STEPHEN H. ANDERSON, Circuit Judge.
The appellants, defendants who have not agreed to settle with TBG, Inc., challenge the district court’s order approving TBG’s settlements with three other defendants. We do not have jurisdiction to review the court’s order approving TBG’s settlement with Robert Mann and John Pappajohn, but we do review the order approving TBG’s settlement with Shook, Hardy & Bacon (“Shook”). We vacate that order because the court imper-missibly barred the nonsettling defendants’ contribution claims against the settling defendants, as well as independent claims by Bendis against Shook.
BACKGROUND
TBG acquired Continental Healthcare Systems, Inc. in 1986. In 1988, TBG sued Richard Bendis, the former president of Continental, Terrance Schreier, the former executive vice president, Robert Mann and John Pappajohn, former outside directors, Shook, Hardy & Bacon, Continental’s outside counsel, and Ernst & Whinney, Continental’s outside auditor. TBG claimed that these defendants had misrepresented Continental’s financial status when TBG acquired it, and sought relief under sections 10(b) and 20(a) of the Securities Exchange Act of 1934, 15 U.S.C. §§ 78j(b), 78t(a), and Rule 10b-5, 17 C.F.R. § 240.10b-5.
In January 1992, Mann and Pappajohn agreed to settle with TBG for $200,000 and the release of their claims for payment of their legal expenses. In September 1992, Shook also agreed to settle with TBG for a confidential sum. Bendis, Schreier, and Ernst & Whinney have not agreed to settle.
Both the Shook and the Mann and Pappa-john settlements were contingent on the district court entering an order barring all related claims against them by, the nonsettling defendants and ordering that the judgment at trial be reduced by the settlement amounts. The district court approved these agreements in a memorandum opinion dated December 30, 1992. 811 F.Supp. 596. On [920]*920January 4, 1993, the court signed separate orders approving each settlement, making the required orders, and certifying each as final and appealable under Fed.R.Civ.P. 54(b). The court entered judgment on the Shook settlement on January 5, then entered the Mann and Pappajohn judgment on January 6.
On January 11, Bendis filed a motion to reconsider the court’s “Memorandum and Order dated December 30, 1992, and its Order and Judgment dated January 4, 1993.” Appellant’s App. at 355. The court denied this motion on February 18.
The appellants did not receive a copy of the court’s order denying the motion to reconsider, and did not learn of it until April 14. On April 15, they filed a joint motion for an extension of time to file notices of appeal. The next day the court gave the appellants fourteen days to appeal, and all three nonset-tling defendants appealed both of the settlement approvals within that time.
On April 23, however, Mann and Pappa-john moved to reconsider the court’s order extending the time for appeal of the order approving their settlement with TBG. On May 4, the court granted their motion because they had already completed their settlement in reliance on the passage of time for appeal. The court subsequently denied for lack of jurisdiction Bendis’s motion for leave to amend his notice of appeal to include an appeal of the court’s withdrawal of this extension. Besides appealing the original orders approving the settlements, Bendis also appeals the court’s order rescinding the extension of time to appeal and the court’s order denying leave to file an amended notice of appeal. The other appellants did not appeal the court’s refusal to extend the time for appealing the approval of the Mann and Pap-pajohn settlement.
DISCUSSION
I. Jurisdiction
Bendis did not appeal the district court’s approval of the Mann and Pappajohn settlement within the thirty days permitted by Fed.R.App.P. 4(a)(1). Therefore we do not have jurisdiction over his appeal unless the court validly extended the time for appeal. See Oda v. Transcon Lines, 650 F.2d 231, 233 (10th Cir.1981) (per curiam). Ben-dis argues on appeal that his notice of appeal was effective because the court abused its discretion when it withdrew the extension of time to appeal the approval of the Mann and Pappajohn settlement. See Jones v. W.J. Servs., Inc. (In re Jones), 970 F.2d 36, 39 (5th Cir.1992) (noting that appellate court must affirm district court’s decision whether to extend time under Rule 4(a)(6) unless the court abused its discretion).
Because Bendis and the other nonsettling defendants did not receive notice of the court’s judgment denying the motion to reconsider, Rule 4(a)(6) allowed the court to extend the time for them to appeal if doing so would not prejudice any party. The district court rescinded the extension of time to appeal because it would prejudice Mann and Pappajohn. We must accept this finding of prejudice because it is not clearly wrong. See In re Marchiando, 13 F.3d 1111, 1114 (7th Cir.1994) (treating prejudice under Rule 4(a)(6) as a factual finding).
Mann and Pappajohn had already completed their settlement with TBG before Bendis asked for an extension of time to appeal. They told the district court that they had made their settlement payment relying on the passage of time for anyone to appeal the court’s approval of their settlement. Extending the time for appeal would prejudice Mann and Pappajohn if they did settle in reliance on the finality of the court’s approval. See Fed.R.App.P. 4 advisory committee’s note to 1991 amendment (“Prejudice might arise, for example, if the appellee had taken some action in reliance on the expiration of the normal time period for filing a notice of appeal.”). Even though they could get their settlement payment back if the settlement was reversed on appeal, they still would lose the settlement itself as well as some of their money’s value.
