dissenting:
I. Introduction.
The district court and the majority undoubtedly are driven by a commendable desire to resolve voluminous personal injury claims against an otherwise strong American company and to ensure an orderly transfer of funds from the company’s insurers to its victims. In order to accomplish this result, however, they have extinguished claims over which they have no jurisdiction and deprived thousands of asbestos victims of basic constitutional rights. The result is the first no-opt-out, mass-tort, settlement-only, futures-only class action ever attempted or approved.
Ironically, the willingness to jettison centuries-old legal precepts hurts the very victims they intend to help: The settlement forces asbestos victims to surrender their claims in exchange for a meager $10 million of Fibreboard’s $225-250 million net worth. They also benefit from Fibreboard’s settlement with its insurers, but Fibreboard and the insurers had powerful incentives to settle that dispute by themselves; in fact, they did so for $2 billion.
There was no need even to involve the class in those negotiations, much less to sacrifice its interests. “Thus, the class members appear to have traded Fibreboard’s liability for nothing to which they did not already have a right.”1
On the other hand, the district court and the majority have bailed Fibreboard’s shareholders out of a mammoth liability and awarded $43.7 million to class counsel. This suit was supposedly brought on behalf of Fibreboard’s victims, but of the four entities directly affected by the settlement — Fibreboard, class attorneys, courts, and asbestos victims — the victims were the only entity absent from the bargaining table. Perhaps for that reason, they also were the only losers.
How could well-intentioned judges sanction — indeed, compel — such an untoward result? Apparently this is simply a case of judges — both trial and appellate — trying too hard to solve the vexing problems posed by unending asbestos litigation. Having certified at least two other high-profile asbestos class actions,2 then-Chief District Judge Parker was acutely aware of the problems posed by asbestos litigation. In the end, he appears to have become too close to bojh the overall problem and the instant settlement to continue to act in a judicial capacity in this case.3
*994When Fibreboard and class counsel announced at a court hearing that they had reached a settlement, Chief Judge Parker referred to “extensive negotiations between counsel that the Court has participated in.” Also at that time, and long before the fairness hearing, he said, “We will trust in the scholarship, the good judgment and common sense of the ... courts of appeal in the event this comes to their attention.” In short, Chief Judge Parker tried his best to solve a perplexing problem, and it is our task to figure out whether that solution is legally sustainable.
There are two primary problems: (1) Fibreboard, class counsel, and Fibreboard’s other creditors have combined to profit at the expense of absent class members; and (2) this case is an affront to the integrity of the judicial system. As we observed when reversing Chief Judge Parker’s certification of another class action against Fibreboard: “The Judicial Branch can offer the trial of lawsuits. It has no power or competence to do more.” Fibreboard, 893 F.2d at 712.
A. Importance and Uniqueness.
This case is extraordinarily important. Prior to the filing of this suit, no one had ever attempted a no-opt-out, mass-tort, settlement-only, futures-only class action. Ever since the district court’s certification order, however, corporate America has been “anxiously awaiting” a decision in this case. Richard B. Schmitt, The Deal Makers: Some Firms Embrace the Widely Dreaded Class-Action Lawsuit, Wall St.J., July 18,1996, at Al. The majority’s unequivocal approval of Fibreboard’s litigation strategy undoubtedly will lead “other financially threatened companies throughout the nation [to] utilize it as a road map for sheltering their assets and improperly restricting the rights of their present and future victims.” Amicus Br. of Trial Lawyers for Public Justice at 2-3.
Thus, the majority’s reliance upon the “unique facts” of this case, see maj. op. at 993, is ironic: The unique fact of the insurance dispute is simply irrelevant, and the other unique facts — a corporate defendant’s hand-picking class counsel, cutting a side deal, reaching a “global settlement” affecting only “future” plaintiffs, and choosing a sympathetic judge to approve the settlement— likely will become far too common now that the majority has approved of them. “[W]hat was meant to provide a remedy for those who would otherwise lack one, enabling them to pool their voices and finances, will become a device to take away remedies from those who could otherwise invoke them.” John Leubsdorf, Co-Opting the Class Action, 80 Cornell L.Rev. 1222,1223 (1995).
B. The Need for Procedural Protections.
Two primary errors led the district court and the majority astray. These are, first, underestimating the importance of jurisdictional and procedural protections for absent class members, and second, departing from the judiciary’s exclusive area of authority and competence — the resolution of lawsuits.
We must keep in mind that it was the defendant — Fibreboard—who selected the class that was to “sue” it and the class action lawyers who were to do the dirty work. Fibreboard hand-picked a class that was uniquely vulnerable to exploitation, class counsel who were widely reported to have sold out a similar class, and a court with a reputation for favoring a global settlement. Class counsel then cut a side deal with Fibreboard before agreeing to the class settlement, and the district judge presided at the fairness hearing on the very settlement he had helped to craft.
The settlement extinguishes claims of people over whom we lack jurisdiction, some of whom have not yet been injured and others of whom have not even been born. It also prevents such future claimants from opting out, because of a supposed need to divide a limited fund among a large number of claimants, but it grants automatic opt-outs to all those who already had filed suit. Coincidentally or not, this gerrymandered class definition includes those most vulnerable to abuse while excluding those most likely to inter*995vene, to monitor class counsel, and to oppose the settlement.
It is fair to question for whom class counsel really worked. Fibreboard picked them, the district court approved them, the insurers paid them, and in exchange, they bailed out Fibreboard’s shareholders and relieved district courts of potentially thousands of eases- — at the expense of the absent asbestos victims whom class counsel purportedly represent.
If all that was at stake for individual class members was some nominal compensation for having been charged an extra five cents on a bag of potato chips, one might not be too concerned with how the courts enforced class counsel’s duties to these people. But often much more is at stake, such as whether a plaintiff will recover for a fatal illness caused by a defective product, and if so, how much. Today, such a person may have her rights adjudicated by a court without actual notice of the action and before she even .knows she has been injured.
Fantastic as this may seem ..., it is true.4 But it need not be. Even rudimentary constitutional protections — such as according absent class members adequate representation and adjudicating only their presently-existing, legally cognizable injuries — -would have prevented Fibreboard from perpetuating this unfortunate miscarriage of justice.
C. Legislated Tort Reform.
The district court legislated a bold and novel tort reform proposal thinly disguised as the settlement of a lawsuit. Of course, there never was a lawsuit: Fibreboard and its hand-picked class counsel agreed to file a suit only if they already had settled it. Thus, Chief Judge Parker began his opinion by stating, “This action was filed to obtain judicial approval of a class settlement.” Ahearn v. Fibreboard Corp., 162 F.R.D. 505, 507 (E.D.Tex.1995).
The class complaint alleges exposure-only claims for which the settlement provides no compensation. Class counsel even conceded that, as a matter of practice, they do not pursue such claims on behalf of their own clients; instead, they wait and file suit after a plaintiff actually has suffered an injury. The only reason to include those claims in the complaint was to manufacture jurisdiction over class members who have not yet manifested symptoms of asbestosis or otherwise suffered a legally cognizable injury. Even that attempt to trump up jurisdiction should fail, however, as many states do not recognize an exposure-only cause of action.
Moreover, the settlement does not resolve the rights of individual class members. The only genuinely judicial aspect of approving the settlement is the release of Fibreboard from liability to the class, or more specifically, the transfer, from Fibreboard’s shareholders to its victims, of the risk that Fibreboard’s insurance assets are inadequate.
The remainder of the settlement is purely legislative: Class members’ causes of action are repealed in favor of the equivalent of a workers’ compensation regime.5 The Association of Trial Lawyers of America summed *996up this point nicely in an amicus brief opposing the settlement: “The alchemy of the [instant] settlement ... had the effect of transforming the common law damage claims of asbestos victims, which were clearly safeguarded by the right to trial by jury, into administrative claims without that right.” Amicus br. at 9.
Even if exchanging state tort law for this private, alternative dispute resolution mechanism were the boon to class members that the majority holds it out to be — and I doubt that it is, see infra part IX — such a policy decision is “better addressed to the representative branches — Congress and the State Legislature.” Fibreboard, 893 F.2d at 712. In addition, Fibreboard hardly deserves more than $200 million for drafting the legislation.
“[Traditional ways of proceeding reflect far more than habit.” Fibreboard, 893 F.2d at 710. As judges are trained and equipped to adjudicate, not legislate, it is understandable that the courts have fallen prey to powerful special interest groups — a wealthy defendant and the class action bar — and unwittingly disserved the very victims the courts were intended to help.
D.Constructive Bankruptcy.
Nor does Fibreboard’s “constructive bankruptcy” justify abridgment of absent class members’ substantive state law rights. In bankruptcy, the claims of all of Fibreboard’s creditors, not just its “future” personal injury victims, would be crammed-down. Permitting Fibreboard to effect a reorganization bankruptcy proceeding in the guise of a futures-only class action circumvents the detailed protections of the Bankruptcy Code for the express purpose of imposing the entire cost of the bailout on Fibreboard’s most vulnerable creditors, to the betterment of its shareholders.
The Second Circuit decertified a similar settlement class for precisely that reason:
Evasion of bankruptcy is ... not without costs or other perils.... [C]lass members in eases such as this would have no say in the conduct of the court-appointed class representatives and, unlike creditors in bankruptcy, are not able to vote on a settlement. For them, it would be “cram-down” from start to finish.
Keene Corp. v. Fiorelli (In re Joint E. & S. Dist. Asbestos Litig.), 14 F.3d 726, 732 (2d Cir.1993) (citation omitted). The amicus brief of the Trial Lawyers for Public Justice puts the point more forcefully: “[I]nstead of protecting class members from the risk that their ability to obtain relief from Fibreboard will be ‘substantially impaired,’ certification of the proposed settlement class here ensures that the class members’ ability to obtain relief from Fibreboard will be totally eliminated,.” Amicus br. at 6.
E.Creating a Circuit Split.
Our sister circuits have rejected all other actions that came even close to attempting what Fibreboard has done here. The Ninth Circuit has squarely held that opt-out rights are available in all class actions seeking predominantly monetary damages, regardless of the subsection under which they were certified. See Brown v. Ticor Title Ins. Co., 982 F.2d 386, 392 (9th Cir.1992), cert. dismissed, 511 U.S. 117, 114 S.Ct. 1359, 128 L.Ed.2d 33 (1994). Two other circuits appear to agree with the Ninth, and none has expressly disagreed. See infra note 16. Without even citing that authority, however, the majority arbitrarily limits opt-out rights to actions certified under Fed.R.Civ.P. 23(b)(3), see maj. op. at 987 n. 16, exalting an irrelevant technicality over the underlying reality and creating a circuit split in the process.
Earlier this year, the Third Circuit firmly held that class counsel cannot adequately represent both extant and latent claimants in a futures-only asbestos class action. Georgine v. Amchem Prods., 83 F.3d 610, 630-31 (3d Cir.1996). The court explained that while extant claimants — those who have already incurred injuries — desire immediate, unlimited recovery from the trust, latent claimants — those who have yet to suffer an injury — desire that recovery be capped or delayed to ensure that extant claimants will not deplete the fund. Id.
The majority mentions Georgine only in a brief footnote, distinguishing it on the ground that the Georgine settlement provides a de*997tailed claims resolution schedule, while the Aheam settlement does not. See maj. op. at 976 n. 8. Class counsel still served conflicting interests, however, and postponing some distributional issues until after certification and appeal hardly makes them disappear. The majority may prefer the devil it does not know to the devil it does, but I am loath to make that decision for an entire class of people who are not even aware that we are “adjudicating” their rights.
On the other hand, the majority is correct that Keene is easy to distinguish, for the defendant in that action was forthright: Instead of retaining plaintiffs’ counsel and having them file a complaint asserting claims they had no intention of pursuing (as occurred here), the asbestos manufacturer asked the court to oversee the negotiation of a settlement. See Keene, 14 F.3d at 728-29. The Second Circuit dismissed the action, finding that “it is a self-evident evasion” of the Bankruptcy Code, “the exclusive legal system established by Congress for debtors to seek relief.” Id. at 732. Future defendants presumably will draw one of two conclusions: Involve the court as little as possible in settlement class actions, or file in the Fifth Circuit.
In sum, the settlement fails either a customary legal analysis or a common-sense smell test. I respectfully but vehemently dissent from all but part III of the majority opinion.
II. Facts and Procedural History.
Though the background to this case is somewhat complicated, the key facts are hard to overlook. Fibreboard approached four plaintiffs’ lawyers, including Ron Motley and Joe Rice, and suggested that they negotiate a “global settlement” of all of Fibreboard’s asbestos liabilities. The negotiations initially failed, perhaps because of the massive scope of the undertaking.
Fibreboard then adopted a risky strategy of assigning claims against its insurers in settlement of individual suits. The danger was that these settlements arguably violated the insurance policies. Fortunately for Fibreboard, a California court approved the deals.
Then something odd happened: Fibreboard settled a large number of cases with Ness Motley — Motley and Rice’s law firm— by assigning insurance assets, and brought an action in the Eastern District of Texas seeking approval of the settlement. Why would Fibreboard, a California company, roll the dice in Texas when it had already won in California? Because something important had happened in Pennsylvania.
The Judicial Panel on Multidistriet Litigation had transferred all pending asbestos cases not yet on trial to a district court in Pennsylvania. See Georgine v. Amchem Prods., 83 F.3d 610, 619 (3d Cir.1996). Then-Chief Judge Robert Parker of the Eastern District of Texas wrote a letter to the transferee judge, telling him that he (the Pennsylvania judge) was “the Eisenhower of this D-Day operation” and encouraging him to prod the parties to a global settlement. See Coffee, supra note 1, at 1390.
When the plaintiffs’ steering committee rejected such a proposal, twenty defendants approached a minority faction of the committee — Motley and Gene Locks — and reached a global settlement with them. See id. at 1391-92, 1457. Actually, they made a series of deals: a class action settlement for future asbestos victim claimants and separate settlements for the lawyers’ pre-existing, individual clients.
The separate settlements were significantly more lucrative than the class one. See id. at 1392-93; Koniak, Feasting, supra note 4, at 1052. In fact, Motley received fifty percent more for his own clients than he did for those in the class. See Coffee, supra note 1, at 1397; Koniak, Feasting, supra note 4, at 1067. The Third Circuit rejected the settlement, finding that class counsel — including Motley — were hopelessly conflicted. See Georgine, 83 F.3d at 630-31.
As the Georgine negotiations concluded, Fibreboard and Ness Motley settled a number of cases and, as noted above, filed an action in Chief Judge Parker’s court. Fibreboard thereby secured class counsel with a track record of making global settlements and a judge with a demonstrated commitment to them.
*998Following the Georgine pattern, class negotiations reached an impasse over the future of Ness Motley’s remaining cases against Fibreboard. The court-appointed “Settlement Facilitator,” Judge Patrick Higginbotham, then suggested that they settle those cases before attempting further negotiation of a global settlement. After concluding the Ness Motley deal — which settled the individual claims for higher-than-average amounts, contingent upon successful completion of a global settlement — Fibreboard and class counsel resumed negotiation of such a settlement.
As those negotiations drew to a close, Chief Judge Parker intervened, taking counsel to his house for a final mediation session. It appears that he was successful, for defense counsel eventually increased their offer to an amount that class counsel later accepted.
Class counsel then filed a complaint in Chief Judge Parker’s court, along with motions to certify the class and approve the settlement. The judge certified the class and found that the settlement was fair.
And Fibreboard’s stock soared.6
III. The Forest for the Trees.
The majority commits two fundamental errors: first, treating justiciability, due process rights, and certification criteria as mere annoyances to be brushed aside in pursuit of what it believes to be the greater good; and second, failing to assess the aggregate effect of its restrictions on asbestos victims’ due process rights.
Justiciability and certification requirements are indispensable in any class action. Justiciability looks, among other things, to whether a person has suffered a legally cognizable injury. If an individual has not been legally injured, it is unlikely that he would receive notice of the action or realize that he is a member of the class. Even if he became aware of the action’s potential effect on his legal rights, he would likely remain apathetic: Any effect on him is remote in time and contingent on the future development of a disease or other damage or injury. Thus, such class members are especially vulnerable to abuse by class counsel.
Similarly, certification criteria such as commonality, typicality, and adequacy of representation ensure that representative litigation is truly representative. If class counsel stand to gain from selling out the class or from benefiting one subgroup of claimants over another, some or all class members are deprived of a meaningful opportunity to be heard — one of the most fundamental of all due process rights.
