it created through its tortious conduct.

Second, Brisboy illustrates the difficulties of relying on comparative fault or negligence as the basis for apportioning damages. In Michigan, as in many jurisdictions, the jury essentially must find that the plaintiff's conduct constituted contributory negligence or assumption of the risk. The comparative fault system carries a lot of baggage that can. be circumvented by applying an apportionment model that looks only to the rational divisibility of the harm and the risks contributing to that harm. Section 433A of the Restatement (Second) of Torts(131) makes clear that apportionment can be predicated on innocent causes, including the plaintiff's conduct, even when the plaintiff was not negligent or otherwise at fault.(132) While smoking could be regarded as "negligence," why require such a finding by the jury? Many aspects of lifestyle, such as moderate or excessive consumption of alcohol or unhealthy dietary habits, can cause or contribute to harm but should not necessarily be denominated "fault." Further, in some jurisdictions juries can assign fault only to tortfeasors who are before them,(133) and thus no fault is assigned to absent tortfeasors.(134) In contrast, the risk contribution model of apportionment would allow the jury to divide the harm among all risk contributors, parties and nonparties,(135) settling and nonsettling defendants.(136)

Reliance on comparative fault as the basis for apportionment has another problem. In some states the application of comparative fault modifies or eliminates joint and several liability.(137) While that fact, standing alone, might not preclude apportionment of harm, some states have explicitly carved out an exception for environmental tort litigation, declaring that joint and several liability will continue to be applied in such cases even though the courts or legislatures eliminated or modified it for other tort cases. For example, the New Jersey Comparative Negligence Act(138) altered joint and several liability so that only defendants determined to be sixty percent or more responsible for damages would be liable for the total amount of an award. A defendant found to be more than twenty percent but less than sixty percent responsible for the damages would be liable for the entire amount of economic loss, but only for that percentage of noneconomic loss directly attributable to its fault. A defendant found to be responsible for twenty percent or less of any damages would be liable only for the percentage of the award directly attributable to its fault. Consequently, defendants found only one percent responsible can no longer be held liable for the entire amount of an award. This limitation is significant in toxic tort and environmental claims, in which the value of damages can be enormous. The statute, however, contains an exception for environmental tort actions: "With regard to environmental tort actions, the party so recovering may recover the full amount of the damage award from any party determined to be liable."(139) An "environmental tort action" is defined as "a civil action seeking damages for personal injuries or death, where the cause of the damages is the negligent manufacture, use, disposal, handling, storage or treatment of hazardous or toxic substances."(140) New Jersey's Product Liability Act,(141) enacted at the same time as the Comparative Negligence Act, similarly exempts environmental tort actions.(142) In that Act, an "environmental tort action" is defined as "a civil action seeking damages for harm where the cause of the harm is exposure to toxic chemicals or substances, but ... not ... actions involving drugs or products intended for personal consumption or use."(143)

In Stevenson v. Keene Corp.,(144) the New Jersey Superior Court held that asbestos cases constituted "environmental tort actions" and hence the defendants were not entitled to the abolition of joint and several liability.(145) However, the New Jersey Supreme Court, in a per curiam affirmance, observed that the "environmental tort action exception" should not be enlarged to include every conceivable injury involving a toxic product.(146)

Apparently the court in Dafter was not concerned with this statute because only one defendant was involved. However, it is not implausible that in multiple-defendant toxic tort cases a court, looking only at fault percentages, would have to retain joint and several liability. If the court based the apportionment not on comparative negligence or fault but on the principles described here, these statutes would arguably have no relevance. In order for them to apply, the defendants must be regarded as joint tortfeasors; that, in turn, implies the existence of indivisible harms.(147) By apportioning the harm, courts have no occasion to conclude that the defendants are joint tortfeasors.(148) For all of these reasons, therefore, courts should avoid depending on comparative fault as the mechanism on which to base apportionment.

The decisions in this section vividly demonstrate that courts have struggled with the concept of apportionment. Until Dafler, no court had offered a cohesive rationale that could effectively resolve the problems created when multiple toxic sources or causes contribute to the harm. Dafler at least serves as one example that a risk contribution and causation model of apportionment can function effectively.

IV. The Multiple-Defendant Paradigm

A. The Now-Traditional Approach to Multiple-Defendant,

Joint-Cause Cases

Two cases illustrate the courts' treatment of multiple-defendant cases: Michie v. Great Lakes Steel Division(149) and Landers v. East Texas Salt Water Disposal Co.(150) In Michie, thirty-seven plaintiffs living in Canada sued three corporations operating seven plants across the Detroit River from Canada. The plaintiffs alleged that the plants discharged pollutants that air currents then carried onto their properties, injuring their persons and property.(151) The question on appeal, as the Sixth Circuit framed it, was whether these multiple defendants - acting independently in discharging pollutants and creating a nuisance - could be jointly and severally liable for the plaintiffs' injuries "where said pollutants mix in the air so that their separate effects in creating the individual injuries are impossible to analyze."(152) The court, applying Michigan law, rejected the rule of the Restatement (First) of Torts(153) and older Michigan cases(154) that required the plaintiff to apportion the harm among multiple causes and instead chose to rely on later Michigan cases involving automobile collisions(155) that applied joint and several liability in cases of single indivisible injury. Although recognizing a conflict in the decisions, the court ruled that once the plaintiff proves the harm is indivisible, the net effect of "the plaintiff's right to recover for his harm should not depend on his ability to apportion damage but that this is a problem which is properly left with the defendants themselves."(156) Thus, once the plaintiff's proved both injury and tortious conduct by the defendants, "Michigan's new rule [was] to shift the burden of proof as to which one was responsible and to what degree from the injured party to the wrongdoers."(157)

The one non-Michigan decision that the Sixth Circuit quoted in Michie was Landers v. East Texas Salt Water Disposal Co.,(158) which the district court also relied on. In Landers, the plaintiff alleged that each of the two defendants, one a saltwater disposal firm and the other an oil company, had independently polluted the plaintiff's lake, killing fish. The saltwater company negligently permitted ten to fifteen thousand gallons of salt water to escape from its pipeline and flow over the plaintiff's land and into his lake, while the oil company negligently permitted "large quantities' of oil and saltwater to escape from its pipeline into a stream and thence into the plaintiff's lake.(159) The trial court dismissed the plaintiff's claims for damages because he refused to maintain two separate suits, one against each defendant;(160) joint and several liability against both defendants was not permissible under existing Texas law.(161) The Texas Supreme Court recognized that prior precedents required the plaintiff to prove what portion of his damages each defendant caused(162) and acknowledged that "the courts of the country seem to be virtually unanimous in refusing to impose joint and several liability on multiple wrongdoers whose independent tortious acts interfere with a landowner's interest in the use and enjoyment of land by interfering with his air or water."(163)

Nevertheless, the Texas Supreme Court, motivated by a strong sense of corrective justice, observed that placing the burden on the plaintiffs to prove what share each of two wrongdoers contributed to the harm was just as "intolerable" as it was in cases of concert of action and common design, in which courts have applied joint liability for decades.(164) In addition, the court was struck by the injustice of the plaintiff recovering no damages at all because of the inability to prove with sufficient certainty, under controlling damages jurisprudence, the portion of his injuries attributable to each defendant.(165) The court concluded that under then-prevailing rules, the law "embraced the philosophy ... that it is better that the injured party lose all of his damages than that any of several wrongdoers should pay more of the damages than he individually and separately caused."(166) The court then overruled prior law and stated its new rule:

Where the tortious acts of two or more wrongdoers join to produce an

indivisible

injury, that is, an injury which from its nature cannot be apportioned with

reasonable certainty to the individual wrongdoers, all of the wrongdoers will

be held jointly and severally liable for the entire damages and the injured

party

may proceed to judgment against any one separately or against all in one

suit.

If fewer than the whole number of wrongdoers are joined as defendants to

plaintiff's suit, those joined may by proper cross action under the governing

rules bring in those omitted.(167)

B. Exploring Their Rationale

Michie and Landers each pronounced new law, at least as to nuisance cases, that authorized joint and several liability where injuries are indivisible or cannot be apportioned with reasonable certainty.(168) In both cases, the principal rationale was that if the court did not create such joint liability, the plaintiff would recover nothing. In other words, they were driven by a compelling need for fairness that could not be fulfilled in any other feasible manner: either each defendant must be jointly and severally liable, and thereby liable for more harm than it caused, or the plaintiff would be without remedy.

However, the risk contribution model of toxic apportionment advanced here provides an alternative to these inequitable results. Under the risk contribution model the plaintiff would recover all of his proven damages (for example, in Landers, all of the fish killed, water polluted, and economic consequences of such harm), but the presumption would be that each defendant would pay only for that portion of the total harm commensurate with the risks it created. As noted earlier, section 433A of the Restatement (Second) of Torts(169) would permit apportionment on the facts in Landers based on the volume of salt water (or oil) each defendant contributed.(170) As to Michie, the opinion contains so few facts that it is not possible to opine on what basis the court might have established apportionment; the opinion identifies neither the nature or quantity of the emissions nor the nature of the plaintiffs' injuries. The Landers opinion reveals the quantity and content of the toxic substances and the nature of the harm suffered, so it is much easier to offer in hindsight a feasible basis for undertaking an apportionment in that case. Pretrial discovery in Michie might have produced more detail respecting such matters. The critical point is that the rigorous application of apportionment principles avoids the Hobson's choice facing the courts in Michie and Landers: to impose joint and excessive liability on each defendant or no liability at all.

Moreover, neither the Landers nor the Michie court ever said they would impose joint and several liability even if apportionment were feasible. These decisions never went so far as the Restatement (Second) has done, to hold that "justice" requires imposing joint liability if one of the tortfeasors is absent or insolvent even though apportionment is otherwise entirely feasible and reasonable under subsection 433A(1)(b).(171)

C. Taking Plaintiffs and Defendants as They Are Found: Of Thin

Skulls and Insolvency

In cases comprising a single plaintiff and single defendant, one defendant's status as uninsured, underinsured, insolvent, immune, or absent will affect the plaintiffs recovery, because tort law makes no guarantee regarding any party's ability to satisfy a judgment. This fact is well known to courts and legislatures, which rarely require a significant level of liability insurance to engage in a particular business (or even to drive an automobile). (172) Thus, in these cases the plaintiff faces the risk that the defendant may be unable to satisfy a judgment. Similarly, both legislatures and courts have created immunities that permit some entities to engage in tortious conduct while shielding them from liability for damages, with the consequence that an injured plaintiff will recover nothing.

Should multiple-defendant cases differ in this regard? The law's desire to make the plaintiff whole is no different in either case, yet the realities of the accident "marketplace" - the lottery of a given defendant's wealth or immunity - compels the plaintiff to take the defendant as she finds him, economically speaking.(173) If defendants must respond in damages to the thin-skulled plaintiff, who sustains an extent of loss not within reasonable expectations,(174) should not plaintiffs also bear damages for the thin-pocketed defendant? In multiple-defendant litigation, the plaintiff is always better off because the odds are that at least one defendant will be able to satisfy its portion of the plaintiffs damages. However, tort law should not encourage a suit only against the well-heeled defendant, which is what happens when toxic apportionment is unavailable and joint and several liability is imposed. Instead, the law should provide plaintiffs with incentives to sue all of the responsible parties - all who created risks that were material to the resulting harm. Indeed, if the plaintiff sues all responsible parties, she will increase the amount of damages collected under the apportionment model. Requiring a single defendant to bring third-party actions against other potential tortfeasors and thereby spread the loss among more of the risk creators is inefficient. Instead, judicial efficiency increases if, in one unitary proceeding, one judge and one factfinder hear all of the evidence on every party's contribution to the harm and render a comprehensive apportionment on the basis of that evidence. The more parties the plaintiff joins, the greater the likelihood that most will be able to satisfy a judgment for their respective allocated share of the plaintiff's total damages.(175) When plaintiffs' lawyers believe there is some possibility of apportionment, which might serve to limit their clients' recoveries, they will be more likely to join all risk creators whose conduct contributed to the plaintiffs' damages.

Under existing practice, because apportionment is the exception rather than the rule in tort litigation generally and toxic tort cases specifically, the plaintiffs' sound strategy is to sue the deep-pocketed defendants, knowing that joint and several liability will afford a full recovery. If toxic apportionment were the rule in cases involving multiple contributors to the harm, plaintiffs' lawyers would be less willing to risk suing fewer than all contributors because apportionment would limit each defendant's share of damages to the portion fairly attributable to it.

With joint and several liability less available, both sides would have an incentive to offer evidence on the apportionment of the harm: each defendant would argue that there is a reasonable basis for comparing the risks that each created, with each also seeking to minimize its share, while the plaintiff would maintain that no reasonable basis for apportionment exists, or failing that, would argue for an apportionment that maximizes recovery. In a sense, such apportionment is akin to market-share liability, in which many courts apply solely proportionate liability, creating incentives for plaintiffs to sue as many of the potentially responsible parties as possible to maximize the aggregate "market share" that the defendants represent.(176) With a doctrinal preference for apportionment, parties would have an interest in joining all those responsible, because potential parties' absence would reduce the plaintiff's recovery and make it more difficult for the factfinder to apportion the entire harm. Moreover, to assure even greater incentive to the plaintiff to join all responsible parties, a defendant could offer proof of the risks absent parties created, and the jury could assign proportionate shares to such "phantom" nonparties. Thus, this apportionment scheme would differ from those comparative fault statutes that allow the jury to assign percentages of fault only to those before the court. The current Restatement (Second) of Torts(177) authorizes precisely this assignment of a portion of the harm to absent causes, be they tortious or innocent.(178) The plaintiff would aim to offer evidence seeking to minimize the absent shares, while the defendants would aim to maximize such shares.

V. Multiple Defendants: The CERCLA Experience

A. The First Decade

CERCLA is a tort-like statute, so far as its liability scheme is concerned, and it provides a wealth of jurisprudence on the application of Restatement(179) principles to toxic harms in general and to multiple-defendant cases in particular. CERCLA is nevertheless not entirely "tort-like" in its application because courts interpret it to dispense with tort law's demand for strict causation proof(180) and because of a judicially gleaned congressional preference for retroactive, strict, and joint and several liability.(181) As a result, whereas tort law requires that a plaintiff satisfy either the "but-for" or "substantial factor" test of cause-in-fact,(182) CERCLA contains no such explicit requirement and the courts have consistently so ruled.(183) If the courts require strict causation as part of the plaintiffs' proofs in CERCLA cases, satisfying that burden of proof would be extremely difficult for at least four reasons: 1) hazardous waste sites often contain many different kinds of hazardous substances difficult to identify;(184) 2) the migration and commingling of such wastes compound that difficulty;(185) 3) hazardous waste, may come from a multitude of sources, some of which may no longer exist;(186) and 4) the passage of years between the time of disposal and cleanup often results in unavailable relevant documents and knowledgeable witnesses.(187) Because of these practical difficulties and the statutory silence regarding a causation requirement, it is not surprising that federal courts, especially when faced with a governmental plaintiff seeking to reimburse the Superfund for monies expended on site remediation,(188) conclude that the government is not required to trace how hazardous waste from a particular defendant incurred response costs. For example, in United States v. Monsanto Co.,(189) the Fourth Circuit observed that "Congress knew of the synergistic and migratory capacities of leaking chemical waste, and the technological infeasibility of tracing improperly disposed waste to its source."(190) In addition, the inclusion of three statutory causation-based affirmative defenses, in which the defendants bear the burden of proving that they were not the cause of environmental harm at the site, implies the absence of any causation requirement.(191)

Some of the same considerations influencing judicial interpretation regarding causation also influence courts to hold defendants jointly and severally liable, thereby rejecting defendants' efforts to limit their liability through apportionment of the harm.(192) For example, in Monsanto, three defendant generators whose waste was shipped to a disposal site argued that although the release of the various hazardous wastes at the site produced a "single harm," there nevertheless existed a rational basis for apportioning harm based on the volume each shipped to the site.(193) The Fourth Circuit, finding the Restatement (Second) of Torts section 433A to provide the governing tort principles, rejected the defendants' position because they failed to satisfy their burden of proof:

To meet this burden, the generator defendants had to establish that the environmental harm at [the site] was divisible among responsible parties. They presented no evidence, however, showing a relationship between waste volume, the release of hazardous substances, and the harm at the site. Further, in light of the commingling of hazardous substances, the district court could not have reasonably apportioned liability without some evidence disclosing the individual and interactive qualities of the substances deposited there. Common sense counsels that a million gallons of certain substances could be mixed together without significant consequences, whereas a few pints of others improperly mixed could result in disastrous consequences. Under other circumstances proportionate volumes of hazardous substances may well be probative of contributory harm. In this case, however, volume could not establish the effective contribution of each waste generator to the harm at the ... site.(194)

Finally, the Monsanto court held that the district court acted within its discretion in refusing to apportion liability among all the defendants pursuant to the contribution provisions of CERCLA section 113(f)(195) and instead choosing to defer the contribution action until "plaintiff has been made whole."(196) Thus, the court held the generator defendants jointly and severally liable because of the practical difficulties in apportioning the harm and CERCLA's statutory preference to fully reimburse the government, forcing the defendants to exercise their contribution rights against all other potentially responsible parties. In other words, courts interpret CERCLA so that apportionment is the exception, rather than the rule, believing that such an approach would facilitate cleanups by enabling the government to recover entire cleanup costs from one or a few responsible parties without having to sue all of the liable parties, thereby foregoing the litigation expenses of proving each party's contribution to the harm at the site.(197)

After twelve years of uniform judicial rejection of apportionment, however, courts in CERCLA actions are finally beginning to be considerably more receptive to apportionment. There are at least two reasons for this attitudinal change: first, the practical difficulties in finding a rational basis for apportionment may not always be insurmountable; and second, compelling one or a few defendants to pay the entire cost of removal or remediation, especially when contribution rights do not exist, is unfair.(198)

B. Recognition of Apportionment: The Alcan Cases

Two recent decisions involving the same defendant cracked the government's string of successes with joint and several liability. However, the decision in United States v. Alcan Aluminum Corp. (Alcan-New York),(199) in which the Second Circuit adopted a divisibility of harm affirmative defense to CERCLA liability, does not quite meet the standard the Third Circuit set in United States v. Alcan Aluminum Corp. (Alcan-Butler).(200) While the government often uses the threat of joint and several liability as a sword to bludgeon defendants into settlement, both Alcan decisions make clear that defendants can also use common-law principles, on which joint and several liability is based, as a shield to limit their liability for cleanup costs.

Both the Second and Third Circuit Alcan decisions arose from CERCLA cost recovery claims by the United States against Alcan Aluminum. At each disposal site, Alcan arranged for disposal of an emulsion used in its aluminum sheet and plate products manufacturing process.(201) The Alcan material was commingled with various hazardous substances generated by other potentially responsible parties (PRPs). Alcan's emulsion was mostly water and mineral oil, but the Government maintained that it contained small amounts of aluminum shavings with less than background concentrations of lead, copper, chromium, zinc, and cadmium compounds.(202)

Given that the same material was at issue in both cases, it is not surprising that Alcan made identical arguments in the two courts. In Alcan-New York, the district court granted the Government's motion for summary judgment, holding Alcan jointly and severally liable for the entire harm at the site. The Second Circuit reversed, finding that genuine issues of material fact existed as to whether Alcan's material contributed to a divisible harm, if any, or whether it contributed at all to the release or cleanup costs. Both decisions, however, reaffirmed the principle that the Government is not obligated to prove that a specific defendant's waste caused the incurrence of response costs.(203)

Turning to Alcan's divisibility arguments, the Second Circuit in Alcan-New York tellingly focused first on the potentially "harsh result" that "there is no limit to the scope of CERCLA liability."(204) To avoid that "harsh result," the court looked to the Restatement (Second) of Torts section 433A for the proposition that when joint tortfeasors cause a single harm for which there is a reasonable basis for division, each is liable for damages only for its own portion of the harm.(205) Holding first that Alcan could avoid all (or a portion of) the liability if it could "prov[e] that its oil emulsion, when mixed with other hazardous wastes, did not contribute to the release and the clean-up costs that followed, or contributed at most to only a divisible portion of the harm"(206) the court placed the burden on Alcan, as the defendant, to establish a reasonable basis for apportioning liability and noted that "apportionment . . . is an intensely factual determination."(207)

The court "candidly admit[ted]" that, after having concluded that the Government need not prove that a specific PRP's waste caused the incurrence of cleanup costs, "causation is being brought back into the case - through the backdoor."(208) In seeking to explain this apparent contradiction, it cautioned that a defendant must show that "its pollutants did not contribute more than background contamination and also cannot concentrate."(209) It also noted that "this limited exception [applies] only in the absence of any EPA thresholds" for the substance in question.(210)

The Second Circuit explicitly rejected the Government's position that commingling of wastes is synonymous with indivisible harm. To avoid summary judgment on divisibility, Alcan "need only show that there are genuine issues of material fact regarding a reasonable basis for apportionment of liability."(211) Based on its analysis of expert affidavits presented by both the Government and Alcan, the court found that questions of fact precluded summary judgment on the divisibility issue. Finally, it recognized the "common sense approach" of resolving divisibility claims during the initial liability phase (the approach adopted in Alcan-Butler), but ultimately left the timing question to the trial court's discretion.(212)

The Second and Third Circuit Alcan decisions are similar in that both establish the appropriateness of causation-based affirmative defenses and both rely on section 433A of the Restatement regarding apportionment of liability. However, there are several subtle but important differences between the cases. First, under Alcan-Butler, so long as the defendant's material "did not or could not, when mixed with other hazardous wastes, contribute to the release and the resultant response costs, then [the defendant] should not be responsible for any response Costs."(213)

Furthermore, in Alcan-Butler the Third Circuit quoted from the Restatement section 433A, comment d, to show that "the drafters of the Restatement found that joint pollution of water is typically subject to the divisibility rule."(214) Although both circuit courts held that commingling is not synonymous with indivisible harm, the Third Circuit's opinion may provide somewhat more support for an argument that a single harm can be reasonably apportioned based on volume contributed by the various

Finally, the Second Circuit in Alcan-New York did not establish a rule as to when during the proceedings a trial court should address the divisibility issue, whereas the Third Circuit indicated that the trial court should decide divisibility at the initial liability phase of the case, before a determination of joint and several liability.(215) The court expressed concern that "the logical consequence of delaying the apportionment determination may well be drastic, for it seems clear that a defendant could easily be strong-armed into settling where other defendants have settled in order to avoid being held liable for the remainder of the response costs."(216) Although it preferred the Third Circuits "common sense approach," the Second Circuit thought that fixing liability first for enforcement purposes and then later litigating the contribution from other responsible parties "may be contrary to the statutory dictates of CERCLA."(217)

Four major points emerge from the Alcan decisions that bear on the availability of apportionment. First, two federal circuit courts flatly rejected the Government's position that the commingling of hazardous wastes necessarily precludes apportionment. Significantly, both courts afforded Alcan the opportunity on remand to prove that the harm caused by its wastes was distinguishable from the harm caused by others' wastes, through the introduction of evidence respecting "the relative toxicity, migratory potential, degree of migration, and synergistic capacities of the hazardous substances at the site."(1218) Second, both courts expressed a preference for trial courts to address the divisibility issue at the initial liability phases of the litigation, before any determination of joint and several liability.(219) Third, both courts suggested that Alcan might avoid all liability if it could show that its wastes, even when combined with those of other responsible parties, "did not contribute to the release and the clean-up costs that followed."(220) Finally, the two courts also recognized the "intensely factual nature of the divisibility issue"(221) and the need for district courts to reject summary judgments in favor of full hearings to resolve this issue before authorizing the imposition of joint and several liability.

C. Further Recognition of Apportionment: Bell Petroleum

The Alcan decisions are not alone in expressing a willingness to allow apportionment in a CERCLA case.(222) In In re Matter of Bell Petroleum Services (Bell Petroleum),(223) the Fifth Circuit Court of Appeals issued a lengthy opinion in which a two-member majority ruled that when a defendant can show a "rough approximation" of its contribution to a single injury, the imposition of joint and several liability is inappropriate.(224) It held that "[t]he fact that apportionment may be difficult, because each defendant's exact contribution to the harm cannot be proved to an absolute certainty, or the fact that it will require weighing the evidence and making credibility determinations, are inadequate grounds upon which to impose joint and several liability."(225) Moreover, the court ruled that Sequa Corporation, one of the defendants, had met its burden of proving, as a matter of law, that such a reasonable basis existed because it demonstrated that it was possible to approximate the volumetric contribution of each of three successive operators of the plating shop from which chromium waste originated.(226)

This case was somewhat unusual because it involved three successive operators of a chrome-plating shop, with each operator discharging chromium-contaminated rinse water that eventually contaminated ground-water.(227) EPA designated a twenty-four-block area around the facility a Superfund site, conducted a remedial investigation, and prepared a "focused" feasibility study to evaluate the need to provide an alternate water supply, which was constructed.(228) The Fifth Circuit, making the same observation as the Alcan courts, stated that

CERCLA, as a strict liability statute . . . can be terribly unfair in certain

instances

in which parties may be required to pay huge amounts for damages to

which their acts did not contribute. Congress recognized such possibilities

and

left it to the courts to fashion some rules that will, in appropriate

instances,

ameliorate this harshness.(229)

The court further noted that the nature of the harm is key in deciding whether apportionment is appropriate under section 433A of the Restatement, including consideration of whether there are distinct harms, "successive" harms, or single harms that are nonetheless divisible.(230)

After analyzing in detail several cases addressing joint and several liability under CERCLA, the court chose to follow what it called the Chem-Dyne approach.(231) From its review of CERCLA jurisprudence, it distilled several "basic principles": 1) joint and several liability is not mandated, but courts should impose it when appropriate under common-law principles; 2) all cases rely on the Restatement in resolving joint and several liability issues; and 3) even when there are commingled wastes of unknown toxicity, migratory potential, and synergistic effect, PRPs can attempt to prove that a reasonable basis for apportionment exists ("although they rarely succeed," the court noted), and when such factors are not present, volume may be a reasonable means of apportionment.(232) Further, the Fifth Circuit said it preferred early resolution of the divisibility inquiry, but left the matter up to the trial court's discretion.(233)

Applying those considerations, the court held that the district court erred in determining that there was no reasonable basis for apportionment because Sequa had met its burden of proving by a preponderance of the evidence that a "rough approximation" was possible.(234) In contrast to many CERCLA actions, Bell Petroleum involved only one hazardous substance - chromium - and "no synergistic effects."(235) Because the chromium "entered the groundwater as the result of similar operations by three parties who operated at mutually exclusive times," and because "it [was] reasonable to assume that the respective harm done by each of the defendants [was] proportionate to the volume of chromium-contaminated water each discharged into the environment," the harm could be reasonably apportioned.(236) Examining the evidence, the court said that while the exact contribution of each operator could not be known, a "reasonable and rational approximation of each defendant's individual contribution to the contamination can be made."(237) In a footnote, responding to the dissenting opinion,(238) the Court clarified further that "evidence sufficient to permit a rough approximation is all that is required under the Restatement."(239) It said Sequa could rely on the calculations of its expert in making its case, even in the face of EPA's competing expert analysis.(240)

The decision in Bell Petroleum should be applauded for its recognition of the harshness of joint and several liability and its conclusion that the record presented a reasonable basis for apportionment based on a volumetric analysis of the defendants' contribution to the harm at the site. The Bell Petroleum case, the two Alcan decisions, and a couple of district court opinions(241) refusing to authorize joint and several liability appear to reveal a growing open-mindedness among the federal judiciary regarding the feasibility of apportionment.(242)

D. Some Observations on Apportionment Under CERCLA

CERCLA liability is analogous to toxic tort liability, and the foregoing CERCLA cases breathe life into apportionment in toxic tort litigation by demonstrating that arguments for divisibility can succeed.(243) By allowing defendants to show what harm, if any, their wastes cause to the site's environment and what response costs are attributable to such wastes, the courts are returning to Restatement principles. For example, the Alcan cases show that synergistic properties of toxic waste do not ipso facto preclude divisibility of the harm. Of course, Dafler itself vividly illustrates this point - smoking and asbestos exposure function synergistically, but the factfinder could nevertheless rationally apportion the harm based on testimony regarding the substances' relative toxicity. In addition, these CERCLA cases clearly indicate that apportionment may require consideration of the effects of comparative toxicity, migratory potential, and actual migration in determining the contribution of each defendant's hazardous waste to the harm at the site. These cases also demonstrate that whether to permit divisibility of liability and reject joint and several liability is entirely a technical, scientific inquiry having nothing to do with culpability, cooperation, fault, or any other such conduct factors that may bear on the allocation of costs in a contribution action.(244)

The real message of the Alcan and Bell Petroleum decisions is that courts are willing to look seriously at evidence of apportionment and not assume, on the basis of the Government's allegations, that joint and several liability is required. And while the language in the opinions often examines the "nature of the harm," rather than the contributions of each source to the harm, these cases imply at least that evidence of the magnitude of the causal contribution attributable to each source of a toxic substance is becoming the gravamen of divisibility.

E. Some Further Comments on Deterrence and Fairness

Joint and several liability can create problems in achieving optimal levels of deterrence.(245) Even with strict liability, the courts encourage a defendant who is solely responsible for the harm to invest in pollution control technology that will reduce its liability commensurately. Thus, even recognizing the difficulty of estimating costs and the high transaction costs involved in applying CERCLA's liability structure, firms still retain, under strict liability rules, some incentive to achieve the proper balance between accident costs and accident avoidance costs. However, CERCLA's joint and several liability rules destroy this symmetry: an enterprise that contributes a small amount of toxic waste to a site could be liable to the government for the entire remediation cost, as the Alcan cases vividly illustrate. Professor Epstein demonstrates the difference that joint and several liability produces:

[S]uppose that a defendant is the only firm whose waste causes total losses of

$2000 at the dumpsite. Suppose further that this defendant can, at a cost of

$500, eliminate half of his toxic substances, which themselves would cause

harm at $1000. As a single defendant, that firm would normally be expected to

undertake the precautions, because the $500 in safety translates into a $1000

reduction of liability for a net gain to the firm of $500.