Bendis argues that Mann and Pappajohn could not have relied on the passage of time for appeal because they actually completed their settlement before the time for appeal had passed. Bendis claims that his motion to [921]*921reconsider extended the time for appeal to March 20, thirty days after the court denied the motion on February 18. Since Mann and Pappajohn completed their settlement on March 18, Bendis reasons, they could not have relied on the finality of the settlement approval.
However, Bendis’s motion to reconsider did not extend the time for appealing the court’s approval of the Mann and Pappajohn settlement. The court separately approved the two settlements in two different judgments, both of which it certified as final and appealable under Rule 54(b). As the district court later found, Bendis moved to reconsider only the approval of the Shook settlement. His motion only referred to a single “Order and Judgment dated January 4, 1993.” Appellant’s App. at 855. The motion itself indicates that this refers to the Shook order, not the Mann and Pappajohn order, because it specifically challenges the court’s “Order barring Bendis from pursuing any potential state law negligence action against the defendant Shook, Hardy & Bacon.” Id. The motion also referred to the court’s December 30 memorandum, which discussed both settlements, but clearly did so only because that memorandum discussed the Shook settlement, not because Bendis intended to imply a challenge to both judgments of January 4. Furthermore, Bendis’s supporting memorandum challenges only the Shook approval, and the district court’s opinion rejecting the motion discusses only the Shook settlement.
Since Bendis moved to reconsider only the Shook approval, the motion extended the time to appeal only that judgment. Bendis correctly observes that a motion to reconsider a judgment allows the court to alter that judgment on any grounds and extends the time for any party to appeal that judgment on any issue. See Fed.R.App.P. 4(a)(4); Diaz v. Romer, 961 F.2d 1508, 1510 (10th Cir.1992); Varley v. Tampax, Inc., 855 F.2d 696, 699-700 (10th Cir.1988). But a motion to reconsider one final judgment does not extend the time to appeal another final judgment just because they are part of the same litigation. See Wielgos v. Commonwealth Edison Co., 892 F.2d 509, 511-12 (7th Cir.1989) (explaining that a motion to reconsider a separately appealable award of costs does not effect the time for appealing the judgment on the merits); Martin v. Campbell, 692 F.2d 112, 114-16 (11th Cir.1982) (observing that a motion for a new trial by one defendant also extends the time to appeal a verdict in favor of a second defendant, unless the district court certifies the judgments as separate and final under Rule 54(b)).
Apparently Bendis also misinterprets the statement in Rule 4(a)(4) that “the time for appeal for all parties runs from the entry of the order” denying a Rule 59(e) motion. This simply means that the motion extends the time in which “all parties” can appeal the judgment to which the Rule 59(e) motion applies, not that it extends the time to appeal any other separate judgment relating to some other party.
The time to appeal the Mann and Pappajohn settlement therefore expired on February 5, thirty days after the court entered its judgment approving the settlement. Mann and Pappajohn did not complete their settlement until March 18, well after the time for appeal had passed. Therefore the court’s finding that Mann and Pappajohn relied on the finality of the approval is not clearly erroneous, and extending the time for appeal would have prejudiced them.
Bendis also points out that the district court originally granted the extension of time under both Rule 4(a)(5) and Rule 4(a)(6), yet the court did not discuss Rule 4(a)(5) when it rescinded the extension. However, the court could not have granted the extension under Rule 4(a)(5) in the first place, since Rule 4(a)(5) would only permit an extension if Bendis requested it within thirty days after his time for appeal expired. As we have explained, the time for appealing the approval of the Mann and Pappajohn settlement expired on February 5. The court therefore could grant an extension under Rule 4(a)(5) only if Bendis requested it by March 7. Bendis first asked for an extension on April 15. Therefore, despite the court’s reference to Rule 4(a)(5), the court could not properly grant the extension under that rule, and the court’s failure to discuss it when rescinding [922]*922the extension is irrelevant. Nor could the motion for an extension of time itself qualify as a proper notice of appeal because it was not “filed within the time constraints of an approved Rule 4(a)(5) motion.” Listenbee v. City of Milwaukee, 976 F.2d 348, 350 (7th Cir.1992); see also Smith v. Barry, 502 U.S. 244, -, 112 S.Ct. 678, 682, 116 L.Ed.2d 678 (1992) (“If a document filed within the time specified by Rule 4 gives the notice required by Rule 3, it is effective as a notice of appeal.”).