The majority addresses each class protection device in isolation, always finding that the protection does not apply because of a legal rule developed in a different context or an historical analogy that fails to recognize the novelty of this action. Such tunnel vision obscures the fact that while a particular protection might not always be necessary, some combination of protections is. When courts remove all meaningful safeguards — as the majority does here — class members suffer dramatically.
A. The Danger Inherent in Representative Litigation.
“It is a principle of general application in Anglo-American jurisprudence that one is not bound by a judgment in personam in a litigation in which he is not designated as a party....” Hansberry v. Lee, 311 U.S. 32, 40, 61 S.Ct. 115, 117, 85 L.Ed. 22 (1940). This “deep-rooted historic tradition that everyone should have his own day in court,” 18 Charles A. Wright, Arthur R. Miller, Edward H. Cooper, Federal Practice and Procedure (hereinafter “Wright & Miller”) § 4449, at 417 (1981), creates the core of due process: the rights to notice, to control one’s own case, and to an opportunity to be heard.
The class action device is an equitable exception to this bedrock principle. See Hansberry, 311 U.S. at 41, 61 S.Ct. at 117-18. Not surprisingly, such a “fundamental departure from the traditional pattern in Anglo-American litigation generates a host of problems.” Mars Steel Corp. v. Continental Ill. Nat’l Bank & Trust Co., 834 F.2d 677, 678 (7th Cir.1987). More bluntly, “class ac*999tions are extraordinary proceedings with extraordinary potential for abuse.” General Motors Corp. v. Bloyed, 916 S.W.2d 949, 953 (Tex.1996).
Accordingly, the Constitution’s guarantee of due process requires us to use this joinder device carefully: In exchange for losing the right to prosecute his own action, a class member must receive a variety of substitute protections. See Phillips Petroleum Co. v. Shutts, 472 U.S. 797, 811-12, 105 S.Ct. 2965, 2974-75, 86 L.Ed.2d 628 (1985); Hansberry, 311 U.S. at 45, 61 S.Ct. at 119-20. Though the Supreme Court has largely refrained from determining the scope of those protections, its scant jurisprudence establishes two related principles: First, the extent to which due process requires procedural protections necessarily depends upon the extent to which class members’ interests are infringed;7 and second, we must meet new uses of the device with new protections.8
In short, the safeguards required by due process necessarily differ according to the type of action, and when confronted with a new animal, we must analyze those safeguards anew. Reliance on strained analogies to inapposite traditional actions leaves us in what one commentator has aptly labeled a “due process quandary.”9 Thus the irony: The majority eviscerates well-established due process protections because of the “unique facts” of the case, see maj. op. at 993, but fails to recognize that those novel facts may actually call for enhanced, not lessened, protections for vulnerable asbestos victims.10
B. Vulnerability of the Class.
This case is indeed such a new animal. The district court certified a class
(1) including people who have not yet been injured or do not yet know that they have been injured;
(2) excluding all present claimants;
(3) for settlement purposes only;
(4) in a mandatory action seeking predominately monetary damages.
Certification of futures-only actions creates a massive potential for abuse. Many putative future claimants have manifested no symptoms and do not even know they were exposed. Others are the future spouses and children of asbestos victims, most of whom either do not exist or could not possibly know that they are class members. Thus, the due process standbys — notice and an opportunity to be heard — are meaningless to countless future claimants.11
*1000Moreover, future claimants are, by definition, persons who have not yet developed a sufficient interest in their claims to file suit; thus, they are likely to be passive and particularly vulnerable to exploitation.12 Finally, courts have a rotten track record with futures-only actions: Of the two largest such actions, one is Georgine, and the other settlement fell apart because the parties radically underestimated the number of claimants. See Coffee, supra note 1, at 1417-18 (discussing failure of Dow Corning settlement).
These concerns are mitigated somewhat by the breadth of the “futures” class. Some members are presently injured and aware of their injuries, and some have even spoken with lawyers. While most of these claimants might not have retained counsel for the sole purpose of intervening, some might have, and others at least might have chosen to opt out, had that protection not been removed as well.13 Of course, any such intervenors protect only their own interests, which differ dramatically from those of class members who are not presently injured. See infra part VI.A.2. Thus, futures-only classes are still highly vulnerable to abuse.
This case also presents a radical extension of the mandatory class action. The class complaint seeks, and the settlement provides, predominately monetary relief. While historically we have permitted mandatory actions when a class sought to litigate joint rights regarding a common fund, the only common fund in this case is the settlement proceeds. To the extent that there is a limited fund, it is a contrived one, created by the litigation and settlement strategies of Fibreboard and its insurers. See infra note 17.
The concept of a futures-only mandatory action is also self-contradictory. If we must bind victims in the class in order to protect their rights and ensure an equitable distribution, then we must bind all such victims, not just some. Limiting the class to future claimants grants the equivalent of an automatic opt-out to present claimants, and there is simply no principled way of distinguishing the one group from the other.14
Arbitrariness in the class definition might not present a problem by itself, but future claimants, unlike present claimants, are particularly vulnerable. Exclusion of all present claimants has the effect (if not the purpose) of excluding all those who are likely to receive notice, monitor the class action, and oppose the class attorneys’ conflicts and other inadequacies.15
Finally, this is a settlement class action. Permitting such actions creates an unparalleled opportunity for collusion between defendants and class counsel, as both stand to gain from negotiating a deal providing generous fees for counsel and meager recovery for the class. See In re Gen. Motors Corp. Pick-Up Truck Fuel Tank Prods. Liab. Litig., 55 F.3d 768, 788 (3d Cir.), cert. denied, — U.S. -, 116 S.Ct. 88, 133 L.Ed.2d 45 (1995); Crampton, supra note 4, at 826-27. Moreover, a defendant may pick his opposing counsel and then negotiate with absolutely nothing to lose from walking away from the deal; class counsel, on the other hand, work pro bono unless they consent to a settlement.
If nothing else, use of a mandatory settlement action with an automatic opt-out for all those likely to intervene, and no opt-out for anyone else, raises a red flag. To the best of my knowledge, no one has ever attempted to do such a thing before: Even the now-discredited Georgine settlement permitted opt-outs.
C. Diminished Protection for the Class.
A novel action laden with such an extreme potential for abuse certainly demands a close *1001look, but the majority accepts the settling parties’ distortion of that background reality and actually ratchets down the degree of protection accorded absent class members.
The Aheam settlement is really two agreements: one between Fibreboard and the insurers to settle their policy disputes for $1.525 billion, and another between the futures class and Fibreboard to limit the class’s recovery to insurance proceeds plus $10 million of Fibreboard’s $225-250 million net worth. Fibreboard and the insurers did not need class proceedings to reach the first agreement. Avoiding an all-or-nothing judgment in the California coverage litigation gave them a powerful incentive to settle, regardless of whether they could extinguish future claims at the same time. As the majority observes, “None of the parties was prepared to take the enormous risk inherent in that litigation.” Maj. op. at 993 (emphasis added). In fact, Fibreboard and the insurers did reach such a settlement. See maj. op. part III (unanimously approving that agreement).
With that cloak removed, the second half of the Aheam settlement is wholly baseless. The class members surrendered their claims against Fibreboard, submitted to an arbitration procedure, and agreed to a total cap on damages, individual caps on damages, an absolute ban on punitive damages, and other restrictive provisions. In exchange, Fibreboard gave the class a mere $10 million — less than five percent of its net worth. Fibreboard sought certification based upon a constructive bankruptcy theory, but it walked away with barely a scratch. Not surprisingly, Fibreboard’s stock skyrocketed when the settlement was announced. See Coffee, supra note 1, at 1402 & n. 232.
Even accepting the settling parties’ mischaraeterization of the settlement does little to justify certification, however, for traditional class protections still prevent it.
TV. Due Process and the Right To Opt Out.
The majority’s treatment of opt-out rights is a paradigmatic example of its erroneous reasoning. Though the Court plainly held in Phillips Petroleum Co. v. Shutts, 472 U.S. 797, 812, 105 S.Ct. 2965, 2974-75, 86 L.Ed.2d 628 (1985), that class members have a right to opt out of actions (such as this) seeking primarily monetary relief, the majority refuses to recognize that right, on the ground that it was not historically available in traditional “common fund” litigation.
The majority’s holding that all rule 23(b)(1) class actions are immune from Shutts is in direct conflict with the holding of a sister circuit.16 Moreover, this case presents anything but a traditional common fund,17 and the majority’s attempt to analo*1002gize it to an action to settle a trust is entirely unjustified: Far from adjudicating equitable rights in a preexisting fund, the settlement creates a common fund by extinguishing personal rights of action. In short, the majority rationalizes its evisceration of a right that we have already recognized with a call to historical rigidity.
A. Basic Requirements of the Shutts Case.
Shutts could not be more unambiguous:
[W]e hold that due process requires at a minimum that an absent plaintiff be provided with an opportunity to remove himself from the class by executing and returning an “opt out” or “request for exclusion” form to the court.
472 U.S. at 812, 105 S.Ct. at 2974-75 (citations and footnotes omitted). The Court, however, specifically limited its holding to claims “for money damages or similar relief at law,” as opposed to actions seeking “equitable relief.”18 Thus, Shutts teaches that in an action seeking money damages, an absent plaintiff is entitled to the right to opt out.19
Following Shutts, a mandatory class action is viable in two cases: (1) where the court has jurisdiction over all the plaintiffs or (2) where the plaintiffs seek equitable relief. Unlike the majority, I believe that whether a class action seeks equitable relief or money damages can be determined only by focusing on the underlying remedy the plaintiffs seek. See, e.g., Findley I, 982 F.2d at 735 (affirming certification of a mandatory class action but noting that it would violate due process if the request for relief were for money damages rather than division of a trust).
If one focuses on the plaintiffs’ remedy, the Aheam class cannot be characterized as one “seeking equitable relief.” The complaint alleges only personal causes of action against Fibreboard for money damages.20 The prayer for relief seeks general and special compensatory damages, punitive damages, costs of the suit, appropriate declarations and orders, and other relief as may be deemed just and proper. These remedies *1003represent paradigmatic legal remedies.21 1 DaN B. Dobbs, Dobbs Law of Remedies (hereinafter “Law of Remedies”) §§ 3.1-3.2 (2d ed. 1993). The majority, maj. op. at 976, agrees: “The central remedial and legal theory of each of the named plaintiffs [is] that Fibreboard is liable in tort for damages incurred due to exposure to Fibreboard asbestos.”
B. The Majority’s Circumvention of Shutts.
The majority circumvents Shutts by calling Aheam an equitable action.22 Instead of focusing on the plaintiffs’ remedies, the majority characterizes the action based on which subsection of rule 23 the court invoked to certify the class. From the historical fact that rule 23(b)(1)(B) originated in courts of equity, the majority concludes that all actions certified under rule 23(b)(1)(B) are equitable and thus immune from Shutts.
1. Conflict with Other Circuits.
The majority’s holding is in direct conflict with decisions of the Third and Ninth Circuits. See Brown; In re Real Estate Title. The Ninth Circuit has directly held that Shutts applies to rule 23(b)(l)-(b)(2) actions when the plaintiffs’ underlying claims are for money damages. Brown, 982 F.2d at 392. The Third Circuit also has recognized that a mandatory class action does not have a res judicata effect on absent, class members when the underlying claims are for money damages and the court did not have personal jurisdiction over the plaintiffs. In re Real Estate Title, 869 F.2d at 768-69. The majority fails to discuss either case. Both cases arose out of the same antitrust class action. See Brown, 982 F.2d at 388-89. The original class complaint sought money damages and injunctive relief and was certified under rule 23(b)(1) & (b)(2). In re Real Estate Title, 869 F.2d at 760, 763. The action settled on terms favorable to the defendant, and a number of absent plaintiffs were left without a settlement check.
The first ease challenging the no-opt-out class action arose by way of injunction in the Third Circuit. Following settlement of the class action, a number of absent plaintiffs filed a new suit against'the same defendants for the same activity. The defendants asked the district court that heard the original class action to enjoin the Arizona state court suit. In re Real Estate, 869 F.2d at 762. The district court granted the injunction, but the Third Circuit reversed, classifying the original class action as a hybrid suit seeking both money damages and equitable relief. Id. at 768-69. The court applied Shutts and held that the district court did not have personal jurisdiction over the absent plaintiffs in the original action and that it would violate due process for the court to enjoin a collateral attack on the class action. Id.
The other shoe dropped in a subsequent action filed in federal court in Arizona. The defendants raised a res judicata defense to the plaintiffs’ claim, pointing to the class action settlement. The Ninth Circuit rejected the defendants’ arguments, holding that the class action did not have a res judicata effect on absent plaintiffs over whom the district court in the original class action did not have personal jurisdiction. Brown, 982 *1004F.2d at 392. The court relied on Shutts and held that the absent plaintiffs’ money damage claims could not be foreclosed by a rule 23(b)(1) class action, because the action did not include a right to opt out.23
2. Conflict with Shutts.
The majority’s approach to classifying this action is also fundamentally at odds with the approach used by the Shutts Court. The exception in Shutts is for actions seeking “equitable relief,” not for class actions that have their origins in equity.24 The Court explicitly recognized that class actions originated as an equitable joinder device, and all class actions are “equitable” in that limited sense. Shutts, 472 U.S. at 808, 105 S.Ct. at 2972-73. See also Hansberry v. Lee, 311 U.S. 32, 41, 61 S.Ct. 115, 118, 85 L.Ed. 22 (1940) (“The class suit was an invention of equity____”); 7A Wright & Miller § 1751, at 7. Despite that fact, the Court classified the action in Shutts as one for money damages. 472 U.S. at 811, 105 S.Ct. at 2974.
The only possible conclusion that can be drawn from the Court’s reasoning is that the distinction between damage remedies and equitable remedies turns on what remedies the plaintiff seeks, not on his choice of joinder devices. To hold otherwise, as the majority does in this case, would create an internal inconsistency in Shutts: If the origin of the class action is determinative of the classifieation of an action, then every class action, including the one in Shutts, would be an equitable action.
S. Misunderstanding Equitable Remedies.
The majority’s assertion that rule 23(b)(1)(B) actions are unique in that they involve only equitable claims is based on an apparent misunderstanding of “equitable remedies” and the history of the class action. Prior to the merger of law and equity, the majority would have been correct in classifying all rule 23(b)(1)(B) class actions (or more appropriately their historical antecedents) as equitable actions, but for reasons different from those the majority offers.
Before law and equity were merged, every class action would have been equitable, as the class action was a procedural device that was available only in courts of equity.25 As a result, every class action aggregated claims seeking equitable remedies or applying equitable substantive law. The fact that the device was once used exclusively by equity courts does not foreclose the possibility, however, that the device currently is used in conjunction with legal remedies.26
Even the majority’s own authorities recognize that class actions are now used to aggregate both legal and equitable claims. See 1 Newberg on Class Actions § 1.18, at 1-46 (recognizing that suits certified under rule *100523(b)(1) “predominately (but not exclusively) involve suits seeking equitable relief’). When Professor Newberg posits that Shutts will not apply to classic limited fund cases, he does so because those cases aggregate equitable claims, not because the device had its origins in equity.
It is unlikely that Shutts will be construed to change the highly focused nature of the equitable relief addressed in these limited fund class actions by requiring opt-out rights which might serve to frustrate the equitable ends of these suits. There are other types of equitable class actions, such as suits to declare a dividend or suits for injunctive relief against a defendant who has acted generally against a class, which similarly ... will not be challenged.
Id. § 1.21, at 1-49 (emphasis added, footnotes omitted).
L Formalistic Distinctions.