Now suppose there are ten defendants, of whom each contributes $2000

to a total loss of $20,000. If one defendant now spends $1000 to clean up its

own waste, it reduces (if the level of activity by other parties remains

constant)

the total social cost from $20,000 to $19,000. That defendant should

therefore

be encouraged to undertake the act. Nonetheless, under a rule of joint and

several liability, that defendant's anticipated share of the loss is reduced

from

$2000 (its one in ten chance of being held for the entire loss) to $1900, for

an

anticipated savings of only $100. At this point, the private gain is far lower

than the social gain, so the party has an incentive to forego the expenditure,

ninety percent of which benefits other parties who ship wastes to the

dumpsites.(246)

Under CERCLA's joint and several liability rules, a firm would find solace only if it reduces its wastes to zero, when nonliability would presumably occur. Thus, faced with the stark realities of joint and several liability, a fim contemplating waste avoidance or waste reduction technology must ask itself whether those investments are worth making since it faces the risk of full liability regardless of whether it contributed at the level of one percent or ninety-nine percent. Why make the investment when the real beneficiaries (to the tune of ninety percent in Epstein's hypothetical) are the other contributors to the waste disposal site? Arguably, therefore, the incentive that joint and several liability creates is not the incentive to invest in risk-reducing technologies but the incentive to reduce such investments and let others bear the remediation costs.(247)

In contrast, apportionment or divisibility of the harm offers a means to minimize such disincentives. By permitting each defendant to limit its liability to only the harm that its risk-creating activities produce, apportionment again encourages that party to invest in risk-reducing technologies. Courts facing Government efforts to impose joint and several liability should become more receptive to defendants' efforts to demonstrate that the harm, and the comparative contributions of each to that harm, are capable of apportionment on a reasonable, albeit imperfect or inexact, basis. The decisions in Alcan and Bell Petroleum should at least demonstrate that apportionment is feasible in many cases and should serve as an example that carries over to non-CERCLA toxic tort litigation.

The final concern in these cases is the potential overdeterrence of defendants with greater wealth because a rule of joint and several liability encourages the Government to name the "deep-pocket" generators as defendants in order to recover its full response costs. To the extent that such defendants can predict the existence of judgment-proof or low-visibility defendants or predict the Government's enforcement strategy, this increased expected liability may result in overdeterrence. In contrast, the application of apportionment rules will not lead to larger liabilities (with attendant overdeterrence) for defendants of greater wealth solely because of their wealth. The Government, rather than the wealthy defendants, will bear the burden of judgment-proof defendants, through reduced recovery. Here, too, apportionment should reduce the risks of overdeterrence by signaling to prospective defendants, especially deep-pocket firms, that they can limit their liability for the harm that their waste activities actually cause or for their comparative contribution to the harm. By developing a continued receptivity to apportionment, courts will be telling firms that if they can present reasonable bases for dividing the response costs, they can anticipate real limits on their liabilities.

What about fairness concerns? The courts' statements in Alcan and Bell Petroleum respecting the "harshness" of imposing joint and several liability were directed precisely at that concern. In Bell Petroleum, recognition that CERCLA could be "terribly unfair" in cases in which defendants must reimburse the Government for "huge amounts for damages to which their acts did not contribute"(248) influenced the court's willingness to fashion rules "to ameliorate this harshness."(249) Similarly, the Third Circuit in Alcan-Butler was expressing a fairness concern when it commented that "delaying the apportionment determination" [to the contribution phase] may have "drastic" consequences because defendants may be "strong-armed" into settling in order to avoid the prospects of liability for the remainder of the response costs.(250) Finally, in Alcan-New York the Second Circuit also expressed concern with the "harsh result" that derives from the fact that "there is no limit to the scope of CERCLA liability."(251)

These fairness concerns motivated all three courts to fashion and apply rules of apportionment that would mitigate harsh and unfair extremes. Apportionment of response costs, or apportionment of tort liability in a toxic tort case, will yield results that are more compatible with fairness objectives than the imposition of disproportionate liability. If courts are sufficiently cognizant of the unfairness associated with joint and several liability, when Congress is presumably setting the broad policy goals, then surely courts faced with difficult issues in toxic tort litigation can arrive at the same accommodation of all parties' interests.

VI. The Scientific Principles

In toxic tort litigation, perhaps more than in any other genre of tort litigation, the role of science is paramount. Scientific principles are critical to the issues of causation, apportionment, and the nature of the harm sustained. This section is devoted to a brief analysis of the toxicological and epidemiological principles that must be considered in resolving these issues.

The principal types of toxicological evidence are 1) short-term screening assays, such as in vitro laboratory experiments that examine bacteria or cells; 2) animal bioassays that involve exposure of groups of animals - usually rodents - to the chemical under investigation to test for the onset of various adverse effects; and 3) epidemiological studies that compare populations of exposed and nonexposed humans in order to draw inferences respecting possible causal associations.

Scientists use short-term in vitro (that is, test-tube) tests to identify various kinds of toxins by examining the biochemical effects of substances on cells, bacteria, organs, or embryos.(252) For example, researchers have studied cultured groups of cells seeking to define which cells within a group become malignant.(253) Researchers can also perform mutagen assays on cultured mammalian cells.(254) In another form of toxicity study, researchers examine and compare the molecular structure of the agent under investigation with those of known carcinogens, believing that if the structures are similar, similar toxic effects may be produced as well.(255)

All of these forms of short-term and in vitro assays focus on the biochemical mechanism. They examine the smallest building blocks of life and subject them to intensive investigation and experimentation in hopes of developing data useful in understanding the cellular mechanics of the disease process in humans. Yet, despite their increasing sophistication, the value of such tests in establishing causal relationships in humans is problematic.(256) The extrapolation from test tube exposures to human exposures requires a considerable degree of confidence in the probative value of such testing. Therefore, such studies may have little utility in toxic tort litigation: they are admissible but never sufficient on the generic causation question of whether a toxic agent has the capability of producing a disease.

The intermediate step in the hierarchy of toxicological evidence in support of causal relationships between a chemical and human pathology is animal toxicity studies. As with all toxicity studies, the objective is to identify the nature of the adverse health effect that a chemical agent produces and the range of doses over which such effects occur. Animal toxicity studies begin with the acute, single-dose toxicity of a chemical.(257) Toxicologists determine the lethal properties of a chemical and estimate its "LD50" dose, a dose which on average is lethal for fifty percent of the animals tested.(258) Once researchers generate acute toxicity data, they may undertake subchronic and chronic studies, which involve exposing animals to the agent repeatedly or continuously for several weeks, months, or the animal's remaining lifetime.(259) Additionally, toxicologists undertake metabolism or pharmacokinetics studies that seek to identify at what rates a chemical enters, flows through, and leaves the body and what changes it undergoes in the body.(260) These studies demonstrate that one of the principal reasons why different animal species experience varying effects from the same dose of a chemical is because they possess differing absorption, distribution, excretion, and metabolic processes.(261)

The most controversial aspect of animal studies in terms of their value in predicting the consequences of human exposure is the selection of dose administered to the animals,(262) which is significantly higher (on a milligram-per-kilogram basis) than the exposure levels for humans. Beginning with the LD50 dose, investigators study the effects of lower doses administered over longer periods of time in order to determine the lowest level at which adverse effects exist.(263)

The single largest purpose for undertaking animal bioassays is to enable regulatory agencies to determine which substances to regulate and to set permissible exposure levels.(264) While bioassays investigate substances for a wide array of health effects, the largest single disease objective is to identify possible human carcinogens. As a general rule, scientists assume that substances that are carcinogenic in animals are also carcinogenic in humans.(265) That assumption, in turn, rests on a number of other assumptions, each of which is potentially controversial when applied to a specific risk assessment. Not surprisingly, the consensus of the scientific and regulatory communities is that risk assessments predicated on animal bioassays represent worst-case presumptions rather than best-case estimates of the potential cancer incidence.(266) Accordingly, reliance solely upon animal data to draw inferences about causality in humans or the risk magnitude associated with human exposures errs on the side of conservative regulatory po]icy.(267) There exists an ongoing debate in the scientific community respecting whether high-dose animal experiments may properly be generalized to humans.(268) Some scientists maintain that animal studies are most properly used to identify the mechanisms of known carcinogenic chemicals, rather than either to answer the threshold question of whether chemicals have carcinogenic potential in humans or to quantify the risk they pose to humans.(269) Some argue that it is the massive dose of the chemicals that account for their carcinogeriic potential.(270) On the other hand, there are many dissenters who contend that animal studies are extremely valuable in evaluating risks to humans,(271) and that such studies should be used in legal decisionmaking, including to prove causation in toxic tort litigation.(272)

In addition to short-term screening assays and animal bioassays, the third type of toxicological study is the epidemiological study. This is the most important type. of toxicology study for toxic tort purposes. Epidemiology is the study of the distribution and determinants of disease in human populations.(273) Epidemiologists test biological inferences by combining the disciplines of statistics (more precisely, biostatistics), sociology, and demography. They look for unusual incidences of human disease and endeavor to identify those factors which distinguish the affected group from other groups. Epidermiological studies can be either experimental(274) or observational.

In the context of pobutants and the kinds of toxic substances involved in most toxic tort litigation, observational studies are the norm. These studies are made possible by "unplanned experiments" in which certain persons are already exposed to a chemical agent (for example, persons voluntarily choose to smoke cigarettes, or nonsmoking persons choose to live with persons who smoke), while others are not exposed, enabling investigators to compare the incidence of a particular disease among the two groups.(275)

The cohort or follow-up study is regarded as the most powerful observational type in terms of its ability to identify causal associations. These studies compare a group of exposed persons with a group of individuals who were not exposed and track them-prospectively to determine the incidence over time of a specific disease within the two groups.(276) The cohort study enables the investigator to calculate the comparative rates of disease within the exposed and nonexposed groups and to compare those two rates. This comparison of rates yields the relative risk, or the risk in the exposed population relative to the risk in the nonexposed population. This relationship is expressed as

[Mathematical Expression Omitted]

where [R.sub.1] is the risk of disease in the exposed population and [R.sub.2] is the risk of disease in a nonexposed population.(277) If RR = 1, there is no suggestion of any association between the factor and the disease in question. If RR > 1, there is a positive association between the exposure and the disease. Conversely, if, RR < 1, then the risk in exposed individuals is less than the risk in nonexposed individuals, suggesting 'a protective effect, such as in the case of vaccines. A study of workers exposed to asbestos in connection with their employment by asbestos product manufacturers revealed that these workers died from lung cancer at a rate of 64 per 100,000 per year, whereas in the general population of males during the same period and at the same ages, the death rate from lung cancer was only 31 per 100,000 per year.(278) The relative risk for this study would be expressed as

[Mathematical Expression Omitted]

In epidemiology, the probability of causation in a given case is determined by the attributable risk (AR), defined as the difference between the risk in the exposed and the risk in the nonexposed and expressed as follows:

[Mathematical Expression Omitted]

where P is the proportion exposed in the study group and RR is the relative risk of the study group. If the entire group is exposed, P = 1 and

[Mathematical Expression Omitted]

Based on the above study on asbestos workers,

[Mathematical Expression Omitted]

or fifty-one percent, which means that the probability that a given case of lung cancer in the exposed group was attributable to the asbestos exposure would be about even, or fifty-one percent. This percentage is sometimes referred to as the etiologic probability or etiologic fraction.(280)

In Dafler,(281) the relative risks of asbestos exposure of between 5.0 and 7.0 would yield etiologic probabilities of roughly between 80% and 86%. The relative risks of smoking were between 10.0 and 12.0, producing etiologic probabilities ranging between 90% and 92%. Finally, recall that asbestos exposure and smoking acted synergistically, so that relative risks of 5.0 for asbestos exposure and 10.0 for smoking yielded a combined relative risk of 50.0, and an etiologic probability of 98%.(282) This means that there existed only a two-percent chance that a person such as Dafler(283) would have contracted lung cancer as a "background" case, unrelated to asbestos or smoking. Therefore, within the exposed group there will always be "background" cases of lung cancer that are not attributable to asbestos exposure or smoking.(284)

Epidemiological evidence showing that exposure to a chemical creates a relative risk greater than one does not end the inquiry into causality. The enhanced relative risk reveals only an association between two factors - the exposure and the disease. The association is not necessary a causal relationship because other factors may be at work. To determine if an association is causal, epidemiologists have developed criteria that treat the statistical association as the starting point of the analysis. Sir Austin B. Hill proposed the most accepted of these criteria in 1965,(285) and federal regulatory agencies frequently use them in the risk assessment process.(286)

The criteria include the following:

1) The strength of the association. The greater the relative risk, the more likely the association is causal. Thus, a relative risk of 10.0 is more likely to suggest a causal relationship than a relative risk of 1.25.(287)

2) Consistency. Consistent observations made in different populations over different observation periods, utilizing different study designs and under different circumstances, imply that associations are causal, because the chance that all associations are the result of error or fallacy is simply too remote.(288)

3) Specificity. This criterion refers to the correspondence of exposure to a specific disease rather than more generalized effects on health. Is the association restricted to specific workers, a specific disease, a certain cancer site, or a specific histological type? The argument for a causal relationship weaker is if the relationship is nonspecific. However, this criterion has been de-emphasized in recent years because many causes have multiple effects and many diseases have multiple causes.(289)

4) Temporality. This seemingly simple criterion refers to the requirement that exposure to the causal factor must precede the disease in order to support a causal association.(290) Clearly if the "cause" does not precede the effect, then it cannot be the cause.

5) Biological gradient (dose-response relationship). If a dose-response relationship exists, the severity or frequency of disease increases with the level or duration of exposure. If the association reveals such a dose-response relationship or biological gradient, the argument for causal association becomes very persuasive. For example, the observation that lung cancer risk increases with the number of cigarettes smoked lends great support to the argument that smoking is a causal factor in lung cancer.(291)

6) Biological plausibility. This criterion looks to the compatibility of the association between the chemical agent and its health effect with current biological knowledge, including information from animal studies, pharmacokinetics, genotoxicity, and in vitro studies. However, given the, ever-increasing state of biological knowledge with new theories frequently advanced, rejected, and modified to fit new evidence, too much reliance on consistency with existing knowledge runs counter to basic exploratory and creative scientific principles.

7) Coherence. This criterion, closely related to plausibility, considers whether the associational data seriously conflict with the natural history and biology of the disease.(292) For example, do animal studies, histopathological studies, or other evidence cohere with the epidemiological findings?(293)

The next section returns to toxic tort litigation, where the relevance and persuasiveness of toxicological evidence is considered, to determine how courts apply such data. As was seen in Dafler, epidemiogical evidence can be extremely probative to the apportionment issues.

VII. Causation and Risk Contribution

The risk contribution model presupposes that each defendant contributed to toxic-related risks in a material way and that the resulting harm was consistent with such risks. For example, if the personal injury is cancer, the plaintiff must offer proof that the defendant is responsible for a human carcinogen. This generic causation - that the substance is capable of causing the harm that materialized - is an essential predicate in every toxic tort case. In cases involving asbestos, cigarette smoking, mercury, or benzene, the evidence of carcinogenicity is powerful. In other cases, the evidence is weaker, but so long as the plaintiff offers sufficient probative evidence of generic causation and individual exposure, it can satisfy that threshold. The proofs will include not only general toxicity data, but also dose-response and exposure data. The plaintiff must show that each defendant is a source of or responsible for the toxic substance that increased the plaintiff's risk of harm. Conversely, if a plaintiff demonstrates that one toxic substance from one source is the sole cause of her injury, there is no occasion to sue others who made only a de minimis contribution to the risk.

Where relative risks exceed 2.0,(294) a plaintiff could demonstrate that a particular substance was more probably than not an actual cause of the harm. Nevertheless, proof that one toxic substance - for example, asbestos - was more likely than not the cause of the plaintiffs disease does not preclude apportionment, as Dafler demonstrates. In Dafler, the evidence demonstrated that smoking was also, more likely than not, the cause of the plaintiff's cancer.(295) Both smoking and asbestos exposure were causes of the plaintiffs cancer, and each party independently satisfied traditional causation burdens of proof.(296) However, just as such high relative risks do not preclude apportionment when other substances also contribute to the risk of harm, so, too, the inability to reach a 2.0 relative risk threshold would not necessarily excuse a risk contributor from apportioned liability in a personal injury case. If the plaintiff offers sufficient evidence of what that risk is, above a threshold, then it would be "in play" and the factfinder could apportion some of the harm to the defendant responsible for that toxic substance. Because knowledge continues to accumulate respecting the toxicity of multitudes of chemicals and other substances, neither the plaintiff's nor the defendant's burden is insurmountable.(297) The toxic substances most likely to be implicated in toxic tort litigation are the ones for which numerous federal agencies dedicate their greatest resources to developing toxicity information.

One mass exposure model of toxic causation seeks to show that courts demand strong epidemiological proof before permitting the causation question to reach a jury or to sustain a verdict.(298) Indeed, a review of the cases and basic scientific principles demonstrate that, at least in mass toxic exposure contexts, courts require plaintiffs to offer proof of causality in the form of corroborating epidemiological studies and compelling toxicological proof, in addition to particularistic evidence of individual exposure and relevant medical testimony. This evidence would form the basis for a causation-in-fact finding and would also serve to provide the foundation for a determination of apportionment.

For example, some decisions hold that the plaintiff must offer epidemiological proof in order to make out a submissible issue on causation. In Brock v. Merrell Dow Pharmaceuticals, Inc.,(299) the Fifth Circuit opined that epidemiological proof was "the most useful and conclusive type of evidence" in to3dc tort cases,(300) and, after examining statistical principles, concluded that the lack of "statistically significant epidemiological proof" was fatal to the plaintiff's case.(301) Another Bendectin case, DeLuca v. Merrell Dow Pharmaceuticals, Inc.,(302) stated that if on remand the plaintiffs proffer of epidemiological proof failed to establish a relative risk in excess of 2.0, the district court could grant the defendant's motion for summary judgment.(303) However, the most ringing endorsement for the primacy of epidemiological proof in toxic tort cases is found in the Agent Orange litigation, as Judge Jack Weinstein declared that epidemiological studies "are the only useful studies having any bearing on causation."(304)

A recent New Jersey Supreme Court opinion, Landrigan v. Celotex Corp.,(305) takes a less stringent position toward the role of epidemiological evidence in proving causation between asbestos exposure and colon cancer. The court described epidemiological principles, pointing out that the two most important criteria in drawing causal inferences "are the strength of the association and the consistency of any such association with other knowledge."(306) The trial court held that when the plaintiff's medical records and interrogatory answers indicated the absence of other risk factors, such as a family history of colon cancer, a high-fat diet, and excessive alcohol consumption, the expert's opinion on causation was admissible and sufficient to satisfy the plaintiffs burden of proof.(307) The court rejected the view that a 2.0 relative risk was essential before a factfinder could conclude that the exposure was, more likely than not, a cause of the plaintiff's condition.(308) Instead, it viewed the relative risk "as one piece of evidence, among others, for the court to consider in determining whether the expert has employed a sound methodology in reaching his or her conclusion."(309)

But other courts flatly reject any rule of causation that would require a plaintiff to offer epidemiological proof to prove causation between exposure and harm.(310) For example, in Ferebee v. Chevron Chemical Corp.,(311) the District of Columbia Circuit Court of Appeals declared:

[A] cause-effect relationship need not be clearly established by animal or epidemiological studies before a doctor can testify that, in his opinion, such a relationship exists. As long as the basic methodology employed to reach such a conclusion is sound, such as use of tissue samples, standard tests, and patient examination, products liability law does not preclude recovery until a "statistically significant" number of people have been injured or until science has had the time and resources to complete sophisticated laboratory studies of the chemical.(312)

The view expressed in Ferebee runs contrary to the majority of decisions that explicitly consider what quantum of proof should be sufficient to establish a prima facie case of causation. Nevertheless, whether epidemiological evidence is deemed essential or not, if the evidence either party presents is insufficient to satisfy cause-in-fact requirements, then it necessarily will fail to demonstrate that the harm is capable of apportionment. Therefore, the evidence cannot sustain the factfinder's factual determination of what that division ought to be among multiple sources and parties.(313) Clearly, the same epidemiologic proof that bears on the causation issue will also be relevant to the issue of apportionment.(314)

In ascertaining whether a toxic exposure caused a plaintiff's injury, the first step is to assess whether the exposure is scientifically capable of producing that kind of harm. The evidence to answer that question will consist largely of toxicological studies.(315) Generic causation uncertainties are common to all cases involving exposure to the same toxic substance. For example, the proof of whether cigarette smoking can cause lung cancer or whether asbestos exposure can produce mesothelioma should not vary substantially from case to case. Nearly every Bendectin case tried during the last ten years demonstrates the same record: the same plaintiffs' experts, defendants' experts, and studies introduced as evidence.(316)

In addition, in order for the jury to reach the causation question, the plaintiff must offer particularistic evidence that addresses her unique circumstances of exposure and medical and personal history that bear on whether exposure is the most likely explanation for her disease. Thus, the ultimate question is not simply whether substance X is capable of causing disease Y, but whether substance X, at exposure level Z and dose D, can cause disease Y. While the threshold question is whether any toxicological evidence exists to support an association between the substance in question and the kind of injury the plaintiff experienced, regardless of dose levels, the secondary, narrower, and more precise question is at what level of exposure and dose such associations are found to be present.(317)

While exposure levels are reasonably ascertainable in pharmaceutical product cases, in other toxic product cases involving exposure to a nongeneric product, such as asbestos,(318) each plaintiff may have experienced uniform exposures.(319) The Agent Orange litigation illustrates the absence of uniform exposure levels. Depending upon where and when servicepersons were situated in Vietnam, Agent Orange exposure varied significantly in intensity and frequency.(320)

Moreover, some toxic substances produce so-called "signature diseases" - rare diseases associated with exposure to a particular substance that rarely occur in the nonexposed population - that can influence causation determinations.(321) The incidence of the background risk for signature diseases is virtually zero; for example, asbestosis and mesothelioma are signature diseases of asbestos exposure, and clear cell adenocarcinoma of the vagina is a signature disease of DES exposure.(322) Scientists discovered these by cluster analysis, and their presence enables the plaintiffs exposed to those substances to establish causation.

In contrast to toxic product cases, occupational and nuisance exposure cases typically involve a more complex and open-ended level of inquiry. For example, in nuisance model cases,(323) the threshold causation questions are more diffuse and intractable.(324) First, there are problems in identifying who has been exposed to chemicals from a hazardous waste site.(325) Typically, the plaintiffs experience different pathways of exposure, diverse levels of exposure, and multiple kinds of toxic exposures.(326) Groundwater contamination will rarely implicate only one toxic substance; rather, it involves the permeation or leaching of many substances that escape from the site and reach the plantiffs' environment by unforeseeable and poorly understood means.(327) Nuisance cases frequently require consideration of the synergistic effects of many toxic substances.(328) Even identifying the site or the substances is extremely difficult because of the absence of records, site boundaries, and knowledge of whit materials were dumped.(329) Furthermore, the plaintiffs' interactions with substances will be more diverse, making the causation questions more difficult.(330) These cases will rarely be resolvable on the basis of the single generic causation question: Does chemical X cause condition Y? The number and diversity of other explanatory causes or contributing factors will be greatest in this genre of cases.(331)

Finally, the substances that are the suspected causal agents of disease or injury are less well known and understood than the agents involved in toxic products exposure litigation. Unlike toxic consumer products that have been regulated for decades and that contain toxic substances more common to commerce and science, the substances deposited at waste sites are industrial byproducts that until recently have not been regulated.(332) Science knows less about the toxic properties of substances implicated in nuisance model litigation because there has been less need to learn the answers to the toxicity questions: human exposures occur less frequently and are unintended. The research investments to determine the causal relationships between a given toxic substance and human condition will be proportional to the anticipated need for society to know those answers.(333) This unremarkable proposition illustrates nothing more than a simple cost-benefit analysis. However, because of an increased focus of regulatory and public attention over the past fifteen years on the health effects from hazardous waste sites, and especially because of the efforts of the Agency for Toxic Substances and Disease Registry (ATSDR), researchers are generating greater informational databases relevant to nuisance cases. Those informational databases not only facilitate proof of causation but also provide the basis for apportioning harm.

Once the case record addresses these causation questions,(334) the inquiry then shifts to the apportionment question. The same evidence that suffices to establish a submissible issue of causation will go a long way toward providing the record essential to apportionment. The plaintiff win not likely offer evidence on the causative effects of smoking, alcohol consumption, consumption of drugs with adverse side effects, and other potential risk-creating behaviors in its case-in-chief Instead, the defendant will have to introduce such evidence in sufficiently probative form so a judge can find the harm capable of apportionment. However, if the parties recognize that courts are amenable to the process of apportioning harm each will have considerable incentive to develop the necessary data. Parties will pursue discovery with an eye toward developing apportionment-related evidence, especially the plaintiff's lifestyle behaviors that have created toxic-related risks that contributed to the harm. While law encourages, defendants to look for evidence of negligence or assumption by plaintiffs, the apportionment inquiry may stimulate discovery other areas not traditionally targeted by fault-based defenses. If that is true, expanding application of apportionment principles may result in some incremental transaction costs. However, more extensive use of apportionment produce savings that may more than offset those incremental costs. Moreover, real property injury cases probably will require little or no incremental discovery.

VIII. Impediments to Adoption of the Model

So, if the risk contribution model is such a good idea, why have parties and courts not rushed to make use of it in toxic tort cases? Several explanations are possible but may not fully account for the judicial reluctance to move toward aggressive application of apportionment principles.

A. Dominance of Comparative Fault

Nationwide investment in comparative fault or negligence systems has chilled the reception of a risk contribution model. Since Mississippi enacted the first general comparative negligence statute in the early twentieth cent",(335) forty-six states have either legislatively or judicially jumped on the comparative fault bandwagon.(336) Of all the major tort movements since World War II, none has arguably had more widespread effect than the comparative fault movement.(337) The hallmark of these comparative fault systems is to supersede the all-or-nothing common law rules of contributory negligence with a system that apportions liability on the basis of more refined gradations of fault. While this system did not, as a matter of legal theory, preempt other forms of apportioning liability, as a practical matter it is seen as the exclusive means by which liability can be apportioned. In addition, the preeminence of comparative fault results in tangential consequences, the most important for this discussion being its impact on joint and several liability.

The complete bar of contributory negligence and the doctrine of joint and several liability evolved together and grew out of the common law concept of the unity of the cause of action; the jury could not be permitted to apportion the damages since there was but one wrong."(338) If a plaintiff was partially at fault, the court treated him as being at fault for the whole, and thus his contributory negligence defeated recovery. Similarly, if a defendant were partially at fault, the court assigned her the fault for the whole and, thus, she was jointly liable for the entire judgment. With the advent of comparative negligence, however, the unifying principle began to unravel. Comparative negligence is based on allocating fault between the plaintiff and the defendant. Thus, one conceptual underpinning for joint and several liability began to erode when comparative fault systems displaced contributory negligence. Courts soon abandoned the concept of indivisibility of liability.(339)

The early comparative fault statutes did not address whether or to what extent the divisibility of fault ought to influence the divisibility of liability among multiple defendants. Joint and several liability soon emerged as a principle based on the indivisibility of harm(340) and differed from liability based on apportionment of harm. As the burden of establishing the basis for apportionment proved intolerable for plaintiffs in the multiple-cause cases, courts' commitments to fairness led to increasingly greater recognition of indivisible harms and joint liability. However, with comparative fault came a potential conflict: Because courts no longer precluded juries from dividing liability on the basis of fault, did that capability imply that the harm was also divisible in multiple-defendant cases? If fault were allocated between the plaintiff and the defendant, why not allow the same allocation among the defendants?(341) The early cases that struggled with the intersection of comparative fault and joint liability for indivisible harms split as to the proper accommodation.(342) Some decisions concluded that fault divisibility barred joint liability.(343) A majority of others concluded that divisible fault and indivisible harm should co-exist, thereby preserving joint and several liability.(344)

The critical point is that the principles of apportionment do not demand a resolution of this controversy.(345) It has always been the law that if the harm was divisible, then no basis existed for finding that the defendants were joint tortfeasors. In an ongoing debate, Professors Wright, Twerski, Schwartz,(346) and others all take as the underlying premise that the multiple defendants are joint tortfeasors. This risk contribution model, however, eliminates precisely that relationship among the defendants. Because the harm,is no longer indivisible, there exists no jointly caused, indivisible, single injury, but rather multiple injuries (or multiple causes contributing to one injury) that create individual proportionate liability.

Nevertheless, in the 1980s, as part of the tort reform movement, dozens of legislatures forced uneasy compromises between the two doctrines of comparative fault and joint and several liability.(347) Legislatures modified joint and several liability in numerous ways, such as preserving it for economic damages but not for non-economic damages,(348) applying it only to fault-free plaintiffs,(349) applying it to those plaintiffs who were less at fault than the defendant,(350) or not applying it to the defendants whose fault was below some prescribed level.(351) This smorgasbord of partial abolitions and modifications required tremendous political capital. Courts, too, have invested heavily in this system, promulgating standard jury instructions and issuing countless opinions addressing the ambiguities remaining after legislatures adopted comparative fault. Faced with these considerable investments and accommodations, would courts now be receptive, with emerging toxic tort caseloads, to add another scheme of percentages for trial courts and juries to struggle with? In addition, would legislatures or courts be willing to apply apportionment and thereby appear to take an end-run around the statutorily or judicially promulgated rules governing joint and several liability.?

The importance of comparative fault systems during the past twenty years has dimmed the prospects for judicial application of apportionment principles that would divide the harm, and necessarily the liability, and would appear to supplant, the governance of comparative fault. In fact, however, apportionment of harm would not replace comparative fault because a jury might still assign fault between the plaintiff and the defendant for the segregated harm. Nevertheless, the judiciary has, unsurprisingly, simply not embraced apportionment of harm. After all, it is easy to declare that most injuries to persons and property are inherently indivisible and therefore not subject to apportionment. Moreover, because comparative fault in no way depends on the apportionability of the harm itself, courts are free to apply it to all cases. It represents one uniform and comprehensive approach.

