In re: Agent Orange Product Liability Litigation, 597 F. Sup. 749 (1984) (excerpts). B. The Indeterminate Plaintiff Summers v. Tice. 33 CaL. 2d 60. L99 P2d 1 (1946) "There is evidence that both defendants, at about the same time or on immediateLy after the other, shot at a quail and in doing so shot toward plaintiff who was uphill from them, and that they knew his location. That is sufficient from which the trial court could conclude that they acted with respect to the plaintiff." In RE Agent Orange Product Liability Litigation cite is 597 F.Sup 740 (1984) by Weinstein. J.: b. The Problem of the Indeterminate Plaintiff The preceding discussion assumed that although a plaintiff would be unable to identify the manufacture of the Agent Orange to which the veteran was exposed he or she would be able to prove by a preponderance of the evidence that the specific injuries from which he or she suffers were caused by Agent Orange. It is likely, however, that even if plaintiffs as a class could prove that they were injured by Agent Orange, no individual class member would be able to prove that this or her injuries were caused by Agent Orange. Fore example, plaintiffs as a class may be able to show that statistically, X% of the population not exposed to Agent Orange could have been expected to develop soft-tissue sarcoma, but that among those veterans who were exposed be Agent Orange. X + Y% suffer from soft tissue sarcoma. If Y is equal to or less than X and there is no meaningful "[illegible] or anecdotal proof as to vast majority of plaintiffs, virtually no plaintiff would be able to show by a preponderance of the evidence that this or her cancer is attributable to the Agent Orange rather than being part of the "background" level of cancer in the population as a whole. the probability of specific cause would necessarily be less than 50% based upon the evidence submitted. (1) Scope of the Problem The problem just noted is one that has received a significant amount of scholarly discussion a well as some attention from the United States Congress. Because of the rarity of the situation until recently scant attention has been given to the issue by the courts. There has apparently been only one other mass exposure decision that has discussed the indeterminate plaintiff problem explicitly. viz, Allen v. United States, 588 F. Sup 247 (D. Utah 1964), a case where [illegible] was claimed as a result of radioactive exposure from testing of atomic explosive devices. In our complex industrialized society it is unfortunately possible that some products used on a widespread scale will cause significant harm to the public. While it may be possible to prove, through the use of such proof as laboratory tests on animals and epidemiological evidence, that such harm -- for example cancer -- can be "caused" by a particular substance, it may be impossible to pinpoint which particular person's cancer would have occurred naturally and which would not have occurred but for exposure to the substance. Epidemiological statistics, which constitute the best (if not the sole) available evidence in mass exposure cases, can only attribute a proportion of the disease incidence in the population to each potential source . . . . But . . . it is impossible to pinpoint the actual source of the disease afflicting any specific member of the exposed population. Rosenberg. The Causal Connection in Mass Exposure Cases: A "Public Law" Vision of the Tort System, 97 Harv. L. Rev. 849 856-57 (1984) (footnotes omitted). In two of the largest and most widely publicized mass tort litigations, those involving DES and asbestos, the problem outlined above does not pose a serious obstacle since at least some of the damage caused by the harmful substance was it has been claimed, unique to that substance. Adenosis and clear cell adenocarcinoma of the vagina and uterus, the conditions associated with DES, are, it is said, almost unknown among women whose mothers had not taken DES. Note, DES and a Proposed Theory of Enterprise Liability, 46 Fordham L.Rev. 963, 965 (1978). The situation is similar, in the asbestos litigation, albeit to a lesser extent. Although lung cancer is associated with cigarette smoking and other factors as well as asbestos exposure and mesothelioma may have causes other than asbestos is alleged to be uniquely associated with asbestos exposure. Barel v. Fiberboard Paper Products Corp., 493 F.2d 1076,1083 (5th Cir. 1973), cert denied, 419 U.S. 869, 96 S.Ct. 127, 42 L.Ed.2d 107 (1974). In most other mass exposure cases, however, the harm caused by the toxic substance is indistinguishable from the naturally occurring disease or condition. See Dore. A Commentary on the Use of Epidemiological Evidence in Demonstrating Cause-in-Fact, 7 Har.Envtl.L.Rev 429, 437 (1983); Solomons, Workers' Compensation for Occupational Disease Victims: Federal Standards and Threshold Problems, 41 Alb.L.Rev. 195, 199 (1977), The recent case of Allen v. United States, 588 F. Sup. 247 (D.Utah 1984), illustrates the problem well. Plaintiffs claimed that they developed various forms of cancer as a result of their exposure to radiation from nuclear explosions. While some forms of cancer can, it is contended, with some certainty be attributed to factors other than exposure to radiation, many others "cannot be distinguished from cancer of the same organ arising from . . . unknown cases". i.e., the 'background' cancers. J. Gofman, Radiation and Human Health 59 (1981), quoted in Allen v. United States, 588 F. Sup at 406. The statistical evidence in Allen apparently made it clear that there was a strong positive association between exposure to low-level ionizing radiation, presumably the result of atomic explosions, and various forms of cancer suffered by plaintiffs. Thus, the Allen case has some of the characteristics of the DES and asbestos cases in addition to persuasive statistical correlations. (2) Preponderance Rule Even if there were near certainty as to general causation, if there were significant uncertainty as to individual causation, traditional tort principles would dictate that causation be determined on a case-by-case basis using the preponderance-of-the-evidence rule. Santosky v. Kramer, 455 U.S. 745, 755, 102 S. Ct. 1388, 1395 71 L.Ed.2d 599 (1982); W. Prosser, Handbook of the Law of Torts, 206-09 (1971); Kaplan, Decision Theory and the Fact Finding Process, 20 Stan. L. Rev. 1065, 1072 (1968). The rule provides as "'all or nothing' approach, whereby [assuming all other elements of the cause of action are proven], the plaintiff becomes entitled to full compensation for those . . . damages that are proved to be 'probable' (a greater than 50 percent chance), but is not entitled to any compensation if the proof does not establish a greater than 50 percent chance." Jackson v. John-Manville Sales Corp., 727 F.2d 506, 516 (5th Cir. 1984). Under the "strong" version of the preponderance rule, statistical correlations alone indicating that the probability of causation exceeds fifty percent are insufficient some "particularistic" or anecdotal evidence, that is, "proof that can provide direct and actual knowledge of the causal relationship between the defendant's tortious conduct and the plaintiff's injury." is required. As Professor Jaffee has put it, If all that can be said is that there are 55 chances of negligence out of 100, that is not enough. There must be a rational, i.e., evidentiary basis on which the jury can choose the competing probabilities. If there is not, the finding will be based . . . on mere speculation and conjecture. Jaffee, Res. Ipea Loquitur Vindicated, I Buffalo L.Rev. 1, 4 (1951). The "weak" version of the preponderance rule would allow a verdict solely on statistical evidence; the "all-or- nothing" approach converts the statistical probability into a legally absolute finding that the causal connection did or did not exist in the case. C. McCormick, Handbook on the Law of Damages, 118 (1935). The justification for not requiring "particularistic" or anecdotal evidence is trenchantly and accurately stated by Professor Rosenberg: [T]he entire notion that "particularistic" evidence differs in some significant qualitative way from statistical evidence must be questioned. The concept of "particularistic" evidence suggests that there exists a form of proof that can provide direct and actual knowledge of the causal relationship between the defendant's tortious conduct and the plaintiff's injury. "Particularistic" evidence, however, is in fact no less probabilistic than is the statistical evidence that courts purport show . . . "Particularistic" evidence offers nothing more than a basis for conclusions about a perceived balance of probabilities. Rosenberg, supra, 97 Har.L.Rev. at 870 (footnotes omitted). Except where it appears that the absence of anecdotal evidence may be due to spoliation, probabilities based upon quantitative analysis should support a recovery. See e.g., E.M. Morgan & J.M. Maguire, Cases and Materials on Evidence, 39 (7th ed. 1983). There would appear to be little harm in retaining the requirement for "particularistic" evidence of causation in sporadic accident cases since such evidence is almost always available in such litigation. In mass exposure cases, however, where the chance that there would be particularistic evidence is in most cases quite small, the consequence of retaining the requirement might be to allow defendants whom, it is virtually certain, have injured thousands of people and caused billions of dollars in damage, be [illegible] liability. Because of this [illegible] fact that "particularistic evidence . . . is . . . no less probabilistic than . . . statistical evidence," the "weak" version of the preponderance rule appears to be the preferable standard to apply in mass exposure cases--particularly where, as here, all claimants and defendants are jointed in one suit. (a) Application of the Preponderance Rule to Mass Exposure Cases Conventional application of the "weak" version of the preponderance rule would dictate that, if the toxic substance caused the indigence of the injury to rise more than 100% above the "background" level, each plaintiff exposed t the substance could recover if he or she is suffering from that type of injury. If, however, to put it in somewhat graphic, albeit artificial terms, the incidence rose only 100% of less, no plaintiff could recover--i.e., the probability of specific causation would not be more than 50%. Where a plaintiff's injuries result from a series of unrelated sporadic accidents, this "all-or-nothing" rule is justifiably rationalized on the ground that it is the fairest and most efficient result. In mass exposure cases, however, this