The majority’s unyielding mantra that rule 23(b)(1) actions are equitable actions makes due process turn on an untenable formalistic distinction between rule 23(b)(1) and rule 23(b)(3), whereunder class actions under rule 23(b)(3) are subject to Shutts, and those certified under rule 23(b)(1) are not. The distinction finds no support in Shutts, which involved a state equivalent of rule 23(b)(3) and not the federal rule itself. The Court’s analysis, which centered on the need to protect an absent plaintiff from the inherent dangers of representative actions, did not focus on what type of class action was involved, for relying on such a distinction ignores the fact that many class actions can be certified under more than one subsection of rule 23. 1 NewbeRG on Class Actions § 4.01, at 4-4 through 4-5; 3B J.W. Moore Et Al., Moore’s Federal Practioe ¶23.31[3], at 236-37 (2d ed. 1995); 7A Wright & Miller § 1772, at 425. In fact, a vast majority of classes that meet the requirements of rule 23(b)(1)(B) inevitably will satisfy the rule 23(b)(3) requirements. 1 Newberg on Class Actions § 4.01, at 4-5. Due process does not turn on such formalistic distinctions.27
C. The Distinction Between Law and Equity.
The majority’s argument that applying Shutts to rule 23(b)(1) would render all mandatory class actions unconstitutional, is also premised on a faulty understanding of the distinction between law and equity. See maj. op. at 987 n. 16. What the majority fails to acknowledge is that the distinction between money damages and equitable remedies preserves mandatory class actions in the vast majority of cases, including traditional common fund cases from equity. See Newberg on Class Actions § 1.18, at 1-46 (recognizing that the majority of actions that meet the requirements of rule 23(b)(1) seek equitable relief). The only cases affected by Shutts are modern eases, such as Aheam, that fall within the broad scope of the rule but do not involve a true common fund.
Take for example Smith v. Swormstedt, 57 U.S. (16 How.) 288, 14 L.Ed. 942 (1853), and Hartford Life Ins. Co. v. Ibs, 237 U.S. 662, 35 S.Ct. 692, 59 L.Ed. 1165 (1914), two cases that represent the paradigmatic mandatory class action. Both survive Shutts because neither involved money damages. In Smith, the class representatives filed a bill in chancery for the division of a trust, a traditional equitable remedy.28 57 U.S. (16 How.) at *1006312. See 1 Law of Remedies § 2.3(1), at 75. In lbs, the mandatory class action at issue also involved distributions from a trust.29 237 U.S. at 671, 35 S.Ct. at 695. Although both actions resulted in a distribution of money, neither involved a legal claim for money damages. See 1 Law of Remedies § 3.1 (discussing money awards available in equity, but distinguishing between money damages and equitable remedies that award money).
The difference between legal and equitable actions, while subtle, is important. In an equitable action such as an action by the beneficiary of a trust, a plaintiffs pro rata rights to the trust corpus are already established. Whether the action proceeds individually or as a class action is irrelevant to absent plaintiffs, because in both cases their rights in the corpus of the trust will be affected. See In Re Joint E. & S. Dist. Asbestos Litig. (“Findley II”), 878 F.Supp. 473, 530-33 (E. & S.D.N.Y.1995) (explaining why the already severe limitations on a trust beneficiary mean that a representative action will do little to diminish a beneficiary’s rights). In fact, in some eases the class device enhances an absent plaintiffs rights by providing him a voice; the classic limited fund is such a case. The same is not true in a case for money damages.
The decision to proceed individually or as a class makes all the difference in a case for money damages, because only in a class action are a plaintiffs rights controlled by another party. The most obvious consequence of such control is that class representatives may settle to the detriment of absent class members in order to benefit themselves. The inherent risk of such a result is exactly what drove the Court to adopt due process protections in such cases. See Shutts, 472 U.S. at 809-11,105 S.Ct. at 2973-74.
D. Unconstitutional Use of the Federal Rules.
Finally, the majority’s reliance on the strong presumption of constitutionality that the Federal Rules of Civil Procedure receive is misplaced. The novelty in this case is not the insistence on opt-out rights but on the extension of class actions, particularly rule 23(b)(1)(B) class actions, to mass torts and the expansion of rale (b)(1)(B) to “constructive bankruptcy.” Such a use of the federal rules is novel, and in particular the rise of mass torts was unforeseen by the drafters of the rule. Castano v. American Tobacco Co., 84 F.3d 734, 746 & n. 23 (5th Cir.1996).
The majority’s attempt to hide behind the traditional acceptance of mandatory class actions is unavailing in this extension of rule 23(b)(1), for the presumption of constitutionality enjoyed by the Federal Rules of Civil Procedure disappears when the rules are applied in a way unanticipated by their drafters. See 19 Wright & Miller § 4509. In short, while this action falls within the broad scope of rule 23(b)(1)(B), that fact does not immunize it from due process protections, including opt-out rights.30
*1007 V. Commonality and Typicality.
The majority is correct that our precedents preclude us from decertifying the class for lack of commonality and typicality. The majority’s discussion of these criteria is misleading, however, and its readiness to skip over the protections without considering the consequences is yet another manifestation of its tunnel vision.
A. Circuit Precedent.
Long before defense attorneys even dreamed of mandatory, mass-tort, futures-only class actions, we held that “tentative or temporary settlement classes are favored when there is little or no likelihood of abuse, and the settlement is fair and reasonable and under the scrutiny of the trial judge.” Meat Price Investigators Ass’n v. Iowa Beef Processors (In re Beef Indus. Antitrust Litig.), 607 F.2d 167, 174 (5th Cir.1979), cert. denied, 452 U.S. 905, 101 S.Ct. 3029, 69 L.Ed.2d 405 (1981). We even went so far as to “agree” with Professor Newberg that parties may “compromise their differences, including class action issues, through this means.” Id. at 177-78 (quoting 3 Newberg on Class Aotions § 5570c, at 476). We took Beef Industry to its logical conclusion two years later, finding that evaluation of the fairness of a settlement entails a consideration of “the strength of plaintiffs’ case on the merits,” including “the risks of class decertification.” Adams Extract Co. v. Pleasure Hours, Inc. (In re Corrugated Container Antitrust Litig.), 643 F.2d 195, 216 (5th Cir.1981).
Note the bizarre effect of these opinions: Certification criteria are designed in part to protect the class against inadequate representatives, but Beef Industry permits those very representatives to compromise the requirements. . Corrugated Container compounds the effect by holding that certification criteria are part of the merits of a class action. Thus, when it is less likely that certification is proper, and therefore more likely that an individual plaintiff has a right to prosecute his own action, a smaller settlement is necessary to extinguish that right.
The majority is more subtle, finding that all class members have a common interest in reaching a settlement that includes certain terms. See maj. op. at 975. Even so, a generalized common interest is not a “question,[] of law or fact common to the class.” Fed.R.CivJP. 23(a)(2) (emphasis added).
Similarly, the majority finds that the legal and remedial theories of the class representatives are typical of those of the class because all elass members (1) claim that “Fibreboard is liable in tort for damages incurred due to exposure to Fibreboard asbestos” and (2) possess a common interest in settling for a large amount. See maj. op. at 975-76. Once again, the latter category of “interests” has nothing to do with the underlying legal theories.
The majority’s characterization of the plaintiffs’ claims is also true only at an extreme level of generality. If a man raped a woman in California and exposed himself to *1008another in Texas, both women might claim that he is “liable in tort for damages incurred due to [sexual misconduct],” but no one would seriously claim that the legal theory of battery is similar to that for, say, intentional infliction of emotional distress.
Nor have the settling parties even attempted to bootstrap themselves into typicality by crafting a settlement that postpones resolution of non-typical issues. The settlement’s prohibition on punitive damages affects plaintiffs from different states differently; its caps on individual recovery discriminate against the most seriously injured class members; and its spendthrift provisions alter the distribution of funds between extant and latent claimants. Thus, the settling parties have negotiated an alteration of the class members’ substantive rights vis-avis one another. In short, the majority dutifully follows our precedent by treating commonality and typicality as mere procedural obstacles to be cleared out of the way.
B. Problems with Circuit Precedent.
Three factors make our duty to follow Beef Industry and Corrugated Container a regrettable one. First, rule 23(a) has a purpose: to protect the due process rights of absent class members.31 Compliance with the rule might prevent us from lumping people with divergent interests into a single class, but the rule is meant to do just that.32
Second, the difficulty inherent in assessing the substantive fairness of a settlement makes prophylactic procedural protections crucial to the protection of absent plaintiffs.33 The range of “fair” settlements is wide in any case,34 and settlement classes present unique opportunities for collusion between class counsel and defendants.35 Once the machinery of a settlement class action is set in motion, the judiciary has a very hard time controlling it; thus, it is critical that we craft that machinery well.
Finally, the majority once again fails to consider the interaction of these due process protections with the others it denies. Professor Newberg, the primary proponent of settlement classes, justified relaxing the rule 23(a) criteria by looking to other protections, most notably opt-out rights, to protect the class. See 2 Newberg ON Class Actions § 11.28, at 60-61. Similarly, the Beef Industry court observed that opt-out rights would help “offset” the loss of other protections. See 607 F.2d at 175. Thus, relaxing rule 23(a) criteria and denying opt-out rights make for a lethal combination.36
*1009 VI. Inadequate Representation.
That leaves two significant due process requirements, the first of which is adequate representation. Instead of vigorously enforcing this requirement, however, the majority lets it slide as well.
A. Standard of Review.
In ordinary litigation, a plaintiff does not need the court to ensure that his attorneys are performing adequately; he can monitor and control them himself. In class litigation, however, class members have no practical control over their attorney.37 Their only remedy against shoddy representation is to opt out of the class, but the district court denied them even that protection. Having left class members defenseless against their own counsel, the court’s duty to ensure adequate representation became all the more critical. Rule 23(a)(4) prohibits a court from certifying a class without first ascertaining that it will receive adequate representation. Rule 23’s concerns are not merely prudential, but constitutional: Due process demands that class members receive adequate representation before they are bound by a judgment.38 An attorney who labors under a conflict of interest cannot satisfy the requirements of rule 23(a)(4).39
We must not confuse rule 23(a)(4) with rule 23(e), which scrutinizes only the substantive fairness of a settlement.40 Rule 23(a)(4) guarantees due process, and a substantively adequate result cannot cure procedural injustice.41
More importantly, substantive adequacy does not necessarily mean that a settlement is truly fair; an attorney who negotiated an adequate settlement might have negotiated a better one but for a conflict of interest.42 *1010Indeed, it would be odd if rule 23(a)(4) did require a court to examine the terms of a settlement; rule 23(a)(4) is a certification requirement, and certification usually precedes settlement. Rules 23(a)(4) and 23(e) are complements, not substitutes; a court applies rule 23(a)(4) before certification to insure procedural fairness and rule 23(e) after settlement to guarantee substantive adequacy.
We apply a mixed standard of review to a district court’s determination that a conflict of interest does not exist. The existence of a particular set of circumstances is a factual determination that we review for clear error. Whether those circumstances amount to a conflict of interest is a legal question that we review de novo. FDIC v. United States Fire Ins. Co., 50 F.3d 1304, 1311 (5th Cir.1995) (“we interpret the controlling ethical norms governing professional conduct as we would any other source of law”) (quoting Dresser, 972 F.2d at 543); see also id. (stating that “we will perform a ‘careful examination,’ or de novo review, of the district court’s application of the relevant rules of attorney conduct”). No factual dispute exists in this case, and therefore we must review the district court’s decisions de novo.
B. Class Counsel’s Financial Conflict.
Ness Motley represented the class during settlement negotiations despite a financial interest in reaching a settlement that diverged from the class’s interest. A divergent financial interest is a paradigmatic conflict of interest, yet the majority refuses to concede that this conflict may have denied the class adequate representation.
The conflict stemmed from Ness Motley’s simultaneous representation of both the class and individual plaintiffs who were also suing Fibreboard. While the class settlement negotiations were proceeding, Fibreboard and Ness Motley agreed to settle all of Ness Motley’s pre-existing individual lawsuits for a higher-than-average amount. That settlement was partly contingent, however, on Ness Motley and Fibreboard’s settling the class claims.
If the class did not settle its claims, Fibreboard would still pay half of the individual plaintiffs’ settlement, but they would receive the entire settlement if the class settled.43 Because Ness Motley represented the individual plaintiffs, but not the class, on a contingency basis, it had a financial interest in settling the class claims on any terms whatsoever, even if those terms were unfavorable to the class.
The majority dismisses this conflict by pretending that Ness Motley, the individual plaintiffs, and the class had similar interests in reaching a class settlement. This is simply wrong.
Ness Motley and the individual plaintiffs had a financial interest in seeing the class settle on any terms. In contrast, the class desired a settlement only if it was more lucrative than the alternative — awaiting the outcome of the coverage ease. The expected value of that alternative was significant. If the California Supreme Court affirmed the two lower courts that had ruled favorably for the class, the class would enjoy a practically unlimited compensation fund. Accordingly, it is easy to imagine a proposed settlement beneficial to Ness Motley and the individual plaintiffs but unfavorable to the class.
The district court never should have allowed Ness Motley to represent the class. Rule 23(a)(4) simply does not permit an attorney to represent a class if he suffers from a conflict of interest. It does not matter whether Ness Motley negotiated a favorable settlement; rule 23(a)(4)’s concerns are procedural, not substantive. Even if, arguendo, the settlement was favorable, an unconflicted attorney might have negotiated a better one. Fibreboard’s attorneys knew that Ness Motley had an interest in settling on any terms, and they undoubtedly converted this knowledge into negotiating leverage.
The conflict of interest in this case was direct and egregious. But I would go a step *1011further and prohibit class counsel from ever simultaneously representing individual plaintiffs in eases such as this, as there is too great an opportunity for corruption.
Consider the different fee structures a plaintiffs’ attorney faces in class and individual litigation. As class counsel, he will be compensated under the lodestar formula,44 leaving him no financial stake in the settlement. When representing an individual plaintiff, he will receive a contingency fee, vesting him with a significant financial stake in the outcome of the litigation.
A shrewd but unethical attorney will accept a significantly smaller settlement in a class action in exchange for a more modest increase in an individual settlement. And even an attorney who would refuse such an outright bribe might be tempted to make concessions during the class negotiations, hoping to develop goodwill that will pay dividends when negotiating over the individual lawsuits.
This is not simply a theoretical problem. Consider the Georgine case, in which Ness Motley also simultaneously represented the class and individual plaintiffs. While negotiating the class settlement, Ness Motley and the Georgine defendants settled the individual claims for a total of about $138 million. Koniak, Feasting, supra note 4, at 1067. Under the terms of the class settlement, on the other hand, the individual plaintiffs would have been entitled to a maximum of about $90 million. Id.
At the least, in Georgine Ness Motley treated the individual plaintiffs preferentially. At worst, Ness Motley obtained a concession from the defendants at the expense of the class. In either case, it acted improperly and to the detriment of the class, and it is to prevent such behavior that I would prohibit simultaneous representation altogether45
C. The Conflict Between Extant and Latent Claimants.
Another conflict of interest arose from the class attorneys’ simultaneous representation of both class members with extant claims and those with latent claims.46 Because they may seek recovery from the settlement fund immediately, extant claimants would profit from high ceilings on recovery or none at all, as there was little danger that the fund would be depleted before it paid their claims. Latent claimants, on the other hand, cannot seek recovery until some triggering event occurs.47 Because they may be unable to file a claim for a number of years, they would benefit from damage caps low enough to ensure that the settlement fund will not dry up before it pays their claim.
The majority fails to recognize this obvious conflict of interest. In fact, it makes no mention of the Third Circuit’s holding in Georgine that an impermissible conflict of interest occurs whenever extant and latent claimants are represented by the same attor*1012ney.48 Georgine’s analysis is worth quoting at length:
The most salient conflict in this class action is between the presently injured and futures plaintiffs. As rational actors, those who are not yet injured would want reduced current payouts (through caps on compensation awards and limits on the number of claims that can be paid each year). The futures plaintiffs should also be interested in protection against inflation, in not having preset limits on how many cases can be handled, and in limiting the ability of defendant companies to exit the settlement. Moreover, in terms of the structure of the alternative dispute resolution mechanism established by the settlement, they should desire causation provisions that can keep pace with changing science and medicine, rather than freezing in place the science of 1993. Finally, because of the difficulty of forecasting what their futures hold, they would probably desire delayed opt out like the one employed in Bowling v. Pfizer, Inc., 143 F.R.D. 141, 150 (S.D.Ohio 1992) (heart valve settlement allows claimants who ultimately experience heart valve fracture to reject guaranteed compensation and sue for damages at that time).
In contrast, those who are currently injured would rationally want to maximize current payouts. Furthermore, currently injured plaintiffs would care little about inflation-protection. The delayed opt out desired by futures plaintiffs would also be of little interest to the presently injured; indeed, their interests are against such an opt out as the more people locked into the settlement, the more likely it is to survive. In sum, presently injured class representatives cannot adequately represent the futures plaintiffs’ interests and vice versa.
Georgine, 83 F.3d at 630-31 (emphasis added).
The majority’s only explanation for its decision is that the district court’s determination that a conflict did not exist was supported by the testimony of the settling parties’ expert, Geoffrey Hazard, and the district court’s own expert, Eric Green. That explanation is hardly compelling.
The district court’s determination that the facts before it did not amount to a conflict of interest was not a finding of fact to which we should defer. Instead, it is a conclusion of law, see United States Fire Ins. Co., 50 F.3d at 1311 (holding that we review de novo a district court’s application of ethical norms governing attorney conduct), and it is not enough simply to note that two hand-picked law professors — one chosen by the settling parties and the other by the district court— had the same opinion of the law as did that court.
Nor is the reasoning behind the legal opinions of Professors Hazard and Green persuasive. According to the majority, both concluded that a conflict did not exist, essentially because both extant and latent claimants would be better off with a settlement than without one. If that was their reasoning, they asked the wrong question.
Even if there was no conflict between extant and latent claimants as to whether a settlement was a good idea, there was undeniably a conflict between the two groups over the terms of the settlement. It may well have been appropriate for the same attorneys to represent extant and latent claimants when negotiating with the defendants over the total amount of the settlement. Once the parties agreed to an amount, however, the interests of extant and latent claimants diverged, and they should have been represented by different attorneys for determination of the terms under which the settlement would be distributed to the class.
D. The Conflict Between Pre- and Post-1959 Victims.
The global settlement achieved a massive redistribution of wealth. Before the settlement, persons who had been exposed to as*1013bestos before 1959 had far more valuable claims than those who had been exposed after 1959. Fibreboard’s insurance policies had expired in 1959, so pre-1959 claimants could seek damages from an insured Fibreboard, with virtually unlimited assets, while post-1959 claimants could only hope to recover from the non-insurance assets of Fibreboard, which amounted to very little. As late as 1993, the settlement value of a pre-1959 claim was three times that of a post-1959 claim.49 The global settlement, however, eliminated the privileged status of pre-1959 claimants and placed them on equal footing with their post-1959 counterparts.
Because the settlement deprived pre-1959 claimants of a substantial right, they were entitled to separate representation. The majority contends that the class as a whole benefited by sacrificing the special rights of pre-1959 claimants, and I emphatically agree. That is precisely the problem: It was in the best interest of one part of the class to give up something that belonged to another part of the class, and that created a conflict of interest.
Unitary class counsel was bound to consider the interests of the class as a whole, and those interests required giving up the rights of pre-1959 claimants. But the rights of pre-1959 victims were theirs alone, and they were entitled to counsel who would not relinquish those rights unless it was in their own best interest. Only separate counsel, representing pre-1959 claimants alone, could have done so.
Nor am I persuaded by the majority’s contention that Fibreboard would never have negotiated separately with pre-1959 claimants and that unitary representation maximized the value of the settlement. So long as the class was negotiating with Fibreboard over the total value of the settlement, there was no need for separate representation, as pre- and post-1959 claimants had the same interest in obtaining as large a settlement as possible. It was only after the parties had agreed on the total amount of the settlement and had begun to bargain over how it should be divided among the class that pre-1959 claimants needed separate representation. There is no reason to suppose this would have any effect on Fibreboard’s willingness to make concessions; it was interested only in the total amount of the settlement, not the allocation of the settlement within the class.
VII. Impartial Judging.
The second due process requirement that the majority fails to enforce is impartial judging. “A fair trial in a fair tribunal is a basic requirement of due process.” In re Murchison, 349 U.S. 133, 136, 75 S.Ct. 623, 625, 99 L.Ed. 942 (1955). Thus, a litigant is entitled to try his case before a judge who is free of any personal stake in the outcome. Tumey v. Ohio, 273 U.S. 510, 532, 47 S.Ct. 437, 444, 71 L.Ed. 749 (1927).
This protection is vitally important in a mandatory class action, for the judge is the only person able to protect class members from an unjust settlement. The class members themselves cannot reject the settlement, and they cannot even reject certification by opting out of the class. They can only present their objections at the fairness hearing and hope the judge will reject the settlement as inadequate. If the judge himself is biased in favor of an unfair settlement, then the class has no protection against it.
The intervenors raise serious questions regarding Chief Judge Parker’s ability to conduct an impartial fairness hearing.50 He had mediated the settlement negotiations, undoubtedly urging the parties to settle on terms he believed to be fair, before the case was even filed. In his lengthy opinion ap*1014proving the settlement, he referred to his “assistance in the discussions” leading to final agreement. 162 F.R.D. at 516.
Having helped to craft the settlement, Chief Judge Parker had a personal stake in finding it to be fair: A determination that the settlement was unfair would imply that he had acted unfairly in helping to craft it. In light of these facts, a person might reasonably conclude that Chief Judge Parker, despite his dedicated and painstaking efforts, probably developed a natural bias in favor of his own work.51
The majority concludes that the judge need not have been recused from the fairness hearing, because the settlement negotiations he mediated were a judicial proceeding. I am not so certain. I agree that we are governed by Liteky v. United States, 510 U.S. 540, 114 S.Ct. 1147, 127 L.Ed.2d 474 (1994), and that if the settlement negotiations were a judicial proceeding, recusal was not required. I also agree that if the negotiations were a judicial proceeding, we must make a case-specific determination: A judge may become so involved in settlement negotiations as to warrant recusal even though the negotiations were judicial proceedings. See id. at -, 114 S.Ct. at 1157 (noting that recusal may be warranted where bias stems from a judicial source if it amounts to “a deep-seated favoritism or antagonism that would make fair judgment impossible”). As the majority points out, Chief Judge Parker’s involvement in this case was not substantial, and recusal is therefore not required unless the settlement negotiations were extrajudicial proceedings.
I part ways with the majority, however, when it comes to deciding whether the settlement negotiations the judge mediated were judicial proceedings. The majority concludes that they were, though its only explanation is to state that “Judge Parker’s role in the negotiation process ... stemmed from three cases filed in his court [i.e., the assignment litigation].” I agree that if the negotiations were a legitimate part of the proceedings in the assignment litigation, they were judicial proceedings. Chief Judge Parker’s judicial authority was limited to the assignment litigation, however, and I do not believe that he could conduct judicial proceedings with respect to Aheam simply because it was somehow related to the assignment litigation.
Remember that the parties had not even filed Aheam yet. Thus, any time the judge spent mediating that dispute — as opposed to the insurance cases filed in his court — was community service, not judging.52
The rules of civil procedure allow judges to participate in settlement conferences. See Fed.R.Civ.P. 16(a)(5) (allowing courts to schedule conference “facilitating the settlement of the case”). Rule 16 settlement conferences are judicial proceedings and generally are not an adequate basis for recusal. Cf. Liteky, 510 U.S. at-, 114 S.Ct. at 1157 (noting that judicial proceedings “can only in the rarest circumstances” form an adequate basis for recusal). The question is whether Chief Judge Parker mediated the settlement negotiations in the course of a rule 16 conference.
If the settlement negotiations he mediated were chiefly negotiations over the assignment litigation, and settling Ahearn was inci*1015dental to settling the assignment litigation, the negotiations probably amounted to a rule 16 conference. If, on the other hand, the primary object of the negotiations was to settle Aheam, the negotiations do not fall within the bounds of rule 16. That rule allows judges to schedule conferences only to settle “the case,” which surely must mean the case filed and pending before the judge. Rule 16 cannot permit a judge to order settlement conferences over cases that are not before him — that would be an exercise of power the Constitution does not permit. See U.S. Const, art. Ill, § 2 (limiting judicial power to actual cases or controversies).
Federal courts are not roving engines of justice careening about the land in search of wrongs to right. Rather, federal courts were designed to be much like all other courts: passive entities resolving only the quarrels which are properly put before them by interested parties and which are within the competence of courts in a tripartite system of constitutional government.
Haitian Refugee Ctr. v. Civiletti, 503 F.Supp. 442, 461 (S.D.Fla.1980), modified, 676 F.2d 1023 (5th Cir. Unit B 1982).
I seriously doubt that Chief Judge Parker was conducting settlement negotiations over the assignment litigation and only incidentally settled Aheam along the way. If so, this is one of the more amazing examples of a tail wagging a dog. I recognize, however, that the content and nature of the settlement negotiations the judge mediated present a question of fact, and I therefore would remand to the district court for findings in this regard.
VIII. Article III Justiciability.
Finally, the majority simply fails to consider a number of difficult justiciability issues. While questions of Article III standing and variances in state law may seem somewhat rarefied, it is important to remember the central purpose of jurisdictional requirements — -to keep us in our place. The constitutionally-assigned task of the federal judiciary is to resolve eases and controversies that Congress and the Constitution have authorized us to adjudicate. When we depart from that role by considering generalized grievances or private dispute resolution mechanisms, we quickly find ourselves acting more as legislative policymakers than as judges.
Failure to honor standing requirements, in particular, can result in devastating consequences for those we intend to help. As standing turns in large part on whether an individual has suffered an injury-in-fact, those who lack standing may be unaware that our legislative dalliances eventually will affect them; thus, they are unlikely to speak up for themselves.
This case presents a particularly egregious example,, for the parties and the district court gerrymandered the class to exclude all those who were sufficiently concerned with their injuries to have filed suit. Thus, the class consists largely of people who are unlikely to monitor class counsel’s performance or challenge the settlement. In a justice system that depends on robust adversarial presentations, that dynamic leaves the judiciary ill-equipped to evaluate the procedural and substantive fairness of the negotiations and eventual settlement.
This case shatters the constitutional limits placed on the authority of the federal courts by Article III, Section 2 of the United States Constitution. We are obligated to consider these issues, sua sponte if necessary, to assure ourselves that the district court properly exercised jurisdiction.53 It is fairly obvious that here, the district court exceeded its jurisdiction in approving the class action settlement.
*1016 A. The Causes of Action.
The class action complaint alleges only-personal causes of action against Fibreboard for money damages.54 The majority agrees: “The central remedial and legal theory of each of the named plaintiffs [is] that Fibreboard is liable in tort for damages incurred due to exposure to Fibreboard asbestos.” Maj.Op. at 975.55 The district court also recognized that the complaint alleges in personam claims.56 Thus, Aheam is a case— admittedly an extraordinary one — involving multiple, in personam, state-law causes of action aggregated by means of rule 23, the federal class action device.
B. The Basis for Subject-Matter Jurisdiction.
The majority does not discuss the basis for subject-matter jurisdiction. The plaintiffs and the defendants, in their joint appellate brief, assert that federal jurisdiction was available under 28 U.S.C. § 1332 (diversity) and 28 U.S.C. § 1335 (interpleader). The district court’s published opinion relies on diversity alone as a basis for federal jurisdiction. See Ahearn v. Fibreboard Corp., 162 F.R.D. 505, 522 (E.D.Tex.1995). In supplemental conclusions of law filed on the same day, however, the district court claimed that federal jurisdiction also is proper under the interpleader jurisdiction statute.
Diversity of citizenship was an arguable basis for subject-matter jurisdiction. Statutory interpleader jurisdiction is not available here, because there is no res or common fund. See Wausau Ins. Cos. v. Gifford, 954 F.2d 1098, 1100-01 (5th Cir.1992).57 Even if interpleader jurisdiction were present, the weighty Article III justiciability issues would still exist.
These Article III issues arise out of the application of Erie R.R. v. Tompkins, 304 U.S. 64, 58 S.Ct. 817, 82 L.Ed. 1188 (1938), to this multi-state class' action. As in cases brought under the ordinary diversity jurisdiction statute, cases brought under statutory interpleader are governed by state substantive law, pursuant to the mandate of Erie. Thus, Erie applies to this class action regardless of whether federal jurisdiction is proper under ordinary diversity, interpleader, or both.58
*1017 C. Rule 23(e) Approval and Article III.
Federal Rule of Civil Procedure 23(e),59 which requires that the dismissal or compromise of a class action be approved by a district court, triggers the inquiry into federal jurisdiction. The act of approving — or disapproving — a class action settlement is an exercise of judicial power by a district court.60
Unlike a private settlement in an ordinary case, a class action settlement requires that a federal district court affirmatively exercise its adjudicative authority. Thus, a judicially approved federal class action settlement is a “judgment.”61 The underlying authority to enter such a judgment is, of course, expressly circumscribed by Article III, which limits the judicial power to “cases” and “controversies.” U.S. Const, art. Ill, § 2.
In approving the class action settlement, the district court purported to release — i.e., extinguish — all claims by class members based on Fibreboard’s liability for its asbestos products. See Ahearn, 162 F.R.D. at 517 (stating that Fibreboard and the insurers would receive “total peace” in return for providing the global settlement monies). Because the court was exercising its Article III judicial power, it did not have the authority to release claims that did not constitute a valid case or controversy at the time it gave its approval.
There are cases either holding or suggesting that in a settlement, courts may release claims that they would lack “jurisdiction” to try. See, e.g., Adams Extract Co. v. Pleasure Hours, Inc. (In re Corrugated Container Antitrust Litig.), 643 F.2d 195, 221 & nn. 39-40 (5th Cir.1981); Epstein v. MCA Inc., 50 F.3d 644, 661-64 (9th Cir.1995), rev’d on other grounds sub nom., Matsushita Elec. Indus. Co. v. Epstein, — U.S. -, 116 S.Ct. 873, 134 L.Ed.2d 6 (1996); Matsushita, — U.S. at-,---, 116 S.Ct. at 877, 879-90. All of these cases, however, refer to dimensions of jurisdiction other than the case-or-controversy requirement. For example, many of them involved federal courts’ approving the release of state-law claims, or state courts approving the release of exclusively federal claims. See, e.g., Corrugated Container, 643 F.2d at 221 & nn. 39-40; Epstein, 50 F.3d at 661-64; Matsushita, *1018— U.S. at-, 116 S.Ct. at 877. None of these cases has advanced the radical proposition that a federal court, in approving a class action settlement — ie., in the exercise of its Article III jurisdiction — can release claims not presenting a case or controversy.
Furthermore, courts often have used the “common nucleus of operative facts” test to determine whether unpled claims can be released. See, e.g., Class Plaintiffs v. City of Seattle, 955 F.2d 1268, 1288 (9th Cir.), cert. denied, 506 U.S. 953, 113 S.Ct. 408, 121 L.Ed.2d 333 (1992); Nottingham Partners v. Trans-Lux Corp., 925 F.2d 29, 34 (1st Cir. 1991). This test is used for determining whether a court may exercise supplemental jurisdiction, and a federal court may exercise supplemental jurisdiction only over claims that are part of the same Article III case or controversy. See 28 U.S.C. § 1367(a) (1994); United Mine Workers v. Gibbs, 383 U.S. 715, 725, 86 S.Ct. 1130, 1138, 16 L.Ed.2d 218 (1966).
Thus, the release-of-claims cases assume that the released claims are justiciable cases or controversies. Those cases do not challenge the key conclusion that a portion of a settlement is void if it purports to release non-justiciable claims.
D. The Class.
The plaintiff class of natural persons certified in this case — the Global Health Claimant Class — is indeed sweeping.62 By its terms, the class definition includes, inter alia, the following categories of persons:
(1) future children of persons exposed to asbestos (“future children”);63
(2) future spouses of persons exposed to asbestos (“future spouses”);64 and
(3) persons who have been exposed to asbestos, who have not manifested an asbestos-related disease, and who had not filed or had dismissed a lawsuit by August 27, 1993 (“exposure-only” claimants).
*1019No person included in the first two categories meets the irreducible constitutional minimum of standing for any claim. Furthermore, some of the persons in the third category lack standing with respect to some or all of their claims. Thus, several of the claims asserted by these categories of persons present no justiciable case or controversy, and the district court therefore erred in deciding to hear them.
The future children, for example, include “persons” who had not been conceived at the time the complaint was filed.65 Common sense alone dictates the conclusion that nonexistent persons cannot have standing to assert in personam causes of action for money damages. In more formal terms, these unconceived plaintiffs cannot prove injury-in-fact and thus cannot establish standing to assert their in personam tort claims. Cf. Lujan v. Defenders of Wildlife, 504 U.S. 555, 560, 112 S.Ct. 2130, 2136, 119 L.Ed.2d 351 (1992) (defining injury-in-faet).
Similarly, future spouses — who may or may not already be born — are, by definition, not yet married to the exposed persons. Common sense once again dictates the conclusion that these persons — who have not been exposed to asbestos themselves — have not yet been injured in fact because they currently have no relation to exposed persons.