And therein lies its beauty and its flaws. To repeat the points made earlier, at least three problems exist in applying comparative fault to toxic tort cases such as Dafler. First, the rubric of "fault" or negligence" gets in the way of looking at actual, lifestyle risks that are not based on fault, and risks that are not created by any party. The, concept of fault is narrower than the concept of causes,' and it is the causes we are after, not the degree of culpability. Second, and relatedly, is the exclusion of nonparties and innocent causes from the comparative fault system. Absent parties, immune parties, insolvent parties, acts of God, are equally relevant to the apportionment of the harm, even though they may be excluded from the comparative fault inquiry.(352) Third, comparative fault carries the baggage of judicial and legislative compromises on joint and several liability, which renders the courts inhospitable to any division of the harm that would defeat joint liability in cases to which it would otherwise apply. This in-hospitability would be most marked in those cases where the legislatures have explicitly carved out an exception to the abolition or modifications of joint liability-that is, the environmental tort exceptions discussed earlier.(353) If the legislature has seen fit to declare that joint liability is to be especially preserved in environmental tort litigation, a court will be reluctant to conclude that the harm is apportionable and therefore not subject to joint liability. In reality, of course, if the harm is reasonably capable of apportionment, then the parties (defendants) are not joint tortfeasors and there would be no occasion to even consider the environmental tort preservation of joint and several liability. Because Dafler did not involve multiple defendants at trial, there was no occasion for that court to address this potential theoretical, but not actual, conflict.

For all of these reasons the structures of the comparative fault system simply are perceived to block the road to more widespread use of apportionment principles.

B. Movement Toward Collective Responsibility

A second but less significant explanation for the reluctance of courts to take up apportionment is the trend toward greater application of collective responsibility.(354) Professor Kenneth S. Abraham(355) and others(356) have documented this trend, sometimes critically and sometimes with applause; it is not the aim of this Article to revisit this controversy or take any position on whether it is a good trend or an undesirable one. The point is that this willingness to apply collective responsibility is especially manifest in toxic product cases and in the application of joint and several liability. The DES market-share litigation has become the quintessential example of group responsibility because market-share liability disavows any effort to connect an individual plaintiff to an individual defendant.(357) The likelihood-of-actual-causation rationale created in Sindell v. Abbott Laboratories(358) works only for groups: if a firm possesses ten percent of the relevant market, its share win converge with the harm it actually caused only by applying the rule across many cases. In other words, market-share liability functions fairly only in mass tort contexts. Indeed, New York's version of market-share liability-a risk contribution to the public at-large model-does not even attempt to achieve fairness between plaintiffs and defendants, because a m that never sold the risk-causing product in New York, nor during the relevant time periods, is precluded from exculpating itself.(359) Thus, the defendants are held proportionately liable to the plaintiffs whose injuries they could not have caused because they created risks to others within the nationwide group of exposed persons.(360)

Another example of collective responsibility is the judicial endorsement of joint and several liability. That endorsement of joint liability(361) was a response to the difficult burdens of proof respecting causation and apportionment, especially in toxic tort cases. The understandable judicial reluctance to preside over a plaintiff's failure to sustain the traditional burdens of proof has led to a relaxation of causal standards, as the market-share cases illustrate. Abraham points out the collectivizing effects of joint and several liability:

Joint and several liability imposed on independent tortfeasors-another method of shifting the burden of proof-has even more collective effects for two reasons. First, the amount of liability imposed bears no necessary relation to the degree of responsibility of any particular defendant for the plaintiff's injury. In certain cases the amount of liability imposed may not even reflect the probability that a particular defendant is responsible at all. Second, because each defendant bears the risk of the others' insolvencies, it may have to pay both its own non-causal share and a portion of the shares of codefendants. Shifting the burden of apportioning damages to a group of defendants, each of whom has caused an undetermined part of a plaintiff's injuries, also is a collective response to the causation problem, because if defendants are unable to meet their burden, they are held jointly and severally liable. Moreover, even if defendants are able to apportion damages the shift has collectivizing effects, because most of the cost of apportioning has been spread among defendants instead of shouldered by the plaintiff.(362)

Abraham also posits two examples of cases in which courts understandably moved toward collective responsibility. In one example, two or more toxic substances combine arithmetically and the plaintiff develops a disease consistent with exposure to both substances. In a second example, the two substances combine synergistically. If the arithmetic or synergistic effects still fall below the more probable than not threshold, the plaintiffs face a difficult causation burden that can only be relieved by some version of market-share liability or other surrogate, such as recovery based on proportional causation or recovery for increased risks, both of which can function only in a mass tort context in which liability is spread over many cases. However, under an aggressive approach to apportionment, proportionate recoveries might be achievable as against each defendant without resort to collective models in cases in which the plaintiff can demonstrate that the combined effects of multiple toxic substances were more probably than not the cause of her injury.

Moreover, to the extent that cases like Landers and Michie are resolvable based on the apportionment of the harm, rather than by application of joint liability, they come much closer to the model of individual responsibility. If the willingness to adopt principles of joint liability is not difficult to identify and the very reasons that led to their adoption can be fulfilled, then legislatures and courts should revert back to the more traditional and tailored model of individual responsibility that apportionment achieves. However, in cases in which the plaintiff cannot satisfy her burden of proof on causation, whether exposed to one or multiple toxic substances, neither collective responsibility nor apportionment can come to her aid. The real problem with alternative approaches to the burden of proof problem, such as proportional causation and recovery for risk, is that they can only achieve a proper measure of deterrence or fairness when applied to groups.

The risk contribution model advanced here is driven by the principle of individual responsibility, and to that extent, it may be moving upstream against the current of collective responsibility . Nevertheless, while this may be one factor influencing a resistance to apportionment, it probably is not a major one.(363)

C. The Substantial Factor Test of Causation

A third impediment to greater reliance on apportionment is the rampant use of a substantial factor test of causation, especially in toxic tort cases. The substantial factor test obfuscates the real relationship between the defendant's tortious conduct and the plaintiff's injury. In Dafler,(364) each party built its strategy on the assumption that the substantial factor test governed the case: the defendant attempted to defeat. the plaintiff's recovery by offering expert testimony that smoking was the sole cause of the lung cancer, thereby rendering asbestos no cause at all, and the plaintiff sought to satisfy the test by offering testimony that asbestos exposure was :i" cause of the cancer, albeit contributing lower relative risks than his smoking. Neither party apparently had as its objective the apportionment of the harm between these two causes on the basis of the relative risks each created. If the substantial causal factor test represents merely a sufficient cause test, then its application has real content, as illustrated in Anderson,(365) in which the defendant's fire was a sufficient cause of the plaintiff's property damage.(366) The problem comes in cases in which the substantial causal factor test lacks any such threshold, as illustrated in Brisboy v. Fibreboard Corp.,(367) in which six to nine months of asbestos exposure was found to be a substantial factor in the plaintiff's fatal lung cancer. The jury assigned the plaintiff s forty years of a two-packs-a-day smoking habit a fifty-five percent fault and the defendant's conduct, forty- five percent fault. The test has become a default, resorted to when nothing else works, and juries are afforded virtually no guidance as to how much of a causal connection is necessary to satisfy the test. The point, however, is not to challenge the substantial causal factor test, but to point out that parties play to it because of the all-or-nothing stakes involved. In Dafler(368) and Martin,(369) barring apportionment, the plaintiff would have obtained a full recovery despite the obvious relevance of other toxic risks in the mix that brought about the cancer.

Parties continually battle over the application of the substantial causal factor test and debate whether the but-for test should control, with defendants generally preferring the latter and plaintiffs the former.(370) Courts encourage these preferences by failing to clarify the content of the substantial factor test.(371) With a more extensive commitment to apportionment principles, the courts would remove some of the high stakes from the causation controversy and allow parties to adjust their respective trial strategies accordingly.

D. The Availability of the Evidence Needed to Apportion

One possible impediment to the greater use of apportionment in toxic tort cases is that the scientific evidence necessary to undertake the comparison of the toxic-related risks is not available.(372) However, am problem may not be insurmountable. First, the law does not demand absolute precision; so long as the evidence is sufficient to afford rough approximations of the harm each toxic source caused or the risks it created that contributed to the harm, the law can ask no more. The Fifth Circuit Court of Appeals emphasized the reality that apportionment can only be achieved roughly, not exactly, and that such estimations should suffice.(373)

Second, and very importantly, the epidemiological and toxicological studies essential to proving causation and apportionment in toxic tort cases are increasingly available as a result of major federal initiatives. For example, Congress created the Agency for Toxic Substances and Disease Registry (ATSDR), a federal public health agency that is part of the Department of Health and Human Services, under CERCLA to address its concern with the adverse health effects of hazardous waste sites.(374) The ATSDR must maintain a registry of persons who were exposed to hazardous substances and who have serious illnesses, conduct studies on the health effects of toxic substances found at waste sites, and conduct periodic screening programs to determine the relationship between exposure to these substances and human illnesses.(375) The ATSDR evaluates data and information on the release of hazardous substances,(376) Conducts public health assessments to assess the impact on the health of those exposed,(377) and identifies studies or actions needed to evaluate or prevent human health effects.(378) To Carry out these objectives, the ATSDR must work with the EPA to develop and prioritize a list of hazardous substances found at National Priorities List (NPL) sites. In developing this hazardous substance priority list, the ATSDR and EPA use the following criteria to rank the substances by potential human health risk: 1) frequency of occurrence at NPL sites, 2) toxicity, and 3) potential for human exposure to the substance.(379) These criteria are significant for toxic tort litigation because the most frequently litigated substances are the ATSDR's highest priority.

Under CERCLA, as amended in 1986,(380) the ATSDR must prepare the following: 1) health assessments and pilot health effects studies;(381) 2) epidemiological studies;(382) and 3) toxicological profiles of all ATSDR priority substances. Moreover, the ATSDR must assure that a research program is initiated to fulfill information needs associated with these substances.(383) CERCLA requires that each toxicological profile include an examination, summary, and interpretation of all available toxicological information and epidemiological studies.(384) The profiles must also include a determination of whether adequate information on the health effects of each substance is available and, if not, the ATSDR, in cooperation with the National Toxicology Program, must assure the initiation of research to determine these health effects.(385)

By 1991, EPA had completed over 130 of the toxicological profiles required under the 1986 CERCLA amendments. The profiles include an array of data relevant to toxic tort litigation, including the physical and chemical properties of the substance in the air, water, and soil; the pharmacokinetics of the substance; its toxicity to humans and animals; and a bibliography of important studies. These toxicological profiles may be of great long-term significance to toxic tort litigation of all kinds - nuisance and occupational, and even toxic product cases. Additionally, the ATSDR must report any assessments of increased incidence of adverse health effects that may be associated with released hazardous substances. Thus, through epidemiological, surveillance, and toxicological studies, the ATSDR will increase the understanding of the relationship between exposure(386) to hazardous substances and adverse human health effects,(387) and it will provide a wealth of data relevant to causation and apportionment issues.

Apart from the ATSDR, several federal agencies undertake risk assessments to carry out statutory mandates. EPA performs risk assessments, such as the one recently completed on environmental tobacco smoke,(388) in order to determine if a substance is a carcinogen and to recommend appropriate actions to other agencies. The National Institute of Occupational Safety and Health similarly performs studies to determine the effects of workplace exposure to potentially toxic substances, and the Occupational and Health Safety Agency must fix exposure limits to such substances based on such studies.(389)

Closely related to the unavailability of the toxicological and epidemiological evidence necessary to perform toxic apportionments is the reluctance of courts and counsel to make use of such data. Courts historically have been uncomfortable with the admission of statistical evidence and with the relevant statistical principles. Nevertheless, in recent years courts are exhibiting increasing familiarity with and understanding of such principles and are willing to examine such evidence critically. Indeed, the recent controversy over the admissibility of scientific evidence culminated in Daubert v. Merrell Dow Pharmaceutical, Inc.,(390) Which articulated standards for the admissibility of scientific evidence under the Federal Rules of Evidence.(391) The Federal Judicial Center recently issued a Reference Manual on Scientific Evidence,(392) which contains several chapters dedicated to toxicology and epidemiology, for use in the federal courts to assist the federal judiciary in carrying out its responsibilities. In addition, CERCLA litigation arguably has forced the federal judiciary to struggle with cleanup standards and the apportionment issues,(393) and has added to the overall knowledge of and comfort level with toxicological issues. Consequently, while parties, their counsel, and the courts have not demonstrated a tremendous willingness to press for apportionment in the first decade of toxic tort litigation, those attitudes seem to be changing. As the availability of the necessary data grows, so too does the receptivity to such information.

IX. Rationales for the Model

Many of the rationales and arguments in favor of more aggressive application of apportionment principles are identified in the sections above. This section represents a summary of those points and contains a few additional observations.

A. Corrective Justice or Fairness

1. The Generic Causation Question: Corrective Justice Requires Such

Proof

Do principles of corrective justice or fairness support application of the risk contribution model of toxic apportionment? To answer this question, one must first acknowledge that the requirement of causation is indispensable to corrective justice. That is, to define the meaning of corrective justice is to incorporate an element of causality. The universality of causation to principles of fairness is illustrated by the diverse legal theorists who either explicitly advocate(394) or implicitly assume the presence of causation in their corrective justice models.(395) The reasons for their doing so are clear: fundamental fairness requires that a defendant compensate for those losses (and only those losses) that his tortious conduct caused and that liability not be imposed solely on the basis of the defendant's fault (absent causation) or the ability to spread the risk of injuries more efficiently.(396)

These generalizations respecting the causal requirement in corrective justice do little, however, to address the search for fair outcomes in toxic tort cases, because these cases necessarily involve at least some causal indeterminacy. These cases are troublesome precisely because the causal questions are intractable and elusive. Some commentators sidestep the problem by advocating other approaches to these cases.(397) For example, Professor Epstein has noted that the "most vexing questions of causation"(398) arising in the Bendectin cases would permit the relationship between drug companies and users to be governed by contract law. In this way, tort law and juries would be spared the array of conflicting scientific data that is incapable of precisely resolving the ambiguous causation question.

In contrast, Professor Wright takes an entirely different tack. He roundly criticizes those commentators who would depend entirely on probabilistic evidence or "probabilistic causation" for confusing ex ante risk with ex post cause-in-fact.(399) He would instead preserve the strict causality requirement by redefining the nature of the harm of which the defendant's conduct was the cause-in-fact by recognizing a new type of injury - risk exposure per se or risk exposure that possibly led to the subsequent injury:

[I]f each defendant is held liable only for her share of the risk exposure, there is no conflict with the corrective-justice view. It still must be proven that each defendant caused the risk exposure that possibly led to the manifested injury, and liability is for such risk exposure, rather than the manifested injury. If the defendant can establish that she did not contribute to the manifested injury - that is, that the risk that she created could not have led to the manifested injury - the courts absolve her from any liability.(400)

Professors Rosenberg(401) and Robinson(402) advocate another approach to the causation problem, maintaining that probabilistic causation is the fairest system to apply in mass toxic exposure cases. However, these approaches achieve fairness across groups of cases but are troublesome in individual toxic tort cases.

The plaintiffs in toxic tort cases must initially offer some epidemiological or other persuasive toxicological evidence in order to demonstrate that the toxic substance was capable of causing the plaintiffs condition.(403) The answer to the generic causation question is not measured solely, or even largely, in probabilistic terms. The answer is either "yes" or "no" (or "probably yes" or "probably no") and turns on drawing a causal inference based on a constellation of evidence, some of which is statistical in nature.(404)

In cases presenting generic causation questions, principles of corrective justice operate on fairly familiar terrain. Here, a demand for strict proof of the substance's ability to produce the kind of harms alleged is wholly compatible with corrective justice. It would be unfair to impose tort liabilities when the plaintiffs' evidence is devoid of any epidemiological proof that the substance was a human carcinogen or similarly harmful substance. The standard of causation that governs toxic tort cases is compatible with notions of fairness because it requires plaintiffs to offer reliable and persuasive generic causation evidence. Absent such evidence, no basis exists for believing that the defendant exposed the plaintiff to any increased risk of disease.

2. Individual Causation and Alternative Causes: Corrective Justice Is

Compatible with Multiple Causes

Once the plaintiff provides sufficient evidence that she was exposed to a substance capable of causing her injury, the question whether the exposure caused this plaintiffs lung cancer remains. Evidence to answer this question involves not only epidemiological and toxicological evidence, but also particularistic evidence respecting medical history and other considerations. In addition, the magnitude of the excess risk attributable to the toxic exposure is of considerable importance. In Dafler, both of the plaintiffs experts testified that asbestos and smoking produced quite substantial relative risks, to the tune of 5.0 to 7.0 for asbestos and 10.0 to 12.0 for smoking.(405) Apportionment explicitly recognizes the relevance of the excess risks that each toxic substance created. If corrective justice accepts the requirement of causation-in-fact, and that inquiry is only so close because of some causal indeterminacy,(406) then it ought to reach out with open arms for an apportionment of that harm based on the very same evidence that was relied upon to establish causation. The plaintiff in Dafler proved, to the extent possible, that smoking was a cause of his cancer, and he also proved, to the extent anyone could, that asbestos exposure was also a cause of his cancer. Each toxic substance was a cause of his cancer.

It does not violate principles of corrective justice to require the defendant to ante up in these cases so long as the plaintiff has met her burden of proof by offering sufficient evidence on the magnitude of the excess risk, epidemiological and toxicological evidence, and more particularistic evidence of her medical and personal history to help eliminate alternative causes. Even so, the plaintiff in Dafler did not eliminate one major alternative cause - smoking. Moreover, smoking was proved to be a cause just as persuasively - even more so - than asbestos exposure was a cause. Corrective justice principles should be compatible with declaring each as a cause. Once the different exposures are declared as causes, however, how should tort law act upon them? The fairest way to factor in the existence of dual causes is to apportion the harm, and that is precisely what the black-letter law of section 433A of the Restatement would require.

To illustrate the principle another way, the market-share cases demonstrated the fairness of using a defendant's share of the market as a surrogate for proof of actual causation. Under Sindell,(407) the plaintiff bears the burden of proving each defendant's market share.(408) In other words, a strong analogy exists between the defendant's share of the market in DES cases (especially since virtually no background cases of these types of vaginal cancer existed), which represents its contribution to the incremental risk of causing the plaintiffs injury, and the defendant's contribution to the excess risk in these toxic tort cases. In the proposed approach, the plaintiffs are asked to offer persuasive evidence of the defendant's "market share" of the cancers suffered in the exposed population when the total market consists of all cancers - background and incremental cases combined.

However, the defendant has also offered evidence of the "market share" of cancers suffered in the exposed population based on the evidence of the relative risks for smoking. In Dafler, the asbestos share of the "market" was less than the smoking share of the "market." Each is relevant, and should, on fairness reasons alone, be used to determine the appropriate share of the harm for which each party is legally responsible.

3. The Comparative Fault Analogy. Apportionment Captures Fairness

Gains Similar to Comparative Fault

To move to another basis for developing the fairness rationale, why did courts and legislatures adopt comparative fault systems? Was not the principal motivation the unfairness of the all-or-nothing rule of contributory negligence? Courts adopting comparative negligence were striving to create a system that permitted apportionment of fault, in which each party's liability was commensurate to the percentage of fault attributable to it. However, the same all-or-nothing unfairness that compelled the abandonment of the bar on contributory negligence should equally compel the abandonment of a rule of indivisible harm or indivisible causation. Professor Wright has declared that causation, unlike fault, can never be a matter of degree,(409) but in toxic tort cases it is always a matter of degree. If the law allows juries to apportion fault on less than a scientific basis, surely it should allow juries to apportion harm on the basis of causes when the evidence offers a reasonable measurement from which a factfinder can render a meaningful comparison of those causes and the resultant harm.

4. The Avoidable Consequences Analogy: Apportionment Captures

Fairness Gains Similar to the Avoidable Consequences Doctrine

Another analogy is the avoidable consequences doctrine. Under this doctrine of mitigation, a jury may segregate a plaintiffs harm between two causes: the portion of the harm the defendant's tortious conduct caused and the portion by which the plaintiffs post-tort conduct increased the total harm. This doctrine, too, rests on a fairness rationale: it is unfair for the defendant to compensate for damages that the plaintiffs own conduct caused or contributed to. For example, when the plaintiffs failure to wear a seatbelt enhanced his injuries, fairness requires a reduction in his damages to reflect that fact.(410) If a plaintiffs continued smoking after a treatment increases the risks of an amputation, which eventually becomes necessary, then the jury may reduce the plaintiffs damages because of that fact.(411) The avoidable consequences rule seems unremarkable. Even though the defendant's tortious conduct was a but-for cause of the harm, the plaintiffs conduct nevertheless increases that harm, and courts willingly recognize the propriety of a commensurate adjustment in the damages awarded.

5. The Multiple-Defendant Cases: Irrelevance of Joint and Several

Liability

Up to this point, the discussion has centered on plaintiff-defendant cases such as Dafler. However, just because apportionment in cases in which the plaintiff is a source of a toxic risk satisfies fairness concerns does not necessarily imply that the same is true for the multiple-defendant, innocent plaintiff cases. These cases are more troublesome because the plaintiff is responsible for no toxic-related risks and the defendants are responsible for all such risks; moreover, these cases implicate the propriety of joint and several liability.

As a threshold matter, however, apportioning harm among defendants is not necessarily an attack on the doctrine of joint and several liability. If the harm sustained is divisible on some reasonable basis, there is no occasion to even reach the doctrine, or to challenge its wisdom when applied to truly indivisible, single harms. As Professor Wright repeatedly stresses, the doctrine applies only when there is a single, indivisible injury and each defendant's tortious conduct was a cause-in-fact and proximate cause of that entire harm.(412) In the multiple-defendant toxic tort cases in which the harm is apportionable on the basis of the magnitude of the excess risks that each defendant created, then the doctrine does not apply. The judicial decisions that struggled with the intersection of comparative fault and joint and several liability are distinguishable, because in those cases the harm remained indivisible. For example, in Coney v. J.L.G. Industries,(413) the Illinois Supreme Court, in holding that joint liability was unaffected by comparative fault, stated that "[t]he feasibility of apportioning fault on a comparative basis does not render an indivisible injury 'divisible' for purposes of the joint and several liability rule."(414) If the harm is divisible, as it will be in many toxic cases, on a reasonable basis that compares the quanta of risks and causation each is responsible for, then the conceptual basis for joint liability varnishes. In the comparative fault debate, the true conceptual basis for joint and several liability remains intact - indivisible harm - but in the apportionment-of-harm debate that conceptual basis is, by definition, now eliminated. Defendants are simply no longer joint tortfeasors. Instead, each defendant is held liable for only the harm that it caused, and because cause in toxic tort cases is a matter of degree and not absolute, the apportionment provides the fairest result achievable.

6. Risk-Based Methodology Is Essential to Fairness in Toxics Cases

Just as causation is essential to corrective justice, causation is also a limiting principle: liability should not exceed the proportion of the harm caused. Toxic harms and their relationship to the causes or forces that contributed to them are different in that respect from automobile collisions, in which the tort system can fix with considerable precision what forces caused the collision. With toxic harm, the most the tort system can do is represented by the most that medical science can do, and the most that medical science can offer respecting the cause of many toxic harms is to identify contributing causes, those that substantially increase the risk of that harm occurring. When those contributing causes can be identified and when their contribution can be measured in some rough manner, the harm ought to be apportioned on that basis.

Even Professor Wright, the leading advocate of the preservation of joint and several liability and an ardent opponent of risk-based or causation-based apportionment, acknowledges the relevance of such methods in toxic exposure cases:

However, the risk-based apportionment methods may be useful for another purpose. Although they are inappropriate for apportioning liability among the multiple responsible causes of some actual harm, these methods may be appropriate for distinguishing the causes of injuries which consist not in some actual harm, but rather in exposure to the risk of such actual harm. Recovery for such risk exposure, at least when the risk exposure possibly contributed to some actual harm, seems necessary if tort law is to keep up with modern scientific knowledge and technology which, by increasing individuals' awareness of and exposure to a much broader range of significant risks from multiple sources, have made it increasingly difficult to establish a causal connection between any particular source of the risk exposure and subsequent actual harm. In such cases of insoluble uncertainty about causation of the actual harm, many courts have been willing to allow recovery for the risk exposure itself, as a new form of legal injury, when the plaintiff can establish that the defendant tortiously caused such risk exposure and that the risk exposure possibly contributed to the actual harm.(415)

Nevertheless, while Professor Wright would redefine the harm as risk exposure and cite as authority the loss-of-chance medical malpractice cases, the courts adjudicating toxic tort cases retain the traditional harms - death, cancer, contaminated property. As such, the only rational means of implementing an apportionment in these cases is to apply a risk-based methodology that compares the causal contribution of each toxic source to the harm.

When the harm is divisible, the multiple-defendant case in essence becomes a plaintiff's claim against each defendant individually because each defendant's liability will be based on its proportionate share of the harm. As a result, the plaintiff will recover full compensation by suing all responsible parties and holding each liable for its share, much like marketshare theory operates in those jurisdictions that hold each defendant proportionately liable only. But what about insolvent defendants or those that are unavailable or can't be located? Why is it fair for the plaintiff to bear those defendants' proportionate share of liability, which equates to their divisible portion of the harm? Admittedly, in these situations corrective justice notions are the most compelling against allowing proportionate liability only. Nevertheless, even in cases of insolvent tortfeasors, the principle of apportionment should not be abandoned.

7. No "Innocent" or "Cause-Free" Plaintiffs in Toxics Cases

There will almost always be background or baseline risks for the disease. This distinguishes toxics cases from accident cases and challenges the "innocent plaintiff" presumption. In Dafler, even if the plaintiff had not smoked and had sued multiple asbestos manufacturers, some chance would have remained that his lung cancer was not caused by inhalation of asbestos fibers. Even with the high relative risks for asbestos exposure in that case - 5:1 to 7:1 - which translate into an etiologic probability of between eighty to eighty-five percent, a fifteen percent likelihood remained that plaintiff's was a background case of cancer. In other words, although the plaintiff is fault-free, the plaintiff is never "cause-free," in toxics disease cases. Once the plaintiff is acknowledged as not "cause-free," it is easier to accept the Dafler-like division of harm in the multiple-defendant cases.

8. Fairness Dictates Treating Insolvencies the Same in Distinct Harm

and Divisible Harm Cases

It is an unsatisfactory resolution to reject apportionment when it should otherwise apply and instead reinstate joint and several liability because a defendant is insolvent. In every tort case, the plaintiff faces the risk that a defendant may be uninsured, underinsured, insolvent, or otherwise incapacitated from satisfying a judgment against him. In Dafler, if the one nonsettling defendant could not satisfy the judgment against it, the plaintiff would have recovered nothing, but with multiple defendants the prospects that the plaintiff will recover all or most of her judgment increase. Each defendant pays for the divisible portion of the harm it caused; the plaintiffs action is similar to suing one defendant (D1) for negligently breaking her leg and another defendant (D2) for breaking her arm. In that case, insolvency of D1 would not result in D2 being jointly liable for the total damages from two broken appendages. Whether the apportionment is predicated on distinct harms or a reasonably divisible harm should make no difference in the effect of defendant insolvency.

9. Defendant Is Not A Cause of the Entire Harm

One persuasive argument for joint and several liability from a fairness perspective is that, after all, each defendant was "a" cause of the total harm, and therefore when one defendant is insolvent it remains fair and reasonable to impose on the other defendants financial responsibility for the insolvent party's proportionate share. However, in toxics cases, causation is a matter of degree, of gradations, that are generally measurable. It simply cannot be demonstrated that each was an independent cause of all the harm. For that reason, some of the fairness rationale for transferring one defendant's proportionate responsibility to another defendant is lacking in toxic tort cases.

10. Fairness Decries Disproportionate Liability: The Market Share

Analogy

The CERCLA experience led several federal appeals courts to bemoan the harsh and unfair effects accompanying the imposition of joint and several liability. In effect, those courts were unwilling to further countenance the disproportionate liabilities that would otherwise attach to the defendants in those cases. Concern for disproportionate liability is the basis for many of tort law's rules, such as no-duty and proximate-cause limitations.(416) For example, the economic loss rule, which shields a negligent actor from liability for purely economic losses, is predicated on the apprehension that the ripple effects of negligent conduct may produce staggering economic losses, even where no physical harm to persons or property results.(417)

In fact, it seems pretty clear that the harshness of disproportionate liability is the single most powerful argument favoring the abolition or modification of joint and several liability. Whether is CERCLA, toxic tort, or even sporadic accident litigation, the concern of legislatures is to soften the excessive liability imposed on so-called "deep-pocket" firms. The principal argument against this position is that each defendant is responsible for the entire harm, a cause in fact and proximate or legal cause of all the harm sustained. However, that characterization, while valid in describing the causal nexus in sporadic accident cases, simply does not hold in toxic tort cases.

If causation is a limiting principle of corrective justice, as most legal theorists of that school acknowledge, then the special properties of toxic causation require legislatures and courts to modify how tort law can best achieve its goals. When the all-or-nothing notions of causation no longer govern, then necessarily the legal rules built on that assumption must be modified. Even if liability created by joint and several liability may not be disproportionate or excessive when defendant's tortious conduct truly is a but-for or sufficient cause of all the harm suffered, when a defendant's tortious conduct is not a legal cause of all the harm or such a relationship cannot be demonstrated, then joint and several liability does produce disproportionate and excessive liability.

In toxic tort cases, the imposition of joint and several liability for divisible harms compels a defendant to compensate for the divisible portions of the harm properly attributable to others. From that realization flows the conclusion that principles of fairness should not sanction such a result.(418)

Relatedly, several courts have rejected the application of joint and several liability to defendants held liable under the market-share theory due to concerns over disproportionate liability. When market-share theory is predicated on the probability of actual causation,(419) a defendant's share of the market will converge with the harm it actually caused over many cases (but not in individual cases) but it is unfair to hold it liable for a percentage of harm it did not cause. The proposed new Restatement (Third) of Torts that governs products liability explicitly rejects joint and several liability in market-share cases:

However, if a court does adopt some form of proportional liability, the liability of each defendant is properly limited to the individual defendant's share of the market. The rules of joint and several liability are incompatible with a market share approach. Unlike the case of concurrent tortfeasors, where several parties contribute to a single plaintiff's entire harm, in the imposition of market share liability it is not established that all the defendants contributed to the plaintiff's injury. Instead, each defendant should pay for harm in proportion to the risk that it caused in the market at large. Joint and several liability would impose liability on each defendant for the entirety of the harm based on its presence in the market with other defendants. In the absence of some concerted conduct among the defendants, such liability is inappropriate.(420)

Furthermore, because market shares are a known percentage, applying joint and several liability, coupled with absent defendants, will always result in disproportionate liability. Toxic tort cases also involve proportionate causation, based not on one's sales of a product, but on the relative risks created by the defendants' conduct. When joint and several liability is applied, and not all defendants are joined or solvent, some will necessarily end up paying beyond their responsibility. Although this may not be an unfair result when no basis for apportionment can be found, it is unfair when rational bases do exist.