all-or-nothing rule results in either a tortious defendant being relieved of all liability or overcompensation to many plaintiffs and a crushing liability on the defendant. These results are especially troublesome because, unlike the sporadic accident cases, it may be possible to ascertain with a fair degree of assurance that the defendant did cause damage, and, albeit with somewhat less certainty, the total amount of that damage. The problem is both illustrated and further compounded by the fact that lack of precision in the data and models used may cause the range of the probabilities estimated by the statistical proof to lie on either or other sides of the 100% line. Because the statistical proof will almost never be as complete or as free from confounding factors as desirable, it may be possible to infer, for example, that the toxic substance caused the incidence to rise over the background level somewhere between 80 and 120%. See e.g., Allen v. United States, 588 F.Sup 247, 438, 439 n. 197 (D.Utah 1984) (noting significant variation in experts interpretations of statistical evidence relating to the likelihood that plaintiff's cancer was caused by exposure to radiation). Moreover, issues of credibility and varying inferences drawn by the trier based upon varying assessments of probative force may cause reasonable people to assess these percentages in a range from almost zero to well over 120. See, e.g., E.M. Morgan & J.M. Maguire, Cases and Materials on Evidence, ch. 1 (7th ed. 1983). Under the traditional application of the preponderance rule, whether individual plaintiffs recover will depend on where the probability percentage line is drawn despite the fact that a reasonable trier would conclude that a large proportion of the plaintiffs were injured by the defendant and a large number were not. Even if the statistical increase attributed to the substance in question is just a few percentage points, if statistical theory supports a finding of correlation there is no reason why the industry as a whole should not pay for the damages it probably caused. A simple hypothetical will illustrate why too heavy a burden should not be placed on plaintiffs by requiring a high percentage or incidence of a disease to be attributable to a particular product. Let us assume that there are 10 manufacturers and a population of 10 million persons exposed to their product. Assume that among this population 1,000 cancers of a certain type could be expected, but that 1,100 exist, and that this increase is "statistically significant," permitting a reasonable conclusion that 100 cancers are due to the product of the manufacturers. In the absence of other evidence, it might be argued that as to any one of the 1100 there is only a chance of about 9% (100/1100) that the product caused the cancer. Under traditional tort principles no plaintiff could recover. (b) Inadequacy of Individualized Solutions Any attempt to resolve the problem on a plaintiff-by- plaintiff basis cannot be fully satisfactory. The solution that would most readily suggest itself is a burden shifting approach, analogous to that used in the indeterminate defendant situation already discussed. Allen v. United States provides a good example of how burden-shifting would be applied in an indeterminate plaintiff case. A plaintiff must show that the defendant, in that case the United States, negligently put "an identifiable population group" of which he was a member at "increased risk" and that his injury is consistent with having been caused by the hazard to which he has been negligently subjected, such consistency having been demonstrated by substantial, appropriate, persuasive and connecting factors . . . . Allen, 588 F. Sup. at 415. At that point, the burden shifts to the defendant which will be held liable unless it can offer "persuasive proof" of noncausation. Id. Generally courts have shifted the burden to the defendant to prove that it was not responsible for plaintiff's injury only in sporadic accident cases where it was certain that one of a very limited number of defendants injured the plaintiff, see, e.g., Summers v. Rice, 33 Cal.2d 80, 199 P.2d (86, 154 P.2d 687 (1944), or in mass exposure cases where general causation was certain and liability was apportioned in accordance with some market share theory. See. r.g., Sindell v. Abbott Laboratories, 26 Cal.3d 588, 607 P.2d 924, 163 Cal.Rptr. 132, cert. denied, 449 U.S. 912, 101 S.Ct. 285, 66 L.Ed.2d 140 (1980), Copeland v. Celotex Corp., 447 So.2d 908 (Fla.App. 1984). But see Abel v. Eli Lilly & Co., 418 Mich. 311, 343 N.W.2d 164 (1984) (joint and several liability in DES cases). Shifting the burden of proof in such cases will, at least theoretically, not result in crushing liability for the defendant either because the litigation only involves a sporadic accident, as in Summers and Ybarra, or because the defendant will only be held liable for the amount of damage it caused based on market share--although as indicated above, there may be practical problems in defining market share. By contrasts, shifting the burden of proof in the indeterminate plaintiff situation could result in liability far out of proportion to damage caused. It is not helpful in most situations to say that the defendant will not be liable for "those harms which [he] can reasonably prove were not in