Finally, Bendis claims that we can hear his appeal because he appealed late in reliance on the district court’s order extending the time for appeal. Bendis blatantly misrepresents the “unique circumstances” doctrine, which permits a late appeal if the appellant received an extension before the original appeal period expired and filed an appeal within that extended period, even though the court later rescinded the extension. See Bernstein v. Lind-Waldock & Co., 738 F.2d 179, 182-83 (7th Cir.1984) (“[I]f before the time for filing the notice of appeal has expired the district judge grants an extension of time ... and the appellant relies on the extension, the notice of appeal is timely if filed within the extended time.”); Stauber v. Kieser, 810 F.2d 1, 1-2 (10th Cir.1982) (per curiam); National Indus. v. Republic Nat’l Life Ins., 677 F.2d 1258, 1264 (9th Cir.1982). But Bendis did not request an extension before his original thirty-day appeal period expired. Even by his own account, his initial appeal period expired on March 20, but he did not request an extension until April 15. He therefore did not let his original appeal period expire in reliance on an extension, and the unique circumstances doctrine does not excuse his late appeal.
Consequently, we do not have jurisdiction over Bendis’s appeal of the order approving the Mann and Pappajohn settlement. We affirm both the district court’s order granting the motion to reconsider its order extending the time for appeal and the court’s denial of leave for Bendis to file an amended notice of appeal. As for Mann and Pappajohn’s motion to dismiss Bendis’s appeal, they filed their motion well beyond the “favored” fifteen-day period without explaining why filing within fifteen days of the notice of appeal was “impracticable.” 10th Cir.R. 27.2.1. Because we do not have jurisdiction regardless of Mann and Pappajohn’s motion to dismiss, we deny that motion without considering when we should accept late motions to dismiss.
Since none of the other defendants timely appealed the approval of the Mann and Pap-pajohn settlement, we do not have jurisdiction to review the $200,000 judgment reduction ordered by the court in its judgment approving the Mann and Pappajohn settlement. Nor may we review the scope of the bar order in that order and judgment.
II. Bar of Contribution Claims
The court’s order approving the Shook settlement barred any of the other defendants from bringing contribution or other related claims against Shook. The court compensated the nonsettling defendants for the loss of their claims against Shook by ordering a pro tanto judgment reduction. That is, the court ordered any judgment against the nonset-tling defendants to be reduced by the amount Shook pays in settlement to TBG.
We have not previously considered whether courts can bar contribution and other related claims in order to facilitate partial settlement in federal securities cases. Nor have we decided whether and how courts should reduce the trial award to reflect the settlement. See FDIC v. Geldermann, Inc., 975 F.2d 695, 698, 700 (10th Cir.1992). The district court inferred from an earlier Tenth Circuit decision that we would adopt the pro tanto method. See Hess Oil Virgin Islands Corp. v. UOP, Inc., 861 F.2d 1197, 1208-10 (10th Cir.1988). That decision only considered state law claims, however, and dealt with the very different issue of whether the court should deduct the plaintiffs percentage of fault before or after a pro tanto settlement credit. Id.
We review de novo whether ordering a pro tanto judgment reduction permits the court to bar contribution claims based on Rule 10b-5 liability. The appellants have said that they do not deny the court’s power [923]*923to bar claims; they only argue that the court must order a proportional fault credit rather than a pro tanto credit. Regardless of how the appellants state the issue, however, they cannot ask us or the district court to order a different judgment credit than the pro tanto credit to which the settling parties agreed. The settlement is contingent on the court ordering a pro tanto credit; if the court orders some other credit, there is no settlement. The appellants therefore cannot claim that the court erred because it did not order a proportional credit. The court could not have made such an order. The issue can only be whether the court erred by approving the settlement and ordering the requested contribution bar and pro tanto credit.
The appellants do not argue that the court erred simply because it ordered a pro tanto judgment reduction. In fact, they would have no reason to object to the pro tanto credit if the court left them free to bring contribution claims for any difference between the credit and the settling defendants’ share of the liability. Despite their apparent misunderstanding of how the court might have erred, the appellants essentially maintain that the court erred by barring their contribution claims. They argue that the bar order was an error because the court did not also order adequate compensation for their barred contribution claims. We agree that the bar order was impermissible, but for a different reason. We conclude that orders barring contribution claims are permissible only because a court or jury has or will have properly determined proportional fault and awarded the equivalent of a contribution claim, not because of the compensatory award alone. Since the court did not decide the settling defendants’ proportional fault and order a credit in that amount, the court had no power to bar the nonsettling defendants’ contribution claims.
