SCHROEDER, Circuit Judge:

The principal issue in this appeal is whether the district court erred in holding that under federal maritime law in a personal injury case, a good faith settlement between a plaintiff and defendant bars an action for contribution by a co-defendant against the settling defendant. The secondary issue is, if a good-faith settlement does constitute such a bar, did the district court err in holding that the settlement in this particular case was in good faith, and hence that the contribution action could not be maintained. The case presents a conundrum that the American Law Institute's Restatement (Second) of Torts has been unable to solve, implicating issues that have divided the circuits and district courts, prompted much scholarly discussion, and spawned differing provisions in three separate model statutes. The issues pertain to the effect of the release of one tortfeasor upon a plaintiff's right to recover against a second tortfeasor, and the latter's rights to contribution from the settling tortfeasor.

BACKGROUND

This case arises from a boating accident that injured the plaintiff, Sally Eisenger, who is not a party to this appeal. At the time of the accident, she was a passenger on a boat owned by appellee Douglas Christopher, the Pandora III. The Pandora III was towing the Mahia, a disabled boat owned by appellant Irwin Miller. Christopher was at the time acting as a good samaritan and had no insurance. The accident occurred when a two rope broke and struck plaintiff Eisenger, causing serious eye injuries. There is no question that the owners of both boats were partially at fault.

Eisenger sued both Miller and Christopher. Christopher settled by transferring to Eisenger a Porsche automobile and a half interest in real property. Miller, acting through his insurance company, eventually settled for policy limits, paying Eisenger $300,000.

Miller then instituted the proceedings leading to this appeal by filing a third party complaint against Christopher. The third party complaint sought contribution for the amount of the Miller insurance settlement that Miller alleged exceeded his proportionate share of liability based on fault. Christopher defended on the ground that he had settled in good faith and that the settlement constituted a bar to the contribution action. He asked the court to look to the tort law of California, where such a good faith settlement bar has been created by statute. Cal.Civ.Proc.Code § 877 (West Supp. 1989).

The district court properly recognized that it was not to apply the state law of the forum, but was to look to applicable principles of federal admiralty law. See Pope Talbot, Inc. v. Hawn, 346 U.S. 406, 409-10, 74 S.Ct. 202, 205, 98 L.Ed. 143 (1953); Daughtry v. Diamond M Co., 693 F. Supp. 856 (C.D.Cal. 1988) (acknowledging that section 877 is not controlling under federal maritime law). The court's problem, however, was that neither the Supreme Court nor this court has decided any cases similar to this one. Further, district court cases seem in conflict, and there is no relevant federal statute. When faced with the choice of either applying a good faith bar rule, or else holding that the settlement and release did not bar an action for contribution under any circumstances, the district court chose the good faith bar approach.

The district court then held an evidentiary hearing on the issue of good faith. The court had to determine the total value of the plaintiff's claim, the relative fault of settling defendant Christopher, and the monetary value of the settlement. Given the unusual nature of the settlement, physical property rather than money, the troublesome point was putting a value on the automobile and the interest in real property. The court found that the total value of the claim was $500,000, Christopher's potential liability ranged from one-third to two-thirds, and the value of the settlement was $100,000. The court concluded that the settlement was not collusive and was in good faith.

In this appeal, Miller challenges both the district court's ruling that a good faith settlement bars a subsequent contribution action, and also its ruling that this particular settlement was in good faith.

DISCUSSION

The Good Faith Settlement Bar Against Contribution

We sympathize with the district court's difficulties in finding guidance from controlling authority on the settlement bar issue. There is none. Congress has traditionally given great leeway to the Supreme Court in establishing the guiding principles of federal maritime law in the personal injury field. Fitzgerald v. U.S. Lines Co., 374 U.S. 16, 20, 83 S.Ct. 1646, 1650, 10 L.Ed.2d 720 (1963). In the past fifteen years two Supreme Court decisions, while not deciding the issues presented here, have set forth significant principles relevant to our consideration of this case.

One such principle is that liability of two or more wrongdoers for damages should be apportioned according to the comparative degree of their fault. United States v. Reliable Transfer Co., 421 U.S. 397, 95 S.Ct. 1708, 44 L.Ed.2d 251 (1975). The Court in Reliable Transfer obliterated the last vestiges of the previous admiralty doctrine of "divided damages," which provided that damages were to be allocated equally between parties causing an accident regardless of their respective degrees of fault. The court looked to more modern principles of comparative negligence in abandoning the old rule, which it termed "archaic." Id. at 408, 95 S.Ct. at 1714. The Court recognized that the goal of a just and equitable allocation of damages can be more nearly attained by allocating damages according to proportion of fault. Id. at 411, 95 S.Ct. at 1715-16.

Four years after Reliable Transfer, the Court decided Edmonds v. Compagnie Generale Transatlantique, 443 U.S. 256, 99 S.Ct. 2753, 61 L.Ed.2d 521 (1979). There a longshoreman had been injured through the fault of both his stevedore employer and a shipowner. The employer was immune to common law suit by virtue of the Longshoremen's and Harbor Workers' Compensation Act. The injured longshoreman sued the shipowner. The shipowner, relying on Reliable Transfer, argued that its liability for damages should be limited to its proportionate share of fault. Rejecting that argument, the Supreme Court reaffirmed the principle that the longshoreman could require the shipowner to pay all the damages not attributable to the injured longshoreman's own negligence. The Court applied the traditional tort principles of joint and several liability which permit a plaintiff to elect to sue and recover in a common law action all damages from one wrongdoer. See also Self v. Great Lakes Dredge and Dock Co., 832 F.2d 1540, 1546 (11th Cir. 1987) (applying Edmonds), cert. denied, ___ U.S. ___, 108 S.Ct. 2017, 100 L.Ed.2d 604 (1988).