fat a consequence of his risk-creating, negligent conduct," Allen, 588 F.Sup. at 415, since, were such individualized proof available, there would have been no need to shift the bur- (3) Possible Solution in Class Action Since the problem results from plaintiff-by-plaintiff methods of adjudication, one solution is to try all plaintiffs' claims together in a class action thereby arriving at a single, class-wide determination of the total harm to the community of plaintiffs. Given the necessarily heavy reliance on statistical evidence in mass exposure cases, such a determination seems feasible. The defendant would then be liable to each exposed plaintiff for a pro rate share of that plaintiff's injuries. This approach can be illustrated using the hypothetical given above. Suppose all 1,100 of those who were exposed to the harmful substance and who developed the cancer in the example join in a class action against all 10 manufacturers. Let us say that damages average $1,000,000 per cancer. A recovery of $100,000,000 (100 x $1,000,000) in favor of the class would be allowed with the percentage of the ward to be paid by each manufacturer depending on the toxicity of its product. For example, if a company produced only 20% of the substance in question but, because of the greater toxicity of its product, likely caused 60% of the harm, it would contribute 60% of the total amount. If accurate records are available on the composition of each defendant's product, that analysis should be possible. Since no plaintiff can show that his or her cancer was caused by any one of the defendants [illegible] giving each a recovery of about $90,000. While any plaintiff might feel that his or her recovery denigrated the degree of harm, the alternative of receiving nothing is far worse. The latter is, of course, the necessary result in any plaintiff's individual suit. Moreover, the deterrent effect of this result on producers would be significant. See Delgado, Beyond Sindell: Relaxation of Cause-in-Fact rules for Indeterminate Plaintiffs, 70 Cal.L.Re. 881, 893 (1982). If the number of cases were only 1,050, most statisticians would say the difference was not statistically significant and a court using the pro rata approach might find that the defendant is not legally responsible for any of the increased incidence. But we Allen v. United States, 588 F.Sup. at 416-17 (noting that although increased incidence might be deemed "insignificant" by a scientists or statistician, it may well be that it "is still far more likely than not" that "the observed increase is related to its hypothetical cause rather than mere chance"). (In the numbers used, the standard deviation could be computed by square root as the square root of 1000 equals 31.6. Two standard deviations--a rough test of statistical significance often used in the law--is 63.2; 50 is less than two standard deviations. Yet, assuming the 100 variation of the hypothetical (more than 3 standard deviations), the law should nonetheless attribute to the defendants al 100 additional cancers, not just 100 minus 63 or 37 (the total variation less two standard deviations). As a matter of rough justice, once legal responsibility for the increased incidence is found, it is sounder to attribute all 100 cases to the defendant, even though a substantial number of these cases may be random variations with no reasonable assurance that they are attributable to the defendants' activities. There is, of course, also the possibility that more than 100 cases were "caused by" defendant's activities since the figure properly attributable to background cancer incidence might have been less than 1000. We are in a different world of proof than that of the archetypical smoking gun. We must make the best estimates of probability that we can using the help of experts such as statisticians and our own common sense and experience with the real universe. Putting a dollar amount on the damages suffered by individual plaintiffs is, from a real-world standpoint, a critical part of the solution. If the judicial and monetary economies of the class action are not to be lost through lengthy and expensive individual trials on damages, some mechanism must be devised to decide damage claims without the need for a full-fledged trial of reach plaintiff. As Professor rosenberg points out, "[p]ossibly the greatest source of litigation expense [in mass exposure tort litigation] is the individual assessment and distribution of damages that must follow trial of common liability question." Rosenberg, supra, 97 Harv.L.Rev. at 916. How individualized such a mechanism would have to be depends on (1) the size of the individual claim and (2) what the variations between plaintiffs are in the nature of the claims: the smaller the individual claim and the less the variation, the more generalized the process can be. If their claims are for one type of injury, a compensation schedule to calculate average loss could be developed based on sampling techniques. See Manual for Complex Litigation 2.712 at 116-18 (1982). If a number of different [illegible] Every effort should be made to reduce questions of fact to a bare minimum. A preferred solution is to pay claims on a fixed and somewhat arbitrary schedule using a ministerial agency as is donee with the Medicaid and Medicare programs where disbursements are made