Imagine, for example, the absurd hypothetical of a “person” from each of the first two categories filing individual lawsuits today in federal court — i.e., without resort to rule 23. I do not doubt that any federal court would dismiss their claims immediately for want of justiciability. We must not allow the aggregation of individual claims through the class action device to divert our attention from such a fatal defect.66
Aside from these glaring defects, a more subtle problem exists with respect to the third category of persons that I defined above — i.e., the exposure-only claimants. Properly addressing the problem, however, requires a review of certain fundamental principles of Article III jurisdiction and their application to diversity cases.
E. Standing.
1. Standing in Relation to Causes of Action.
It is impossible to analyze standing without reference to a cause of action: “[Standing is gauged by the specific common-law, statutory or constitutional claims that a party *1020presents.”67 Suppose, for example, that a public employer has discriminated against an applicant for employment on the basis of race. Undoubtedly, such an allegation of injury would meet Article Ill’s standing requirements for, e.g., a Fourteenth Amendment equal protection claim.
Nevertheless, that injury may not serve as the predicate for standing to assert any conceivable cause of action. For example, such an injury does not operate to confer standing on the applicant for a violation of the right to be free from unreasonable searches and seizures under the Fourth Amendment. For while he could show injury-in-fact with respect to his equal protection rights, he cannot show injury-in-fact with respect to his rights under the Fourth Amendment. See Lewis v. Casey, — U.S.-,-- n. 6, 116 S.Ct. 2174, 2181-82 n. 6, 135 L.Ed.2d 606 (1996).68
Thus, standing analysis assumes, at a minimum, that the plaintiff has alleged a cause of action created to vindicate a legally protected interest. The cause of action alleged serves as a necessary frame of reference for the standing inquiry — hence the Court’s explicit connection of standing and causes of action in International Primate Protection League v. Administrators of Tulane Educ. Fund, 500 U.S. 72, 111 S.Ct. 1700, 114 L.Ed.2d 134 (1991):
Standing does not refer simply to a party’s capacity to appear in court. Rather, standing is gauged by the specific common-law, statutory or constitutional claims that a party presents. “Typically, ... the standing inquiry requires careful judicial examination of a complaint’s allegations to ascertain whether the particular plaintiff is entitled to an adjudication of the particular claims asserted.”69
In other words, standing analysis does not operate in a vacuum: The allegation of a cause of action frames the inquiry. See Catholic Social Serv. v. Shalala, 12 F.3d 1123, 1125 (D.C.Cir.1994).70
*1021In diversity cases, pursuant to the Ene doctrine, state law defines the substantive causes of action. In an ordinary case, we rarely have to confront the interaction between Erie and federal standing analysis. Regardless of what else one can say about Aheam, however, no one can fairly accuse it of being an ordinary ease.
2. Injury-in-Fact.
In Lujan v. Defenders of Wildlife, the Supreme Court set out the test for Article III standing as follows:
Over the years, our eases have established that the irreducible constitutional minimum of standing contains three elements: First, the plaintiff must have suffered an “injury in fact” — an invasion of a legally protected interest which is (a) concrete and particularized; and (b) “actual or imminent, not ‘conjectural or hypothetical.’ ” Second, there must be a causal connection between the injury and the conduct complained of — the injury has to be “fairly ... traee[able] to the challenged action of the defendant, and not ... th[e] result [of] the independent action of some third party not before the court.” Third, it must be “likely,” as opposed to merely “speculative,” that the injury will be “redressed by a favorable decision.”
504 U.S. at 560-61, 112 S.Ct. at 2136 (footnote and citations omitted). For the purposes of this case, we need focus only on the first prong of this standard: injury-in-fact.
The Lujan injury-in-fact test consists of two sections:
(1) the “invasion of a legally protected interest”; and
(2) the criteria of
(a) concreteness and particularity; and
(b) actuality or imminence.
The failure of a plaintiff fully to meet the requirements of both sections constitutes the failure to establish the irreducible constitutional minimum of standing.
Thus, to survive the Article III inquiry, a plaintiff must make- — inter alia — a showing that he had a legally protected interest that has been invaded. The failure to make such showing would be fatal to his attempt to establish Article III standing.
3. Effect of State Law.
The second part of the Lujan injury-in-fact test is governed purely by federal law (even in diversity cases), for no pronouncement of state law can affect a federal court’s determination of whether an invasion of a legally protected interest is, e.g., sufficiently concrete and particularized to satisfy Article III. Similarly, no decision of state law can affect a federal court’s determination of whether such an invasion is sufficiently actual or imminent to satisfy Article III.
In diversity cases, however, the first part of the Lujan injury-in-fact test is not completely independent of state law. The concept of “invasion of a legally protected interest” immediately invites the question of who or what protects the interest. In an ordinary, non-diversity case, federal law is the source of substantive rights and of causes of action designed to vindicate those rights. In diversity cases, it is state law that creates legally protected interests and provides causes of action to vindicate them.
This conclusion cannot be avoided by chanting the mantra that federal law cannot depend on state law. Under Erie, the judgment of a federal court depends on state substantive law in a diversity case. The definition of a legally protected interest — as well as the provision of a cause of action to vindicate that interest — is a matter of substantive law.71 If the federal injury-in-fact *1022test articulated in Lujan depends on the definition of a legally protected interest — and it does — then the outcome of that federal inquiry depends in part on how state law defines that interest.
Moreover, when a state has declined to create a given cause of action, it has decided not to accord legal protection to a particular interest. Therefore, if a given exposure-only cause of action embraced in the settlement is not founded on a substantive right,72 it is impossible for a person asserting that claim to prove injury-in-fact. This is so because the injury-in-fact test requires proof of the invasion of a legally protected interest. See Lujan, 504 U.S. at 560, 112 S.Ct. at 2136; cf. International Primate, 500 U.S. at 77, 111 S.Ct. at 1704 (stating that substantive claim is a predicate for standing analysis).73
j. Choice of Law Analysis.
The causes of action alleged in Aheam are state-law causes of action brought on the basis of diversity jurisdiction. Standing must be measured by the cause of action pled, see International Primate, 500 U.S. at 77, 111 S.Ct. at 1704; therefore, the standing of the exposure-only plaintiffs must be gauged by the causes of action they assert.
All of the exposure-only claimants have a common predicate for their causes of action: Their claims, by definition, are based on mere exposure to asbestos, not on an extant injury caused by exposure to asbestos. The states vary, however, in the degree to which they recognize a cause of action predicated on exposure alone — i.e., they vary in the degree to which they legally protect an interest in being free from exposure to asbestos.
In Pennsylvania, for example, “asymptomatic pleural thickening is not a compensable injury which gives rise to a cause of action.” Simmons v. Pacor, Inc., 543 Pa. 664, 674 A.2d 232, 237 (1996). Pennsylvania does not recognize, in the absence of manifest injury, a legally protected interest either.in (1) being free from the increased risk of cancer because of exposure to asbestos or in (2) being free from present emotional distress resulting from exposure to asbestos (i.e., distress caused by the fear of contracting cancer). See id. 674 A.2d at 237-38.74 Consequently, *1023because Pennsylvania does not legally protect those interests, a plaintiff asserting such causes of action could not possibly establish injury-in-fact under the standing test announced in Lujan — i.e., he cannot demonstrate the invasion of a legally protected interest.
States other than Pennsylvania also have declined to recognize all of the exposure-only causes of action embraced in the instant global settlement.75 Such unauthorized claims are not founded upon legally protected interests. Thus, because those claims cannot properly be before an Article III court, they cannot be extinguished by the Ahearn global settlement. Therefore, because the global settlement purports to extinguish those claims, reversal or vacatur is required.
It was necessary to conduct a choice of law analysis to determine the answer to the threshold question of standing in this case. More specifically, the district court should have conducted the choice of law analysis to determine whether every class member alleged the invasion of a legally protected interest — per the Lujan injury-in-fact test— with respect to each claim they asserted. The failure to do so resulted in the district court’s exercising its powers over claims not presenting a case or controversy — i.e., in the absence of Article III jurisdiction.76
*1024 F. Erie, Diversity, and Article III.
The majority’s failure to raise and address the Article III issues is directly contrary to Guaranty Trust Co. v. York, 326 U.S. 99, 65 S.Ct. 1464, 89 L.Ed. 2079 (1945), and Angel v. Bullington, 330 U.S. 183, 67 S.Ct. 657, 91 L.Ed. 832 (1947). In Guaranty Trust, the Court faced the question of whether a state statute of limitations barring a class action was to be applied by a federal court sitting in diversity. See Guaranty Trust, 326 U.S. at 100-01, 65 S.Ct. at 1465-66. The Court expressly stated that Erie implicates the jurisdiction of Article III courts: “Our starting point must be the policy of federal jurisdiction which Erie R. Co. v. Tompkins embodies.” Id. at 101, 65 S.Ct. at 1466 (citation omitted).
Recognizing that the Erie doctrine “[ijnevitably” applies to suits in equity, the Court framed the issue presented as follows:
Is the outlawry, according to State law, of a claim created by the States a matter of “substantive rights” to be respected by a federal court of equity when that court’s jurisdiction is dependent on the fact that there is a State-created right, or is such statute of “a mere remedial character” which a federal court may disregard?
Id. at 107-08, 65 S.Ct. at 1469 (citation omitted) (emphasis added). Thus, the Court explicitly recognized that the viability of a state-law cause of action was an essential predicate of diversity jurisdiction.
The Court’s holding in Guaranty Trust was unambiguous:
Plainly enough, a statute that would completely bar recovery in a suit if brought in a State court bears on a State-created right vitally and not merely formally or negligibly. As to consequences that so intimately affect recovery or non-recovery a federal court in a diversity case should follow State law.
Id. at 110, 65 S.Ct. at 1470. Thus, a state’s decision to limit the life of a state-created cause of action must be respected by federal courts sitting in diversity. It is equally *1025plain — indeed, it follows a fortiori — that when a state has declined to give life to a cause of action at all, federal courts sitting in diversity also must refuse to entertain that cause of action.
In fact, we have interpreted a decision of the Supreme Court as specifically stating that federal courts sitting in diversity must not entertain diversity actions that are unavailable under state law. See Kuchenig v. California Co., 350 F.2d 551, 556 (5th Cir. 1965) (interpreting Angel), cert. denied, 382 U.S. 985, 86 S.Ct. 561, 15 L.Ed.2d 473 (1966). In deciding whether indispensability of a party was a matter governed by state or federal law in a diversity case, we applied an analysis in Kuchenig that closely parallels appropriate the Erie /standing analysis in Aheam.
We first noted that “every court that has dealt with th[at] issue at all seems to have treated it, appropriately in our view, as a run-of-the-mine [sic] Erie problem, requiring the usual balancing of substantive and procedural elements.” Kuchenig, 350 F.2d at 555. We stated further:
[I]ndispensability, while not properly regarded as a jurisdictional issue, is closely related to jurisdiction, and may act to defeat diversity jurisdiction .... [R]ules of joinder depend on the substantive rights and liabilities of the parties, present and absent. In diversity actions, these substantive rights and liabilities are creatures of state law.
Id. at 555-56 (footnotes and citations omitted, emphasis added). We thus recognized an inevitable relationship, in diversity cases, between state substantive law and federal jurisdiction. We also noted that this relationship was not confrontational, but symbiotic: “Professor [Charles Alan] Wright characterizes the conflict in these cases as ‘more apparent than real’: state law determines the interests of the parties; federal law determines whether these state-created rights render a missing person indispensable.” Id. at 556.77
This approach to indispensability of a party — a matter “closely related to jurisdiction” that “may act to defeat diversity jurisdiction,” id. at 555 — parallels the approach I have applied above in relation to standing. State law determines the substantive rights and interests of the exposure-only claimants. When a state has conferred a substantive right, federal law determines whether persons asserting such a right meet Article III requirements.
When a state has not conferred such a right, it is not within the power of a federal court to entertain that action in diversity. See id. at 556. The district court’s diversity jurisdiction in this multi-state class action depended on the existence of state-created causes of action. See Guaranty Trust, 326 U.S. at 107-08, 65 S.Ct. at 1469 (stating that a federal court’s diversity jurisdiction is “dependent on the fact that there is a State-created right”). In order to determine whether it had jurisdiction over all of the claims released in the class action settlement, the district court had to conduct a choice of law inquiry to determine which states’ laws applied to which causes of action. Any claims not authorized by state law should have been explicitly excised from the settlement. The adjudication by the district court of those claims violated Article III limits on that court’s jurisdiction.
Under these circumstances, we have no discretion: We must reverse or vacate the judgment approving the class action settlement. In reviewing the approval of a class action settlement, we may not modify the terms of the settlement but must approve or disapprove it in its entirety. Cotton v. Hinton, 559 F.2d 1326, 1331-32 (5th Cir.1977).
*1026The reason for this rule is straightforward: Class action settlements often include delicate compromises, the disruption of which would lead one or more of the parties to reconsider the wisdom of the settlement. Therefore, if an appellate court negates part of a class action settlement — e.g., because of justiciability problems — it would have to vacate or reverse and remand for further proceedings. It could not simply excise the objectionable portion of the settlement and uphold the remainder as a viable settlement. See id.
IX. Summary and Conclusion.
Unencumbered by legislative safeguards and shedding nearly every judicial protection, the district court enacted a novel, untested tort reform package. As a result, Fibreboard’s victims find themselves guinea pigs in a dubious (and legally unfounded) experiment.
The only protection accorded the class was a rule 23(e) fairness hearing. The district court also appointed a guardian ad litem to represent absent class members, but he did so only after class and defense counsel had completed the settlement. Thus, class members received absolutely no structural or procedural protections; instead, they had to rely on an after-the-fact review of the settlement’s substance.
The district court and the guardian ad litem undertook that task diligently, but an after-the-fact substantive review is far too little, far too late. The court cannot conduct a trial in order to avoid one; nor can it turn back the clock and appoint different counsel to renegotiate the settlement fairly. Thus, the extent to which class counsel’s numerous conflicts and Fibreboard’s stacking of the deck actually affected the final settlement is unknowable. As Fibreboard entered the negotiations in constructive bankruptcy and left with more than ninety-five percent of its assets intact, however, there is reason to be skeptical.
The effect of replacing the tort system with an administrative processing center is equally hard to ascertain, for judges lack legislative fact-finding and investigative capabilities. If the trust proves to be funded adequately and managed fairly, it might process claims more efficiently than the courts, reducing transaction costs and providing plaintiffs with faster and more reliable recovery. As such a reduction in transaction costs would generate a surplus for Fibreboard and the class,- Fibreboard might deserve to walk away with over $200 million in remaining assets.
On the other hand, the trust might attempt to impose arbitrary limits similar to those of the Georgine trust and stonewall plaintiffs’ counsel who protest, forcing them to endure a tedious series of procedural delays before their clients finally receive a day in court. The trust might also be inadequately funded, as was the Dow Corning settlement, leaving plaintiffs scraping for what little they can get while bureaucrats struggle to hold on to their jobs.
In short, we simply do not know what the courts have wrought. What we do know is that this “reform” involves denial of established constitutional rights; relaxation of already lax ethical rules; extinguishing of claims that we have no power to adjudicate, much less abolish; and a significant likelihood of collusion between the defendant and the class counsel.
I respectfully dissent from the majority’s refusal to reverse — or veto — certification of this no-opt-out, mass-tort, settlement-only, futures-only class action.
. John C. Coffee, Jr., Class Wars: The Dilemma of the Mass Tort Class Action, 95 Colum L.Rev. 1343, 1420 (1995).
. See In re Fibreboard Corp., 893 F.2d 706 (5th Cir. 1990) (granting writ of mandamus); Jenkins v. Raymark Indus., 782 F.2d 468 (5th Cir.1986) (affirming certification).
.For example, certain of the appellants make much of a gathering Chief Judge Parker arranged at his house during which, allegedly, *994counsel — especially the insurers' — were hounded into settling. I reach no conclusion regarding the details of this episode except that it demonstrates both that the proceedings in this case were unusual and that Chief Judge Parker was aggressively involved in the settlement.