We disagree with the concurrence that we could reverse “more economical[ly]” by analogizing to McDermott, Inc. v. AmClyde, — U.S. -, 114 S.Ct. 1461, 128 L.Ed.2d 148 (1994). Concurring Op. at 932. Since McDermott only required proportional fault credits in admiralty cases, see McDermott, — U.S. at -, 114 S.Ct. at 1464-65, we cannot say that approving a different credit in this ease was reversible error. Furthermore, the court in McDermott had the freedom to weigh and choose from various possible credits because the settlement itself did not mandate a credit. See id. at -, 114 S.Ct. at 1463. The court in this case, however, could only approve or disapprove a settlement with a pro tanto credit. McDer-mott explains why courts should choose a proportional fault credit when they are free to choose, and orders them to do so in admiralty, but it does not explain why approving a different credit on which a settlement depends would be an error where the Supreme Court has not mandated a proportional fault credit. Agreeing with McDermott that a pro tanto credit does not fully compensate for lost contribution rights does not lead to the conclusion that approving a pro tanto credit itself is a reversible error. It does suggest that barring contribution claims to recover any difference between the credit and the settling defendants’ proportional fault would be inappropriate, but it does not explain why, since no bar order was at issue in McDer-mott. We explain here why barring contribution claims in such a case is an error.
Defendants have a federal right to contribution in Rule 10b-5 actions. Musick, Peeler & Garrett v. Employers Ins., — U.S. -, -, 113 S.Ct. 2085, 2090-91, 124 L.Ed.2d 194 (1993). Although related to an implied cause of action, this contribution right is statutory, not merely equitable. The Supreme Court recognized the right to contribution because Congress would have authorized contribution claims if it had expressly authorized the 10b-5 action. Id. The Court did not create a federal common law contribution right, and refused to consider the equity or efficiency of the contribution right. Id. at -, 113 S.Ct. at 2090. Courts therefore have no power to take away that right for equity or policy reasons, even if such reasons might justify modifying or extinguishing common law contribution rights.
A Encouraging Settlement
Nevertheless, courts often have barred claims for contribution in Rule 10b-5 actions when the parties to a partial settlement have [924]*924requested a bar order. Almost every court that has identified the source of its power to enter a bar order has relied entirely on precedent and the federal policy encouraging settlements. See, e.g., Wald v. Wolfson (In re U.S. Oil & Gas Litig.), 967 F.2d 489, 494 (11th Cir.1992); Kovacs v. Ernst & Young (In re Jiffy Lube Sec. Litig.), 927 F.2d 155, 160 (4th Cir.1991); In re Atlantic Fin. Mgmt., Inc. Sec. Litig., 718 F.Supp. 1012, 1016 (D.Mass.1988). These courts typically reason that defendants will never agree to a partial settlement if the court does not protect them from further exposure by barring contribution and similar claims against them. See, e.g., Jiffy Lube, 927 F.2d at 160 (“The justification for imposing a bar to contribution ... is that ... the right to contribution removes the incentive to settle....”); In re Nucorp Energy Sec. Litig., 661 F.Supp. 1403, 1408 (S.D.Cal.1987) (asserting that without a bar order, “partial settlement of any federal securities case before trial is, as a practical matter, impossible”).
We disagree that the interest in settlement can give courts the power to bar statutory contribution claims. Although federal policy generally encourages settlement, that policy does not override statutory rights. Courts may not extinguish such rights in order to facilitate settlement unless the statute authorizes them to do so. Nor do judgment reductions or other compensation alone justify barring statutory claims. Courts do not have the general power to bar statutory claims just because they offer fair compensation. The absence of a statutory provision forbidding such orders does not suggest that courts have the power to issue them. But see Franklin v. Kaypro Corp., 884 F.2d 1222, 1229 (9th Cir.1989) (reasoning that since “nothing in either the statute or our prior decisions ... says contribution cannot be satisfied prior to a full trial,” the “overriding public interest” in settlements permits bar orders).
Even if courts could assume this power when necessary to permit settlement, bar orders are not necessary to partial settlements. Although a bar order may be more convenient than the alternatives, the court and the plaintiff can protect settling defendants from further exposure without a bar order. For example, the plaintiff can agree to indemnify the settling defendants against claims by the nonsettling defendants that are based on their liability to the plaintiff. See, e.g., In re San Juan Dupont Plaza Hotel Fire Litig., 907 F.2d 4, 5 & n. 2 (1st Cir.1990); Resolution Trust Corp. v. Evans, No. 92-0756, 1993 WL 354796, at *1 (E.D.La. Sept. 3, 1993); In re Washington Pub. Power Supply Sys. Sec. Litig., 720 F.Supp. 1379, 1398-99 (D.Ariz.1989), aff'd, Class Plaintiffs v. City of Seattle, 955 F.2d 1268 (9th Cir.), cert. denied, — U.S. -, 113 S.Ct. 408, 121 L.Ed.2d 333 (1992); In re National Student Mktg. Litig., 517 F.Supp. 1345, 1346 (D.D.C.1981). The plaintiff also may agree to defend the settling defendants, so that the defendants do not face more litigation expenses. Such an indemnity agreement encourages defendants to settle just as much as a bar order does. See McDermott, — U.S. at -, 114 S.Ct. at 1467. The settling defendant does not have to worry about further exposure, while the plaintiff assumes no greater risk than it would with a bar order and appropriate compensation for barred claims. Sometimes this may add to the burden on the courts, id., but most of the time the nonsettling defendants will have brought cross-claims for contribution that the court can decide in the same trial even though the settling defendants have settled the primary claim against them. This burdens the court no more than a proportional fault credit would. Whatever the added burden, necessity cannot justify bar orders. They may be helpful, but they are not necessary.