11. Fairness Rejects Having Defendants Bear the Portion of Harm

Attributable to Immune Actors

While the reference in the text is to insolvent tortfeasors, Professor Twerski has pointed out that the real culprit is not tortfeasors who are insolvent, but rather those on whom some immunity has been conferred, often by legislatures.(421) Even if there is some unfairness in having a plaintiff bear a loss attributable to a defendant without financial resources, there is no unfairness in refusing to have one solvent defendant bear the share of an immune defendant. If the legislature has conferred workers' compensation system tort immunity to an employer, or governmental immunity, there is no reason based on fairness why another nonimmune defendant, rather than the plaintiff, should absorb that portion of the loss.

B. Achieving Optimum Levels of Deterrence

In my view, the efficiency and deterrence theory is unsatisfactory as an explanatory model of tort rules and therefore is not addressed at length. Toxic apportionment is preferable from a deterrence perspective because plaintiffs are given an incentive to minimize voluntary toxic-related risks to themselves. Additionally, defendants who compensate for the portion of the harm commensurate with the toxic risks they created, will be appropriately deterred from tortiously creating such risks. But the risk contribution model of apportionment includes more than tortious conduct of the parties. It includes any causally related toxic risks whether or not they are characterized as fault. To the extent that the model incorporates conditions or forces that are not subject to control of the parties, deterrence is not relevant. hi the plaintiff-defendant paradigm, a plaintiff who is a source of a toxic risk that is simply a lifestyle behavior, one not universally recognized as creating very significant health risks, such as moderate consumption of alcohol, is unlikely to be deterred by having to bear a portion of the harm she sustains.

Or take the twin fires case again. If the plaintiff was nontortiously responsible for one of the fires that combined with the fire tortiously started by the defendant, deterrence-based rules will not support apportionment.

The deterrence issues are also affected by the nature of the causation proof. Professors Rosenberg and Robinson(422) maintain that adoption of a proportionality rule of causation based on purely statistical and probabilistic evidence in mass exposure cases will optimize the level of care taken by defendants. Because the defendants will pay only for the excess risk of injury that their toxic substances created, firms win have an incentive to invest in safety up to the point of preventing that excess risk.(423) The apportionment model advanced here is not a proportionality rule, although the content of scientific proofs will obviously rely heavily on probabilistic evidence. The epidemiological and toxicological proofs win enable experts to render opinions on the likelihood that the plaintiffs exposure to a toxic substance was a cause or contributing factor in the plaintiff's injury. Coupled with particularistic evidence respecting the plaintiff, the experts' opinions may not necessarily be expressed as a fixed percentage. While relative risks derived from epidemiological studies may yield such a fixed number, in fact causal analysis looks to a variety of information including confidence intervals, ranges, and a wealth of other data. Noted epidemiologists such as Kenneth Rothman suggest that it is more informative to present an array of data, some of which may not be statistical at all.(424)

The advocates of a proportionality, rule are motivated by the belief that a preponderance of the evidence rule will not adequately deter wrongful behavior because it permits defendants to escape some liabilities for injuries that they have caused, especially those where the relative risks exceed 1.0 but are not greater than 2.0.(425) Professor Fischer has demonstrated the weakness of a pure proportional causation rule from a deterence standpoint.(426) Because the risk contribution model of toxic apportionment does not alter the underlying causation standards, it should achieve deterrence objectives.

The causation standard in toxic exposure litigation combined with apportionment will achieve optimum levels of deterrence. Firms will invest in safety research and precautionary measures commensurate with the risks potentially posed by the hazardous substances they generate. Moreover, the risks of underdeterrence are minimized by the very nature and content of the scientific proofs that the model requires. By eschewing exclusive dependence on statistically significant studies' yielding a unitary relative risk greater than 2.0, the plaintiffs will recover (or at least present a prima facie case of causation sufficient to reach a jury) when the totality of evidence suggests that the substance was a likely contributing factor to the plaintiffs harm. To the extent that other factors also contributed to the harm (for example, the plaintiffs smoking history or use of drugs), juries are competent to apportion the harm among those contributing causes.

The proportionality rule is premised on the apprehension of underdeterrence resulting when a relative risk or etiological probability (say 1.5 and 330%) fail to surmount the preponderance of the evidence burden of proof. The reality is, however, that a plaintiff who could demonstrate a relative risk of 1.5 and who could couple that with evidence of other nonstatistical causal criteria, such as biological plausibility and consistency and related toxicological studies, and sufficiently eliminated differential causes, may satisfy the causation standard. The application of a comprehensive and more informative causation standard should be more hospitable to plaintiffs who possess deserving claims. The rigorous standard of causation is not so rigid as to declare insufficient an abundance of epidemiological and toxicological evidence reveling a plausible causal connection between a plaintiffs exposure to a toxic substance and his injury. Rosenberg's dual assumptions - that the data would permit finding a fixed statistical probability and that if the probability were below fifty-one percent, the plaintiff would invariably lose - have not been supported in the decade of litigation since his article.(427)

In addition, other sources of deterrence, exist apart from tort law. The growing role of regulatory agencies in the risk assessment process as it relates to the environment, the workplace, consumer products, and pharmaceuticals further minimizes the potential for underdeterrence. These regulatory programs, which are targeted at public health and safety concerns identified by Congress, are extremely conservative in estimating risks and liberal in imposing limitations and restrictions on allowable exposures for toxic substances. Such regulatory systems promulgated or administered by the Environmental Protection Agency, Food and Drug Administration, the Agency for Toxic Substances and Disease Registry, the Occupational and Health Safety Agency, the National Institute of Occupational Safety and Health, the Consumer Products Safety Commission, and the National Toxicology Program currently offer considerable protection against and deterrence of high-risk or injury-creating conduct.

Finally, turning to the deterrent effects of apportioning harm versus applying a rule of joint and several liability, several points are relevant. Joint and several liability discourages firms from making sufficient investments in toxic waste reductions technology. If other firms can capture one firm's investment benefits by application of joint and several liability, then why make the investment? Apportionment or divisibility of the harm minimizes such disincentives by permitting each defendant to limit its liability to- only the harm that its risk-creating activities produce. A firm must balance in the decision-making process, ex ante, a smaller probability of a potentially large liability, representing the effect of joint and several liability, against a higher probability of a lesser liability because of the apportionment of the harm among defendants. It is not altogether obvious how a firm would weigh the competing options, and it is certainly plausible that it would regard them to be equivalent. Moreover, even if one assumes that theoretically apportionment yields better incentive effects, will a firm have available to it the raw information required to reach that conclusion? Will defendants know about the presence of other defendants who may have contributed to the harm? What will they know about the toxicity of those firm's toxic substances? Will they possess any information respecting those firm's capacities to manage those substances safely? And finally, will they be able to estimate what the total liabilities are, such as remediating a waste site or paying wrongful death damages? It seems that the net effect of these considerations is to render the underterrence argument of modest weight only. Certainly if that were the sole basis for advocating adoption of a risk contribution model of toxic apportionment, it would not be worth the trouble.

But what if firms can anticipate the imposition of greater liabilities because of joint and several liability? If the information were available and the firm was a "deep- pocket" firm, the risks of overdeterrence are not trivial. If firms can predict disproportionate and excessive liabilities because of their wealth and their ability to anticipate the presence of insolvent or immune firms that would not pay their divisible portion of the harm, then overdeterrence will result. The same disproportionate liability that flew in the face of corrective justice considerations likewise raises overdeterrence concerns. Here too, toxic apportionment should also reduce the risks of overdeterrence by signaling to prospective defendants, especially deep-pocket firms, that they can limit their liability to the harm which their risk-creating activities actually cause.

X. Conclusion

The work represented here is only a beginning. Considerably more effort is needed in order to assess the feasibility of this proposal. Moreover, some issues of obvious importance were not addressed, such as who should bear the burden of proof on apportionment. Should there be a burden of going forward on the defendant to show some basis for apportionment, and then a requirement for the plaintiff to rebut that evidence by demonstrating that no apportionment is possible? Should we return to the position of the Restatement (First) of Torts that placed the burden on the plaintiff in nuisance cases to prove the harm caused by each of multiple defendants?

Second, how should the synergism case be handled? Is it fair to the plaintiff to apportion the harm solely on the basis of the underlying relative risks, when a synergistic effect has resulted in a substantially greater risk than he bargained for?

These and other matters deserve further consideration. Hopefully this Article will stimulate further discussion on these issues. * Professor of Law, Thomas M. Cooley Law School. J.D. 1967, Harvard University School of Law; B.A. 1964, Ohio Wesleyan University. (1) Toxic torts comprise harms to persons, property, or the environment due to the toxicity of a product, substance, or process. The term "toxic tort" was first used in a judicial opinion in In re Agent Orange Product Liability Litigation, 506 F. Supp. 737, 743 (E.D.N.Y. 1979), revd, 635 F.2d 987 (2d Cir. 1980). An injury may be a personal, physical injury to a worker, patient, product user or consumer, or bystander. See, e.g., Shaw v. Dow Brands, Inc., 994 F.2d 364, 365 (7th Cir. 1993) (personal injury claim based on fumes from defendant's mildew and stain remover); Shorter v. Champion Home Builders Co., 776 F. Supp. 333, 334-35 (N.D. Ohio 1991) (claim against mobile home builders alleging exposure to formaldehyde vapors, and alleging personal injuries, including headaches and emotional harm). An injury may also include property damage, ranging from rendering an individual premises uninhabitable to the contamination of a watershed. See, eg., Sterling v. Velsicol Chem. Corp., 855 F.2d 1188, 1192-93 (6th Cir. 1988) contamination of 242 acres used as a landfill site for by-products of manufacturing chlorinated hydrocarbon pesticides); Renaud v. Martin Marietta Corp., 972 F.2d 304, 305 (10th Cir. 1992) (improperly disposed wastewater from missile plant contaminating residential water supply). (2) The types of toxic exposures that are the subject of this Article are well developed. For example, one commentator defines hazardous (i.e., toxic) substances in terms of exposure: they are "those occupational or environmental toxins or consumer products" that 1) feature low dose and chronic exposure; 2) include hidden toxic effects, at least at the beginning of exposure; 3) have a latency period between exposure and harmful effect; 4) cause irreversible injury; and 5) leave no evidence that provides a causal chain between exposure and disease. Troyen A Brennan, Causal Chains and Statistical Links: 7he Role of Causation in Hazardous Substances Litigation, 73 Cornell L. Rev. 469, 501-02 (1988). A few comments about this definition are in order. First, the requirement of low dose and chronic exposure stems from the fact that high dose, acute exposures simply do not engender the same kinds of intractable causal problems that have frustrated courts and commentators. Similarly, items 2), 3), and 4) engender the same difficulties in proof that are not present when the effects are obvious, immediate, and reversible. Finally, item 5) represents the prototypical situation in diseases such as cancer that leave no evidence or "fingerprints" of etiology. See Bernard D. Goldstein & Mary Sue Henifin, Reference Guide on Toxicology, in Reference Manual on Scientific Evidence 181 (1994). (3) For a history of asbestos litigation in the United States, see Paul Brodeur, Outrageous Misconduct. The Asbestos Industry on Trial 6-70 (1985). Asbestos is the generic term used to describe a group of fibrous metasilicate minerals that occur naturally in the environment. The chemical composition and structure of asbestos confer remarkable properties of flexibility, tensile strength, and resistance to acid or fire. Because of asbestos' unique properties and varied uses, many industrial and commercial workers have been occupationally exposed to it. See generally Agency for Toxic Substances & Disease Registry, U.S. Dep't of Health & Human Services, Toxicological Profile For Asbestos (1990) [hereinafter Profile For Asbestos]. The precise number of occupationally exposed workers in the United States is uncertain, but estimates have ranged from approximately 8 to 14 million alive in 1980. Estimates of "substantial" exposures range from 70,000 to 4 moon. D.E. Lilienfeld et al., Projection of Asbestos Belated Diseases in the United States, 1985 to 2009: I Cancer, 45 Brit. J. Indus. Med. 283, 285-86 (1988) (reviewing earlier estimates of exposure and projected mortalities . (4) For examples of authorities discussing the nature of the causal uncertainties in cancer and other diseases, see U.S. Environmental Protection Agency, Guidelines for Carcinogen Risk Assessment, 51 Fed Reg. 33,992 (Sept. 24, 1986); Bert Black, A Unified Theory of Scientific Evidence, 56 Fordman L. Rev. 595 (1988); Troyen A. Brennan, Helping Courts with Toxic Torts: Some Proposals Regarding Alternative Methods for Presenting and Assessing Scientific Evidence in Common Law Courts, 51 U. Pitt. L. Rev. 1 (1989); E. Donald Elliott, The future of Toxic Torts: Of Chemophobia, Risk as a Compensable Injury and Hybrid Compensation System, 25 Hous. L. Rev. 781 (1988); Daniel A. Farber, Toxic Causation, 71 Minn. L. Rev. 1219 (1987); Ora F. Harris, Toxic Tort Litigation and the Causation Element: Is There Any Hope of Reconciliation?, 40 Sw. L.J. 909 (1989) (attempting to reconcile traditional causation concepts with modern toxic tort litigation); Peter Huber, Safety and the Second Best: The Hazards of Public Risk Management in the Court, 85 Colum. L. Rev. 277 (1985); Allan Kanner, Future Trends in Toxic Tort Litigation, 20 Rutgers L.J. 667 (1989); Francis E. McGovern, Management of Multiparty Toxic Tort Litigation: Case Law and Trends Affecting Case Management, 19 Forum 1 (1983); Glen O. Robinson, Multiple Causation in Tort Law: Reflections on the DES Cases, 68 Va. L. Rev. 713 (1982); David Rosenberg, The Causal Connection in Mass Exposure Cases: A "Public Law" Vision of the Tort System, 97 Harv. L. Rev. 849 (1984); Steve Gold, Note, Causation in Toxic Torts: Burdens of Proof, Standards of Persuasion, and Statistical Evidence, 96 Yale L.J. 376 (1986); Note, Tort Actions for Cancer: Deterrence, Compensation, and Environmental Carcinogenesis, 90 Yale L.J. 840 (1981). (5) See W. Page Keeton et al., Prosser and Keeton on the Law of Tort [sections] 41, at 265-67 (5th ed. 1984). (6) See Restatement (Second) of Torts [sections] 431(a) (1965); See also Keeton et al., supra note 5, [subsections] 4142, at 267-68, 278; Richard W. Wright, Causation, Responsibility, Risk, Probability, Naked Statistics, and Proof. Pruning the Bramble Bush by Clarifying the Concepts, 73 Iowa L. Rev. 1001, 1018-19 (1988) (setting forth a test described as the Necessary Element of a Sufficient Set (NESS) test of causation). (7) See Profile For Asbestos, supra note 3. The most significant epidemiological study establishing the causal relationship between asbestos and asbestosis and lung cancer is Irving J. Selikoff et al., The Occurrence of Asbestosis Among Insulation Workers in the United States, 132 Annals N.Y. Acad. Sci. 139 (1965). For a list of over 400 studies relating to asbestos-related disease and health problems, see Profile for Asbestos, supra note 3, at 97-135. Following the decision in Borel v. Fibreboard Paper Prods. Corp., 493 F.2d 1076 (5th Cir. 1973), cert. denied, 419 U.S. 869 (1974), courts began to take judicial notice of the causal relationship of asbestos exposure to asbestosis, lung cancer, and mesothelioma and held that causation was established as a matter of law. See Karjala v. Johns-Manville Prod. Corp., 523 F.2d 155, 160 (8th Cir. 1975); see also Bertrand v. Johns-Manville Sales Corp., 529 F. Supp. 539, 544 (D. Minn. 1982) ("[I]t is clear that it is appropriate to estop litigation on the issue of whether asbestos dust can cause diseases such as asbestosis and mesothelioma. This proposition is so firmly entrenched in the medical and legal literature that it is not subject to serious dispute."); Hardy v. Johns-Manville Sales Corp., 509 F. Supp. 1353, 1354 (E.D. Tex. 1981), rev'd in part on other grounds, 681 F.2d 334 (5th Cir. 1982) ("Ten years after Borel, it cannot be seriously argued that asbestos exposure causes disease. So comfortable are we with that assertion, that a former Secretary of Health, Education and Welfare estimated that more than 67,000 human lives are taken each year by asbestos-related cancers. Thus far in the litigation, asbestos has been found to be a competent producing cause of asbestosis and mesothelioma." (footnote and citation omitted)); Flatt v. Johns-Manville Sales Corp., 488 F. Supp. 836, 841 (E.D. Tex. 1980) (As a matter of law asbestos dust is a producing cause of certain lung diseases, including asbestosis and mesothelioma."). (8) Beginning with two major studies published in 1950 and 1952, the epidemiological evidence of a causal association between smoking and lung cancer is impressive. See Richard Doll & A. Bradford Hill, A Study of the Aetiology of Carcinoma of the Lung, 2 Brit. Med. J. 1271 (1952); Ernest L. Wynder & Evarts A Graham, Tobacco Smoking as a Possible Etiologic Factor in Bronchiogenic Carcinoma, 143 JAMA 329, 336 (1950); see also U.S. Dep't of Health & Human Services, The Health Consequences of Smoking: Cancer: A Report of The Surgeon General vi; 5-6 (1982) [hereinafter Health Consequences]; U.S. Dep't of Health, Educ. & Welfare, Smoking and Health: Report of the Advisory Committee to the Surgeon General of the Public Health Service (1964). For judicial statements regarding that causal relationship, see Cipollone v. Liggett Group, Inc., 893 F.2d 541, 576-77 (3d Cir. 1990), rev'd in part, aff'd in part, 112 S. Ct. 2608 (1992). (9) For discussions of the consequences of toxic substance exposure and the effects of several variables - such as exposure duration; types of exposure consequences; exposure levels; and substance toxicity, proximity, and concentration - on such exposure, see Environ Corp., Counsel in Health & Envtl. Sci., Elements of Toxicology and Chemical Risk Assessment, A Handbook for Nonscientists, Attorneys and Decision Makers (1986) [hereinafter Elements of Toxicology]; Ernest Hodgson & Patricia Levi, Modern Toxicology (1987); Michael A. Kamrin, Toxicology: A Primer on Toxicology Principles and Applications 36-38 (1988); Ellen K. Sibergeld, The Role of Toxicology in Causation. A Scientific Perspective, 1 Cts., Health Sci. & the Law 374 (1991); Chris F. Wilkinson, Toxicology and the Law, in The Role of Science in Toxic Tort Litigation: Evaluating Causation and Risk (ABA Torts & Insurance Practice Section ed. 1988). (10) See, eg., Brisboy v. Fibreboard Corp., 418 N.W.2d 650 (Mich. 1988). (11) By "fairer and more efficient results," or "corrective justice," I mean that tort rules should endeavor to achieve justice as between a plaintiff and defendant by emphasizing their individualized relationship; focusing on the plaintiffs actual harm and the defendant's causal role in that loss, through the creation of risks, the rules seek to "correct" that appropriation by the defendant's payment of damages. Their roots are moral, a "just desserts" idea that it is fair that a defendant pay damages for the loss that its conduct caused.

There are several alternative conceptions of corrective justice. Professor Richard Epstein has advocated an approach that seeks to "develop a normative theory of torts that takes into account common sense notions of individual responsibility." Richard A. Epstein, A Theory of Strict Liability, 2 J. Legal Stud. 151, 151 (1973). He maintains that with actors responsible for the harm they cause, causation is the central ingredient of corrective justice. Causal assertions initially necessitate a presumption of liability and even compel the question of "whether there are in fact any means to distinguish between causation and [legal] responsibility, so close is the connection between what a [person] does and what he is answerable for." Id. at 169. By preferring strict liability over negligence and refusing to accept any affirmative duties to act, Epstein perceives corrective justice so that cause creates the basis and boundary of liability. Thus, apportionment of harm that is tied to causation is compatible with the basic idea of fairness. However, Epstein's notions of causation are very individualized, and he seems to disavow any room for probabilistic ideas of causation that are endemic to toxic torts. Indeed, he advocates the disposition of many toxic tort cases by contracting out rather than through tort law. Richard A. Epstein, Causation in Context: An Afterward, 63 Chi.-Kent L. Rev. 653, 677 (1987).

Professor George Fletcher advances a theory of fairness premised on the principle that "a victim has a right to recover for injuries caused by a risk greater in degree and different in order from those created by the victim and imposed by the defendant - in short, for injuries resulting from nonreciprocal risks." George Fletcher, Fairness and Utility in Tort Theory, 85 Harv. L. Rev. 537, 542 (1972). His paradigm of reciprocity holds that in al communities of reciprocal risks - that is, background, shared risks - those who cause damage ought not to be held liable. He premises his idea of fairness on the belief that individuals in society have a common right to roughly the same degree of security from risk, but "interests of the individual require us to grant compensation whenever this disproportionate distribution of risks injures someone subject to more than his fair share of risk." Id. at 550-51. Fletcher's ideas of fairness incorporate a principle that fits quite comfortably with the apportionment model advanced herein, which seeks to compare the magnitude of the risks created by each actor, coupled with causation, and to render an allocation of liability based on that comparison.

Jules Coleman, another legal philosopher who has written extensively on the notion of corrective justice, observes in describing the views of Epstein and others that

there is a general principle of responsibility which holds that in order to be just liability must depend on the responsibility of the injurer. In this view, responsibility is analyzed in terms of causality or causal powers, not in terms of fault. Because injurers are liable for all and only the causal results of their agency, the principle of responsibility renders liability both strict and just.

Jules L, Coleman, Tort Law and the Demands of Corrective Justice, 67 Ind. L.J. 349, 350 (1992) (emphasis added). However, unlike Epstein and Fletcher, Coleman perceives corrective justice as demanding that only "wrongful (or unjust) gains and losses be rectified, eliminated, or annulled," because "[n]ot every loss ... is the concern of corrective justice." Id. at 357. While fault is important in determining what is a wrongful gain or loss, strict liability is too blunt a tool for such a purpose. Part I contains further discussion of corrective justice principles as applied to the causation and risk contribution models of apportionment. See infra notes 19-28 and accompanying text (12) See, e.g., Elam v. Alcolac, Inc., 765 S.W.2d 42, 174 (Mo. Ct. App. 1988), cert. denied, 493 U.S. 817 (1989). In affirming the plaintiff s verdict, the court applied a substantial factor test:

[A]lthough our law requires proof of cause to recover in tort, it does not require proof of a single cause. The substantial factor standard - which ascribes liability to a cause which has played an important part in the production of the harm, even though the harm may have occurred absent that cause - is particularly suited to injury from chronic exposure to toxic chemicals where the subsequent manifestation of biological disease may be the result of a confluence of causes.

Id. (citations omitted); see also Tragarz v. Keene Corp., 980 F.2d 411, 424-25 (7th Cir. 1992). With most of the defendants settling, the court approved the exclusion of evidence regarding the plaintiff's exposure to products made by the settling defendants. Tragarz, 980 F.2d at 424-25. Applying a substantial factor test of "whether each contributing cause, standing alone, is a substantial factor in causing the alleged injury," the court nevertheless rejected the argument that exposure to one defendant's product might not be substantial in comparison to others because it might result in verdicts at odds with "the purposes of the substantial factor test which is aimed at alleviating the inequities that result when applying the but-for test in a multi-defendant case." Id. In Welch v. Keene Corp., 575 N.E.2d 766 (Mass. App. Ct. 1991), the court confused the substantial factor test of causation with apportionment of damage. The plaintiff had installed asbestos for 22 years, but also was exposed to the defendant's asbestos products for a shorter period. The court stated that once an expert testified that the cumulative effect of all asbestos exposure over the 22 years caused asbestosis and that "each inhalation of asbestos dust can result in additional damage to lung tissue," any defendant to whose products the plaintiff was exposed becomes a "substantial factor" in the disease. Id. at 770. (13) See Borman v. Raymark Indus., Inc., 960 F.2d 327 (3d Cir. 1992) (refusing to apportion damages between smoking and asbestos exposure, even with extensive epidemiological data on relative risks of each); Martin v. Owens-Corning Fiberglas Corp., 528 A.2d 947, 948 (Pa. 1987) (reversing trial court and intermediate appeals court for allowing apportionment when the plaintiff was exposed to the defendant's asbestos products for 6 years and had smoked for 45 years). (14) For a more complete review of causation issues in toxic tort litigation, see generally Gerald W. Boston, A Mass-Exposure Model of Toxic Causation: The Content of Scientific Proof and the Regulatory Experience, 18 Colum. J. Envtl. 181 (1993). (15) See, e.g., Glen O. Robinson, Probabilistic Causation and Compensation for Tortious Risk, 14 J. Legal Stud. 779 (1985); Rosenberg, supra note 4, at 925; Christopher H. Schroeder, Corrective Justice and Liability for Increasing Risks, 37 UCLA L. Rev. 439 (1990); Gregory L. Ash, Comment, Toxic Torts and Latent Diseases: The Case for an Increased Risk Cause of Action, 38 U. Kan. L. Rev. 1087 (1990); Brent Carson, Comment, Increased Risk of Disease from Hazardous Waste. A Proposal for Judicial Relief, 60 Wash. L. Rev. 635 (1985); Gold, supra note 4, at 393; Dwight C. Harvey, Comment, Epidemiologic Proof of Probability: Implementing the Proportional Recovery Approach in Toxic Exposure Torts, 89 Dick. L. Rev. 233 (1984); Barton C. Legum, Note, Increased Risk of Cancer as an Actionable Injury, 18 GA. L. Rev. 563 (1984); David S. Pegno, Note, An Analysis of the Enhanced Risk Cause of Action (or How I Learned to Stop Worrying and Love Toxic Waste), 33 Vill. L. Rev. 437 (1988). (16) See, eg., Boston, supra note 14, at 241-48; David A. Fischer, Proportional Liability: Statistical Evidence and the Probability Paradox, 46 Vand. L. Rev. 1201 (1993). (17) See Wright, Supra note 6, at 1072-74 (arguing for a strict causation requirement by modifying the definition of the harm). (18) 42 U.S.C. [subsections] 9601-9675 (1988 & Supp. V 1993). (19) Keeton et al., supra note 5, at 322. (20) Id. at 328-29, 347-48. (21) Id. at 347; (22) See, e.g., Maddux v. Donaldson, 108 N.W.2d 33 (Mich. 1961); Arnst v. Estes, 8 A.2d 201 (Me. 1939). (23) Keeton et al, supra note 5, at 348-49. (24) As discussed in part II, the Restatement (Second) of Torts [sections] 433A provides for apportionment only when the harm are distinct or divisible on some reasonable basis. (25) See infra note 171. (26) See discussion infra part II. (27) See, e.g., Reynolds v. Texas & Pac. Ry., 37 La. App. 693 (1885) (holding that when a corpulent plaintiff fell down a flight of stairs because of a misstep while the defendant was negligent in failing to properly light the stairway, the defendant could be liable because "where the negligence of defendant greatly multiplies the chances of accident to the plaintiff, and is of a character naturally leading to its occurrence, the mere possibility that it might have happened without the negligence is not sufficient to break the chain of cause and effect"). If direct evidence is unavailable, circumstantial evidence may establish factual causation because "to require precision of proof would impose an insurmountable burden"). (28) See, e.g., In re Agent Orange Prod. Liab. Litig., 597 F. Supp. 740 (E.D.N.Y. 1984), affd, 818 F.2d 145 (2d Cir. 1987), cert. denied, 484 U.S. 1004 (1988) (approving a class action settlement, Judge Jack Weinstein, in dicta, describes a class solution to the causation problem in which Agent Orange manufacturers would pay damages that correspond to the overall increased risk caused by the group's exposure to the herbicide, and the plaintiff class would divide the total equally even though for most the disease has not yet appeared or had not run its complete course). (29) 248 S.W.2d 731 (Tex. 1952). (30) 495 F.2d 213 (6th Cir.), cert. denied, 419 U.S. 997 (1974). (31) As the court stated, We believe the principal question presented by this appeal may be phrased thus: Under the law of the State of Michigan, may multiple defendants, whose independent actions of allegedly discharging pollutants into the ambient air thereby allegedly create a nuisance, be jointly and severally hole to multiple plaintiffs for numerous individual injuries which plaintiffs claim to have sustained as a result of said actions, where said pollutants mix in the air so that their separate effects in creating the individual injuries are impossible to analyze. Id. at 215. (32) Landers, 248 S.W.2d at 733 (noting that the plaintiff could not apportion damages with reasonable certainty between two defendants). (33) See Borman v. Raymark Indus., Inc., 960 F.2d 327 (3d Cir. 1992) (refusing to apportion damages between smoking and asbestos exposure, even with extensive epidemiological data on the relative risks of each); Martin v. Owens-Corning Fiberglas Corp., 528 A.2d 947 (Pa. 1987) (reversing trial court and intermediate appeals court for allowing apportionment when plaintiff was exposed to defendant's asbestos products for 6 years and had smoked for 45 years). (34) As will be discussed below in part II, section 433A of the Restatement (Second) of Torts provides for apportionment in which harms are distinct or divisible on some reasonable basis. (35) Restatement (Second) of Torts [sections] 433A cmt. i (1965) ("By far the greater number of personal injuries, and of harms to tangible property, are thus normally single and indivisible.") (36) See discussion infra part II. (37) Restatement (Second) of Torts [sections] 433A (1965). (38) 179 N.W. 45 (Minn. 1920). (39) Id. at 46. (40) Id. at 49. (41) Id. (42) While Anderson is not a toxic torts case, fire certainly bears many of the same risk factors. Interestingly, the Resource Conservation and Recovery Act, 42 U.S.C. [subsections] 6901-6987 (1988 & Supp. V 1993), views flammability as one of the principal categories for determining whether waste is hazardous. See Identification and Listing of Hazardous Waste, 40 C.F.R. [sections] 261.21 (1994) (ignitability one of four characteristics for classifying waste as hazardous). (43) Sections 431 and 432 of the Restatement (Second) of Torts are relevant: [sections] 431. What Constitutes Legal Cause The actor's negligent conduct is a legal cause of harm to another if (a) his conduct is a substantial factor in bringing about the harm, and (b) there is no rule of law relieving the actor from liability because of the manner in which his negligence has resulted in the harm. [sections] 432. Negligent Conduct as Necessary Antecedent of Harm (1) Except as stated in Subsection (2), the actor's negligent conduct is not a substantial factor in bringing about harm to another if the harm would have been sustained even if the actor had not been negligent. (2) If two forces are actively operating, one because of the actor's negligence, the other not because of any misconduct on his part, and each of itself is sufficient to bring about harm to another, the actor's negligence may be found to be a substantial factor in bringing it about. Restatement (Second) of Torts [subsections] 431-432 (1965). Professor Wright treats the twin fire cases as an example of overdetermined or duplicative cause, and he would hold the defendant liable for the entire loss. Richard W. Wright, The Efficiency Theory of Causation and Responsibility: Unscientific Formalism and False Semantics, 63 Chi.-Kent L. Rev. 553, 571 (1987) [hereinafter Wright, Efficiency Theory]; Richard W. Wright, Causation in Tort Law, 73 Cal. L. Rev. 1735, 1791-94, 1796 (1985) [hereinafter Wright, Causation in Tort Law].