. Susan P. Koniak, Through the Looking Glass of Ethics and the Wrong with Rights We Find There, 9 Geo. J. Legal Ethics 1, 13 (1995). The Aheam and Georgine settlements have received significant attention in the academic literature, most of it extremely negative. See, e.g., Coffee, supra note 1, at 1393-1404; Roger C. Crampton, Individualized Justice, Mass Torts, and “Settlement Class Actions": Ah Introduction, 80 Cornell L.Rev. 811, 825-35 (1995); Susan P. Koniak, Feasting While the Widow Weeps: Georgine v. Amchem Products, Inc., 80 Cornell L.Rev. 1045 (1995); Richard L. Marcus, They Can’t Do That, Can They? Tort Reform via Rule 23, 80 Cornell L.Rev. 858, 898-900 (1995).
. The purported "back-end opt-out right" likely will prove to be no right at all. Before he may even file a lawsuit, a victim must (1) file a claim with the trust and wait for it to evaluate his claim; (2) engage in settlement discussions; (3) proceed to mediation; and (4) participate in nonbinding arbitration.
Even after securing a court judgment in his favor, the claimant may not enforce that judgment; instead, he must accept installment payments over a number of years. His recovery is capped at a pre-set dollar amount, and he is barred from receiving punitive damages or pre- or post-judgment interest. In short, the settlement ensures that the trust can make trial an impracticable method of recovery, forcing class members to settle within the confines of the administrative procedure devised by Fibreboard and class counsel.
. See. Coffee, supra note 1, at 1402 & n. 232.
. See Shutts, 472 U.S. at 808-11, 105 S.Ct. at 2972-74 (finding that opt-out right, rather than opt-in requirement, adequately protects class members in light of burdens imposed on them).
. See Mulleme v. Central Hanover Bank & Trust Co., 339 U.S. 306, 313-14, 70 S.Ct. 652, 656-57, 94 L.Ed. 865 (1950) (balancing interests of class members and efficient operation of modem investment trusts). The majority's assertion that “[t]he rule that adequate representation is all that due process requires for the traditional mandatory class action in equity was not challenged by Shutts," maj. op. at 986, is erroneous. The Hansberry court reserved judgment on what other procedures might be required, see 311 U.S. at 43-44, 61 S.Ct. at 118-19, and the Court later found that members of mandatory classes have an additional due process right to adequate notice, see Mullane, 339 U.S. at 318-19, 70 S.Ct. at 659-60. Nor did Shutts construe Hansberry so narrowly. See Shutts, 472 U.S. at 808-09 n. 1, 105 S.Ct. at 2972-73 n. 1 ("The holding in Hans-berry, of course, was that petitioners in that case had not a sufficient common interest with the parties to a prior lawsuit such that a decree against those parties in the prior suit would bind the petitioners.”).
. We need not consider the outer limits of due process in this case, however, as the district court failed to employ even basic protections such as opt-out rights and adequate representation.
. See Ivy v. Diamond Shamrock Chems. Co. (In re Agent Orange Prod. Liab. Litig.), 996 F.2d 1425, 1435 (2d Cir.1993) (observing that "providing individual notice and opt-out rights to persons who are unaware of an injury would probably do little good”), cert. denied, 510 U.S. 1140, 114 S.Ct. 1125, 127 L.Ed.2d 434 and cert. denied, 510 U.S. 1140, 114 S.Ct. 1126, 127 L.Ed.2d 434 (1994); see generally Marcus, supra note 4, at 889 (explaining that the notice given to Aheam class members was particularly hard for them to understand).
. See Crampton, supra note 4, at 828.
. Two groups of plaintiffs intervened in this action. Thus, while exclusion of present claimants appears to have limited the opposition, it did not completely eliminate it.
. See Koniak, Feasting, supra note 4, at 1058; Crampton, supra note 4, at 829-30.
.In addition, the settlement’s silence regarding the actual compensation that claimants can expect — other than various caps and limitations on recovery — makes it difficult for class members and courts to evaluate the settlement. That lack of information might be one of the reasons that the intervenors chose not to attack the settlement's substantive fairness on appeal.
. The Ninth Circuit has held that a class action certified under rule 23(b)(1) and (b)(2) cannot bind absent plaintiffs unless they are allowed to opt out. Brown v. Ticor Title Ins. Co., 982 F.2d 386 (9th Cir. 1992) (holding that absent plaintiffs were not bound by a rule 23(b)(1) — (b)(2) class action for money damages, because the original class action court did not have personal jurisdiction over the plaintiffs and did not provide them with an opt-out right), cert, dismissed as improvidently granted, 511 U.S. 117, 114 S.Ct. 1359, 128 L.Ed.2d 33 (1994). See also In re Real Estate Title & Settlement Servs. Antitrust Litig., 869 F.2d 760 (3d Cir.) (reversing an injunction and allowing a collateral attack against a rule 23(b)(1)— (b)(2) class action to proceed in a different jurisdiction), cert, denied, 493 U.S. 821, 110 S.Ct. 77, 107 L.Ed.2d 44 (1989). In dictum, the Second Circuit has recognized that a rule 23(b)(1) action that aggregates claims seeking legal remedies requires an opt-out right. In re Joint E. & S. Dist. Asbestos Litig. (Findley I), 982 F.2d 721, 735 (2d Cir.1992) (upholding a mandatory class action by beneficiaries of a trust but recognizing that "[i]f the members of the plaintiff class were not all beneficiaries of the Trust, we would think that the applicable standards for personal jurisdiction would be drawn more from Shutts than from Hansberry"); In re Joint E. & S. Dist. Asbestos Litig. (Findley III), 78 F.3d 764, 777-78 (2d Cir. 1996) (distinguishing the class action from that in Shutts because the restructuring of a trust is an equitable remedy).
. The paradigmatic use of rule 23(b)(1)(B) is for a common (or limited) fund, which exists
when a fixed asset or piece of property exists in which all class members have a preexisting interest, and an apportionment or determination'of the interests of one class member cannot be made without affecting the proportionate interests of other class members similarly situated. Classic illustrations include claimants to trust assets, a bank account, insurance proceeds, company assets in a liquidation sale, process of a ship sale in a maritime accident suit, and others.
*10021 Herbert Newberg & Alba Conte, Newberg on Class Actions (hereinafter "Newberg on Class Actions”) § 4.09, at 4-32 through 4-33 (3d ed. 1992). The language of rule 23(b)(1)(B) is broad enough to encompass more than the traditional common fund, id. at 4-31, and the fact that an action meets the requirements of rule 23(b)(1)(B) does not necessarily transform it into one for the division of a fund.
This case deviates from the traditional common fund in a number of ways. First, there is no fund to speak of. The insurance proceeds are not common or limited, but are simply all that the defendants are willing to provide to the settlement. A settlement offer is far from "a fixed asset ... in which all class members have a preexisting interest.” The fact that Fibreboard's asbestos liabilities are greater than its assets is also insufficient to create a common fund, even though it is sufficient to meet the requirements of rule (b)(1)(B). See Arthur R. Miller & David Crump, Jurisdiction and Choice of Law in Multistate Class Actions after Phillips Petroleum Co. v. Shutts, 96 Yale LJ. 1, 42 (1986) (describing the “constructive bankruptcy” theoiy for certifying mass torts under rule (b)(1)(B)). See also Marcus, supra note 4, at 877-81 (suggesting that a common fund theory does not work for mass torts). Second, the plaintiffs do not have a preexisting interest in Fibreboard's assets; the putpose of the suit is to establish those rights and not to divide preexisting rights.
.472 U.S. at 811 & n. 3, 105 S.Ct. at 2973-74 & n. 3 ("Our holding today is limited to those class actions which seek to bind known plaintiffs concerning claims wholly or predominately for money judgments. We intimate no view concerning other types of class actions, such as those seeking equitable relief.”).
. The specific counts in the complaint are (1) negligent failure to warn; (2) strict product liability; (3) breach of express and implied warranty; (4) concert of action and conspiracy; and (5) "all other viable claims." The last count consists of every “claim or cause of action” that the plaintiffs can assert against Fibreboard.
. See also Marcus, supra note 4, at 888 ("|T]he limited fund concept should rarely, if ever, be available in mass tort litigation. Even if it can be sustained in some instances, there is no denying that the claims asserted are for compensatory damages, and there is arguably a constitutional right to opt out.”).
. See maj. op. at 986 ("The limitation of Shutts to claims of unknown plaintiffs that are predominantly for money damages forecloses application of its holding to 23(b)(1)(B) actions which have always been equitable and often involve unknown plaintiffs."). The majority also intimates that Shutts does not apply to rule 23(b)(1)(B) actions because the absent plaintiffs are unknown. See maj. op. at 986. The unknown-plaintiff exception in Shutts does not apply to this case, however.
The court lacks' personal jurisdiction over a large number of known plaintiffs who have manifested injuries but failed to file suit before the class action was filed. Moreover, a large number of the exposure-only plaintiffs are known. For example, shipyard workers who have not yet manifested any sign of disease could easily be identified and notified of the settlement. They then could decide whether to opt out. If they did, their individual and derivative wrongful death claims no longer would be part of the settlement.
. The court also held that the absent plaintiffs’ claims for injunctive relief were barred by res judicata because the class action could bind absent plaintiffs on their claims for equitable relief. Brown, 982 F.2d at 392.
. It is notable that the Shutts Court used the phrase “equitable relief." Shutts, 472 U.S. at 812 n. 3, 105 S.Ct. at 2974 n. 3. Equity courts were distinguished from law courts in that their substantive rules, procedures, and remedies differed from law courts. 1 Law of Remedies § 2.6(3) at 154-55. As one commentator pointed out, procedure "will seldom if ever form a basis for distinguishing law from equity" where the distinction matters. Id. at 155 n. 1.
The question then becomes whether the class action is an equitable remedy or an equitable procedural device. I discuss why the rationale of Shutts supports that latter interpretation at infra part IV.B.3. Further support for the obvious proposition that the class action is a procedural tool comes from the fact that it has never been treated as a "remedy." See 1 Law of Remedies §§ 1.1-1.6 (listing all remedies available at law and equity).
. The class action originated in courts of equity as a procedural joinder device. See generally 1 Law of Remedies § 1.1, at 1-2 (distinguishing procedural law, substantive law, and remedial law); id. § 2.1(1), at 57 (noting that differing procedures were available in equity and legal courts).
. In similar situations, where an equitable procedural device has been extended to actions at law, courts have used at least two distinct tests for determining equitable status. The first test focuses on the remedy: If the claim seeks a coercive remedy, it is deemed equitable. The second test characterizes a claim as equitable if the plaintiff seeks to enforce a right that was originally created in the equity courts, or a right that was traditionally decided according to equitable principles. 1 Law of Remedies § 2.6(3), at 154-55.
. Browning-Ferns Indus., Inc. v. Kelco Disposal, Inc., 492 U.S. 257, 299, 109 S.Ct. 2909, 2933, 106 L.Ed.2d 219 (1989) (O’Connor, J., concurring in part and dissenting in part) (stating that "the applicability of a provision of the Constitution has never depended on the vagaries of state or federal law”); Escobedo v. Illinois, 378 U.S. 478, 486, 84 S.Ct. 1758, 1762, 12 L.Ed.2d 977 (1964) (declining to "exalt form over substance” in determining the temporal scope of Sixth Amendment protections); Crowell v. Benson, 285 U.S. 22, 53, 52 S.Ct. 285, 293, 76 L.Ed. 598 (1932) (opining that “regard must be had, ... in ... cases where constitutional limits are invoked, not to mere matters of form but to the substance of what is required”); Chicago, Burlington & Quincy R.R. v. Chicago, 166 U.S. 226, 235, 17 S.Ct. 581, 584, 41 L.Ed. 979 (1897) ("In determining what is due process of law regard must be had to substance, not to form.”).
. The suit was over proceeds from a book fund maintained by the Methodist Episcopal Church. Smith, 57 U.S. (16 How.) at 303. The fund was established for the care of traveling preachers. Id. When the church divided over slavery, both sides agreed to split church property based upon a prearranged formula. Id. at 305. The travel*1006ing preachers from the south sued the northern church because it refused to relinquish that portion of the fund that belonged to the southern church. Id.
. Ibs involved a suit, for money damages, by a participant in a mortuary fund in state court. The question was whether a prior mandatory class action would have a res judicata effect on the plaintiff. The class action was against an insurance company and sought to compel the company to make distributions from a trust. The court held that the action against the trust must be given res judicata effect in the later proceeding.
. The appellees argue that even if due process requires an opt-out right, the settlement's "back-end” exit satisfies due process. Even if a “back-end” exit could serve as a proxy for an opt-out provision — a suggestion made only by the Fourth Circuit — the procedure provided in Ahearn falls far short of satisfying due process. See, e.g., In re A.H. Robins Co., 880 F.2d 709, 745 (4th Cir.) (finding that a "back-end” exit right provides "everything that an express opt-out provision could give a class member if such right is required under due process”), cert, denied, 493 U.S. 959, 110 S.Ct. 377, 107 L.Ed.2d 362 (1989).
In A.H. Robins, the court could determine with confidence that the "back-end” exit provided absent plaintiffs with the same rights they would have in an opt-out class action. The class action involved a trust established during a chapter 11 bankruptcy. A.H. Robins, 880 F.2d at 717. An absent class member could seek a jury trial if the alternative dispute resolution provided for by the trust did not settle the plaintiff's claim in a satisfactory fashion. Id. at 722. Liability, causation, and damages would be submitted to the jury. Id. Punitive damages were disallowed, *1007but a plaintiff seeking a jury trial would be entitled to an amount from the trust in lieu of the punitive damage award. Id. The exit mechanism left absent plaintiffs a right to a jury trial on precisely those issues on which they had a right to a jury trial before certification. Id. at 744 (“The Plan gives every such class member the right to have her claim settled in a trial with all the procedural rights normally attaching to a jury trial. That is everything that an express opt-out provision could give a class member if such right is required under due process.”).
The “back-end” exit in this case fails to provide absent plaintiffs with a right substantially equivalent ■ to a juty trial. The exit is nothing more than window dressing. While it allows claimants to settle disputes in front of a jury, it limits all damages to $500,000. By contrast, some plaintiffs have been able to recover several million dollars from asbestos defendants at trial. See, e.g., Mechanic’s Family Awarded $14 Million in Auto Brake Asbestos Case, West’s Legal News, June 20, 1996, at 1996 WL 339400 (discussing asbestos verdict for two plaintiffs); Asbestos Litigation — Damages, Chi. Daily Law Bulletin, June 13, 1995, at 1 (discussing $12 million verdict); Notable Verdicts, Nat'l Law Journal, Apr. 15, 1996, at A15 (discussing $2.2 million jury verdict); Reviewing Verdicts for Excessiveness, New York L.J., Apr. 10, 1996, at 1 (discussing verdicts of $64.6 million for four plaintiffs); $18 Million Verdict for Asbestos Claims, New York L.J., Feb. 15, 1996, at 2 (discussing verdict for five plaintiffs). Such a severe limitation on a party’s ability to prove and try damages deprives him of an essential element of a jury trial. I certainly cannot say that a procedure that caps a plaintiff’s damages gives him “everything that an express opt-out provision could give a class member." A.H. Robins, 880 F.2d at 744.
. See In re Gen. Motors Corp. Pick-Up Truck Fuel Tank Prods. Liab. Litig., 55 F.3d 768, 796 (3d Cir.), cert. denied,-U.S.-, 116 S.Ct. 88, 133 L.Ed.2d 45 (1995); 1 Newberg on Class Actions § 1.13, at 37-38.
. See Marcus, supra note 4, at 899-90 (criticizing Aheam settlement).
. See Mars Steel Corp. v. Continental Ill. Nat’l Bank & Trust Co., 834 F.2d 677, 683 (7th Cir. 1987) ("The agency problems which [class] actions create require that the district judge be vigilant in protecting the procedural rights of class members.”); cf. 7B Wright & Miller § 1785, at 103 (“A formal class certification determination plays an important role in assuring adequate protection to the absent class members.”).
. See Koniak, Looking Glass, supra note 4, at 18 ("A poisoned or tainted pudding may taste okay, if not delicious.").
. See supra page 1008; General Motors, 55 F.3d at 788; Coffee, supra note 1, at 1378-82.
. While I would follow Beef Industry and Corrugated Container reluctantly, the majority actually embraces them on the ground that settlement ■negotiations generate information. See maj. op. at 975-76. While it is hard to argue in favor of ignoring information, settlement negotiations generate little information regarding commonality and typicality, and it is the majority that ignores any such information.