If Congress wants to allow bar orders, it can follow the many states that have authorized courts to bar contribution claims if they provide a specified reduction of any judgment against nonsettling defendants. See, e.g., Cal.Code of Civ.Proc. § 877.6 (Supp. 1994); Ill.Ann.Stat. ch. 70, ¶ 302 (Smith-Hurd 1989); Md.Ann.Code art. 50, § 20 (1991); Mass.Gen.L. ch. 231B, § 4 (1986); N.Y.Gen.Oblig.L. § 15-108 (1989); Wash. Rev.Code Ann. § 4.22.060 (West 1988); Jiffy Lube, 927 F.2d at 160 n. 2. Without such a law, ordering a judgment credit alone does [925]*925not give courts the power to bar federal statutory contribution claims.
Admittedly, private agreements do not create the unique incentives to settlement that a bar order with a pro tanto credit creates. The pro tanto method guarantees that the plaintiff will get whatever the jury awards at trial. The plaintiff does not risk making a low settlement and getting less than that amount. The plaintiff therefore can enjoy the benefits of settling, including funding further litigation, without discounting its recovery, forcing the nonsettling defendant to pay most of that discount after trial. See Franklin, 884 F.2d at 1230. Defendants also have a greater interest in settling in order to avoid paying more than their share after trial. See McDermott, — U.S. at -, 114 S.Ct. at 1468. However, the pro tanto method creates these additional incentives by unfairly shifting the risk of low settlements from the plaintiff to the nonsettling defendants. See id. at -, 114 S.Ct. at 1469; Alvarado Partners, L.P. v. Mehta, 723 F.Supp. 540, 553 (D.Colo.1989).
Although private alternatives may not increase the incentives to settle, they generally do not decrease the incentives either. They leave the plaintiff in a position similar to a full settlement. An indemnity agreement lets a plaintiff enjoy the benefits of a favorable settlement and suffer the consequences of a low settlement. These alternatives therefore do not penalize a plaintiff for settling, and give the plaintiff just as much of an incentive to settle with a defendant for an amount approximating the defendant’s likely liability.
Indemnity agreements or other private arrangements may discourage partial settlement with certain defendants, however. The plaintiff has less incentive to settle with defendants whose probable liability significantly exceeds their ability to pay. If the plaintiff settles for what such defendants can pay, and a jury later sets their share of damages at a higher amount, the plaintiff will come up short. But if the plaintiff keeps such defendants in the case, the plaintiff will be able to collect the entire judgment from the other defendants and make them seek contribution from the defendants with fewer resources. See TBG Inc. v. Bendis, 811 F.Supp. 596, 604 n. 16 (D.Kan.1992); FDIC v. Geldermann, Inc., 763 F.Supp. 524, 529-30 (W.D.Okla.1990), rev’d, 975 F.2d 695 (10th Cir.1992).
Nevertheless, the interest in settlement does not justify depriving third parties of their statutory rights. See United States Fidelity & Guar. v. Patriot’s Point Dev. Auth., 172 F.Supp. 1565, 1575 (D.S.C.1991) (“To the extent there is tension between the goal of promoting settlements and fundamental fairness to litigants, the latter must prevail.”); In re Sunrise Sec. Litig., 698 F.Supp. 1256, 1261 (E.D.Pa.1988); cf. United States v. Reliable Transfer Co., 421 U.S. 397, 408, 95 S.Ct. 1708, 1714, 44 L.Ed.2d 251 (1975) (“Congestion in the courts cannot justify a legal rule that produces unjust results in litigation simply to encourage speedy out-of-court accommodations.”). The unique settlement incentives of the pro tanto method therefore do not justify its use.
B. All Writs Act
Although the interest in settlement cannot empower a court to bar contribution claims, the All Writs Act may. At least one court has claimed the power to issue a bar order under the All Writs Act. In re Washington Pub. Power Supp. Sys. Sec. Lit., 720 F.Supp. 1379, 1399 (D.Ariz.1989). The All Writs Act authorizes federal courts to “issue all writs necessary or appropriate in aid of their respective jurisdictions.” 28 U.S.C. § 1651(a). This authority includes enjoining further suits relitigating issues a court has already decided. See, e.g., Farmers Bank v. Kittay (In re March), 988 F.2d 498, 500 (4th Cir.) (“The All Writs Act empowers a federal court to enjoin parties before it from attempting to relitigate decided issues and to prevent collateral attack of its judgments.”), cert. denied, — U.S. -, 114 S.Ct. 182, 126 L.Ed.2d 141 (1993); In re Baldwin-United Corp. (Single Premium Deferred Annuities Ins. Litig.), 770 F.2d 328, 335 (2d Cir.1985).