by insurance agencies acting for the United States. No matter what system is used the purpose is to hold a defendant liable for no more than the aggregate loss fairly attributable to its tortious conduct. As long as that goal is met a defendant can have not valid objection that its rights have been violated. (a) Analogy and Precedent (i) The Employment Discrimination Cases It was noted above that courts have only recently been faced with fact patterns involving mass exposure and indeterminate plaintiffs. That is not to say that there is no relevant guidance or precedent. In the employment discrimination area courts have long dealt with cases involving a defendant who through his acts exposed an entire class to harm resulting in an injured class with indeterminate plaintiffs. The solutions developed and the reasoning used by some courts in those cases parallel that outlined above for the mass exposure fact patterns. The defendant employer has acted wrongfully through his discriminatory acts. As here, however, it is often impossible to prove whether the plaintiff's condition, in that case the lower salary or lack of promotion or employment, is due to the defendant's wrongful acts or whether it would have occurred even absent in the discriminatory acts. For example, if class members outnumber the openings for job promotions it is undeniable that some class members would not have been promoted even if there had been no discrimination. The solution adopted by some courts parallels the pro rate approach suggested above for mass exposure cases: awards are based on the "average" pay and rank of a group of employees, not injured by discrimination, comparable in all other material respects, such as ability and length of employment. In the employment discrimination cases the impossibility of identifying individual class members is recognized. The courts have noted that although "[t]here is no way of determining which jobs the class members would have . . . obtained if [the] discriminatory [system] . . . had not been in existence [because] [c]lass members outnumber promotions cannot be awarded." [i]t does not follow that back pay claims based on promotions cannot be awarded." Pettway v. American Cast Iron Pipe Co, 494 F.2d 211, 260 (5th Cir. 1974). "Doubts about which individuals . . . would have been promoted" call for "a class approach." Id. at 261 n. 150. [T]he total award for the entire class would be determined. At that point, individual claims would be calculated on pro rata shares for those workers of similar ability and seniority claiming the same position, possibly eliminating the necessity of deciding which one of many employees would have obtained the position but for the discrimination. Id. at 263 n. 154. Here, too, the "total [injury] to the entire class [can] be determined" and, "[a]t that point, individual claims could be calculated on pro rata shares for those" exposed plaintiffs "claiming the same" injury, "eliminating the necessity [illegible] which one of many" plaintiffs would not have developed the injury "but for" the exposure. Cf. P.J. Spiegelman, Court-Ordered Hiring Quotas After Stotts: A Narrative on the Role of Feminine And Masculine Coives in the Legal Process, 31ff, (1984). The "Agent Orange" case is especially well-suited to the application of such an approach. Not only are all producers of the allegedly harmful substance before the court, but so is almost every person allegedly injured who could now be before an American court. While, for example, American civilians and Vietnamese residents may have been injured, the statutes of limitations have probable run on their claims. (iii) Consumer Class Actions There is another area in which courts have dealt with fact patterns involving large numbers of injured plaintiffs where it is possible to determine the damage on a class-wide basis, but it is not feasible to determine the amount of damage suffered by each class member and to compensate him for that damage. Such situations arise where defendants have sold a product to or performed a service for, numerous persons, sometimes ranging into the millions and have illegally overcharged many or all of those persons. While each class member's damages are often quite small, the defendant has reaped a substantial wrongful gain. Because most plaintiffs will not come forward to claim their damages due to the small amounts involved, courts are faced with the choice of either allowing the defendant to keep the wrongful gain, or as a number of courts have done, to distribute the damages to as many members of the class as is practicable. See, for example, Eisen v. Carlisle & Jacquelin, 52 F.R.D. 253 (S.D.N.Y. 1971), where the defendant brokerage firms were accused of monopolizing odd-lot trading on the New York Stock exchange and of charging excessively high brokerage commissions. The court in Eisen suggested that one possible method of distributing the damages would be to set up "a fund equivalent to the amount of unclaimed damages . . . and [to reduce] the odd-lot differential . . . in an amount determined reasonable by the court until such time as the fund is depleted." Eisen, 52 F.R.D. at 265. This type of solution is sometimes referred to as a "fluid recovery," i.e., one which avoids the requirement of individual identification of class members and individual distribution of damages, in favor of a remedy for the next best class, e.g., macro-statistical description of the class harm remedied by future reduction in prices or rates charged by the defendant. 