First, no competent counsel would enter negotiations without first divining his clients’ legal theories and the questions of law or fact underlying them. Thus, "the process of negotiation does not reveal anything about commonality and typicality.” General Motors, 55 F.3d at 796.
Second, this justification for relaxing the requirements is inconsistent with the majority's application of them. To the extent that diverging interests emerge in negotiations, those interests militate against mass certification, as they indicate that class counsel negotiated the settlement while suffering from a conflict. The majority relies upon the existence of information in relaxing the criteria, but ignores that information when applying them.
Finally, the majority's emphasis on information is ironic. Conventional wisdom posits that early settlement decreases the amount of information available to counsel and the court. See, e.g., Manual for Complex Litigation 3d § 30.45, at 243 (1995).
. In theory, the class representative should control the attorney. In practice, class counsel often selects the class representative and is likely to pick one who is passive. Cf. In re Dresser Indus., 972 F.2d 540, 545 n. 11 (5th Cir.1992) (noting consent of client in a class action is of “limited utility” as a safeguard against conflicts of interest).
. See Woolen v. Surtran Taxicabs, 684 F.2d 324, 332 (5th Cir.1982) ("The adequacy of representation in.Rule 23(a)(4) is that essential to due process under Hansberry v. Lee], 311 U.S. 32, 61 S.Ct. 115, 85 L.Ed. 22 (1940),] before absent class members can be bound."); Gonzales v. Cassidy, 474 F.2d 67, 75 (5th Cir. 1973) (“The judgment in a class action will bind only those members of the class whose interests have been adequately represented by existing parties to the litigation.’’) (quoting Sam Fox Publishing Co. v. United States, 366 U.S. 683, 691, 81 S.Ct. 1309, 1314, 6 L.Ed.2d 604 (1961)).
. See North Am. Acceptance Corp. v. Arnall, Golden & Gregory, 593 F.2d 642, 644 n. 4 (5th Cir.) (reasoning that attorney's conflict of interest could prevent class certification under rule 23 because of "the important role the attorney plays in protecting the interests of the class”), cert. denied, 444 U.S. 956, 100 S.Ct. 436, 62 L.Ed.2d 328 (1979); see also In re Fine Paper Antitrust Litig., 617 F.2d 22, 27 (3d Cir.1980) (concluding that attorneys with conflict of interest do not meet the adequate representation requirement of rule 23(a)(4)).
. See In re "Agent Orange” Prod. Liab. Litig., 818 F.2d 216, 224 (2d Cir.) (“We also reject the district court’s finding that its authority to approve settlement offers under Fed.R.Civ.P. 23(e) acts to limit the threat to the class from a potential conflict of interest.”), cert. denied, 484 U.S. 926, 108 S.Ct. 289, 98 L.Ed.2d 249 (1987).
. [D]ue process must mean something other than that the result is just, otherwise some lynchings would be consistent with due process. A 'fairness' hearing that appraises a settlement made outside the court's presence only as to the substantive fairness of the terms provides no more process than would be provided by a post-lynching hearing that assessed whether the dead guy really did commit the crime.
Koniak, Feasting, supra note 4, at 1123.
. See Corrugated Container, 643 F.2d at 211 n. 25 ("We hasten to add that the adequacy of settlement terms cannot ordinarily redeem a settlement that was bargained by a party in a conflict position.’’).
The court, to be sure, will not approve a settlement if it is unfair, but “fairness” may be found anywhere within a broad range of lower and upper limits. No one can tell whether a compromise found to be “fair” might not have been “fairer” had the negotiating [attorney] possessed better information or been animated by undivided loyally to the cause of the class. The court can reject a settlement that is inadequate; it cannot undertake the partisan task of bargaining for better terms. The integrity of the negotiating process is, therefore, important.
Haudek, The Settlement and Approval of Stockholders’ Actions — Part II: The Settlement, 23 Sw. L.J. 765, 771-72 (1969), quoted in In re General Motors Corp. Engine Interchange Litig., 594 F.2d *10101106, 1125 n. 24 (7th Cir.), cert. denied, 444 U.S. 870, 100 S.Ct. 146, 62 L.Ed.2d 95 and cert. denied, 444 U.S. 870, 100 S.Ct. 146, 62 L.Ed.2d 95 (1979).
. The individual plaintiffs also would receive the entire setdement if Fibreboard won the coverage case.
. Under the lodestar formula, a court determines a class attorney’s compensation by multiplying the number of hours he has spent .on the litigation by an appropriate hourly wage. While the court may determine the hourly wage in part by looking to how favorable a settlement the attorney negotiated, the number of hours the attorney has worked is usually the chief determinant of his total compensation. See Jonathan R. Macey & Geoffrey P. Miller, The Plaintiffs’ Attorney's Role in Class Action and Derivative Litigation: Economic Analysis and Recommendations for Reform, 58 U.Chi.L.Rev. 1, 50 (1991).
. See Koniak, supra note 4, for a more extensive discussion of the perils of simultaneous representation.
. The majority misleadingly labels these groups as "near” and “far” futures, which suggests that the difference between these groups is simply one of degree. In fact, there is a difference in kind. All of the extant claimants, or " 'near' futures," may file claims as soon as approval of the settlement is final. But none of the " ‘far’ futures” may file suit until some triggering event occurs, and they have no way of knowing when, or even whether, that event will occur.
.State law determines exactly what is the necessary triggering event. Some states permit exposure-only victims to seek recovery, while others require that a claimant develop some illness from his exposure before permitting him to recover. One group of latent claimants, of course, are the future family members of asbestos victims: future spouses and unborn or even unconceived children. For this group, the necessary triggering event is marriage or birth.
. The majority does distinguish the case before us from In re Joint E. & S. Dist. Asbestos Litig., 982 F.2d 721 (2d Cir.1992), on the ground, that certain members of the class in that case had rights under a trust agreement to have their claims paid first, while there is nothing akin to a trust agreement in Aheam. But Georgine also lacked anything similar to a trust agreement, so the distinction does not explain the majority’s failure to distinguish Aheam from Georgine.
. Fibreboard’s pre-1959 claims settled for an average of $12,000 in 1993, while its post-1959 claims settled for only $4,000. See GAF, Fibreboard, Pfizer Report Decline in Asbestos Cases, Mealey's Litig. Rep., June 2, 1995, at 20. In 1994, the discrepancy almost disappeared, with pre1959 claims settling for $8,000 and post-1959 claims settling for $7,000. Id. As the asbestos world became aware that the expected settlement in Aheam would not differentiate between pre- and post-1959 claimants, the settlement values of those claims naturally moved toward a common figure.
. It is well settled that impartiality can reasonably be questioned even where there is no actual bias. See In re Faulkner, 856 F.2d 716, 721 (5th Cir.1988) (per curiam).
. For similar reasons, Congress in 1978 limited the administrative duties of bankruptcy judges. Under the old Bankruptcy Act, bankruptcy judges (then called referees) had been actively involved in the day-to-day affairs of bankruptcy reorganizations. Congress found this objectionable.: "No matter how fair a bankruptcy judge is, his statutory duties give him a certain bias in a case, and the bankruptcy court as a result has been viewed by many as an unfair forum.” H.R.Rep. No. 595, 95th Cong., 1st Sess. 88 (1978), reprinted in 1978 U.S.C.C.A.N. at 6050. Congress went on to note that "[djeeper problems arise because of the inconsistency between the judicial and administrative roles of the bankruptcy judges. The inconsistency compromises his impartiality as an arbiter of bankruptcy disputes.” Id. at 89, reprinted in U.S.C.C.A.N. at 6050. To rectify this problem, Congress limited the role of bankruptcy judges as administrators and mediators. See, e.g., 11 U.S.C. § 341(c) (1994) ("The court may not preside at, and may not attend, any meeting under this section including any final meeting of creditors.”).
. Certainly there can be no criticism of Judge Higginbotham's role. From the start, he served only as a mediator, not a judge. In facilitating agreement, he did precisely what a mediator is supposed to do.
. The Third Circuit pretermitted the jurisdictional issues raised in a nearly identical case because its decision to decertify the class disposed of the case in favor of the intervenors. See Georgine v. Amchem Prods., Inc., 83 F.3d 610, 617, 623 (3d Cir.1996). Even in that case, one judge wrote separately to emphasize, inter alia, that the jurisdictional issues should have been reached. See id. at 635-38 (Wellford, J., concurring). The majority's decision to affirm certification in this case, however, obligates it to address — sua sponte if necessary — the troubling jurisdictional problems raised by the district court's approval of the settlement. Nevertheless, the majority has failed even to raise these issues, let alone consider them.
. The counts alleged are (1) negligent failure to warn; (2) strict product liability; (3) breach of express and implied warranty; (4) concert of action and conspiracy; and (5) “all other viable claims.” This last count consists of every "claim or cause of action” that the plaintiffs can assert “against Fibreboard.”
. The majority also states that "Aheam ... presents us with claims against a healthy company for personal injuries and a proposed settlement of those claims.” Maj. op. at 983.
. The district court actually stated that “[the] claims were, prior to the settlement, in personam in character.” This statement implies that, in the court’s view, the settlement transformed the claims into something other than in personam claims.
. In Wausau, we were faced with “six insurance funds encompassing different periods during a four-year span." 954 F.2d at 1101. Furthermore, the legitimacy of the claims against each fund was not established. Id. Cautioning that "[(Interpleader is not designed to solve all problems associated with multiparty litigation,” id. (citing State Farm Fire & Casualty Co. v. Tashire, 386 U.S. 523, 535, 87 S.Ct. 1199, 1206, 18 L.Ed.2d 270 (1967)), we held: "Because the present case involves six separate funds and because it is unclear whether there are legitimate claims against each fund, we conclude that an identifiable fund as required for interpleader actions is not involved." Id.
In Aheam, we are again faced with multiple funds — the policies issued by the two insurers, Continental and Pacific — and the legitimacy of the liabilities against the policies is not well-established — i.e., the validity of all the class members' claims against Fibreboard has not yet been decided. Moreover, because of the potentially limitless liability under these comprehensive general liability policies, the ultimate scope of coverage cannot be determined until damages are awarded on each successful claim. The terms of the policies in Wausau at least limited coverage to a specific dollar amount; thus, the total possible liability was easily defined. Yet even there we held that interpleader jurisdiction was unavailable. It follows a fortiori from Wausau that such jurisdiction does not exist in Aheam.
.See Griffin v. McCoach, 313 U.S. 498, 503, 61 S.Ct. 1023, 1025-26, 85 L.Ed. 1481 (1941) (applying Erie and Klaxon Co. v. Stentor Elec. Mfg. Co., 313 U.S. 487, 61 S.Ct. 1020, 85 L.Ed. 1477 (1941) to federal statutory interpleader case); Perkins State Bank v. Connolly, 632 F.2d 1306, 1311 (5th Cir.1980) (stating that, with limited exceptions, "substantive state rules of decision generally govern federal interpleader proceed*1017ings’'); Bluff Creek Oil Co. v. Green, 257 F.2d 83, 85 (5th Cir.1958) (applying Erie after characterizing § 1335 interpleader action as "just another diversity suit”); Williams v. McFerrin, 242 F.2d 53, 55 (5th Cir.1957) (applying Louisiana conflict of laws principles to case, per Klaxon, because "interpleader comes under the diversity jurisdiction of the federal courts”); Whirlpool Corp. v. Ritter, 929 F.2d 1318, 1320-21 (8th Cir.1991) (stating in dictum that Klaxon applies to suits brought under § 1335 “because the federal inter-pleader statute is merely a special brand of diversity jurisdiction”) (citing Griffin, 313 U.S. at 503, 61 S.Ct. at 1025-26, and Williams, 242 F.2d at 55); 7 Wright & Miller § 1713 (1986) (stating that the Ene doctrine applies fully in both Fed. R.Civ.P. 22 and § 1335 interpleader actions) (noting exceptions for certain procedural, remedial, and administrative matters — e.g., pleading rules, the availability of injunctive relief, and attorneys' fees).
. This provision reads: "A class action shall not be dismissed or compromised without the approval of the court, and notice of the proposed dismissal or compromise shall be given to all members of the class in such manner as the court directs.” Fed R.Civ.P. 23(e).
. Epstein v. MCA, Inc., 50 F.3d 644, 667 (9th Cir. 1995), rev’d on other grounds sub nom. Matsushita Elec. Indus. Co. v. Epstein,-U.S.-, 116 S.Ct. 873, 134 L.Ed.2d 6 (1996); see Evans v. Jeff D., 475 U.S. 717, 726-727, 106 S.Ct. 1531, 1536-1537, 89 L.Ed.2d 747 (1986) (referring to district court's rule 23(e) "power to approve or reject a [class action] settlement”); see also In re General Motors Corp. Pick-Up Truck Fuel Tank Prods. Liab. Litig., 55 F.3d 768, 793 (3d Cir.) ("A judicially supervised and approved class action settlement, like a judicially supervised trial, is a means of hearing and determining judicially, in other words ‘adjudicating,’ the value of claims arising from a mass tort.") (quoting with approval Roger H. Trangsrud, Joinder Alternatives in Mass Tort Litigation, 70 Cornell L.Rev 779, 835 (1985) (footnote omitted)), cert. denied, -U.S. -, 116 S.Ct. 88, 133 L.Ed.2d 45 (1995).
.See Purcell v. BankAtlantic Fin. Corp., 85 F.3d 1508, 1511 (11th Cir.1996) (referring to “judgment" approving class action settlement), White v. National Football League, 836 F.Supp. 1508, 1510-11 (D.Minn.1993) (describing and entering "judgment” approving class action settlement), off d, 41 F.3d 402, 406-07 (8th Cir.1994) (referring to "judgment decree" and "consent judgment" approving class action settlement), cert. denied, - U.S. -, 115 S.Ct. 2569, 132 L.Ed.2d 821 and cert. denied,-U.S.-, 115 S.Ct. 2569, 132 L.Ed.2d 821 (1995); cf. Matsushita, -U.S. at---, 116 S.Ct. at 877-78 (treating class action settlement approved by state court as a state court judgment).
.The precise class definition is as follows:
(a) All persons (or their legal representatives) who prior to August 27, 1993 were exposed, directly or indirectly (including but not limited to exposure through the exposure of a spouse, household member or any other person), to asbestos or to asbestos-containing products for which Fibreboard may bear legal liability and who have not, before August 27, 1993, (i) filed a lawsuit for any asbestos related personal injury, or damage, or death arising from such exposure in any court against Fibreboard or persons or entities for whose actions or omissions Fibreboard bears legal liability; or (ii) settled a claim for any asbestos-related personal injury, or damage, or death arising from such exposure with Fibreboard or with persons or entities for whose actions or omissions Fibreboard bears legal liability;
(b) All persons (or their legal representatives) exposed to asbestos or to asbestos-containing products, directly or indirectly (including but not limited to exposure through the exposure of a spouse, household member or any other person), who dismissed an action prior to August 27, 1993 without prejudice against Fibreboard, and who retain the right to sue Fibreboard upon development of a nonmalignant disease process or a malignancy; provided, however, that the Settlement Class does not include persons who filed and, for cash payment or some other negotiated value, dismissed claims against Fibreboard, and whose only retained right is to sue Fibreboard upon development of an asbestos-related malignancy; and
(c)All past, present and future spouses, parents, children and other relatives (or their legal representatives) of the class members described in paragraph (a) and (b) above, except for any such person who has, before August 27, 1993, (i) filed a lawsuit for the asbestos-related personal injury, or damage, or death of a class member described in paragraph (a) or (b) above in any court against Fibreboard (or against entities for whose actions or omissions Fibreboard bears legal liability), or (ii) settled a claim for the asbestos-related personal injury, or damage, or death of a class member described in (a) or (b) above with Fibreboard (or with entities for whose actions or omissions Fibreboard bears legal liability).
. The future children purportedly bring "indirect” claims — e.g., claims for wrongful death— derived from the exposure to asbestos of the persons who will become their parents. Some states treat these claims as non-derivative. See Adams v. U.S. (In re “Agent Orange” Prod. Liab. Litig.), 818 F.2d 201, 203 (2d Cir.1987), cert. denied, 484 U.S. 1004, 108 S.Ct. 695, 98 L.Ed.2d 647 (1988). Most of these claims are also unripe because the person exposed to asbestos — the future parent of the future child — is still alive.
. The future spouses also purportedly bring indirect claims — e.g., loss of consortium. Again, some jurisdictions treat these claims as non-derivative. See Adams, 818 F.2d at 203.