We agree with the concurrence, the Ninth Circuit, and most other courts that have decided the issue that the Rule 10b-5 contribution right entitles a defendant to recover the amount of damages attributable to [926]*926another party’s fault. See Smith v. Mulvaney, 827 F.2d 558, 560-61 (9th Cir.1987). Therefore, if a court or jury properly decides the settling defendants’ share of the fault and somehow credits that amount to the nonset-tling defendants, the All Writs Act probably would authorize an order barring future contribution claims because they would necessarily relitigate that issue. However, the finding on which a credit is based precludes relitigation, not the credit itself. If there were only a credit and no finding of relative fault, a nonsettling defendant would be free to file a subsequent contribution action to recover the difference between the credit awarded and the amount he claims is attributable to the settling defendants’ fault. The court in that later action would then have to decide the parties’ proportional fault in order to decide whether the nonsettling defendant paid more than his share of the liability even after the judgment credit. Only the finding of proportional fault, on which a proportional fault credit is based, can estop the nonset-tling defendant from litigating that issue in a later contribution action.
The concurrence suggests that “statutory construction” and “traditional equitable powers” are “ample authority” for a bar order. Concurring Op. at 936. The concurrence does not explain at all how it interprets section 10(b) to authorize bar orders. Such extraordinary statutory construction could infer such authorization from almost any statute. As for “traditional equitable powers,” the early Supreme Court cases cited by the concurrence simply express the authority now codified in the All Writs Act, 28 U.S.C. § 1651(a). Even if traditional equitable powers were distinct from the All Writs Act, the concurrence cites no case indicating that those traditional powers give courts greater freedom to enjoin future litigation than the All Writs Act does. The concurrence has not identified any source of power to enter bar orders other than the one we have identified: the power to bar relitigation of issues already decided by the court. See Root v. Woolworth, 150 U.S. 401, 411-12, 14 S.Ct. 136, 138-39, 37 L.Ed. 1123 (1893) (describing courts’ power to “effectuate their own decrees by injunctions ... in order to avoid the relitigation of questions once settled between the same parties”); Farmers Bank, 988 F.2d at 500; Baldwin-United, 770 F.2d at 335; Wood v. Santa Barbara Chamber of Commerce, 705 F.2d 1515, 1524 (9th Cir.1983), cert. denied, 465 U.S. 1081, 104 S.Ct. 1446, 79 L.Ed.2d 765 (1984). Therefore, when a party challenges a bar order, we logically should ask first whether the court already decided the issues or claims the relitigation of which the order enjoins. If not, the court cannot bar future litigation of those issues.
The district court in this case did not order that the jury would decide the settling defendants’ share of the liability. Nor did the court determine their share in the fairness hearing, because the court relied on many other factors in deciding that the settlement was fair. See TBG, 811 F.Supp. at 605-07. Such a hearing could not possibly estop relit-igation of issues decided therein, and therefore permit a bar order under the All Writs Act, unless the court directly decided the settling defendants’ share of fault and followed procedures that would fully and fairly adjudicate the issue. See Kremer v. Chemical Constr. Co., 456 U.S. 461, 480-81, 102 S.Ct. 1883, 1896-97, 72 L.Ed.2d 262 (1982); Willner v. Budig, 848 F.2d 1032, 1034 (10th Cir.1988) (per curiam), cert. denied, 488 U.S. 1031, 109 S.Ct. 840, 102 L.Ed.2d 972 (1989); Greenblatt v. Drexel Burnham Lambert, Inc., 763 F.2d 1352, 1361 (11th Cir.1985). The district court did not hold such a hearing before ordering the pro tanto credit. In fact, settlements with such hearings would be rare, since defendants could not receive a discount reflecting the uncertainty of trial. See McDermott, — U.S. at -, 114 S.Ct. at 1467.
Furthermore, we doubt that even an extensive hearing that would ordinarily estop relitigation could justify an order barring a nonsettling defendant’s contribution claims against a settling defendant. The All Writs Act only authorizes such orders in aid of the court’s jurisdiction. 28 U.S.C. § 1651(a). It does not authorize a court to assume jurisdiction over claims not otherwise before it. A court therefore cannot bar further contribution claims if it does not have jurisdiction to decide the defendants’ proportional fault in the first place. The district court may not [927]*927have had jurisdiction to decide Shook’s contribution liability, especially since the nonset-tling defendants had not filed cross-claims for contribution. We doubt that the potential contribution defendant’s request to have the court decide its contribution liability could give the court jurisdiction to effectively decide an unwilling potential contribution plaintiffs claim. Cf. Ebanks v. Great Lakes Dredge & Dock Co., 688 F.2d 716, 722 (11th Cir.1982) (holding that trial court erred by requiring the jury to allocate fault between a nonsettling defendant and a settling defendant instead of leaving that issue for the nonsettling defendant’s separate contribution action), cert. denied, 460 U.S. 1083, 103 S.Ct. 1774, 76 L.Ed.2d 346 (1983). But since the district court did not decide the settling defendants’ proportional fault anyway, we do not have to decide in this case if and when a district court has the power to determine a settling defendant’s proportional fault in a fairness hearing or at trial.