One economic model of apportionment would relieve the defendant of all liability, because the other fire of unknown origin already destroyed the present value of the property. Mario J. Rizzo & Frank S. Arnold, Causal Apportionment in the Law of Torts: An Economic Theory, 80 Colum. L. Rev. 1399, 1414 (1980). However, that model also provides for apportionment if the two fires are both tortiously set and would look to the risks created by each tortfeasor. Id. at 1415. (44) See discussion of Landers v. East Texas Salt Water Disposal Co., 248 S.W.2d 731 (Tex. 1952), infra part IV. (45) Restatement (Second) of Torts [sections] 433A cmts. c, d (1965); see discussion infra part II. (46) See, e.g., Lewis A. Kornhauser & Richard L. Revesz, Sharing Damages Among Multiple Tortfeasors, 98 Yale L.J. 831 (1989); William M. Landes & Richard A. Posner, Joint and Multiple Tortfeasor: An Economic Analysis, 9 J. Legal Stud. 517 (1980). (47) See Pierringer v. Hoger, 124 N.W.2d 106 (Wis. 1963) (where plaintiff settled with tortfeasor A, and in accepting payment of settlement, agreed to reduce his claim against tortfeasor B to B's proportionate liability, whatever that turned out to be). (48) Restatement (Second) of Torts (1965). (49) Id. [sections] 433A. (50) Id. [sections] 433A cmt. a. The comment provinces: They apply where each of the causes in question consists of the tortious conduct of a person; and it is immaterial whether all or any of such persons are joined as defendants in the particular action. The rules stated apply also where one or more of the contributing causes is an innocent one, as where the negligence of a defendant combines with the innocent conduct of another person, or with the operation of a force of nature, or with a pre-existing condition which the defendant has not caused, to bring about the harm to the plaintiff. The rules stated apply also where one of the causes in question is the conduct of the plaintiff himself, whether it be negligent or innocent. Id.; see also id. [sections] 433A cmt. e. (51) Id. [sections] 433A cmt. b. (52) Id. (53) Id. [sections] 433A cmt. c. (54) The example is unclear because the Restatement assumes that the toxicity of each polluting substance is comparable and that the relationship between the two sources is additive, as opposed to multiplicative or exponential. However, even in that event, a risk contribution model would be workable. See infra parts IV, V. (55) Restatement (Second) of Torts [sections] 433A cmt. d (1965). (56) Id. (57) See infra part IV. (58) 42 U.S.C. [subsections] 9601-9675 (1988 & Supp. V 1993). (59) Section 881 of the Restatement also reinforces the provisions of section 433A. Section 881 provides: [sections] 881. Distinct or Divisible Harms If two or more persons, acting independently, tortiously cause distinct harms or a single harm for which there is a reasonable basis for division according to the contribution of each, each is subject to liability only for the portion of the total harm that he has himself caused. Restatement (Second) of Torts [sections] 881 (1979). Comment c of section 881 uses private nuisance as an example: The rule stated in this Section finds frequent application in cases of private nuisance in which the interference with the plaintiffs use and enjoyment of his land is capable of apportionment, on the basis of the respective degree of contribution among two or more contributing tortfeasors. (60) Restatement (Second) of Torts [sections] 433A cmt. h (1965). (61) Id.; see also id. [sections] 433A cmt. i. The comments do not use the term "joint and several liability," but state that a solvent defendant is liable for the entire harm. See Restatement (Second) of Tort [sections] 875 (1979) (referring to holding a defendant liable for the entire harm when the harm is "single and indivisible"); id. [sections] 879 (holding a defendant liable for the entire harm when the "legal cause of the harm cannot be apportioned"). (62) Restatement (Second) of Torts [sections] 433A cmt. i (1965). (63) Id. (64) Id. (65) Id. [sections] 433B(1). (66) Id. [sections] 433B(2). (67) Id. [sections] 433B(2) cmt. c. (68) Id. [sections] 433B(2) cmt. d. (69) Id. (70) Id. [sections] 433B(2) cmt. e. (71) Id. (72) See CERCLA, 42 U.S.C. [sections] 9622(g)(1)(a) (1988 & Supp. V 1993) (authorizing early settlement for potentially responsible parties whose contributions "are minimal in comparison to other hazardous substances at the facility). For cases denying apportionment, see, e.g., O'Neil v. Picillo, 883 F.2d 176. (1st Cir. 1989), cert. denied, 110 S.Ct. 1115 (1990); United States v. Monsanto Co., 858 F.2d 160 (4th Cir. 1988), cert. denied, 109 S.Ct. 3156 (1989); United States v. Chem-Dyne Corp., 572 F. Supp. 802 (S.D. Ohio 1984). For recent cases authorizing apportionment, see, e.g., United States v. Alcan Aluminum Corp., 964 F.2d 252 (3d Cir. 1992); United States v. Alcan Aluminum Corp., 990 F.2d 711 (2d Cir. 1993). For a discussion of these cases, see infra part VI. (73) Restatement (Second) of Tort [sections] 433A cmt. i, & illus. 7 (1965). (74) Compare Dafler v. Raymark Indus., Inc., 611 A.2d 136 (N.J. Super. Ct. App. Div. 1992), aff'd, 622 A.2d 1305 (N.J. 1993) (affirming apportionment between smoking and asbestos) with Martin v. Owens-Corning Fiberglas Corp., 528 A.2d 947, 948 (Pa. 1987) (reversing instruction allowing apportionment between smoking and asbestos exposure). For a discussion of this issue, see infra part III. (75) Restatement (Second) of Torts [sections] 433A cmt. h (1965). (76) Sections 875 and 879 also bear on this point: [sections] 875. Contributing Tortfeasors - General Rule Each of two or more persons whose tortious conduct is a legal cause of a single and indivisible harm to the injured party is subject to liability to the injured party for the entire harm. [sections] 879. Concurring or Consecutive Independent Acts If the tortious conduct of each of two or more persons is a legal cause of harm that cannot be apportioned, each is subject to liability for the entire harm, irrespective of whether their conduct is concurring or consecutive. Restatement (Second) of Torts [subsections] 875, 879 (1979). These sections call for joint and several liability when the harm "cannot be apportioned" (section 879) or the harm is "single and indivisible" (section 875). This differs from comment h to section 433A, which assumes the harm and its causes are capable of apportionment. Restatement (Second) of Torts [sections] 433A cmt. h (1965). (77) Kenneth S. Abraham, Individual Action and Collective Responsibility: The Dilemma of Mass Tort Reform, 73 Va. L. Rev. 845, 862 (1987). Professor Abraham continues: Debate about whether the incentive effects created by joint and several liability rules are superior to approaches that decline to impose such forms of collective responsibility is unresolved. Arguably, the more collective responsibility a given rule entails - on its face or because of the collectivizing effects of liability insurance - the less optimal is its effect on the safety incentives of those to whom it applies. Free-riding will confound these incentives, because the cost of failing to opt" safety will be collectively, rather than individually, imposed. Because most enterprises subject to joint and several liability will not receive sufficient benefit to warrant optimal investment, they will underinvest in safety if others can capture some of that benefit. Id. at 863 (footnotes omitted). (78) Id. at 862. (79) Richard A. Epstein, Two Fallacies in the Law of Joint Torts, 73 Geo. L.J. 1377, 1385-86 (1985); see also Landes & Posner, supra note 46, at 543. Landes and Posner offer this example of an inefficient outcome in which strict liability applies: There are ten polluters of a stream, each of whom imposes costs of $1 million through his pollution. Five could avoid polluting at a cost of $600,000 each, the other five at a cost of $1.4 million each. Under a negligence standard the first five will have incentives to avoid polluting and society will be wealthier by $2 million. Under strict liability, with each polluter jointly liable for the total $10 million in pollution costs, the five polluters who can avoid pollution at a cost of $600,000 a piece will not do so, since each will stir have an expected liability of $500,000 (assuming either a no-contribution rule with each equally likely to be sued or a contribution rule in which damages are divided by the number of defendants). Landes & Posner, supra note 46 at 543. (80) Rizzo & Arnold, supra note 43, at 1422. (81) 495 F.2d 213 (6th Cir.), cert. denied, 419 U.S. 997 (1974). (82) 248 S.W.2d 731 (Tex. 1952). (83) See Brown v. Abbott Lab., Inc., 751 P.2d 470, 487 (Cal. 1988) (holding that liability is several, not joint); Sindell v. Abbott Lab., 607 P.2d 924, 937 (Cal. 1980), cert. denied, 449 U.S. 912 (1980); Conley v. Boyle Drug Co., 570 So. 2d 275, 286 (Fla. 1990) (applying several liability); Hymowitz v. Eli Lilly & Co., 539 N.E.2d 1069, 1078 (N.Y. 1994), cert. denied, 493 U.S. 944 (1989) (applying several liability). (84) Brown, 751 P.2d at 485. , (85) 611 A.2d 136 (N.J. Super. Ct. App. Div. 1992), aff'd, 622 A.2d 1305 NJ. 1993). (86) For example, assume that pretrial discovery shows two epidemiological studies, one suggesting a relative risk from exposure to a toxic substance of 2.0 and the other a relative risk of 1.5. In this situation, a mediation panel would likely resolve the difference. (87) 611 A.2d 136 (N.J. Super. Ct. App. Div. 1992), aff'd, 622 A.2d 1305 (N.J. 1993). (88) Id. at 138. (89) Id. at 140. (90) Id. (91) Id. While some studies suggest that the relationship is multiplicative, other studies found the risk is less than multiplicative but more than additive. See U.S. Dep't of Health & Human Services, Toxicological Profile for Asbestos, Update 51-52 (1993) [hereinafter Toxicological Profile for Asbestos, Update]. (92) Dafler, 611 A.2d at 141. (93) Id. (94) Id. at 145-46 (95) Id. (96) "The question [of apportionment] is primarily not one of the fact of causation, but of the feasibility and practical convenience of splitting up the total harm into separate parts which may be attributed to each of two or more causes." Id. at 141-42 (quoting Keeton et al., supra note 5, [sections] 52). (97) See Philip E. Enterline, The Role of Epidemiology and Biostatistics in Toxic Substances Litigation, in The Role of Science in Toxic Tort Litigation: Evaluating Causation and Risk 126-27 (ABA Torts & Insurance Practice Section ed. 1988). (98) The etiologic probability or attributable risk (AR) is derived by the formula:

[Mathematical Expression Omitted]

where RR = relative risk. See Boston, supra note 14, at 236 n. 10. The concept of attributable risk, which is a bit more complex than this example suggests, was developed by Philip Cole and Brian MacMahon, Attributable Risk Percent in Case-Control Studies, 25 Brit. J. Preventive Soc. Med. 242 (1971) and Olli S. Miettinen, Proportion of Disease Caused or Prevented by a Given Exposure Trait or Intervention, 99 Am. J. Epidemiology 325 (1974). (99) Enterline, supra note 97, at 133. (100) This relationship between smoking and asbestos is synergistic, meaning a correlated action or cooperation on the part of two or more agents increases the resultant effect beyond what the two would cause separately. See Joseph A. DiPaolo et aL, Asbestos and Benzo(a)pyrene Synergism in the Transformation of Syrian Hamster Embryo Cells, 27 Pharmacology 65, 71-72 (1983). Dr. Irving Selikoff reported a 90-times increased risk of lung cancer in insulators who smoked. Irving J. Selikoff et al., Asbestos Exposure, Smoking, and Neoplasia, 204 JAMA 106, 111 (1968); see also Toxicological Profile for Asbestos, Update, supra note 91, at 51-52.

For more information on this relationship between smoking and asbestos exposure, see Agency for Toxic Substances & Disease Registry, U.S. Dep't of Health & Human Services, Toxicological Profile for Asbestos: Draft for Public Comment/Update (1993); Profile for Asbestos, supra note 3, at 52; G. Berry et al., Combined Effect of Asbestos and Smoking on Mortality from Lung Cancer and Mesothelioma in Factory Workers, 42 Brit. J. Indus. Med. 12, 17 (1985) (finding death from lung cancer more likely for smokers than for nonsmokers when both groups exposed to asbestos). Indeed, some authors have opined that lung cancer is rare among nonsmoking asbestos workers. See Brooke T. Mossman & J. Bernard L. Gee, Asbestos-Related Disease, 320 New Eng. J. Med. 1721, 1724 (1989).

On smoking alone, see generally U.S. Dep't of Health, Educ. & Welfare, Smoking and Health: Report of the Advisory Committee to the Surgeon General of the Public Health Service (1964); Health Consequences, supra note 8. (101) See supra note 97 and accompanying text. (102) Asbestos and smoking is not the only important synergistic relationship. For example, the relationship between smoking and alcohol consumption also reveals a synergistic effect with respect to oral cancers. See Health Consequences, supra note 8, at 90. Many authors conclude that the relative risk of contracting oral cancer is 2.5 times that expected if the effects of tobacco and alcohol were merely added together. William M. Keane et al., Epidemiology of Head and Neck Cancer, 91 Laryngoscope 2037, 2041 (1981). The effects of alcohol and tobacco may account for 75% of the oral cancers in men in the U.S. Id.

Indeed, laryngeal cancers may be related to all three: asbestos exposure, smoking, and alcohol consumption. See Charles K. Chan & J. Bernard L. Gee, Asbestos Exposure and Laryngeal Cancer. An Analysis of the Epidemiologic Evidence, 30 J. Occupational Med. 23, 26 (1988) (concluding that smoking and drinking are the two most important factors in laryngeal cancer, while asbestos causation is weak). (103) An interesting contrast to Dafler is found in In re Manguno, 961 F.2d 533 (5th Cir. 1992). In Manguno, the jury rendered a verdict for the defendant, which the appeals court reversed because the jury instruction improperly barred three plaintiffs from arguing that cigarette smoking and workplace exposure to asbestos may have combined to cause the plaintiffs' cancer. Id. at 535-36. The court held that the instruction by a federal trial judge wrongly required the plaintiffs to prove that "but for" workplace exposure to asbestos, the long-term smokers would not have developed lung cancer. Id. at 535. The court said it was medically impossible "to distinguish whether cigarettes alone or asbestos alone" caused the workers' cancer, and it remanded the case for a new trial. Id. One of the plaintiffs' medical experts testified that asbestos exposure "was a substantial contributing factor" to a worker's interstitial fibrosis and asbestosis. Id. at 534. The expert also testified that smoking could have played a part in the development of the cancer. Id. Another plaintiff's expert contended that asbestos exposure was a causal factor in the cancer and that the two causes multiplied the risks. Id.

The trial judge instructed the jury that the law recognizes that there may be more than one cause of an injury, but "each may be a cause, so long as it can reasonably be said that, except for the asbestos exposure, the injury complained of would not have occurred." Id. The appeals court explained: The testimony of the plaintiffs' witnesses was to the effect that the asbestos exposure "played a substantial part in bringing about or actually causing the injury . . ." That the jury was not informed that this testimony could have been sufficient, to establish legal causation ... constitutes reversible error. In today's wide world of carcinogens, a conscientious jury would be hard pressed to state unequivocally that a person would not get cancer absent exposure to asbestos. Id. at 536 (citation omitted). (104) See, e.g., Martin v. Owens-Corning Fiberglas Corp., 528 A.2d 947, 948-51 (Pa. 1987). (105) 528 A.2d 947 (Pa. 1987). (106) Id. at 949. (107) Id. at 950. (108) Id. (109) Id. (110) Id. at 951 (Nix, C.J., dissenting). (111) Id. at 954 (Hutchinson, J., dissenting). (112) Dafler, 611 A.2d at 140. (113) Martin, 528 A.2d at 950. One expert testified:

I can't separate these two diseases for you in terms of percentage. My opinion is that both play a significant role in this man's disability, but I have no way of equating them or breaking them down. I can't tell you that asbestos contributed 48% and emphysema 52%, I don't know how to do that. There is no way I know of to separate these two diseases which are so closely intertwined.

I believe that both of them exist to a significant degree and they are both significant factors in this man's disability. That is the best I can do with that. Id. Another testified:

It is not possible for me to separate out the relative contribution of cigarette smoking and asbestos from the cause of his obstructive pulmonary disease and the cause of his total and permanent disability. Both factors are important in producing the effect and the pulmonary disability that he has.

... It is not possible to separate out what fraction of his lung disease is due to cigarette smoking, what factor, what fraction is due to asbestosis. Id. (citations omitted). (114) Keeton et al., supra note 5, [sections] 52. (115) Martin v. Johns-Manville Corp., 502 A.2d 1264 (Pa. Super. Ct. 1985), rev'd sub nom. Martin v. Owens-Corning Fiberglass Corp., 528 A.2d 947 (Pa. 1987). (116) Id. at 1270-71. (117) Id. at 1271. (118) Because different areas of the lungs were affected in Martin, that case arguably presents an even stronger case for apportionment than Dafler. Nevertheless, the risk contribution model does not depend on such facts. For decisions following Martin, see, e.g., Borman v. Raymark Indus., Inc., 960 F.2d 327 (3d Cir. 1992) (declining to find adequate evidence in the record to sustain a jury instruction on apportionment between smoking and asbestos exposure even though evidence on relative risks was offered); Taylor v. Celotex Corp., 574 A.2d 1084, 1098 (Pa. Super. Ct. 1990) (affirming trial court's decision not to grant apportionment instruction when neither party offered evidence about relative risks). (119) 418 N.W.2d 650 (Mich. 1988). (120) Id. at 657. (121) Id. at 651-52. (122) Id. at 651. (123) Id. (124) Id. at 653. (125) Id. at 652. (126) Id. at 654. (127) Id. at 655. (128) Id.; see also Champagne v. Raybestos-Manhattan, Inc., 562 A.2d 1100, 1122 (Conn. 1989) sustaining the jury's finding that plaintiff's smoking was 75% responsible under Comparative Responsibility Act for asbestos and smoking-related lung diseases). In Champagne, the Connecticut Supreme Court concluded that Were is no reason to prohibit use of one's smoking history in determining comparative responsibility in this case involving asbestosis." Champagne, 562 A.2d at 1118; see also Hao v. Owens-Illinois, Inc., 738 P.2d 416, 418-19 (Haw. 1987) (sustaining jury finding that shipyard worker's negligence in smoking was 51% responsible for causing his asbestos-related diseases and manufacturer's responsibility was 2%, with 47% applied to other defendants who were no longer parties in the case). (129) 418 N.W.2d 650 (Mich. 1988). (130) See, e.g., Tragarz v. Keene Corp., 980 F.2d 411 (7th Cir. 1992) (refusing to allow apportionment based on the contributions of the settling defendants and finding one remaining defendant to be a substantial factor, holding it liable for all plaintiff's damages). (131) Restatement (Second) of Torts [sections] 433A (1965). (132) Id. [sections] 433A cmt. a quoted supra note 50). (133) Nowhere are the difficulties of comparative negligence more noticeable than in attempts to deal with the issues of phantom parties and joint and several liability. A "phantom" party is a party not before the court, but who may be partially at fault in the occurrence that gave rise to the action. The party may be unknown, judgment-proof, or protected by governmental, family, or workers' compensation immunity. The problem with phantom parties arises when courts must decide whether to take into account the fault attributable to that nonparty in a trial in which that party is not joined. The answer to this problem can have a dramatic impact on the relative liabilities of the parties. Not surprisingly, states do not follow a consistent approach. The majority position seems to favor allowing the jury to allocate fault to nonparties. See, e.g., Brown v. Keill, 580 P.2d 867, 876 (Kan. 1978) (interpreting a comparative negligence statute that was silent on the issue of nonparty fault allocation and allowing the jury to allocate). Colorado and Arizona are examples of states that adopted statutes that expressly allow the defendant to argue for the allocation of fault to nonparties, provided that defendants give notice to the court prior to trial. Colo. Rev. Stat. [sections] 13-21-111.5(3) (1987) (notice must be given within 90 days of commencement of the action); Ariz. Rev. Stat. Ann. [sections] 12-2506B (Supp. 1994) (notice must be given prior to trial). However, the Uniform Comparative Fault Act, 12 U.L.A. 4 (West Supp. 1994), along with states that follow the Act, takes the opposite view. Id. [sections] 2(a), 12 U.L.A. 4. These states limit fault apportionment only to parties before the court, or to parties already released from liability. Id. (134) See, e.g., Ohio Rev. Code Ann. [sections] 2315.19(B)(4) (Baldwin 1990) (providing that, when contributory negligence is established as an affirmative defense, special interrogatories shall be answered specifying "the percentage of negligence ... that is attributable to each party to the action from whom the complainant seeks recovery"); Eberly v. A-P Controls, Inc., 572 N.E.2d 633, 638 (Ohio 1991) (holding that apportionment of negligence to absent tortfeasors is not allowed); see also Or. Rev. Stat. [sections] 18.480(1)(b) (1993) (providing that, under comparative negligence scheme, trier of fact may be requested to answer special interrogatories indicating "[t]he degree of each party's fault expressed as a percentage of the total fault attributable to all parties represented in the action"); Mills v. Brown, 735 P.2d 603, 605 (Or. 1987) ("The statutory scheme of comparative fault restricts the ... fact-finder[] to consideration only of the fault of the parties before the court at the time the case is submitted to the fact-finder ...."). (135) Some states do allow an assignment of fault to phantom parties-those absent, insolvent, or subject to an immunity. See Ariz. Rev. Stat. Ann. [sections] 12-2506B (Supp. 1994) ("In assessing percentages of fault the trier of fact shall consider the fault of all persons who contributed to the alleged injury,.... regardless of whether the person was, or could have been, named as a party to the suit."); Colo. Rev. Stat. [sections] 13-21-111.5(3)(a) (1987) ("[T]he finder of fact in a civil action may consider the degree or percentage of negligence or fault of a person not a party to the action, based upon evidence thereof, which shall be admissible, in determining the degree or percentage of negligence or fault of those persons who are parties to such action."). In some jurisdictions the comparative negligence or fault statute is silent as to the permissibility of allocating fault to absent parties, but courts interpret the statutes to authorize their inclusion. See, e.g., Connar v. West Shore Equip., 227 N.W.2d 660, 662 (Wis. 1975) (interpreting Wis. Stat. Ann. [sections] 895.045 (West Supp. 1994) to require that, "when apportioning negligence, a jury must have the opportunity to consider the negligence of all parties to the transaction, whether or not they be parties to the lawsuit"); Mills v. MMM Carpets, Inc., 1 Cal. Rptr. 2d 813, 814 (1992) (interpreting Cal. Civ. Code [sections] 1431.2 (West Supp. 1995) to require that apportionment of liability "must take into account the fault of all tortfeasors, whether or not they are named as defendants"); Fabre v. Marin, 623 So. 2d 1182, 1185:86 (Fla. 1993) (holding that, under Fla. Stat. Ann. [sections] 768.81(3) (West Supp. 1995), "party" means all entities regardless of whether named as defendant); Espaniola v. Cawdrey Mars Joint Venture, 707 P.2d 365, 371-73 (Haw. 1985) (holding that, under Haw. Rev. Stat. [sections] 663-31 (1985), fault of immune employer is admissible for determination of appropriate degrees of fault); Doyle v. Rhodes, 461 N.e.2d 382,385 (M. 1994) (holding that, under Ill. Ann. Stat. ch. 735, para. 5/2-1116 (Smith-Hurd 1992), workers' compensation does not protect employer from allocation of fault); Lines v. Ryan, 272 N.W.2d 896, 903 (Minn. 1978) (interpreting Minn. Stat. Ann. [sections] 604.01 (West 1988 & Supp. 1995) to require that "the apportionment must include all whose negligence may have contributed to the arising of the cause of action" (quoting Connar, 227 N.w.2d at 662)); Sullivan v. Scoular Grain Co., 853 P.2d 877, 880 (Utah 1993) (holding that failure to include immune employer in determination of fault is contrary to allocation requirement provided in Utah Code Ann. [subsections] 78-27-38 to 78-2740 (1992) (amended in 1994 to adopt Sullivan)). (136) Here, too, some courts have, under comparative fault systems, held that juries may assign fault to settling tortfeasors. See, e.g., Stacey v. Bangor Punta Corp., 108 F.R.D. 72, 74-77 (D. Me. 1985) (interpreting Me. Rev. Stat. Ann. tit. 14, [sections] 156 (West 1980) to allow nonsettling tortfeasor the option to have liability reduced by settling tortfeasor's proportionate fault); Bartels v. City of Williston, 276 N.W.2d 113, 118 (N.D. 1979) (holding that, under N.D. Cent. Code [sections] 9-10-07 (1987), "the percentage of causal negligence ... of the nonsettling defendant ... can only be determined by a proper allocation of all the causal negligence ... of all the joint tortfeasors" quoting Bielski v. Schulze, 114 N.W.2d 105, 111-12 (Wis. 1962))); Young v. Latta, 559 A.2d 465, 466 (N.J. Super. Ct. App. Div. 1989), aff'd, 589 A.2d 1020 (N.J. 1991) (holding that, under N.J. Stat. Ann. [sections] 2A:15-5.2 (West Supp. 1994), jury may allocate fault between settling and nonsettling tortfeasors, even in absence of claim for contribution); Penzell v. New York, 466 N.Y.S.2d 562, 568 (N.Y. Ct. Cl. 1983) (interpreting N.Y. Civ. Prac. L. & R. 4533-b (McKinney 1990) to require that any award to plaintiff be reduced by relative culpability of released tortfeasor). Some statutes explicitly provide that fault is to be assigned to settling tortfeasors. See, e.g., Alaska Stat. [sections] 09.17.080 (1994); Colo. Rev. Stat. [sections] 13-21-111.5 (1987); Mont. Code Ann. [sections] 27-1-703 (1993). (137) Set directly on top of the confusion over how to deal with the allocation of fault to nonparties is the dilemma of how to apply joint and several liability in a comparative fault system. Comparative fault was, most experts seem to believe, born to soothe the harshness of the contributory negligence rule. See Keeton et al, supra note 5, at 468-70; Gary T. Schwartz, Contributory and Comparative Negligence: A Reappraisal, 87 Yale L.J. 697 (1978). See also discussions in opinions overturning contributory negligence in favor of comparative negligence, such as Scott v. Rizzo, 634 P.2d 1234 (N.M. 1981); McIntyre v. Balentine, 833 S.W.2d 52 (Tenn. 1992). Nevertheless, the goal of comparative negligence was largely unarticulated, and as a result, questions remained as to whether it preempts joint and several liability as well as contributory negligence. Joint and several liability was, after all, the upside of contributory negligence. When contributory negligence was heartless to the plaintiff who could be shown to be even slightly at fault, joint and several liability rewarded the innocent plaintiff without having to worry about whether all of her tortfeasors were solvent. Whatever those who originally conceived the idea of comparative negligence intended, the result has been that joint and several liability survives in one form or another in most comparative fault jurisdictions. See discussion infra part VIII.A. In some states, joint and several liability survives only where the plaintiff is without fault See, e.g., Ga. Code Ann. [sections] 51-12-33 (Supp. 1994); Wash. Rev. Code Ann. [sections] 4.22.070(1)(b) (West Supp. 1995). California and New York retain joint and several liability for economic damages, but abolish the doctrine for noneconomic damages. See Cal. Civ. Code [sections] 1431.2(a) (West Supp. 1995); N.Y Civ. Prac. L. & R. 1601 (McKinney 1990). Still other states compare the fault of the defendant to the fault of the plaintiff and allow joint and several liability only when the plaintiff's fault is less than that of the defendant. See, e.g., Fla. Stat. Ann. [sections] 768.81(3) (West Supp. 1995); Nev. Rev. Stat. [sections] 41.141 (1993); Or. Rev. Stat. [sections] 18.485 (1993). And some simply abolish joint and several liability as to all the plaintiff's damages. See, e.g., Ariz. Rev. Stat. Ann. [sections] 12-2506 (1994). (138) N.J. Stat. Ann. [sections] 2A:15-1.1 to 15-5.3 (West Supp. 1994). (139) Id. [sections] 2A:15-5.3(d). (140) Id. [sections] 2A:15-5.3(f)(1). (141) Id. [sections] 2A:58C-1 to 58C-7. (142) Id. [sections] 2A-58C-1. (143) Id. [sections] 2A:58C-1(b)(4). (144) 603 A.2d 521 (N.J. Super. Ct. App. Div. 1992), aff'd, 620 A.2d 1047 (N.J. 1993). (145) Id. at 527. (146) Id. at 526-28. Similarly, in Tragarz v. Keene Corp., 980 F.2d 411 (7th Cir. 1992), the Seventh Circuit interpreted the Illinois exception to its elimination of joint and several liability, which provides that in any action in which the trier of fact determines that the injury or damage for which recovery is sought was caused by an act involving the discharge into the environment of any pollutant, including any waste, hazardous substance, irritant or contaminant, including, but not limited to smoke, vapor, soot, fumes, acids, alkalis, asbestos, toxic or corrosive chemicals,... any defendants found liable shall be jointly and severally liable for such damage. Ill. Ann. Stat. ch. 735, para. 5/2-1118 (Smith-Hurd 1992). The court held that the discharge of asbestos fibers into the internal environment of a building constituted "the discharge into the environment of any pollutant." Tragarz, 980 F.2d at 430. Other courts have reached similar conclusions. See, e.g., Sofie v. Flbreboard Corp., 771 P.2d 711, 727-28 (Wash. 1989), modified by 780 P.2d 260 (Wash. 1989) (as amended, holding that Wash. Rev. Code [sections] 4.22.070(3)(a) provides an exception for "hazardous waste or substances" encompassing asbestos litigation). (147) The definitions of who are joint tortfeasors are not always clear. See, e.g., Unif. Contribution Among Tortfeasors Act of 1939 [sections] 1, 12 U.L.A. 57 (1975) (defining joint tortfeasors as two or more persons jointly or severally liable in tort for the same injury to person or property, whether or not the plaintiff recovered judgment against all or some of them); N.M. Stat. Ann [sections] 41-3-1 (Michie 1978) (adopting above-cited definition of "joint tortfeasors"). Other jurisdictions statutorily have adopted a definition of "joint tortfeasor" substantially similar to that found in the Uniform Contribution Among Tortfeasors Act. See, e.g., Idaho Code [sections] 6-803(4) (1990) (stating that "joint tortfeasor" means a "person[] jointly or severally liable in tort for the same injury to person or property, whether or not judgment has been recovered against" other tortfeasors); cf. Adams v. Johnston, 860 P.2d 423, 429 n.5 (Wash. Ct. App. 1993) (noting that "[j]oint tortfeasors are those persons acting independently whose conduct concurred and caused an indivisible injury"). Keeton et al., supra note 5, at 322, points out the confusion that has existed in defining who are joint tortfeasors. Although originally confined to "vicarious liability for concerted action," id., tortfeasors are now "joint" when two or more actors owed a common duty to the plaintiff. Moreover, procedural rules governing the joinder of parties in the same action also complicated the meaning of "joint tortfeasor." Id. at 324-27. Keeton et al. states the modern rule that is most relevant for our purposes: "Where two or more causes combine to produce such a single result, incapable of any reasonable division, each may be a substantial factor in bringing about the loss, and if so, each is charged with all of it." Id. at 347. (148) See Arnst v. Estes, 8 A.2d 201, 203 (Me. 1939) (stating that the entire liability rests on the absence of any logical basis for apportionment). Where the damages are capable of apportionment, joint and several liability does not apply. Keeton et al., supra note 5, at 348-52. (149) 495 F.2d 213 (6th Cir.), cert. denied, 419 U.S. 997 (1974). (150) 248 S.W.2d 731 (Tex. 1952). (151) Michie, 495 F.2d at 213-14. (152) Id. at 215. (153) Id. (quoting and rejecting the First Restatement, which reads: "Where two or more persons, each acting independently, create or maintain a situation which is a tortious invasion of a landowner's interest in the use and enjoyment of land by interfering with his quiet, light, air or flowing water, each is liable only for such proportion of the harm caused to the land or of the loss of enjoyment of it by the owner as his contribution to the harm bears to the total harm." Restatement (First) of Torts [sections] 881 (1939)). (154) Id. at 215-16 (citing and rejecting Meier v. Holt, 80 N.W.2d 207 (Mich. 1957); De Witt v. Gerard, 275 N.W. 729 (Mich. 1937); De Witt v. Gerard, 264 N.W. 379 (Mich. 1936); Frye v. City of Detroit, 239 N.W. 886 (Mich. 1932)). (155) Id. (citing Watts v. Smith, 134 N.W.2d 194 (Mich. 1965) and Maddux v. Donaldson, 108 N.W.2d 33 (Mich. 1961)). (156) Id. at 218 (quoting 1 Fowler Harper & Fleming James, Jr., The Law of Torts [sections] 10.1, at 701-02 (1956)). Harper and James also make the statement respecting nuisance cases in which the harm, "while theoretically divisible, is single in the practical sense so far as the plaintiff's ability to apportion it among the wrongdoers is concerned (as where a stream is polluted as a result of refuse from several factories)." Harper & James, supra, at 701-02. (157) Michie, 495 F.2d at 218. On the basis of this rule the plaintiffs were able to satisfy the jurisdictional amount requirement to maintain diversity jurisdiction. Id. (158) 248 S.W.2d 731 (Tex. 1952). (159) Id. at 731-32. (160) Id. at 732. (161) Id. at 732-33. (162) In fact, Texas precedents had permitted joint and several liability in some cases involving simultaneous actions of two tortfeasors. See, e.g., Texas Power & Light Co. v. Stone, 84 S.W.2d 738, 740 (Tex. Civ. App. 1935). (163) Landers, 248 S.W.2d at 733 (citing, among other authorities, the Restatement (First) of Torts [sections] 881 (1939)). (164) Id. (165) Id. at 733-34. (166) Id. at 734. (167) Id. (168) The old rule that Michie and Landers displaced died hard in some jurisdictions. See, e.g., Somerset Villa, Inc. v. City of Lee's Summit, 436 S.W.2d 658 (Mo. 1969). Somerset Villa was a suit to recover damages from an apartment complex flooding, brought against the city for maintaining a nuisance and against a shopping center for negligence. Id. at 660. Both permitted weeds and debris to obstruct drainage culverts, thereby causing banks of a natural watercourse to overflow into the apartment complex. Id. The court held that the plaintiff could not hold the defendants jointly and severally liable, but each could only be liable for the portion of the damage each caused, with the burden of proving such portion resting on the plaintiff. Id. at 665. (169) Restatement (Second of Torts [sections] 433A (1965). (170) Id. [sections] 433A cmts. c-d; see also supra part 1. (171) See Restatement (Second) of Torts [sections] 433A cmt. h. It is not my intention to make a sweeping condemnation of joint and several liability. The academic literature contains an extensive debate on the doctrine's merits. See generally, e.g., John W. Wade, Should Joint and Several Liability of Multiple Tortfeasors Be Abolished?, 10 Am. J. Trial Advoc. 193 (1986); Richard W. Wright Allocating Liability Among Multiple Responsible Causes: A Principled Defense of Joint and Several Liability for Actual Harm and Risk Exposure, 21 U.C. Davis L. Rev. 1141 (1988) [hereinafter Wright, Allocating Liability]; Richard W. Wright, Throwing Out the Baby with the Bathwater. A Reply to Professor Twerski, 22 U.C. Davis L. Rev. 1147 (1989); Aaron D. Twerski, The Baby Swallowed the Bathwater. A Rejoinder to Professor Wright, 22 U.C. Davis L. Rev. 1161 (1989) [hereinafter Twerski, The Baby]; Aaron D. Twerski, The Joint Tortfeasor Legislative Revolt: A Rational Response to the Critics, 22 U.C. Davis L. Rev. 1125 (1989) [hereinafter Twerski Joint Tortfeasor]. (172) See Twerski, Joint Tortfeasor, supra note 171, at 1130-36. (173) A few courts have explicitly acknowledged this point in declining to retain joint and several liability after the adoption of comparative fault. For example, in Brown v. Keill, 580 P.2d 867 (Kan. 1978), the Kansas Supreme Court stated:

The perceived purpose in adopting [Kansas's comparative negligence statute] is fairly clear. The legislature intended to equate recovery and duty to pay to degree of fault. Of necessity, this involved a change of both the doctrine of contributory negligence and of joint and several liability. There is nothing inherently fair about a defendant who is 10% at fault paying 100% of the loss, and there is no social policy that should compel defendants to pay more than their fair share of the loss. Plaintiffs now take the parties as they find them. If one of the parties at fault happens to be a spouse or a governmental agency and if by reason of some competing social policy the plaintiff cannot receive payment for his injuries from the spouse or agency, there is no compelling social policy which requires the codefendant to pay more than his fair share of the loss.... Previously, when the plaintiff had to be totally without negligence to recover and the defendants had to be merely negligent to incur an obligation to pay, an argument could be made which justified putting the burden of seeking contribution on the defendants. Such an argument is no longer compelling because of the purpose and intent behind the adoption of the comparative negligence statute. Id. at 873-74. (174) Keeton et al., supra note 5, [section] 43. (175) In 1992, the Tennessee Supreme Court judicially adopted comparative fault, eliminated joint and several liability, and provided that because a particular defendant will henceforth be liable only for the percentage of a plaintiffs damages occasioned by that defendant's negligence, situations where a defendant has paid more than his "share" of a judgment will no longer arise, and therefore the Uniform Contribution Among Tort-feasors Act will no longer determine the apportionment of liability between codefendants. McIntyre v. Balentine, 833 S.W.2d 52, 58 (Tenn. 1992) (citation omitted). (176) See Sindell v. Abbott Lab., 607 P.2d 924, 937 (Cal.), cert. denied., 449 U.S. 912 1980), modified, Brown v. Superior Court of San Francisco, 751 P.2d 470, 487 (Cal. 1988) (holding that liability is several only, with each defendant liable only for its respective market share); Hymowitz v. Eli Lilly & Co., 539 N.E.2d 1069,1072-78 (N.Y.), cert. denied, 493 U.S. 944 (1989) (holding that liability of each defendant is several, not joint and several, with liability based on national market shares and each firm's contribution to the public-at-large's risk); Conley v. Boyle Drug Co., 570 So. 2d 275, 286 (Fla. 1990) (using only local market shares to apply several liability). But see Collins v. Eli Lilly & Co., 342 N.W.2d 37,49-53 (Wis.), cert. denied 469 U.S. 826 (1984) (allocating joint and several liability based on risk contribution). (177) Restatement (Second) of Torts (1979). (178) Id. (179) Restatement (Second) of Torts (1965 & 1979). (180) The House version of CERCLA would have based liability on causation, imposing liability on "person[s] who caused or contributed to the release or threatened release" of hazardous substance. See H.R. 7020, 96th Cong., 2d Sess. [section] 3701(a)(D) (1980), reprinted in 3 Environmental Law Inst., Superfund: A Legislative History 48 (Helen C. Needham ed. 1983) [hereinafter Superfund History]. The House passed H.R. 7020 in September 1980. H.R. 7020, 96th Cong., 2d Sess. (1980), 126 Cong. Rec. H9, 437-78 (daily ed. Sept. 23, 1980) In contrast, the Senate version of CERCLA contained no such causation provision and in stead based liability on certain responsible parties. See S. 1480, 96th Cong., 2d Sess. [section] 4(a) (1980), reprinted in Superfund History, supra, at 33-36.

Congress enacted the Senate version. The basic principles of liability are set forth CERCLA [section] 107(a):

Notwithstanding any other provision or rule of law, and subject only to the defenses set forth in subsection (b) of this section -

(1) owner and operator of a vessel or a facility,

(2) any person who at the time of disposal of any hazardous substance owned or operated any facility at which such hazardous substances were disposed of,

(3) any person who by contract agreement, or otherwise arranged for disposal or treatment, or arranged with a transporter for transport for disposal or treatment, of hazardous substance owned or possessed by such person, by any other party or entity, at any facility or incineration vessel owned or operated by another party or entity and containing such hazardous substances, and

(4) any person who accepts or accepted any hazardous substances for transport to disposal or treatment facilities, incineration vessels or sites selected by such person, from which there is a release, or a threatened release which causes the incurrence of response costs, of a hazardous substance, shall be liable for -

(A) all costs of removal or remedial action incurred by the United States Government or a State or an Indian tribe not inconsistent with the national contingency plan;

(B) any other necessary costs of response incurred by any other person consistent with the national contingency plan;

(C) damages for injury to, destruction of, or loss of natural resources, including the reasonable costs of assessing such injury, destruction, or loss resulting from such a release; and

(D) the costs of any health assessment or health effects study carried out under section 9604(i) of this title. 42 U.S.C. [section] 9607(a) (1988). (181) The courts uniformly interpret CERCLA as providing for strict liability, despite congressional silence on this critical point. See, eg., United States v. Monsanto, 858 F.2d 160, 171-73 (4th Cir. 1988), cert. denied, 490 U.S. 1106 (1989); New York v. Shore Realty Corp., 759 F.2d 1032, 1042 (2d Cir. 1985); United States v. Chem-Dyne Corp., 572 F. Supp. 802 (S.D. Ohio 1983). CERCLA's legislative history suggests that Congress intended to leave the task of developing the appropriate liability standard to the judiciary rather than requiring courts to impose a potentially inflexible or inequitable standard. Moreover, CERCLA's definition section states that liability under CERCLA "shall be construed to be the standard of liability which obtains under section 1321 of title 33 [Federal Water Pollution Control Act (FWPCA), 33 U.S.C. [sub-section] 1251-1387 (1988 & Supp. V 1993)]." 42 U.S.C. [section] 9601(32) (1988). Although the FWPCA is similarly silent on the liability standard, Congress was aware that courts had interpreted it to be a strict liability standard. See H.R. Rep. No. 253, 99th Cong., 2d Sess., pt. 1, at 74 (1986), reprinted in 1986 U.S.C.C.A.N. 2835, 2856. In enacting the Superfund Amendments and Reauthorization Act in 1986, Congress reaffirmed its intention that courts impose a strict liability standard on potentially responsible parties under CERCLA Id. (182) See, e.g., Keeton et al., supra note 5, [sub-section] 4142; Restatement (Second) of Torts [section] 431(a) (1979). (183) See, eg., United States v. Alcan Aluminum Corp., 964 F.2d 252, 264 (3d Cir. 1992) ("[CERCLA] does not, on its face, require the plaintiff to prove that the generator's hazardous substances themselves caused the release ... [only] that the release or threatened release caused the incurrence of response costs, and that the defendant is a generator of hazardous substances at the facility."); Dedham Water Co. v. Cumberland Dairy Farms, Inc., 889 F.2d 1146, 1152 (1st Cir. 1989) ("A literal reading of [CERCLA] imposes liability if releases or threatened releases from defendant's facility cause the plaintiff to incur response costs; it does not say that liability is imposed only if the defendant causes actual contamination of the plaintiff's property."); Monsanto, 858 F.2d at 168 ("The traditional elements, of tort culpability on which the site-owners rely simply are absent from the statute. The plain language of section 107(a)(2) extends liability to owners of waste facilities regardless of their degree of participation in the subsequent disposal of hazardous waste.").

For a thorough analysis of the causation issues in CERCLA litigation and relevant policy, see generally John C. Nagle, CERCLA, Causation, and Responsibility, 78 Minn. L Rev. 1493 (1994). (184) See, e.g., City of New York v. Exxon, 766 F. Supp. 177, 198 (S.D.N.Y. 1991); United States v. Bliss, 667 F. Supp. 1298, 1309-10 (E.D. Mo. 1987). (185) See, eg., United States v. South Carolina Recycling & Disposal, Inc., 653 F. Supp. 984, 990, 993 n.6 (D.S.C. 1984), aff'd in part, vacated in pail, United States v. Monsanto, 858 F.2d 160 (4th Cir. 1988), cert. denied, 490 U.S. 1106 (1989) (noting that, at a storage facility containing 7,200 55-gallon drums of different hazardous substances, it "would have cost in the range of $2.5 million to attempt through analytical means to identify all waste types in the conglomerate of materials stored at the [site], approximately five times the cost of surface removal itself"); Violet v. Picillo, 648 F. Supp. 1283, 1286 (D.R.I. 1986) (involving 10,000 barrels and containers). (186) See, eg., United States v. Chem-Dyne Corp., 572 F. Supp. 802, 811 (S.D. Ohio 1983) (involving 289 generators and transporters); B.F. Goodrich Co. v. Murtha, 958 F.2d 1192, 1196 (2d Cir. 1992) (involving two hundred parties); Alcan Aluminum Corp., 990 F.2d at 717 (in which 83 parties, including Alcan Aluminum, were involved with the waste disposal center). (187) See, e.g., United States v. Rohm & Haas Co., 2 F.3d 1265, 1268 (3d Cir. 1993); Idaho v. Hanna Mining Co., 882 F.2d 392, 393 (9th Cir. 1989). (188) CERCLA [section] 107(a)(4)(A) permits the recovery of "all costs of removal or remedial action incurred by the United States Government ... not inconsistent with the national contingency plan." 42 U.S.C. [section] 9607(a)(4)(A) (1988). For government actions, "all costs of removal or remediation action" afford compensation to EPA for virtually all expenses in any way related to its efforts to secure cleanup of an area. See id. [section] 9607(a) (4)(B) (extending liability to "any other necessary costs of response"). A clue to the meaning of "response costs" as used in CERCLA [section] 107(a)(4)(B) comes from CERCLA's definition of the terms "respond" and "response." CERCLA [section] 101(25) states: "The terms `respond' or `response' means remove, removal, remedy, and remedial action; all such terms (including the terms `removal' and `remedial action') include enforcement activities related thereto." Id. at [section] 9601(25) (footnotes omitted). In short, "response costs" means, at a minimum, all of the costs relating to the cleanup of a site. (189) 858 F.2d 160 (4th Cir. 1988), cert. denied, 490 U.S. 1106 (1989). (190) Id. at 170. (191) As CERCLA [section] 107(b) provides,

(b) Defenses. There shall be no liability under subsection (a) of this section for a person otherwise liable who can establish by a preponderance of the evidence that the release or threat of release of a hazardous substance and the damages resulting therefrom were caused solely by -

(1) an act of God;

(2) an act of war;

(3) an act or omission of a third party other than an employee or agent of the defendant, or than one whose act or omission occurs in connection with a contractual relationship, existing directly or indirectly, with the defendant ... if the defendant establishes by a preponderance of the evidence that (a) he exercised due care with respect to the hazardous substance concerned, taking into consideration the characteristics of such hazardous substance, in light of all relevant facts and circumstances, and (b) he took precautions against foreseeable acts or omissions of any such third party and the consequences that could foreseeably result from such acts or omissions; or

(4) any combination of the foregoing paragraphs. 42 U.S.C. [section] 9607(b) (1988).

The Monsanto court considered these defenses as evidence that Congress "allocated the burden of disproving causation to the defendant who profited from the generation and inexpensive disposal of hazardous waste." 858 F.2d at 170; see also United States v. Alcan Aluminum Corp., 964 F.2d 252, 265 (3d Cir. 1992) (asserting that "[i]mputing a specific causation requirement would render these defenses superfluous"); New York v. Shore Realty Corp., 759 F.2d 1032, 1044 (2d Cir. 1985) (stating that "a causation requirement makes superfluous the affirmative defenses provided in section 9607(b)"); Violet v. Picillo, 648 F. Supp. 1283, 1292 (D.R.I. 1986) concluding that an interpretation of CERCLA requiring a causation element "would reduce the CERCLA defenses to statutory surplusage"). (192) See, e.g., United States v. Wade, 577 F. Supp. 1326, 1337-38 (E.D. Pa. 1983); United States v. Chem-Dyne Corp., 572 F. Supp. 802, 807-08 (S.D. Ohio 1983). United States v. Ottati & Goss, Inc., 630 F. Supp. 1361 (D.N.H. 1988), was a cost recovery action against operators and former operators of drum reconditioning businesses, property owners, and generators of waste contained in the drums that were sent to the site for reconditioning. The evidence showed that chemical substances leaked or spilled from drums and were mixed together. Although the generators satisfied their burden of proving approximately how many drums each brought to the site, the court nevertheless imposed joint and several liability, because "the exact amount or quantity of deleterious chemicals or other noxious matter [could not] be pinpointed as to each defendant" and "[t]he resulting proportionate harm to surface and groundwater [could not] be proportioned with any degree of accuracy as to any individual defendant." Id. at 1396.

A similar situation existed in O'Neil v. Picillo, 883 F.2d 176 (1st Cir. 1989). The site at issue there was a Rhode Island pig farm that had been used as a waste disposal site. The site was described as having "massive trenches and pits `filled with free-flowing, multicolored, pungent liquid wastes' and thousands of `dented and corroded drums containing a veritable potpourri of toxic fluids." Id. at 177. The defendants argued that it was possible to apportion the removal costs, because there was evidence of the total number of barrels excavated during each phase of the cleanup, the number of barrels in each phase attributable to them, and the cost of each phase. Id. at 181. There was testimony that, of the approximately 10,000 barrels excavated, only 300 to 400 could be attributable to a particular defendant. Id. at 182. The court concluded that because most of the waste could not be identified, and the defendants had the burden of accounting for the uncertainty, the imposition of joint and several liability was appropriate. The court noted that, even if there had been evidence of the number of barrels attributable to each defendant, more would be required to demonstrate that the removal costs were capable of apportionment, because the cost of removing barrels varied depending upon their contents. Furthermore, the costs of removing contaminated soil, in which the waste had commingled, "would necessarily be arbitrary". Id. at 183 n. 11. (193) Mosanto, 858 F.2d at 171-72. (194) Id. at 172-73 (footnotes omitted). (195) 42 U.S.C. [section] 9613(f) (1988). (196) 858 F.2d at 173 (quoting United States v. South Carolina Recycling & Disposal, Inc., 653 F. Supp. 984, 995 & n.8 (D.S.C. 1984)). (197) Some states' mini-CERCLA laws explicitly address this issue by enacting, in effect, section 433A of the Restatement (Second) of Torts. For example, the Michigan Environmental Response Act provides that

(1) If 2 or more persons acting independently cause a release or threat of release that results in response activity costs, or damages for injury to, destruction of, or loss of natural resources, and there is a reasonable basis for division of harm according to contribution of each person, each person is subject to liability ... only for the portion of the total harm that the person caused. However, a person seeking to limit its liability on the ground that the entire harm is capable of division shall have the burden of proof as to the divisibility of the harm and as to the apportionment of liability.

(2) If 2 or more persons cause or contribute to an indivisible harm that results in response activity costs, or damages for injury to, destruction of, or loss of natural resources, each person is subject to liability ... for the entire harm. Public Act 71, [section] 20129(1) (effective June 5, 1995 and codified at Mich. Comp. Laws Ann. [section] 299.612c); see also Elizabeth F. Mason, Contribution, Contribution Protection, and Non-settlor Liability Under CERCLA: Following Laskin's Lead, 19 B.C. Envtl. Am. L. Rev. 73, 83-85 (1991). (198) See, eg., In re Bell Petroleum Servs., Inc., 3 F.3d 889, 896-97 (5th Cir. 1993); United States v. Rohm & Haas Co., 2 F.3d 1265, 1279-80 (3d Cir. 1993); United States v. Alcan Aluminum Corp., 964 F.2d 252, 267 (3d Cir. 1992); Monsanto, 858 F.2d at 173 (noting that the court "shar[ed] the appellants' concern that they not be ultimately responsible for reimbursing more than their just portion of the governments' response costs"); O'Neil v. Picillo, 682 F. Supp. 706, 725 (D.R.I. 1988), aff'd, 883 F.2d 176, 179 (1st Cir. 1989) (citing concerns about burdens on defendants, but refusing to apportion anyway); United States v. Stringfellow, 661 F. Supp. 1053, 1060 (C.D. Cal. 1987) citing concerns about fairness of joint and several liability, but not at the fault stage). (199) 990 F.2d 711 (2d Cir. 1993). Alcan-New York involved a 15-acre waste disposal and treatment center in Oswego County, New York, operated in the 1970s by Pollution Abatement Services (PAS). In 1977, EPA and the State of New York undertook response measures that amounted to over $12 million over the ensuing 10 years. Alcan's wastes consisted of 4.6 million gallons of oil emulsion, containing mostly water and mineral oil, along with small quantities of aluminum ingot shavings containing heavy metals.

The district court granted summary judgment for the Government on joint and several liability, rejecting Alcan's arguments concerning its minimal Contribution of hazardous substances and lack of causation. The court also held that Alcan failed to meet its burden to show that the harm at PAS was divisible, and awarded the Government approximately $4 million in accumulated response costs. The Government had entered into a consent decree with 82 of the 83 defendants, recovering 74% or $9.1 million of the cleanup costs it incurred. (200) 964 F.2d 252 (3d Cir. 1992). Alcan-Butler involved harm to the Susquehanna River in Pennsylvania, at the site of the Butler Tunnel, a network of underground mine-sand tunnels bordering the river. In the late 1970s, companies disposed of various liquid wastes through an open borehole, directly into the mine workings at the site. A hurricane in 1985 flushed approximately 100,000 gallons of water contaminated with hazardous substances from the site into the Susquehanna River, necessitating EPA's response costs.