. For simplicity, I restrict my analysis to persons not yet conceived.
. The rule 23(b)(3) class in the Agent Orange case was carefully defined to avoid future children and future spouses. The- class definition read as follows:
[T]hose persons who were in the United States, New Zealand or Australian Armed Forces at any time from 1961 to 1972 who were injured while in or near Vietnam by exposure to Agent Orange or other phenoxy herbicides.... The class also includes spouses, parents, and children of the veterans born before January 1, 1984, directly or derivatively injured as a result of the exposure.
Ivy v. Diamond Shamrock Chems. Co. (In re "Agent Orange" Prod. Liab. Litig.), 996 F.2d 1425, 1429 (2d Cir.1993) (quoting In re "Agent Orange" Prod. Liab. Litig. MDL No. 381, 100 F.R.D. 718, 729 (E.D.N.Y.1983), aff'd, 818 F.2d 145 (2d Cir. 1987), cert. denied, 484 U.S. 1004, 108 S.Ct. 695, 98 L.Ed.2d 647 (1988)), cert. denied, 510 U.S. 1140, 114 S.Ct. 1125, 127 L.Ed.2d 434 and cert. denied, 510 U.S. 1140, 114 S.Ct. 1126, 127 L.Ed.2d 434 (1994). This definition restricted the class to past and present spouses and already-bom children, thus avoiding the jurisdictional anomalies presented by future spouses and future children.
The appellees would have us believe that the federal courts routinely certify classes including future claimants; the authorities they cite belie that claim, however. In United States Parole Comm’n v. Geraghty, 445 U.S. 388, 393, 100 S.Ct. 1202, 1207, 63 L.Ed.2d 479 (1980), the class definition did include future claimants, but the district court had declined to certify the class; as a result, the Court never had occasion to consider the standing issues raised by the inclusion of such persons. Likewise, the district court in Comer v. Cisneros, 37 F.3d 775, 780, 796 (2d Cir. 1994), had declined to certify a class including future claimants. Thus, the holding in Comer was merely that the purported class representatives — who were present claimants — had standing, not that the future claimants did.
Finally, McArthur v. Scott, 113 U.S. 340, 5 S.Ct. 652, 28 L.Ed. 1015 (1885), simply has nothing to do with this mass toxic tort class action. The portion of McArthur relied upon by the appellees discusses trusts and estates, as do the cases cited therein. See id. at 401-02, 5 S.Ct. at 673-74.
. International Primate Protection League v. Administrators of Tulane Educ. Fund, 500 U.S. 72, 77, 111 S.Ct. 1700, 1704, 114 L.Ed.2d 134 (1991); see Catholic Social Serv. v. Shalala, 12 F.3d 1123, 1125 (D.C.Cir.1994) (“[A] plaintiff's standing must be analyzed with reference to the particular claim made.”); see also Financial Insts. Retirement Fund v. Office of Thrift Supervision, 964 F.2d 142, 146 (2d Cir.1992) (quoting International Primate, 500 U.S. at 77, 111 S.Ct. at 1704).
In Louisiana Landmarks Soc'y v. City of New Orleans, 85 F.3d 1119, 1122 n. 3 (5th Cir.1996), we stated that "[sjtanding is a concept distinct from the concept of private rights of action.” The appellee in that case had failed to brief the implied cause of action issue raised by the appellants,. asserting that the appellants had waived a "standing” argument in the district court. We expressed our puzzlement with the appellee’s waiver argument by noting that standing and implied rights of action are not equivalent concepts and that, moreover, standing can never be waived, because it is jurisdictional. See id. Thus, Landmarks should not be read as implying that standing is unrelated to the causes of action alleged.
. The Court made this point explicitly:
[Sjtanding is not dispensed in gross. If the right to complain of one administrative deficiency automatically conferred the right to complain of all administrative deficiencies, any citizen aggrieved in one respect could bring the whole structure of state administration before the courts for review. That is of course not the law. As we have said, "[njor does a plaintiff who has been subject to injurious conduct of one kind possess by virtue of that injury the necessary stake in litigating conduct of another kind, although similar, to which he has not been subject.” Blum v. Yaretsky, 457 U.S. 991, 999, 102 S.Ct. 2777, 2783, 73 L.Ed.2d 534 (1982).
Casey,-U.S. at-n. 6, 116 S.Ct. at 2181-82 n. 6 (parallel citations omitted).
. 500 U.S. at 77, 111 S.Ct. at 1704 (quoting Allen v. Wright, 468 U.S. 737, 752, 104 S.Ct. 3315, 3325, 82 L.Ed.2d 556 (1984)) (emphasis in original); see Catholic Social Serv., 12 F.3d at 1125; Howe v. Ellenbecker, 8 F.3d 1258, 1261 (8th Cir.1993), cert. denied, - U.S. -, 114 S.Ct. 1373, 128 L.Ed.2d 49 (1994); Financial Insts., 964 F.2d at 146-47; see also United Food & Commercial Workers Int’l Union, Local 751 v. Brown Group, Inc., 50 F.3d 1426, 1428-29 (8th Cir.1995), rev’d on other grounds, -U.S.-, 116 S.Ct. 1529, 134 L.Ed.2d 758 (1996); cf. Idaho Conservation League v. Mumma, 956 F.2d 1508, 1514 (9th Cir.1992) ("Where ... Congress is the source of the purportedly violated legal obligation, we look to the statute to define the injury.”) (quoting International Primate, 500 U.S. at 77, 111 S.Ct. at 1704).
. I do not suggest that we look to the merits of the cause of action or determine whether the cause of action is valid (in the sense of determining whether the plaintiff ultimately would win). *1021Rather, we must examine the pleadings to determine what cause of action the plaintiff has alleged and whether his allegations of injury, in the context of that cause of action, satisfy Article III standards.
. In a recent case interpreting the Warsaw Convention, the Supreme Court had occasion to consider whether the question of who may bring suit is substantive or procedural. See Zicherman v. Korean Air Lines Co., - U.S. -, -, 116 S.Ct. 629, 634, 133 L.Ed.2d 596 (1996). The Court was interpreting Article 24 of the Convention, which leaves to domestic law the issues of who is entitled to sue and what types of damages they may recover. See id. at •---, 116 S.Ct. at 633-34. The Court concluded:
The most natural reading of [Article 24] is that, in an action brought under Article 17, the law *1022of the Convention does not affect the substantive questions of who may bring suit and what they may be compensated for.... It does not seem to us that the question of who is entitled to a damages award is procedural.
Id. at-, 116 S.Ct. at 634 (emphasis added). Thus, the question of who may bring suit is a question answered by substantive law — in a diversity case, state substantive law. This is not to say that standing in federal court is purely an issue of state law in a diversity case; rather, the role of state law is to identify, within federal constitutional limits, the persons who have legally protected interests and what those interests are.
. This is equivalent to saying that the law of the relevant state (determined through a choice of law analysis) does not authorize that exposure-only cause of action.
. Cases that opine in broad language that state law cannot affect the diversity jurisdiction of the federal courts are inapposite. Cf., e.g., Begay v. Kerr-McGee Corp., 682 F.2d 1311, 1315 (9th Cir. 1982). In Begay, the court did opine that state law cannot create or enlarge federal jurisdiction, but that statement must be viewed in context.
The Begay court faced a state statute that granted a substantive right and then purported to state that the sole remedy for violation of the right was an administrative claim over which a state commission had exclusive jurisdiction. The court responded that a state law, having conferred a substantive right, cannot determine whether a federal court sitting in diversity has jurisdiction over a claim based on that right. See id. at 1315-17. Rather, the proper inquiry in such a case is whether the complaint stated a claim upon which relief could be granted. See id. at 1315, 1317, 1320.
This analysis does not apply to the Article III justiciability issues raised in Ahearn. The existence vel non of a cause of action is a factor that the Supreme Court has incorporated into the Article III standing inquiry. See Lujan, 504 U.S. at 560, 112 S.Ct. at 2136. Reinforcing this approach is the unambiguous recognition by the Court that we are to measure standing according to the particular common-law, statutory, or constitutional claim asserted. See International Primate, 500 U.S. at 77, 111 S.Ct. at 1704. Thus, in diversity cases, whether a state-law cause of action exists at all is a jurisdictional issue insofar as it relates to standing and injury-in-fact, not an issue of whether a claim for relief has been stated.
.Despite this unambiguous holding, the Supreme Court of Pennsylvania has allowed plaintiffs to recover medical monitoring costs for exposure to asbestos. See Simmons, 674 A.2d at 239-40. The court stated that such costs were properly awarded for meritorious exposure-only cases but that damages for increased risk and *1023fear of cancer were not authorized. See id. 674 A.2d at 239-40.
There are two ways to analyze the phenomenon of exposure-only causes of action. First is the monolithic view — i.e., that there is only a single cause of action based on exposure to asbestos, for which there are multiple remedies (some of which may not be available). Second is the “polylithic” view (for lack of a better word), that there are multiple causes of action for exposure to asbestos (some of which may not be available) — i.e., there is one cause of action for increased risk of cancer, one cause of action for medical monitoring costs, etc.
It is largely irrelevant which of these two models is more elegant from a conceptual standpoint. What is important — indeed, determinative — for our inquiry is which of the two each state has adopted. Pennsylvania's high court unambiguously has held that Pennsylvania law supports no cause of action for increased risk of cancer or for present emotional distress arising from the fear of cancer. Thus, the court's statement about medical monitoring damages is best viewed as the authorization of a distinct cause of action for medical monitoring expenses. Reading the opinion otherwise would defy its plain language regarding Pennsylvania's nonrecognition of increased risk and emotional distress causes of action.
. See, e.g., Burns v. Jaquays Mining Corp., 156 Ariz. 375, 752 P.2d 28, 29-31 (Ct.App. 1987) (holding that subclinical asbestos-related injury is not sufficient to support cause of action), review dismissed, 162 Ariz. 186, 781 P.2d 1373 (1989); DeStories v. City of Phoenix, 154 Ariz. 604, 744 P.2d 705, 707-11 (Ct.App.1987) (same); Mergenthaler v. Asbestos Corp. of Am., 480 A.2d 647, 651 (Del. 1984) (holding that present physical injury caused by exposure to asbestos is "essential element” of claims for mental anguish and medical monitoring costs); Capital Holding Corp. v. Bailey, 873 S.W.2d 187, 192 (Ky.1994) (requiring manifestation of asbestos-caused injuiy before recognizing existence of cause of action for negligence based on exposure to asbestos); Larson v. Johns-Manville Sales Corp., 427 Mich. 301, 399 N.W.2d 1, 2 (1986) (holding that cause of action for asbestosis accrues upon discovery of disease, not at time of exposure to asbestos); Locke v. Johns-Manville Corp., 221 Va. 951, 275 S.E.2d 900, 904-06 (1981) (holding that injury does not occur upon exposure to asbestos, but rather upon development of disease).
. The appellees cite a legion of cases indiscriminately — and in most cases without sufficient analysis — for the proposition that the exposure-only claims present a case or controversy. These cases, for one reason or another, are insufficient to overcome the force of Lujan and International Primate.
For example, the appellees cite Duke Power Co. v. Carolina Envtl. Study Group, Inc., 438 U.S. 59, 98 S.Ct. 2620, 57 L.Ed.2d 595 (1978), as an example of a case where the Supreme Court found exposure to a harmful substance a sufficient injuiy-in-fact for standing purposes. The plaintiffs sought a declaratory judgment that a federal statute capping liability for nuclear accidents had violated their constitutional rights. See id. at 67, 98 S.Ct. at 2627. Three justices wrote separate concurrences expressing their conclusions that the dispute in that case was not within the district court's jurisdiction. See id. at 94-103, 98 S.Ct. at 2641-46 (concurring opinions of Justice Stewart, then-justice Rehnquist, and Justice Stevens). The majority’s standing analysis appeared to focus on environmental and aesthetic injuries unconnected to the claims asserted by the plaintiffs, which were claims for violations of the Due Process and Takings Clauses of the Fifth Amendment. Compare id. at 69, 98 S.Ct. at 2628 (claims alleged in complaint) with id. at 72-74, 98 S.Ct. at 2629-31 (injuries alleged in complaint).
That methodology — analyzing injuiy unanchored by the substantive claim asserted — -is no longer available to us in light of Lujan and Inter*1024national Primate. In fact, the analysis advanced by the Duke Power majority is logically inconsistent with the Supreme Court's modern standing framework, embodied primarily in Lujan and enhanced by the unanimous directive in International Primate. Given that irreconcilable inconsistency, we have no choice but to follow the Court's more recent pronouncements.
Helling v. McKinney, 509 U.S. 25, 28, 113 S.Ct. 2475, 2478, 125 L.Ed.2d 22 (1993) — a case not cited by the appellees — is not on point, because the plaintiff alleged present injuries based on exposure to an alleged toxin (second-hand tobacco smoke). See also Georgine, 83 F.3d at 636 (Wellford, J., concurring). As a result, the Supreme Court never reached the issue of standing based on mere exposure to a toxin.
In Agent Orange, the Second Circuit did not mention either Lujan or International Primate in stating that the exposure-only plaintiffs had sustained an " ‘injuiy in fact.' " See Agent Orange, 996 F.2d at 1434 (relying only on Duke Power). Although I admire the Second Circuit’s handling of the massive Agent Orange litigation, I must respectfully disagree with its analysis that exposure-only claimants in that litigation met Article Ill’s requirements. Other than Duke Power, the court appeared to rely only on cases involving the interpretation of the term "injury'' in insurance policies. See id. These cases, however, are not authoritative with respect to the manner in which "injury-in-fact” is defined for Article III purposes. But see id. The Supreme Court's recent guidance in Lujan, on the other hand, is dispositive of that definitional issue.
The appellees cite several bankruptcy cases to support their exposure-only theory — e.g., In re UNR Indus., 20 F.3d 766, 770-71 (7th Cir.), cert. denied,-U.S.-, 115 S.Ct. 509, 130 L.Ed.2d 416 (1994); Findley v. Blinken (In re Joint E. & S. Dist. Asbestos Litig.), 129 B.R. 710, 834-37 (E. & S.D.N.Y. and Bankr.S.D.N.Y.1991), vacated, 982 F.2d 721 (2d Cir.1992), modified on reh'g, 993 F.2d 7 (2d Cir. 1993); Lindsey v. Dow Corning Corp. (In re Silicone Gel Breast Implant Prods. Liab. Litig. (MDL 926)), No. CV 92-P-10000-S, Civ.A. No. CV94-P-11558-S, 1994 WL 114580, at *1 (N.D.AIa. Apr. 1, 1994). On its face Article III does not apply to bankruptcy courts, which are Article I courts. See Rohm & Hass Texas, Inc. v. Ortiz Bros. Insulation, Inc., 32 F.3d 205, 210 n. 18 (5th Cir.1994). Because of the statutory scheme requiring that matters be referred to bankruptcy courts by district courts, however, the jurisdiction of the bankruptcy courts may be limited by Article III. See, e.g., In re Interpictures, Inc., 86 B.R. 24, 28-29 (Bankr.E.D.N.Y. 1988); John T. Cross, Congressional Power to Extend Federal Jurisdiction to Disputes Outside Article III: A Critical Analysis from the Perspective of Bankruptcy, 87 Nw.U.L.Rev. 1188, 1197-98 (1988).
If Article III does not apply to bankruptcy courts, the cited cases are plainly distinguishable. If it does apply, the reasoning of those cases is erroneous. The two district court cases do not cite to Lujan or International Primate. The Seventh Circuit's decision in UNR does cite to the former, but only in a conclusionary fashion and without any in-depth analysis.
. Critically, we modified Professor Wright’s apparent conflict formulation in one critical respect: "This short-hand formulation must at least be qualified insofar as it leaves room for a federal court to run afoul of Angel v. Bullington by entertaining diversity actions unavailable under state law.” Id. (citation omitted). Professor Wright had acknowledged this limitation, referring to a prominent district judge's articulation of it: " 'Judge Wyzanski carefully points out the limitation that if, as a matter of substantive law, a state does not recognize that a plaintiff has a particular right of action unless he joins with him certain others, then the federal court in a diversity action is precluded from giving a plaintiff who fails to join those others an opportunity to proceed as though alone he had a substantive right.' ” Id. (quoting 2 Barron & Holtzoff, Federal Practice and Procedure § 511 (1964 pocket part) (Wright ed. 1961)).