The concurrence complains that it is “difficult to understand” why we raise this question. Concurring Op. at 933. The concurrence first observes that a proportional fault credit “is fully consistent with the purposes behind nonsettling defendants’ relative fault contribution rights.” Id. We agree, of course, but courts may not decide claims over which they have no jurisdiction just because doing so is “consistent with the purposes behind” those claims.
The concurrence also points out that the settling defendants’ absence is unimportant because they are voluntarily absent, they do not face further liability, and the plaintiff will litigate their liability fully. Again, we agree, but these observations are beside the point. The settling defendants’ absence is relevant to this question only because it makes the proportional fault decision seem even further beyond the court’s jurisdiction. In a case such as this one, once the settling defendants are gone neither the potential contribution claim nor the parties to that potential claim are before the court.
Our concern that the potential contribution claim was not before the court “mystified” the concurrence because the court should be able “to consider the question as part of its resolution of the judgment credit required,” which is “highly relevant to the remaining parties’ dispute.” Concurring Op. at 934. This reasoning mystifies us. The concurrence assumes the question to be decided, claiming that a court has authority to decide relative fault because it has the authority to enter a judgment credit based on that finding. But there is no specific grant of authority to enter proportional fault judgment credits. A court only has authority to enter a judgment credit because doing so may be part of resolving the dispute before it. The court cannot step outside the boundaries of that dispute to set a judgment credit just because it will be relevant to and prevent a separate dispute in the future between the settling and nonsettling defendants. Without a contribution claim, the court does not have before it a legal claim to which proportional fault is relevant. Liability in Rule 10b-5 cases is strictly joint and several and is never allocated among individual defendants in deciding the plaintiffs claim. See U.S. Indus. v. Touche Ross & Co., 854 F.2d 1223, 1261 (10th Cir.1988); G.A. Thompson & Co. v. Partridge, 636 F.2d 945, 963 (5th Cir.1981); cf. Borden, Inc. v. Florida East Coast Ry., 772 F.2d 750, 753 (11th Cir.1985) (holding that trial court plainly erred by allocating joint and several tort damages according to jury’s determination of defendants’ relative fault). The defendants’ relative fault is therefore irrelevant to the plaintiffs claim. A court cannot make Rule 10b-5 liability comparative by allocating fault among defendants when their contribution claims are not before the court. Cf. Ebanks, 688 F.2d at 722 (holding that court erred by directing jury to decide settling defendant’s proportional fault in plaintiffs case rather than in separate contribution action). Nor can the court accept jurisdiction over a question not relevant to the legal claim before it just because the plaintiff or even both parties ask it to.
The concurrence claims that the Supreme Court’s admiralty decision in McDer-mott implies that the court could allocate proportional fault in this situation. See Concurring Op. at 934. However, unlike securities law, admiralty law is comparative, which [928]*928allows courts to allocate fault among the defendants in deciding the plaintiffs claim, regardless of whether the defendants have filed contribution claims. See United States v. Reliable Transfer Co., 421 U.S. 397, 411, 95 S.Ct. 1708, 1715-16, 44 L.Ed.2d 251 (1975); Leger v. Drilling Well Control, Inc., 592 F.2d 1246 (5th Cir.1979). As the Supreme Court explained in McDermott, an admiralty defendant will be liable only for its proportional share of the liability unless “factors beyond the plaintiffs control” limit other responsible parties’ liability. McDermott, — U.S. at -, 114 S.Ct. at 1471. Since settlement is not such an “outside force[ ],” id., the nonsettling defendants in admiralty remain liable only for their share of the liability. The issue of proportional fault therefore remains relevant to the primary claim in admiralty. Although both admiralty law and securities law seek to allocate fault among defendants, they do so in different ways. Securities law requires a contribution claim before the court can order the jury to divide damages among the defendants. A court cannot assume jurisdiction over a contribution claim that is not before it simply because that claim has the same purpose as a comparative fault or other law that allows the court to allocate fault in the primary claim. The concurrence cites three cases allowing trial courts to order proportional fault credits “without the presence of settling defendants.” Concurring Op. at 934. But none of those cases discusses whether a court has jurisdiction to decide proportional fault, especially when the nonsettling defendants have not filed contribution claims. In fact, none of them even says that the nonsettling defendants had not filed contribution claims, so we doubt that the issue even occurred to those courts. In any event, their failure to discuss the possible jurisdictional problem is hardly persuasive authority.