3B J.W. Moore, Moore's Federal Practice 23.45(4.-4)(1984). As in the solution proposed for the indeterminate plaintiff situation, "the damage issue [is tried] only once," with the result that "the amount of damages arrived at is likely to correspond to the total injury inflicted by defendant or the extent of its "unjust enrichment." 7A C. Wright & A. Miller, Federal Practice and Procedure 1784 (1972). Courts must choose between compensating large numbers of demonstrably injured people in an imperfect manner or not compensating them at all. It should be emphasized that we are not concluding that the analysis and solutions is the employment discrimination and consumer class action cases compel a similar result in the mass exposure cases. There are obviously differences between the fact patterns. It also appears undisputed that no court, except perhaps as part of a settlement, has ever applied the pro rata plan in mass exposure cases. Given the present reluctance of many courts to certify mass tort litigations as class actions, it is likely that courts, at least for a number of years, will not be faced with a request to adopt such an approach. Se the analysis of cases in re "Agent Orange" Product Liability Litigation, 100 F.R.D. 718 (E.D.N.Y. 1983), mandamus denied, 725 F.2d 858 (2d Cir. cert. denied -- U.S. --, 104 S. Ct. 1417, 79 L.Ed.2d 743 (1984). The application of such a scheme, although consistent with the accepted goals of tort law, requires a fundamental rethinking of a number of well established legal concepts. Nevertheless, the employment discrimination and consumer class action cases reflect a pragmatic judicial solution to analogous problems, providing a clue to the direction courts may properly take in resolving mass exposure cases. (b) Practical Advantages of a Class-wide Solution In addition to the strong legal arguments in favor of dealing with the indeterminate plaintiff fact pattern in a class action, there are strong practical considerations as well. The most important of these was already noted in passing--namely, the judicial and economic efficiencies that result from class-wide treatment. This factor has already been alluded to in the opinion certifying plaintiffs as a class. See In re "Agent Orange" Product Liability Litigation, 100 F.R.D. 718 (E.D.N.Y. 1983). Although recent studies suggest that defendants in mass tort cases spend more than plaintiffs do for legal services, see. J. Kahalik, P. Ebener, W. Reistiner & M. Shanley, Costs of Asbestos Litigation, Rand Corp. The Institute for Civil Justice, July 1988), defendants, often large corporations, are usually better able to bear the cost of litigating individual suits than are plaintiffs' attorneys who are generally financing the litigation out of their own pockets. A class action, by allowing plaintiffs' claims at once, should not only reduce the overall cost of the litigation significantly but should tend to equalize the odds between the two sides. Cf. Note. The Inapplicability of Traditional Tort Analysis to Environmental Risks: The Example of Toxic Waste Pollution Victim Compensation, 35 Stan.L.Rev. 575, 586 (1983). A class action would also facilitate settlement by allowing defendants to buy their peace with all of the plaintiffs at once. See, e.g., "Agent Orange,: 100 F.R.D. at 721. One the negative side, making it easier to sue may increase the possibility of obtaining unjustified settlements by a form of legal coercion. By litigating the issue of liability only once courts can avoid the danger of inconsistent verdicts that could easily result in those cases where the increased incidence of the disease hovers around 100%. The danger is especially acute given the number of possible confounding factors and the fact that [illegible] data and statistical studies are rarely as reliable and complete as they ideally could be. It is, therefore, altogether possible that different conclusions on similar or identical evidence despite the fact that their interpretations of the evidence might not differ appreciably. The transaction costs of obtaining a remedy are a proper consideration in determining substantive law. If much of a recovery will go to attorneys and experts rather than to those injured, then traditional tort remedies may be so ineffective as to put in doubt their utility in particular types of cases. Punishment of defendants who cause harm and deterrence of future harmful conduct is by- product of the traditional tort system, but it should not independently furnish the rationale for private civil litigation. Where, therefore, the transaction costs of obtaining a remedy for a class are much less per dollar recovered than they would be in a case-by-case recovery, the class actionmay, as a matter of policy, be the only reasonable route to recovery. Added to the costs to defendants and plaintiffs must be the cost to the public and the courts in burdening the system with large numbers of private, individual claims. Thus, while the class action is deemed procedural and distinct from substantive considerations for most purposes, it may become, in a case like "Agent Orange," the only practicable way to secure a remedy.