In EPA's cost recovery action against 20 defendants, the Government settled with all except Alcan, leaving Alcan with liability for approximately 36% of the Government's response costs, even though Alcan's volumetric contribution was less than 5%. Alcan argued first that its waste oil emulsion did not constitute a "hazardous substance" as defined by CERCLA due to its below-ambient levels of heavy metals, and second that its waste neither caused nor contributed to the release, nor necessitated response costs incurred by the Government. The district court disagreed and granted the Government's motion for summary judgment, holding that Alcan was jointly and severally liable for the response costs, because Alcan's waste contained identifiable levels of hazardous substances and was present at the site from which there was a release. (201) Alcan-Butler, 964 F.2d at 256. (202) Id. (203) Alcan-New York, 990 F.2d at 721; Alcan-Butler, 964 F.2d at 266. (206) Id. (207) Id. (208) id. (209) Id. (210) Id. (211) Id. (212) Id. (213) Alcan-Butler, 964 F.2d at 270. (214) Id. at 269 n.27. (215) Id. at 270 n.29. (216) Id. (217) Alcan-New York, 990 F.2d at 723. (218) Id. at 722. The Alcan-Butler court also stated: We observe in this regard that Alcan's burden in attempting to prove the divisibility of harm to the Susquehanna River is substantial, and the analysis will be factually complex as it will require an assessment of the relative toxicity, migratory potential and synergistic capacity of the hazardous waste at issue. Alcan-Butler, 964 F.2d at 269. (219) Alcan-New York, 990 F.2d at 722; Alcan-Butler, 964 F.2d at 269. (220) Alcan-New York, 990 F.2d at 722. (221) Alcan-Butler, 964 F.2d at 269. (222) In United States v. Chem-Dyne Corp., the court described the nature of the "fairly complex factual determination" involved in deciding whether the defendants were jointly and severally liable as follows: The Chem-Dyne facility contains a variety of hazardous waste from 289 generators or transporters, consisting of about 608,000 pounds of materials. Some of the wastes have commingled but the identities of the sources of these wastes remain unascertained. The fact of the mixing of the wastes raises an issue as to the divisibility of the harm. Further, a dispute exists over which of the wastes have contaminated the ground water, the degree of their migration and concomitant health hazard. Finally, the volume of waste of a particular generator is not an accurate predictor of the risk associated with the waste because the toxicity or migratory potential of a particular hazardous substance generally varies independently with the volume of the waste. 572 F. Supp. 802, 811 (S.D. Ohio 1983). The court concluded that the defendants had not met their burden of demonstrating the divisibility of the harm and the degree to which each was responsible, and it denied their summary judgment motion. Id. (223) 3 F.3d 889 (5th Cir. 1993). At a site in Odessa, Texas, discolored drinking water led to an investigation focusing on a chrome-plating shop operated from 1971 through 1977 successively by three entities: John Leigh, Bell Petroleum Services, Inc., and Sequa Corporation. Id. at 892-93. Some operator or operators pumped rinse water from the finished parts out of the building onto the ground during this time period, eventually leading EPA to designate a 24-block area as a Superfund site. Id. After EPA filed a cost recovery acdtion against the defendants, the district court heard arguments in three phases: Phare I considered liability, Phase II addressed the recoverablity of EPA's response costs, and Phase III evaluated response cost responsibility. After the Phase III hearing, the district court concluded there was no basis for dividing the liability among the defendants that was not speculative. Id. Therefore, the district court imposed joint and several liability for the replacement of private water supply systems and all future costs incurred by EPA in developing permanent remedial actions. Id. at 894. (224) Id. at 904 n.19. (225) Id. at 903. (226) Id. at 903-04. (227) Id. at 892-93, 903-04. (228) Id. at 892-93. (229) Id. at 897. (230) Id. at 895. (231) See United States v. Chem-Dyne Corp., 572 F. Supp. 802 (S.D. Ohio 1983). (232) Bell Petroleum, 3 F.3d at 901. (233) Id. (234) Id. at 904 n. 19. (235) Id. at 903. (236) Id. (237) Id. (238) Chief Judge Robert M. Parker of the Eastern District of Texas, sitting by designation, dissented in part. Although he agreed that determining whether the court can quantitatively apportion the harm in a particular case is a question of law and that "the single chromium harm suffered by the Trinity Aquifer is the sort theoretically capable of apportionment," id. at 909 (Parker, J., concurring in part, dissenting in part), Judge Parker believed Sequa had failed to meet its factual burden relative to apportionment, proving by a preponderance of the evidence the extent of environmental injury that could be attributed to it. Id. By ruling in Sequa's favor, the dissent said, the majority essentially abandoned the requirement of proof by a preponderance of the evidence in a CERCLA apportionment case and thereby "eviscerat[ed] the very concept of joint and several liability." Id. (239) Id. at 904 n. 19. (240) Id. (241) See, e.g., United States v. Chrysler Corp., Nos. 88-341 CMW, 88-534CMW (D. Del. Dec. 9, 1990). In Chrysler, the district court vacated an earlier ruling granting the Government's request for joint and several liability, and it held that Chrysler was entitled to a full hearing on apportionment: The Court believes that, in some circumstances, evidence of volumetric contributions of waste, or evidence that the usages of a site by two parties were temporally distinct, could potentially provide a reasonable basis for apportionment of liability, and Chrysler has raised a genuine issue of material fact on this issue. Chrysler offered evidence that identified its maximum contribution of paint sludge to the site at one truckload per week, whereas General Motors' contribution of the paint sludge to the site was approximately 1600 gallons per day. Chrysler also contended that the evidence shows that General Motors' wastes were taken to the site from 1962 until at least 1968, and possibly until 1971 whereas its sludge was not delivered after 1963. Accordingly, Chrysler argued, these records provide a basis for the apportionment of liability between Chrysler and General Motors by establishing that the harm to the site is "severable in point of time." Id.; see also Kamb v. United States Coast Guard, 869 F. Supp. 793 (N.D. Cal. 1994). In Kamb, the court found "there [was] a reasonable basis for apportioning [the defendants'] CERCLA liability based on the volume of lead each [defendant] contributed to the Site and based on the divisibility of the Site into two discrete sections: a trap/skeet range, not used by the defendants, and a firing range." Kamb, 869 F. Supp. at 799. As a result, the court found that imposition of joint and several liability was not apportioning and that, as the parties agreed, "the proper framework for addressing this issue is set forth in In re Bell Petroleum Servs., Inc." Id. (citation omitted); see also Hatco Corp. v. W.R. Grace Co., 801 F. Supp 1309, 1330 31 (D.N.J. 1992) (holding that the issue of divisibility of CERCIA liability precluded summary judgment). (242) One recent federal appeals court decision raises some doubt about this conclusion. In a post-Alcan decision, United States v. Rohm & Haas Co., 2 F.3d 1265 (3d Cir. 1993), the Third Circuit considered an appeal from the district court's imposition of joint and several liability against one of four defendant generators, Chemical Properties (CP). After a bench trial, the district court in Rohm & Haas imposed jointly and several liability against all four of the defendants. Id. at 1268-69. Only CP appealed from this judgment. Id. at 1269. On appeal, it argued that it should not be held jointly and severally liable because the damages were capable of apportionment. Id. at 1280. In support of this argument, CP asserted: 1) it owned only 10% of the facility's land area, and 2) other defendants disposed of most, if not all, of the waste. Id. The Third Circuit rejected both arguments. In rejecting the first argument, the court stated that "[i]n order to warrant apportionment, a defendant cannot simply provide some basis on which damages may be divided up, but rather it must show that there is a `reasonable basis for determining the contribution of each cause to a single harm.'" Id. Further, according to the court, "CP must prove that there is a way to determine what portion of the `harm' (i.e., the hazardous substances present at the facility and the response costs incurred in dealing with them) is fairly attributable to CP . . ." Id. With regard to CP's second argument in support of apportionment, which, was based upon some volumetric evidence, the Third Circuit stated that this argument "comes closer to warranting apportionment." Id. The court reasoned that [i]ndeed, if CP were able to prove that none of the hazardous substances found at the site were fairly attributable to it, we might well conclude that apportionment was appropriate and CP's apportioned share would be zero. However, it is clear from the factual findings of the district court that CP did not prove that none of the harm was attributable to it. Id. Therefore, CP was not entitled to apportionment. The Rohm & Haas case suggests that some courts consider apportionment appropriate only when a defendant can prove that none of the waste is its fault, a ruling that essentially defeats apportionment's objective. (243) For an excellent analysis of viable means for allocating response costs in CERCLA litigation, see Ridgway M. Hall, Jr. et al., Superfund Response Cost Allocations: The Law, the Science and the Practice, 49 Bus. Law. 1489 (1994). (244) Courts overwhelmingly refuse to apply equitable factors in the apportionment versus the joint and several liability phase of the litigation. See In re Bell Petroleum Servs., 3 F.3d 889, 901-02 (5th Cir. 1993); United States v. Monsanto, 858 F.2d 160, 171 n.22 (4th Cir. 1988) (apparently agreeing with the district court's rejection of the use of equitable factors and stating that, while equitable factors are relevant in an action for contribution, "[t]hey are not pertinent to the question of joint and several liability, which focuses principally on the divisibility among responsible parties of the harm to the environment"). Other courts similarly conclude that equitable factors, such as those listed in the Gore amendment (see discussion below), have, no place in making the decision whether to impose joint and several liability, but are appropriate in an action for contribution among jointly and severally liable defendants. See United States v. Alcan Aluminum Corp. (Alcan-Butler), 964 F.2d 252, 270 n.29 (3d Cir. 1992) (noting that "the contribution proceeding is an equitable one in which a court is permitted to allocate response costs based on factors it deems appropriate, whereas the court is not vested with such discretion in the divisibility determination"); United States v. Western Processing Co., 734 F. Supp. 930, 938 (W.D. Wash. 1990) ("Defendants may ... bring contribution actions for ultimate allocation of damages among the responsible parties where it is entirely appropriate to utilize the Gore Factors to determine the burden each party must bear."); United States v. Stringfellow, 661 F. Supp. 1053, 1060 (C.D. Cal. 1987) ("The Court's discretion in apportioning damages among the defendants during the contribution phase does not effect [sic] the defendants' liability."). Equitable factors are relevant to the contribution phase of CERCLA litigation, which authorizes courts to apply "such equitable factors as the court determines are appropriate" in allocating response costs among PRPs. 42 U.S.C. [subsection] 9613(f)(1) (1988). Among the factors usually relied upon are the Gore factors, named for then-representative Albert Gore, Jr. D-Tenn.), which were not included in the final version of the law. The Gore factors include (1) The ability of the parties to demonstrate that their contribution to a discharge, release or disposal of a hazardous waste can be distinguished; (2) The amount of the hazardous waste involved; (3) The degree of toxicity of the hazardous waste involved; (4) The degree of involvement by the parties in the generation, transportation, treatment, storage, or disposal of the hazardous waste; (5) The degree of care exercised by the parties with respect to the hazardous waste concerned, taking into account the characteristics of such hazardous waste; and (6) The degree of cooperation by the parties with Federal, State or local officials to prevent any harm to the public health or the environment. Western Processing, 734 F. Supp. at 934-35. The only court that has suggested applying these factors to the divisibility of harm issue is United States v. A & F Materials Co., 578 F. Supp. 1249, 1256 (S.D. Ill. M. 1984). (245) See discussion infra pan IX.B. (246) Epstein, supra note 79, at 1385-86. Professor Barbara Ann White describes the problem eloquently: Thus, to ensure that society as a whole pays as little as possible for cleanup costs, the EPA argues for, and the courts impose, joint and several liability. Therein lies the dilemma. On the one hand, Congress has decided that industry and not society should pay for the cost to clean up the toxic wastes. On the other hand, the methodology for allocating these costs is such that any firm who has been connected with any amount of disposal of hazardous wastes is, in effect, holding a lottery ticket that potentially represents millions of dollars of liabilities to the firm for the cleanup of wastes unrelated to its own activities. The question is whether there is another way of assigning cleanup costs that preserves the basic decision to have industry bear the costs and yet does not unduly burden any individual firm with the lion's share of cleaning up particular wastes not including its own. Barbara Ann White, Economizing on the Sins of Our Past. Cleaning Up Our Hazardous Wastes, 25 Hous. L. Rev. 899, 912-13 (1998) (footnotes ommitted). Later she states: Allocating the costs retroactively to individual firms on the basis of joint and several liability cannot be justified on the grounds of efficiency gains. Furthermore, negative impacts on efficiency result from proceeding along the lines of CERCLA in which retroactive liability is applied after a change in standards. Incentives for inefficient and undesirable behavior are created by an atmosphere of not being able to trust the rules. In particular, unpredictability as to the boundaries of liability has made it extraordinarily difficult for insurance companies to insure for toxic wastes liabilities and, in many instances, insurance has become unavailable. Alternatively, faced with the uncertainty whether proper disposal today may be viewed as improper in the future, firms may consider it more pragmatic to risk the sanctions from illegal dumping which are difficult to trace rather than risk facing liabilities with unanticipated penalties (such as joint and several liability) from disposing of the wastes legally. Id. at 920 (footnote omitted). (247) It is possible that potential defendants are indifferent ex ante between a rule which leads to uncertain but potentially large liability and one that guarantees liability that is more modest because an apportionment formula spreads it among all defendants. Considering the practical difficulties of trying to anticipate liability, the differences in incentive effects that exist between apportionment versus joint and several liability may disappear-hazardous waste defendants will probably have little information available on the number of other generators, the relative toxicity of wastes, the ability of other generators to safely dispose of their wastes, and the total potential costs of cleaning a site. Recall that Congress adopted a version of CERCLA that eliminated a causation requirement for the CERCLA plaintiff in order to minimize the risk that the plaintiff would not recover cleanup costs because of the difficulties in establishing a causal nexus between the harm at the site and a particular defendant's waste activities. See supra note 180. Applying joint and several liability for all defendants, unlike a requirement of proving causation, does not shield those who contribute to a waste site simply because the government is unable to leap the causation hurdle and positively link them to the harm done at the site. See infra part VII. (248) Bell petroleum, 3 F.3d at 897. (249) Id. (250) 964 F.2d 252, 270 n.29 (3d Cir. 1992). (251) 990 F.2d 711, 722 (2d Cir. 1993). (252) See Kamrin, supra note 9, at 56-57, James Huff et al., Scientific Concepts: Value and Significance of Chemical Carcinogenesis Studies, 31 Ann. Rev. of Pharmacology & Toxicology 621, 626-28 (1991). (253) Recent theories on the etiology of cancer focus on the cell's genetic machinery. Troyen A. Brennan & Robert F. Carter, Legal and Scientific Probability of Causation of Cancer and Other Environmental Disease in Individuals, 10 J. Health Pol., Pol'y & L. 33, 40 (1985). "Given the Hypothesis that carcinogens affect DNA [deoxyribonucleic acid], attention has long focused on the correlation of carcinogenesis with mutagenesis." Id. Recently developed testing procedures enable investigators to demonstrate that many known carcinogens are also mutagens. Id. at 41. This, in turn, results in treating mutageriic agents as surrogates for carcinogenic ones. Id. at 41 n.92 (citing International Agency For Research on Cancer, Long and Short Term Screening Assays For Carcinogens: A Critical Approach 75-80 (1986)). Short-term assays take advantage of current understanding of the genetic changes associated with mutation. (254) For example, in vitro testing for teratogens involves transplanting animal fetal cells, organs, or embryos into a medium in which the chemical under investigation is present in order to examine the effects on the transplanted tissues. See Felix Beck, Model Systems in Teratology Research, in Developmental Toxicology 11, 24-27 (K. Snell ed., 1982). (255) See Kamrin, supra note 9, at 56 ("It is not clear exactly how to apply these [study] results, except in the qualitative sense that they indicate that a particular chemical has the potential to cause genetic damage. They are certainly not useful in establishing the dose at which such effects might occur in humans and cannot be aplied to long-term exposures."). (256) Indeed, one of the purposes of the in vitro and short term studies is to help nominate those chemicals that might reasonably be carcinogenic and select and subject only them to the more expensive and time-consuming animal studies. Huff et al., supra note 252, at 622. (257) Elements of Toxicology, supra note 9, at 15-16. For an explanation of acute toxicity testing, see Kamrin, supra note 9, at 45-48. (258) Those chemical agents exhibiting lower LD50 values are more acutely toxic than those with higher LD50 values. For example, the LD50 (mg/kg of bodyweight) is 29,700 for sucrose, 1000 for aspirin; and 1 for nicotine. Kamrin, supra note 9, at 46. (259) Elements of Toxicology, supra note 9, at 16-17; see also Kamrin, supra note 9, at 48-50 (recommending assessment of effects over period of 10% of animal's life). (260) Elements of Toxicology, supra note 9, at 17-18. (261) Kamrin, supra note 9, at 23-26. (262) Rats and mice are toxicologists' favorite species because of cost and handling considerations and the greater understanding of their genetic background and disease susceptibility. Moreover, because their lifespans are short, toxicologists can perform lifetime studies within a reasonable period. The National Toxicology Program, which specifies protocols for carcinogenicity bioassay studies, recommends using rats and mice, specifies the number to be tested (50 males, 50 females), the size of doses, the regimen for dosing (begin exposure at 6 weeks, end at 24 months), and the organs and tissues to be examined. See Elements of Toxicology, supra note 9, at 18-19; see also Robert Percival et al., Environmental Regulation: Law, Science, and Policy 481-83 (1992). (263) The lowest level at which adverse effects exist is known as the lowest observed effect level (LOEL) and the level at which no such effects are observed is the Noel. See Elements of Toxicology, supra note 9, at 20-21. In order for a toxicity experiment to be of significant value, at least one dose used must be of sufficient magnitude to cause some type of adverse effect within the duration of the experiment. Id. at 20. The National Toxicology Program recommends using a maximum tolerated dose (MTD), or some fraction of the MTD, defined as the maximum dose administered for a major portion of a laboratory animal's lifetime that can be tolerated without experiencing significant impairment of growth or without suffering observable toxic effects (other than carcinogenicity). Id. at 21; Kamrin, supra note 9, at 51; Huff et al., supra note 252, at 628. (264) See generally Howard Latin, Good Science, Bad Regulation and Toxic Risk Assessment, 5 Yale J. on Reg. 89 (1988) (analyzing EPA's toxic risk assessment guidelines); Percival et al, supra note 262, at 497-512 (discussing the uncertainty of conclusions based on animal bioassays and the problems created for regulation). (265) See Interdisciplinary Panel on Carcinogenicity, Criteria for Evidence of Chemical Carcinogenicity, 225 Sci. 682, 683 (1984); Brennan & Carter, supra note 253, at 42; see also 1 International Agency for Research of Cancer 1972-1990 IARC Monographs on the Evaluation of Carcinogenic Risk To Humans ("In the absence of adequate data on humans, it is biologically plausible and prudent to regard agents and mixtures ... for which there is sufficient evidence of carcinogenicity in experimental animals as if they presented a carcinogenic risk to humans."), quoted in Huff et al., supra note 252, at 649. (266) See Alon Rosenthal et al., Legislating Acceptable Cancer Risk from Exposure to Toxic Chemicals, 19 Ecology L.Q. 269, 280-82 (1992) (discussing the assumptions); see also Bruce N. Ames, Environment Pollution Natural Carcinogens and the Causes of Human Cancer: Six Errors, in Important Advances in Oncology 1989 (V. DeVita et al. eds., 1990); Lois S. Gold et al., Rodent Carcinogens: Setting Priorities, 258 Sci. 261, 264 (1992). (267) See Chemical Carcinogens: A Review of Science and Its Associated Principles, 50 Fed. Reg. 10,371, 10,438 (Mar. 14, 1985); Rosenthal, supra note 266, at 288; Latin, supra note 264, at 103-05; Kamrin, supra note 9, at 52-53. Critical to the risk assessment process is the selection of an extrapolative model that takes data from bioassays and predicts the level of risk in humans. Risk assessments must extrapolate from the results of high-dosage animal tests to low-dose animal risks, and then to long-term human risks at significantly lower doses. For general discussions of competing extrapolative models, see Identification, Classification, and Regulation of Potential Occupational Carcinogens, 45 Fed. Reg. 5001, 5184-88 (Jan. 22, 1980); Elements of Toxicology, supra note 9, at 37-41; Rosenthal, supra note 266, at 287-89. (268) There are a number of reasons for questioning the reliability of animal data to establish causation in humans. The major reason is that species-related differences exist in the metabolism and disposition of chemicals generally. See Rosenthal, supra note 266, at 284-85 citing example in which male kidney tumors may result from a biological mechanism unique to male rats); Brennan & Carter, supra note 253, at 46; Latin, supra note 264, at 103-04. Differences in the balance between bioactivation and detoxification mechanisms and in the dose of the chemical actually delivered to the target organ represent two common problems concerning the extrapolation of animal responses as a means of predicting the outcome of similar human exposures. Moreover, known species' differences in biological processes important in chemical carcinogenesis exist, including differences in the rate of DNA repair, the background incidence of tumors, and the anatomy and biochemical functioning of affected organs or systems. See, eg., Bruce N. Ames & Lois S. Gold, Too Many Rodent Carcinogens: Mitogenests Increases Mutagenesis, 249 Sci. 970 (1990); C. Jelleff Carr & Albert C. Kolbye, A Critique of the Use of the Maximum Tolerated Dose in Bioassays to Assess Cancer Risks from Chemicals, 14 REG. Toxicology & Pharmacology 78 (1991). (269) See Samuel M. Cohen & Leon B. Ellwein, Cell Proliferation in Carcinogens: Mitogenesis Increases Metagenesis, 249 Sci. 1007, 1007 (1990) ("During the past two decades, emphasis has been shifting from the use of animal models primarily for the study of carcinogenic mechanism to the use of animals to assay for carcinogenic potential of chemicals."); Ames & Gold, supra note 268, at 970. (270) See Gold et al., supra note 266, at 261 (pointing out weaknesses in the regulatory process because of reliance on maximum-tolerated doses which "can cause chronic cell killing and consequent cell replacement (a risk factor for cancer that can be limited to high doses)"). (271) See Huff et al., supra note 252, at 621-22. (272) Ellen K Silbergeld, The Role of Toxicology in Causation: A Scientific Perspective, 1 Cts., Health Sci. & The L. 374 (1991). For a summary of the debate, see Boston, supra note 14, at 229-31. (273) There are a number of excellent treatises on the subject See, e.g., Gary D. Friedman, Primer of Epidemiology (3d ed 1987); Abraham M. Lilienfeld & David E. Lilienfeld, Foundations of Epidemiology (2d ed. 1980); Kenneth J. Rothman, Modern Epidemiology (1986). (274) Rothman, supra note 273, at 52-54. (275) There are two principal kinds of epidemiological studies. Case-control studies compare individuals with the disease (cases) to persons who do not have the disease (controls) in an attempt to retrospectively determine commonalities within the diseased group which may reveal a relationship to a chemical agent exposure. Rothman, supra note 273, at 70-72. For example, if an investigator is interested in whether exposure to environmental tobacco smoke (ETS) is related to lung cancer in nonsmokers, she could study nonsmokers who contract the disease (cases) and compare them to those nonsmokers without lung cancer (controls) in terms of whether they had lived with a smoking spouse. One would expect that if there is an association between lung cancer and exposure to environmental tobacco smoke, a history of ET's exposure would be proportionately greater in cases than in controls. This example reveals a number of potential problems affecting the validity of the association. First, it is essential to obtain accurate information about past exposures, including the location and length of exposure. The second problem is of recall bias respecting past exposures in which it is possible that those with the disease (cases) may be more likely to recall exposures than those without the disease (controls). Third, in selecting the control group, care must be taken to assure that persons exposed are not artificially excluded; that is, the cases and controls should be comparable in all respects except their disease status. For an evaluation of ETS studies, see Boston, supra note 14, at 263-69. (276) See Rothman, supra note 273, at 57-62; Elements of Toxicology, supra note 9, at 12-13. The cohort study alleviates one of the major problems with the case-control study (that being memory bias) because the exposure is the starting point for constructing the study. See Boston, supra note 14, at 242, 244. (277) See Elements of Toxicology, supra note 9, at 12; Rothman, supra note 273, at 36-37. (278) Enterline, supra note 97, at 127-28; see also Philip E. Enterline et al., and Cancer: A Cohort Followed Up to Death, 44 Brit. J. Indus. Med. 396-401 (1987). (279) See supra note 98. (280) Id. (281) Dafler v. Raymark Indus., Inc., 611 A.2d 136 (N.J. Super. Ct. App. Div. 1992), aff'd, 622 A.2d 1305 (N.J. 1993). (282) While some studies suggest that the relationship is multiplicative, other studies find that smoking increases the risk of lung cancer from asbestos more than predicted by additivity, but less than predicted by a multiplicative model. See Agency for Toxic Substances & Disease Registry, U.S. Dep't of Health & Human Services, Toxicological Profile for Asbestos: Draft for Public Comment/Update 51 (Oct. 1993). (283) See Dafler, 611 A.2d at 139-40 (describing Dafler as having beenexposed to asbestos for 6 years in engine room and boiler room pipefitting and as having smoked cigarettes for almost 45 years). (284) However, due in part to their rarity, signature diseases may lack significant background cases. See Brennan, supra note 4, at 21; Abraham, supra note 77, at 864. (285) Austin B. Hill, The Environment and Disease: Association or Causation?, 58 Proc. Royal Soc'y Med. 295-300 (1965); see generally Jack D. Hackney & William S. Linn, Koch's Postulates Updated: A Potentially Useful Application to Laboratory Research and Policy Analysis in Environmental Toxicology, 119 Am. Respiratory Diseases 849 (1979). (286) See, e.g., EPA Guidelines for Carcinogen Risk Assessment, 51 Fed. Reg. 33,992, 33,999 (Sept. 24, 1986). (287) Hill, supra note 285, at 296-97. For example, Hill observed that the mortality rates for lung cancer among heavy cigarette smokers (those who smoked more than one pack per day) ranged from 20 to 30 times the normal rates of lung cancer across a large number of epidemiological studies. The stronger the relative risks, the less likely the association is explainable by confounding factors, bias, or other factors. Id. (288) Id. at 296 (noting that the 1964 U.S. Surgeon General's Report on Smoking and Lung Cancer relied on 29 retrospective and 7 prospective studies, an revealing a strong positive association). However, inconsistent associations across studies do not rule out a causal relationship, since differing study methodologies may produce varying relative risks. (289) In the early days, evidence of an association between smoking and lung cancer was criticized. The observation that a wide spectrum of diseases, in addition to lung cancer, were associated with cigarette smoking was used as evidence against a causal interpretation. Joseph Berkson, Smoking and Lung Cancer: Some Observations on Two Recent Studies, 53 J. Am. Stat. Ass'n 28, 36 (1958). Hill argued, however, that since the elevation of lung cancer risk among smokers was more than tenfold, this considerable magnitude of risk strength of association) would be equivalent to specificity - a specificity in the magnitude of association. Hill, supra note 285, at 297. Rothman, however, argues that because the criterion of specificity calls for a single effect, rather than multiple effects, it "seems useless and misleading." Rothman, supra note 273, at 18. (290) Rothman, supra note 273, at 19. This criterion is particularly important for chronic diseases with long latency periods (time between onset of exposure and diagnosis) and for study factors change over time. (291) Id. at 18. A dose-response relationship allows a simple and intuitive explanation and obviously enhances the causal interpretation. Unfortunately, in many epidemiological studies, precise exposure data are absent, making it difficult or impossible to calculate dose-response relationships. The existence of a dose-response curve does not invariably suggest causality because there may be a confounding factor that also yields a separate gradient. (292) Hill, supra note 285, at 298. (293) See Rothman, supra note 273, at 19-20. However, the absence of coherent information, as distinguished from the presence of conflicting evidence, should not be taken as (294) See supra notes 277-82 and accompanying text. (295) Dafler, 611 A.2d at 140. (296) Id. at 144-46. (297) Indeed, various federal statutes have increased knowledge about such substances by mandating that agencies such as the Environmental Protection Agency, the Food and Drug Administration, the National Institute of Occupational Safety and Health, the Agency for Toxic Substances and Disease Registry, and others to undertake studies to assess toxicological risks. (298) See Boston, supra note 14, at 381. (299) 874 F.2d 307 (5th Cir. 1988), modified, 884 F.2d 166 (5th Cir.), reh'g denied en banc, 884 F.2d 167 (5th Cir. 1989) (involving product liability for Bendectin). (300) 874 F.2d at 311. (301) Brock, 884 F.2d 167, 167 (5th Cir. 1989). The court reversed a jury verdict for the plaintiff. (302) 911 F.2d 941 (3d Cir. 1990). (303) Id. at 958 (stating that "the relative risk of limb reduction defects arising from the epidemiological data [plaintiff's witness] relies upon will, at a minimum, have to exceed 2"). The court recognized the importance of considering all toxicological and epidemiological evidence and exhibited considerable sophistication in its analysis of epidemiological methods, including the role of relative risks, statistical significance, and confidence intervals. Id. at 958-59. (304) In re Agent Orange Prod. Liab. Litig., 611 F. Supp 1223, 1231 (D.D.N.Y. 1985) (opt-out plaintiffs), aff'd, 818 F.2d 187 (2d Cir. 1987), cert. denied, 487 U.S. 1234 (1988). (305) 605 A.2d 1079 (N.J. 1992). (306) Id. at 1085. (307) Id. at 1087-88. (308) Id. at 1087. (309) Id. (310) Ferebee v. Chevron Chem. Co., 736 F.2d 1529, 1535 (D.C. Cir.), cert. denied, 469 U.S. 1062 (1984); Intako Aluminum v. Department of Labor & Indus., 833 P.2d 390 (Wash. Ct. App. 1992); Wells v. Ortho Pharmaceutical Co., 615 F. Supp. 262, 295 (N.D. Ga. 1985) aff'd in part, modified in part, 788 F.2d 741 (11th Cir.), cert. denied, 479 U.S. 950 (1986); Osburn v. Anchor Lab., 825 F.2d 908, 915-16 (5th Cir. 1987), cert. denied, 485 U.S. 1009 (1988). (311) 736 F.2d 1529 (D.C. Cir.), cert. denied, 469 U.S. 1062 (1984). (312) Id. at 1535-36. (313) One exception to this statement would be when two separate toxic exposures are independently incapable of causing harm, but are capable when combined. If the plaintiff can demonstrate that synergistic effect, she can satisfy causation requirements. (314) See Dafler v. Raymark Indus., Inc., 611 A.2d 136 (N.J. Super. Ct. App. Div. 1992), aff'd, 622 A.2d 1305 (N.J. 1993). (315) See supra part VI. (316) See generally Joseph Sanders, The Bendectin Litigation: A Case Study in the Life Cycle of Mass Torts, 43 Hastings L.J. 301 (1993). Coordination and networking between the plaintiffs' counsel facilitates the development of a body of toxicological evidence that is used in every trial. Id. at 309; Joseph Sanders, From Science to Evidence: The Testimony on Causation in the Bendectin Cases, 46 Stan. L. Rev. 1 (1993). (317) In fact, the causation inquiry can be separated further, including, at least, the following questions: 1) Has exposure to the substance occurred? 2) Has the exposure resulted in a dose? 3) By what pathway of exposure was the dose received? 4) What are the known effects of the substance? Is the substance capable of causing the kinds of effects involved in the case? 5) Was the effect consistent with the known effects of the substance? 6) Was the dose sufficient to cause such an effect? 7) Does the evidence demonstrate an illness or disease? 8) Is the onset of the disease temporary related to the exposure? 9) Is the effect biologically plausible? 10) Are there other confounding or contributing factors, including other toxic exposures? If so are the sufficiently controlled for? (318) Asbestos workers were exposed in different occupational settings, including insulation application, primary manufacturing, shipbuilding and construction, plant engineering, and automobile maintenance. Depending on each plaintiffs occupation and the duration of that occupation, the exposure levels vary dramatically. (319) See generally Irving J. Selikoff et al., Asbestos Exposure and Neoplasia, 188 JAMA 22 (1964); Irving J. Selikoff & Douglas H.K. Lee, Asbestos and Disease (1978); see also Philip E. Enterline, Mortality Among Asbestos Products Workers in the United States, 132 Annals N.Y. Acad. Sci. 156, 158 (1965) (demonstrating different risks of asbestos-related cancer by the industrial categories listed above). In Cimino v. Raymark Industries, Inc., 751 F. Supp. 649 (E.D. Tex. 1990), Judge Robert M. Parker established a matrix that divided up 2,298 claimants into groupings based on their worksite, trade or craft, and disease category. Because the scientific literature demonstrated how sensitive the dose-response relationship is and because the duration and intensity of exposure for each worksite and craft were different, such categories were essential in determining causality in each grouping. Id. at 653-54. (320) See In re Agent Orange Prod. Liab. Litig., 597 F. Supp. 740, 817-19 (E.D.N.Y. 1984), aff'd, 818 F.2d 145 (2d Cir. 1987), cert. denied, 484 U.S. 1004 (1988) (class action settlement opinion). Judge Weinstein described the different ways servicepersons were exposed:

Members of the armed services were exposed to Agent Orange in a number of different ways. Air Force personnel handled the Agent Orange in preparation for large scale spraying from the air. Hand and mechanical equipment was used locally to clear the perimeter of installations. Troops were exposed to the spraying when they walked and lived in areas with affected vegetation. Some drank water or ate food prepared from crops that had been contaminated with Agent Orange. The exigencies of battle sometimes resulted in fast dumping of large qualities of the herbicide on troops. Id. at 819. Additionally, the toxicity of each manufacturer's Agent Orange herbicide differed because the percentage content of dioxin varied among formulas applied by each, but each producer's product contained at least some dioxin, ranging from 1 part per million (ppm) to 70 ppm. (321) See Abraham, supra note 77,.at 864; Brennan, supra note 4, at 21. (322) See Brennan, supra note 4, at 21 n.96, 22 n.101 (pointing out that recent studies suggest other causes of clear cell vaginal cancers); see also Public Health Service Act, 42 U.S.C. [section] 283a(c) (Supp. V 1993) (mandating that the National Institute of Health conduct longitudinal studies of DES's relationship to diseases in women who consumed it, men and women exposed to it in utero, and their children). (323) See supra part V. (324) See, e.g., Sterling v. Velsicol Chem. Corp., 855 F.2d 1188, 1199-1201 (6th Cir. 1988). (325) See Sandra A. Geschwind et al., Risk of Congenital Malformations Associated with Proximity to Hazardous Waste Sites, 135 Am. J. Epidemiology 1197 (1992) (discussing many of the difficulties inherent in determining relationships to disease in humans and exposure to hazardous waste sites); Leon Gordis, Epidemiologic Approaches for Studying Human Diseases in Relation to Hazardous Waste Disposal Sites, 25 Hous. L. Rev. 837 (1988) (setting forth the array of causation problems faced in nuisance model litigation). Gordis specifically addresses the difficulty in defining what constitutes a hazardous waste site:

What are some of the problems in defining a hazardous waste disposal site for purposes of epidemiological study.? Usually chemical dumping is not an organized activity. Perhaps we have a barren or relatively barren area. We hear that chemical drums were dropped there over a number of years, but as with many sites, there are no records of what was dumped. We often do not even have a clear demarcation of the border of the dump site, and in many situations we do not know what materials were dumped there. The amounts, the timing, and the nature of the specific material dumped are critical in defining exposure for the epidemiologic study. Id. at 844. (326) Gordis lists five possible routes of human exposure from hazardous waste sites: 1) direct skin contact, 2) airborne contact, 3) waterborne contact, 4) explosion or fire, and 5) incorporation into the food chain. Gordis, supra note 325, at 853. (327) See Restatement (Second) of Torts [sections] 832 cmt. f (1979) (noting that groundwater pollution should not be considered intentional because such invasions are unforeseeable); Philip H. Abelson, Inefficient Remediation of Ground-Water Pollution, 250 Sci. 733 (1990); Sterling, 855 F.2d at 1198-99; Geschwind et al., supra note 325, at 1200-02; Shanna H. Swan et al., Congenital Cardiac Anomalies in Relation to Water Contamination: Santa Clara County, California, 1981-1983, 129 Am. J. Epidemiology 885 (1989). (328) See, e.g., Village of Wilsonville v. SCA Servs., Inc., 426 N.E.2d 824, 830 (Ill. 1981); see supra part V. (329) See Gordis, supra note 325, at 843-45; Geschwind et al., supra note 325, at 1200-01. (330) Questions include the following: How much contaminated water did the plaintiff consume? What was the level of the toxic substance in the water? How long was the toxic substance in the water? (331) The problem of specifying what adverse outcomes to expect exacerbates the problem of multiple causes because k makes more uncertain what adverse effects are to be studied. Gordis, supra note 325, at 846. (332) Resource Conservation and Recovery Act (RCRA), 42 U.S.C. [subsections] 6901-6922k (1988 & Supp. V 1993) (enacted in 1976); Comprehensive Environmental Response, Compensation, and Liability Act (CERCLA), 42 U.S.C. [subsections] 9601-9675 (1988 & Supp. V 1993) (enacted in 1980). However, persons may have deposited the materials located at sites years earlier, before these federal regulatory initiatives. (333) In 1986, Congress expanded the Agency for Toxic Substances and Disease Registry component of CERCLA to require the Agency to perform health assessment studies for all sites on the National Priorities List and to create toxicological profiles for all major toxic chemicals involved at such sites. 42 U.S.C. [section] 9604(i) (1988 & Supp. V 1993). (334) See, e.g., Dafler v. Raymark Indus., Inc., 611 A.2d 136, 141 NJ. Super. Ct. App. Div. 1992), aff'd, 622 A.2d 1305 (N.J. 1993) (addressing causation before addressing apportionment). (335) Keeton et al., supra note 5, at 471. (336) DAN B. Dobbs, Torts and Compensation 258 (2d ed. 1994). (337) In addition, the issuance of the Uniform Comparative Fault Act fueled the movement. (338) Bartlett v. New Mexico Welding Supply, Inc., 646 P.2d 579, 584 (N.M. Ct. App. 1982) (quoting William L. Prosser, Joint Torts and Several Liability, 25 Cal. L. Rev. 413, 418 (1936-37)); see also Smith v. Department of Ins., 507 So. 2d 1080, 1090-91 (Fla. 1987). (339) See McIntyre v. Balentine, 833 S.W.2d 52, 58 (Tex. 1992) ("Having ... adopted a rule [of comparative negligence] closely linking liability and fault, it would be inconsistent to simultaneously retain a rule, joint and several liability, which may fortuitously impose a degree of liability that is out of all proportion to fault."); David K. DeWolf, Several Liability and the Effect of Settlement on Claim Reduction: Further Thoughts, 23 Gonz. L. Rev. 37, 40-41 (1988) ("[S]ince juries were now permitted to assess the relative fault of defendants and plaintiffs in percentage terms, there remained no obstacle to doing likewise for the relative shares of defendants' liability."). (340) See supra part I.B. (341) Professor Richard Wright finds considerable importance in whether the plaintiff is at fault or free of fault. He argues that when the plaintiff is free of fault, joint and several liability should continue to govern multiple tortfeasors. See Richard W. Wright, The Logic and Fairness of Joint and Several Liability, 23 Mem. St. U. L Rev. 45, 72-73 (1992) (arguing that when the plaintiff bears zero responsibility for any portion of his injury ... plaintiff's corrective justice claim against each tortfeasor has priority over the tortfeasors" claims for allocation of liability among themselves); Wright, Allocating Liability, supra note 171, at 1186. (342) Professor Wright maintains that corrective justice requires that an innocent plaintiff receive full compensation and that she not bear the risk of receiving less than full compensation because of insolvent or unavailable defendants. Wright, Allocacting Liability, supra note 171, at 1185. Thus, under joint and several liability, those risks are borne by the defendants, who have a right to seek contribution in order to apportion the loss. Id. (343) See, e.g., Bartlett v. New Mexico Welding Supply, Inc., 646 P.2d 579 (N.M. Ct. App. 1982). (344) See, e.g., Coney v. JLG Indus., 454 N.E.2d 197 (Ill. 1983). (345) Professor Wright presumes that "[e]ach tortfeasor ... was a tortious, actual, and proximate cause of the plaintiff's entire injury and thus bears independent full responsibility for the injury." Wright, Allocating Liability, supra note 171, at 1186 (emphasis added). He also views causation as all or none: Causation, unlike the level or foreseeability of the risk or most of the other factors relevant to responsibility, is not a matter of degree. Some condition either was or was not a cause (in the proper scientific sense) of a particular injury. There is no way, based purely on causation, to identify one cause of an injury as more important or significant than any other cause of the same injury. Thus, proponents of causal theories of absolute liability have had to rely on arbitrary "causal priorities" or risk-based "relative causal contributions" to allocate liability when there are multiple causes of the same injury. True "causal apportionment" is conceptually meaningless. Pseudo "causal apportionment" based on relative risk creation is conceptually confused, since, contrary to the assumptions or arguments in much recent scholarship, risk creation is not the same as causation. Id. at 1146 (footnotes omitted). In other articles, Wright considers the special problems presented by toxic tort cases that do not fit his model of causation by redefining the nature of the harm. In order to preserve his causality requirement he redefines the harm as the actual risk exposure per se. See Wright, Efficiency Theory, supra note 43, at 576-77; Wright, supra note 6, at 1073. As Wright explains:

In contrast, there is no conceptual barrier to recovery in these types of [toxic tort] cases under the traditional corrective-justice view. All that is required is that the courts recognize a new type of injury in such cases - risk exposure per se or risk exposure which possibly led to the subsequent injury. The plaintiff would still have to prove that the defendant contributed to the risk-exposure. This approach greatly improves the liability analysis by clarifying the basic policy issue (should risk-exposure be recognized as a legal injury in certain types of cases?) and distinguishing it from the causal issue, whereas the "probabilistic causation" approach, by treating the policy issue as a causal issue, fails to address the policy issue explicitly and confuses mere increased risk with causation. Wright, Efficiency Theory, supra note 43, at 576-77.

In other words, Professor Wright's positions are largely inapposite to this risk contribution model because he 1) sidesteps the toxic tort cases and 2) assumes that joint and several liability applies where the injury itself is truly indivisible. (346) See Twerski, The Baby, supra note 171; Twerski, Joint Tortfeasor, supra note 171; Gary T. Schwartz, Contributory and Comparative Negligence: A Reappraisal, 87 Yale L.J. 697 (1978). (347) See discussion supra part III.C. (348) See, e.g., Cal. Civ. Code [sections] 1431.2(a) (West Supp. 1995); N.Y. Civ. Prac. L & R. 1601 (McKinney 1990). (349) See, eg., Ga. Code Ann. [sections] 51-12-33 (Supp. 1989); Wash. Rev. Code Ann. [sections] 4.22.070(1)(b) (West Supp. 1995). (350) See, e.g., Fla. Stat. Ann. [sections] 768.81(3) (West Supp. 1995); Nev. Rev. Stat. [sections] 41.141 (1993). (351) See, e.g., Ill. Ann. Stat. ch. 735, para. 5/2-1117 (Smith-Hurd 1992). (352) For a discussion of how courts have treated these issues under various comparative fault systems, see Cheri D. Green & Michael K Graves, Allocation of Fault: Joint Tortfeasors in Court and the Ones Who Should Be, 63 Miss. L.J. 647 (1994). (353) See supra notes 139-43 and accompanying text. (354) See Abraham, supra note 77, at 847 (defining collective responsibility as the practice of treating all defendants as one). (355) Id. (356) See, e.g., Robert A. Baruch Bush, Between Two Worlds: The Shift from Individual to Group Responsibility in the Law of Causation of Injury, 33 UCLA L Rev. 1473 (1986); Andrew R. Klein, Beyond DES. Rejecting the Application of Market Share Liability in Blood Products Litigation, 68 Tul. L Rev. 883 1994); Richard Delgado, Beyond Sindell: Relaxation of Causes-in-Fact Rules for Indeterminate Plaintiffs, 70 Cal. L. Rev. 881 (1982); Joseph King, Jr., Causation Valuation, and Chance in Personal Injury Torts Involving Preexisting Conditions and future Consequences, 90 Yale L.J. 1353 (1981); Rosenberg, supra note 4, at 905-08. (357) See Sindell v. Abbott Lab., 607 P.2d 924, 937-38 (Cal.), cert. denied, 449 U.S. 912 (1980). The Washington Supreme Court explicitly acknowledged the trade-off between providing some compensation to victims and rejecting the principles of individual responsibility in favor of collective liability: "We are presented with a conflict between the familiar principle that a tortfeasor may be held liable only for damage that it has caused, and the sense of justice which urges that the victim of this tragedy should not be denied compensation. . . ." Martin v. Abbott Lab., 689 P.2d 368, 381 (Wash. 1984); see also Collins v. Eli Lilly & Co., 342 N.W.2d 37, 52 (Wis.), cert. denied, 469 U.S. 826 (1984) ("[[S]ome of the ... defendants may be innocent, but we accept this as the price the defendants, and perhaps ultimately society, must pay to provide the plaintiff an adequate remedy."). (358) 607 P.2d 924 (Cal.), cert. denied, 449 U.S. 912 (1980). (359) See Hymowitz v. Eli Lilly & Co., 539 N.E.2d 1069,1077-78 (N.Y), cert. denied, 493 U.S. 944 (1989); Klein, supra note 356, at 903-04. (360) In Hymowitz, the court stated

We are aware that the adoption of a national market will likely result in a disproportion between the liability of individual manufacturers and the actual injuries each manufacturer caused in this State. Thus our market share theory cannot be founded upon the belief that, over the run of cases, liability will approximate causation in this State. Nor does the use of a national market provide a reasonable link between liability and the risk created by a defendant to a particular plaintiff. 539 N.E.2d at 1078 (citations omitted). (361) See supra parts IV, V. (362) Abraham, supra note 77, at 862 (footnotes omitted). Not all writers agree. For example, Professor Robert A. Baruch Bush maintains that joint and several liability is consistent with the principle of individual responsibility (IR):

When the actions of several individuals combine, either sequentially or contemporaneously, to produce a particular harm that would not occur in the absence of any one of them, all are held jointly and severally liable as tortfeasors. However, under joint and several liability, each tortfeasor is still seen as individually responsible to the victim for the entire harm. The separate rights of contribution or indemnity that each tortfeasor might have against the others do not change the fact that each one may have to pay the entire judgment. Nor does this result violate the IR principle. Since each tortfeasor's actions constitute a cause sine qua non of the entire harm, each can properly be held fully responsible. In other cases, a number of individuals are held liable for having "acted in concert" in a way that resulted in the victim's injury. Again, the IR principle remains intact. Each individual is considered a necessary link in a "conspiratorial" circle, and each therefore is a cause sine qua non and bears individual responsibility to the victim for the entire harm. Baruch Bush, supra note 356, at 1476-77 (footnotes omitted). Professor Baruch Bush's discussion is not apposite to many toxic tort cases because sine qua non causes often cannot be demonstrated. Even in Dafler it could not be shown that asbestos was a but-for cause of plaintiff's lung cancer. (363) For example, illustrative of the resistance to market share, several courts refused to apply the theory to the lead paint litigation. See Santiago Sherwin-Williams Co., 782 F. Supp. 186 (D. Mass. 1992), aff'd, 3 F.3d 546 (1st Cir. 1993); City of Philadelphia v. Lead Indus. Ass'n, 994 F.2d 112 (3d Cir. 1993). Some courts refused to apply market-share liability even to DES litigation. See Smith v. Eli Lilly & Co., 560 N.E.2d 324, 337 (Ill. 1990). New Jersey refused to apply market share to DPT vaccine litigation. Shackil v. Lederle Lab., Inc., 561 A.2d 511, 529 (N.J. 1989). Many other courts refused to apply market-share liability to asbestos litigation. See, e.g., Case v. Fibreboard Corp., 743 P.2d 1062, 1067 (Okla. 1987) ("[T]he public policy favoring recovery on the part of the innocent plaintiff does not justify the abrogation of the rights of a potential defendant to have a causative link proven between that defendant's specific tortious acts and the plaintiff s injuries."). (364) Dafler v. Raymark Indus., Inc., 611 A.2d 136 (N.J. Super. Ct. App. Div. 1992), aff'd, 622 A.2d 1305 (N.J. 1993). (365) Anderson v. Minneapolis, St. P. & S. Ste. M. Ry., 179 N.W. 45 (Minn. 1920). (366) Section 432(2) of the Restatement (Second) of Torts (1979) states:

(2) If two forces are actively operating, one because of the actor's negligence, the other not because of any misconduct on his part, and each of itself is sufficient to bring about harm to another, the actor's negligence may be found to be a substantial factor in bringing it about.

This is merely a sufficient cause test and does not raise the problems that occur when the test is applied absent any proof of the sufficiency of the toxic exposure to have caused the harm. An example of this is the twin fires case, in which two fires coalesce and cause a single harm. The Restatement notes that the creators of either fire may be found liable because "either fire would have been sufficient to bum the house and timber." Restatement (Second) of Torts [sections] 432(2) cmt. d, illus. 3 (1979). (367) 418 N.W.2d 650 (Mich. 1988). (368) 611 A.2d 136 (N.J. Super. Ct App. Div. 1992), aff'd, 622 A.2d 1305 (N.J. 1993). (369) 528 A.2d 947 (Pa. 1987). (370) See Tragarz v. Keene Corp., 980 F.2d 411, 424 (7th Cir. 1992) (Under Illinois law the question is "whether each contributing clause, standing alone, is a substantial factor in causing the alleged injury."); In re Manguno, 961 F.2d 533, 534 (5th Cir. 1992) (reversing the jury instructions that provided that asbestos could be found a cause only "so long as it can reasonably be said that, except for the asbestos exposure, the injury complained of would not have occurred," because the instruction was too narrow and applied the substantial factor test). (371) See, e.g., Welch v. Keene Corp., 575 N.E.2d 766, 769-70 (Mass. App. Ct. 1991). (372) In rejecting proposals for proportionate causation or recovery for risk, Professor

Abraham argues m part that the essential statistical evidence is unavailable:

Logical though this approach might be, it is unfortunately very impractical, because nothing even remotely as precise a proxy for the probability of responsibility as market share is available in most nonsignature disease cases. Science has progressed to the point where it can identify some substances and exposures that seem to increase the probability of certain diseases, but only infrequently can these probabilities be quantified with any precision.

Abraham, supra note 77, at 865 (footnotes omitted). He later comments:

[T]he use of collective responsibility as a surrogate for individual responsibility does not necessarily have the incentive effects that might otherwise justify its imposition; and in any case, the data necessary to support the probabilistic measures of causation that could serve as surrogates for causal responsibility will rarely be available. The entire effort usually requires more than the legal or scientific state of the art is capable of providing

Id. at 867-68.

(373) Bell Petroleum, 3 F.3d 889 (5th Cir. 1993). For a discussion, see supra part V.C.; see also Dafler v. Raymark Indus., Inc., 611 A.2d 136 (N.J. Super. Ct. App. Div. 1992), aff'd, 622 A.2d 1305 (N.J. 1993); Martin v. Owens-Corning Fiberglas Corp., 528 A.2d 947, 951 (Pa. 1987) (Nix, C.J., dissenting) (stating that "rough approximations" of harm apportionments are acceptable). (374) CERCLA, 42 U.S.C. [sections] 9604(i)(1) (1988 & Supp. V 1993); see 54 Fed. Reg. 33,617 (1989); see generally James A. Rogers, 7he Potential Role of Superfund in Toxic Tort Litigation, in Environmental Law Manual 486 (Theodore L. Garrett ed., 1992). (375) 42 U.S.C. [sections] 9604(i)(1)(a) (1988 & Supp. V 1993). (376) Id. "Hazardous substance" is defined in 42 U.S.C. [sections] 9601(14) (1988). The Administrator has the authority to designate additional substances as hazardous if the Administrator determines such classification appropriate. Interim Petitioned Public Health Assessment Response Procedures and Decision Criteria, 57 Fed. Reg. 37,381, 37,382 (Aug. 18, 1992). (377) The ATSDR defines "health assessment" as "the evaluation of data and information on the release of hazardous substances into the environment in order to: assess any current or future impact on public health, develop health advisories or other recommendations, and identify studies or actions needed to evaluate and mitigate or prevent human health effects." Health Assessments and Health Effects Studies of Hazardous Substances Releases and Facilities, 53 Fed. Reg. 32,259, 32,260 (Aug. 24, 1988) (the term "health effects study" is mentioned but not otherwise defined in CERCLA, 42 U.S.C. [sections] 9607(a)(4)(d) (1988)). (378) Id. (379) The ATSDR and EPA use the reportable quantity (RQ) approach as the toxicity hazard scoring system. The RQ scoring system is described in 51 Fed. Reg. 34,534 (Sept. 29, 1986); 52 Fed. Reg. 8140 (Mar. 16, 1987); 54 Fed. Reg. 35,988 (Aug. 30, 1989). The ATSDR ranks the hazard potential of each candidate substance according to the following algorithm: total score = NPL frequency + toxicity + potential for human exposure. 56 Fed. Reg. 52,165, 52,168 (Oct. 17, 1991). (380) Superfund Amendments and Reauthorization Act of 1986, Pub. L. No. 99-499, 100 Stat. 1613 (1986). (381) In a pilot health effects study, researchers study a relatively small group of people - usually fewer than 1,000 - to determine whether a medical follow-up or an epidemiological study should be conducted, or whether action is required to protect the public from further contact with the hazardous substance. Agency for Toxic Substances & Disease Registry, U.S. Dep't of Health & & Human Services, Biennial Report 83 (1989-1990) [hereinafter Biennial Report]. (382) Some examples of epidemiological studies include the following:

1) California Birth Defects Study: The ATSDR funded the Center for Environmental Health and Injury Control (CEHIC) to determine the rates of adverse reproductive outcomes in census tracts when exposed to Superfund sites. These rates were compared with rates in other census tracts within the San Francisco Bay area and indicated an association between exposure and cardiac birth defects. Id. at 88-90.

2) Massachusetts Environmental Investigations: In 1988, the ATSDR funded CEHIC to study possibly contaminated public wells in the Woburn, Massachusetts area. The objectives of the study were to 1) investigate the rates of selected adverse reproductive outcomes occurring since 1969 for residents of Woburn and a comparison community; 2) establish ongoing surveillance for selected adverse reproductive outcomes in Woburn and other selected Massachusetts communities; and 3) assess present and past environmental contamination in Woburn and link adverse reproductive outcomes with potential environmental exposures. Id. at 89-90. The Woburn well contamination has been the subject of protracted toxic tort litigation. See, e.g., Anderson v. W.R. Grace & Co., 628 F. Supp. 1219 (D. Mass. 1986). (383) 42 U.S.C. [sections] 9604(i)(3) (1988 & Supp. V 1993). (384) Id. (385) Some of the toxicological profiles available include arsenic, nickel, asbestos, radium, benzene, radon, cadmium, toluene, chlorobenzene, trichloroethylene, chromium, uranium, lead, vanadium, mercury, and vinyl acetate. For a discussion of the National Toxicology Program, see Huff et al. supra note 253. (386) The ATSDR states that when an assessor is evaluating the health impacts of exposure to hazardous substances, the assessor should consider data from studies of human exposures as well as the results of experimental animal studies. Similarly, the ATSDR states that "the Health. assessor must rely on the results of experimental animal studies" when inadequate human data exist Agency for Toxic Sibstances & Disease Registry, U.S. Dep't of Health & Human Services, Public Health Assessment Guidance Manual 7-4 (1992). (387) Problems do remain, however. sometimes the populations being studied are not sufficiently large or the increased incidence not sufficiently great to produce results that are statistically significant. For example, one author illustrates the problems of finding any statistically significant results by looking at 10 years of records for a hypothetical town of 5,000 people. If the disease being studied occurs in 1 in every 1,000 people a year, the expected incidence over 10 years, all things being equal, would be 50 cases. Th find statistically significant results at the 9996 level of confidence, there would have to be at least 70 cases of this disease in the community. The community would have a relative risk of 1.4, or a 4096 higher chance of suffering from the disease in question.

If the annual rate of the disease is 1 in 10,000, the expected number of cases during a decade in any community would be 5. For the incidence in this community to be found significantly different there must have been 12 cases. At that rate, the relative risk in this community is 2.4. See Raymond R. Neutra, Reviews and Commentary: Centerpoint from a Cluster Buster, Am. J. Epidemiology 1-8 (1990).

In sum, when there is a demand for proof of health effects, the preferred method for finding such proof is to conduct traditional epidemiological studies. But these studies are severely restrained by common circumstances of community exposure. (388) Office of Health & Envtl. Assessment, Envtl. Protection Agency, Respiratory Health Effects of Passive Smoking: Lung Cancer and Other Disorders (1992). (389) Occupational Safety and Health Act of 1970, 29 U.S.C. [subsections] 651-678 (1988 & Supp. V 1993); see Occupational Safety and Health Standards, 29 C.F.R. [subsections] 1910.1001-.1101 (1994) (containing standards for many substances, including asbestos, coal for tar pitch, vinyl chloride, inorganic arsenic, lead, benzene, cotton dust, and formaldehyde). (390) 113 S. Ct. 2786 (1993). (391) Id. (392) Goldstein & Henifin, supra note 2. (393) See supra part V. (394) See, eg., Richard A. Epstein, Causation - In Context: An Afterword, 63 Chi.-Kent L. Rev. 653 (1987); Richard A. Epstein, A Theory of Strict Liability, 2 J. Legal Stud. 151 (1973); Alan Schwartz, Causation in Private Tort Law: A Comment on Kelman, 63 Chi.-Kent L. Rev. 639 (1987); Wright, supra note 6. (395) See Jules L Coleman, Tort Law and the Demands of Corrective Justice, 67 Ind. L.J. 347, 358 (1992) (arguing that "corrective justice is concerned with the gains and losses that one person causes another" as it "prohibits creating a wrongful loss"). (396) See id. at 349. (397) See Epstein, Causation - In Context, supra note 394, at 677. (398) Id. Professor Epstein further argues that

[t]he theories did not create the factual problems of the modem asbestos, delayed trauma toxic tort case. Rather the difficulty arises because the present legal system is so wedded to its own conceptions of causation and responsibility that it does not permit any form of contracting out, even where it is feasible.

Id. at 678. (399) See Wright, supra note 6, at 1049-77; Wright, Efficiency Theory, supra note 43, at 560. (400) Wright, supra note 6, at 1073. Another formulation is that, in contrast, there is no conceptual barrier to recovery in these types of cases under the traditional corrective justice view. All that is required is that the courts recognize a new type of injury in such cases - risk exposure per se or risk exposure which possibly led to the subsequent injury. The plaintiff would still have to prove that the defendant contributed to the risk exposure. This approach greatly improves the liability analysis by clarifying the basic policy issue (should risk exposure be recognized as a legal injury in certain types of cases?) and distinguishing it from the causal issue. The "probabilistic causation" approach, on the other hand, treats the policy issue as a causal issue and fails to address the policy issue explicitly. This approach confuses mere increased risk with causation. Wright, Efficiency Theory, supra note 43, at 576-77. (401) Rosenberg, supra note 4, at 924-29. (402) Robinson, supra note 4, at 743-44. (403) Boston, Mass Exposure Model, supra note 14, at 189, 381. Evidentiary questions may include: Is asbestos exposure a cause of lung cancer? Is asbestos exposure a cause of colon cancer? Is Bendectin a teratogen? Does Agent Orange (dioxin) cause skin cancer? Do nickel and cadmium cause colon cancer? (404) For the criteria to determine whether an association is casual, see supra notes 285-93 and accompanying text, in which the criteria for determining whether an association is causal are set forth. (405) Dafler v. Raymark Indus., Inc., 611 A.2d 136,140 (N.J. Super. Ct. App. Div. 1992), aff'd, 622 A.2d 1305 (N.J. 1993). (406) When the harm consists of diseases such as cancer, whose precise etiology is as yet poorly understood, there will always remain some element of arbitrariness in assigning responsibility to the defendant. This element will remain regardless of how much persuasive evidence the plaintiff offers. Thus, "despite any amount of scientific information directly on point, the existence of residual baseline risk renders any categorical decision on causation epistemologically arbitrary." Vern R. Walker, 7he Concept of Baseline Risk in Tort Litigation, 80 Ky. L.J. 631, 669 (1992); see also Brennan, supra note 2, at 470-71. Even in tobacco and asbestos litigation, the distinct possibility exists that the plaintiff's lung cancer was a background case and not actually attributable to these toxic exposures. (407) Sindell v. Abbott Lab., 607 P.2d 924, 937 (Cal.), cert. denied, 449 U.S. 912 (1980). (408) This is not true for all states' versions of market-share liability. While California placed the burden on the plaintiff, other states have allowed the plaintiff to sue a single defendant who will bear the entire loss, unless the defendant impleads other firms among whom to distribute the loss. See, e.g., Martin v. Abbott Lab., 689 P.2d 368 (Wash. 1984); Collins v. Eli Lilly & Co., 342 N.W.2d 37, 50-51 (Wis.), cert. denied, 469 U.S. 826 (1984). (409) See Wright, Causation in Tort Law, supra note 43. (410) See Keeton et al., supra note 5, [sectiobs] 52 (stating that apportionment of damages cannot be precluded solely because of the difficulty in doing so, and comparing the avoidable consequences doctrine in that "[t]he difficulty is certainly no greater than in cases where part of the damage is to be attributed to the unreasonable conduct of the plaintiff, and the rule of avoidable consequences is applied to limit recovery"). (411) See, e.g., Ostrowski v. Azzara, 545 A.2d 148 (N.J. 1988). (412) See Wright, supra note 6, at 1073; Wright, Efficiency Theory, supra note 43, at 576-77. (413) 454 N.E.2d 197 (Ill. 1983). (414) Id. at 205 (emphasis added); see also, American Motorcycle Ass'n v. Superior Court of Los Angeles, 579 P.2d 899 (Cal. 1978). The Tennessee Supreme Court recently adopted comparative fault in McIntyre v. Balentine, 833 S.W.2d 52 (Tenn. 1992), and in so doing rejected decisions such as Coney and American Motorcycle and held that joint and several liability was abolished by its action:

[T]oday's holding renders the doctrine of joint and several liability obsolete. Our adoption of comparative fault is due largely to considerations of fairness: the contributory negligence doctrine unjustly allowed the entire loss to be borne by a negligent plaintiff, notwithstanding that the plaintiffs fault was minor in comparison to defendant's. Having thus adopted a rule more closely linking liability and fault, it would be inconsistent to simultaneously retain a nile, joint and several liability, which may fortuitously impose a degree of liability that is out of all proportion to fault. 833 S.W.2d at 58. (415) Wright, Allocating Liability, supra note 171, at 1146-47. (416) See, e.g., In re Kinsman Transit Co., 338 F.2d 708 (2d Cir.), cert. denied, 380 U.S. 944 (1964); Ryan v. New York Cent. R.R., 35 N.Y. 210 (1866); Overseas Tankship (U.K.) Ltd. v. Morts Dock & Eng'g Co. (The Wagon Mound), 1 Eng. Rep. 404 (P.C. 1961). (417) The classic opinion is Louisiana ex rel. Guste v. M/V Testbank, 752 F.2d 1019 (5th Cir. 1985) (en banc), involving a collision of two ships that produced a chemical spill and quarantine of the New Orleans port area. The opinion contains a wealth of arguments respecting unlimited liability in the context of public nuisance liability and also discusses liability for economic losses based on negligence stemming from what could have been a mass exposure. The Fifth Circuit refused to permit, ripple-effect plaintiffs (other than crabbers and shrimpers who fished in those waters) to recover the economic losses they sustained from the cessation of all shipping, marina, and recreational activities for 17 days. Id. at 1032. The court opted for a bright-line rule requiring physical property damage in order to prevent the specter of unlimited and unpredictable losses flowing from the defendant's negligence. Id. at 1029.

One means of addressing the risk of imposing unlimited liabilities on firms in toxic tort cases is to require the rigorous scientific proof of causation and to apply principles of apportionment. (418) Those legislatures that retain joint and several liability for environmental tort actions, while otherwise abolishing or modifying it, have k backwards, except to the extent that the harm is really incapable of apportionment on any reasonable basis. See supra notes 138-45 and accompanying text for a description of such statutes. (419) See Sindell v. Abbott Lab., 607 P.2d 924, 937 (Cal.), cert. denied, 449 U.S. 912 (1980). (420) Restatement (Third) of the Law of Torts: Products Liability 264-65 (Tentative Draft No. 2, Mar. 13, 1995). (421) Professor Twerski states:

The true problem is "institutionally immune" defendants. These defendants have been granted a broad license to act, bearing limited financial responsibility for their conduct. It is not surprising for legislatures to conclude that it is unfair to saddle solvent defendants with the full brunt of damages substantially caused by conduct which society immunized. The decision against joint tortfeasor liability essentially states society's belief that the limited compensation available from recovery-immune defendants reflects the appropriate level of compensation for the immune conduct. For example, motorists who are permitted by law to drive with woefully inadequate liability insurance limits are not simply insolvent tortfeasors. Instead, they are recovery-immune defendants, legally sanctioned to drive with full knowledge that they will only be able to pay a small fraction of the costs of the harm they cause. To shift losses to the solvent joint tortfeasor is to treat him as a "whipping boy" and to require him to bear full responsibility for broad-based immunities that cut a very large swath through traditional tort liability. Twerski, Joint Tortfeasor, supra note 171, at 1132-33. (422) Rosenberg, supra note 4, at 861-62; Robinson, supra note 4, at 740-44. (423) See William M. Landes & Richard A. Posner, Causation in Tort Law: An Economic Approach, 12 J. Legal Stud. 109, 123-24 (1983). (424) Rothman, supra note 273, at 7; see also supra part VI. (425) If a plaintiff must prove causation more probably than not, and only probabilistic evidence is offered, the risk must exceed 2.0 to produce an etiologic probability of greater than 50%. (426) Fischer, supra note 16, at 1205 (stating that proportional liability overdeters defendants because their risk of liability outweighs incentive to act safely). (427) Rosenberg, supra note 4.
COPYRIGHT 1995 Lewis & Clark Northwestern School of Law
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1995, Gale Group. All rights reserved. Gale Group is a Thomson Corporation Company.

Article Details
Printer friendly Cite/link Email Feedback
Author:Boston, Gerald W.
Publication:Environmental Law
Date:Jun 22, 1995
Words:55417
Previous Article:Environmental Cooperation in Europe: The Political Dimension.
Next Article:Running on empty: the case for a sustainable national transportation system.
Topics:


Related Articles
Understanding epidemiological evidence; what to ask the expert witness.
Allocating fault in negligent security cases.
Providing adequate remedies to toxic tort victims.
New apportionment of liability provisions considered at recent ALI meeting.
Fifth Circuit extends Daubert to physicians' causation testimony.
Intentional disregard: remedies for the toxic workplace.
Protecting against toxic mold lawsuits. (Insiders Outlook).
Keep good science in toxic tort cases; plaintiff lawyers can use a new section of the Restatement (Third) of Torts to help judges widen the gates to...
What a difference a word makes.
Making up for lost time: your client could have recovered from an illness, if only his or her physician had made a correct diagnosis and provided...

Terms of use | Privacy policy | Copyright © 2019 Farlex, Inc. | Feedback | For webmasters