Nevertheless, we do not have to decide now whether a court could have the jury allocate proportional fault in a ease such as this one. The district court erred because neither the court nor the jury determined the settling defendants’ proportional fault and therefore the court had no basis for barring the nonsettling defendants’ federal contribution claims.
III. Bar of Bendis’s Independent Claims
Bendis also challenges the court’s order approving the Shook settlement because it bars his independent claims along with his contribution and similar claims. We agree that the court abused its discretion by barring Bendis’s independent state law claims.
State law governs whether the court can bar state law claims and under what circumstances. See In re Granada Partnership Sec. Litigs., 803 F.Supp. 1236, 1239 (S.D.Tex.1992); Sunrise Sec., 698 F.Supp. at 1257. However, apparently neither the Kansas legislature nor the Kansas courts have said whether courts can bar claims by a nonsettling defendant against a settling defendant.
Even if Kansas would permit bar orders of related claims, we do not think Kansas would go beyond any other jurisdiction and allow courts to bar even independent claims. Courts that have allowed bar orders have only barred claims in which the damages are “measured by” the defendant’s liability to the plaintiff. Alvarado Partners, 723 F.Supp. at 554. Besides contribution and indemnity claims, these include any claims in which the injury is the nonsettling defendant’s liability to the plaintiff. See U.S. Oil & Gas, 967 F.2d at 495-96; Alvarado Partners, 723 F.Supp. at 554. No court has authorized barring claims with independent damages.
The district court’s bar order in this case goes beyond claims measured by Bendis’s potential liability to TBG. The order bars Bendis
from asserting or pursuing, against Shook, Hardy & Bacon, ... any and all claims ... whether for contribution, indemnity or otherwise, ... involving or that arise out of or relate to any claim, demand, right or cause of action connected with, arising out of or based in whole or in part on any of the acts, omissions, facts, matters, transactions or occurrences alleged, described ... or involved or connected in any manner with this action or the allegations of or any [929]*929judgment sought or obtained by the TBG Entities in this action....
Order & Judgment, Appellant’s App. at 344. This order is virtually unlimited. It could bar truly independent claims because it bars “all claims ... that ... relate to any claim ... based ip whole or in part on any of the ... facts ... connected in any manner with this action.” This would include independent claims that were related factually but in which the damages were not based on Ben-dis’s primary liability to TBG.
The appellees argue that even if the order is too broad, Bendis has no potential independent claims anyway. Since TBG bought all of the Continental stock, they reason, no third party could sue Bendis for some misconduct in the Continental acquisition that Bendis in turn might blame on Shook. Nevertheless, the district court did not find that Bendis has no independent claim, and we cannot be sure that Bendis might not have some independent claim that the order would bar. Furthermore, the court should not purport to bar claims it has no power to bar, even if it thinks that there really are no such claims.
Although Bendis challenges the scope of the bar order generally, he particularly complains that the order would bar a negligence claim against Shook if he was liable to TBG for misrepresentations. Kansas might permit a court to bar such a claim because it would be based on Bendis’s liability to TBG, not on independent damages. Bendis would be suing Shook to recover the damages that he had to pay TBG, without which he would have no damages to claim from Shook. See, e.g., U.S. Oil & Gas, 967 F.2d at 495-96 (holding that court properly barred fraud and negligence claims against settling defendants because those claims were just another theory for recovering damages defendant had to pay to plaintiff); South Carolina Nat’l Bank v. Stone, 749 F.Supp. 1419, 1433 (D.S.C.1990).
However, we think Kansas would permit barring only claims for which the court provided adequate compensation through a judgment credit or other means. See, e.g., Patriot’s Point, 772 F.Supp. at 1572. The court did not order adequate compensation for this barred claim. Bendis claims that Shook prepared a document limiting his liability, and therefore he would sue for any amount over that limit that he had to pay to TBG. Even with a fairness hearing, a pro tanto credit would not be even roughly equal to the value of this claim except by accident. The district court did not consider at all whether the settlement credit compensated Bendis for the loss of this potential claim when it found the settlement to be fair. We therefore conclude that the court abused its discretion in barring Bendis’s potential malpractice claim and possible independent claims against Shook.
We AFFIRM the district court’s order granting the motion to reconsider its order extending the time for appeal, as well as the court’s denial of leave for Bendis to file an amended notice of appeal. We therefore DISMISS Bendis’s appeal of the court’s order approving the Mann and Pappajohn settlement for lack of jurisdiction. We also VACATE the district court’s order approving the Shook settlement, entered January 5, 1993, insofar as it impermissibly bars federal statutory contribution claims, Bendis’s independent state law claims, and Bendis’s potential malpractice claim without providing adequate compensation.