Clearly, after Reliable Transfer and Edmonds an injured party obtaining a money judgment against two jointly responsible wrongdoers may choose to seek satisfaction of the full amount from either one, regardless of the comparative degree of fault. Contribution provides the appropriate device in such circumstances for adjusting damages to achieve liability based on proportionate fault. Although at least one commentator has tended to read Edmonds more narrowly on its facts, see M. Morris, The Law of Maritime Personal Injuries 3rd § 170 (Supp. 1988), neither Reliable Transfer nor Edmonds modified the maritime rule that a wrongdoer who has paid more than his apportionable share of damages to a plaintiff may seek contribution from another wrongdoer who has not borne his share. See Self, 832 F.2d at 1546.

However, the problem that the district court below and now we face in this case has not been directly addressed by any Supreme Court case, since none of the Court's decisions have involved a settlement and release by one wrongdoer. The parties to this appeal, and several lower federal courts, have recognized that the policies behind these presently well-established admiralty principles, mirrored by developing tort principles of comparative negligence under state law, do not clearly point to one result concerning the wisdom of a settlement bar. This is because the proportional fault and contribution principles of Reliable Transfer and Edmonds do not take into account the policy of promoting settlements that settlement bars are designed to advance. The result is a tension between the policies favoring full recovery and those favoring final and efficient dispute resolution.

Thus the principles of joint and several liability underlying the Edmonds decision have generally led plaintiffs to argue that despite a settlement with one wrongdoer, they may pursue a total recovery of remaining damages from another wrongdoer, regardless of principles of comparative liability. See, e.g., Self, 832 F.2d at 1548 (discussing maritime law's role offering "special protection" to seamen). On the other hand, settling defendants, like appellee Christopher in this case, point to the principle of comparative fault, as well as the policy of encouraging settlements, to contend that a settlement bar is necessary. See, e.g., Leger v. Drilling Well Control, Inc., 592 F.2d 1246, 1249 (5th Cir. 1979) (discussing Reliable Transfer).

These competing policy considerations have led to a considerable degree of consternation on the part of the experts who have grappled with these problems. The consternation is exemplified by the inability of the American Law Institute to arrive at any synthesis or consensus. Restatement (Second) of Torts § 886A on "Contribution Among Tortfeasors" does not address the issue of the effects of one tortfeasor's settlement on the rights of other tortfeasors to contribution, other than to give a special Caveat warning that the Institute takes no position on the issue.

In comment m to section 886A, the Restatement discusses the Caveat and sets forth three possible approaches to the situation of the settling tortfeasor: (1) allowing an action for contribution against a settling tortfeasor by any other tortfeasor who has paid more than his equitable share of the plaintiff's claim; (2) imposing a bar to contribution claims against a settling tortfeasor, perhaps in conjunction with a requirement that the settlement be in "good faith"; and (3) reducing the claim of the plaintiff by the pro rata share of a settling tortfeasor's liability for damages, which has the effect of eliminating any reason to sue a settling tortfeasor for contribution. In this case the appellant asks us essentially to choose the first approach and allow contribution. The appellee urges the second approach, as reflected in California's legislative enactment of a good faith settlement bar. No one urges the third approach; however, this may well be because the possibility of reduction of the plaintiff's recovery was not a factor at the time of the policy limits settlement of the claim against Miller.

Comment m notes that the first approach, allowing for contribution, "has been called the fairest solution, but it has proved to be very discouraging to settlements." The comment notes that tortfeasors have no incentive to settle if they are not sure the settlement will extinguish their liability. The comment observes that the second approach, imposing a settlement bar, favors settlements, but opens the way to collusion between the plaintiff and the settling tortfeasor. The imposition of a prophylactic "good faith" rule can help, but such a rule calls the finality of the settlement into question until a court has passed on the issue of good faith, and so reduces the usefulness to the parties of the settlement. The comment observes that the third approach, the prorata reduction of the plaintiff's claim, "works more strongly against the interest of the injured party and may have the effect of discouraging him from entering into a settlement." It cites as an example the situation where a plaintiff may be dissuaded from settling for maximum insurance policy limits if the settlement will reduce his total claim by a much larger amount.

In the area of state tort law, each of the Restatement approaches is represented by the provisions of a model act. The original version of the Uniform Contribution Among Tortfeasors Act (UCATA), drafted in 1939, adopted the first approach. 12 U.L.A. 99 (1975). Unsatisfactory results with this approach led the drafters of the 1955 revision of UCATA to adopt the second approach instead. Id. at 99-100. The Uniform Comparative Fault Act, drafted in 1977 and meant for adoption in place of UCATA in states that adopt comparative negligence, adopted the third approach, finding it more in line with the aims of the comparative negligence system. 12 U.L.A. Supp. 40, 52-53 (1989).

When federal courts decide settlement bar issues under the Federal Tort Claims Act, we apply state law. See 28 U.S.C. § 2674 (1982 West Supp. 1989); see also Owen v. United States, 713 F.2d 1461, 1462 (9th Cir. 1983) (applying the settlement bar of Cal.Civ.Proc.Code § 877). In federal admiralty law, however, state law is not controlling. Decisions of federal courts have focused, but differed, on the relative merits of the second and third approaches. The case law is in some disarray.

The Fifth Circuit adopted the third approach in Leger v. Drilling Well Control, Inc., 592 F.2d 1246 (5th Cir. 1979), a case that predates Edmonds. The Eleventh Circuit, on the other hand, has refused to follow Leger, holding that it is inconsistent with Edmonds, and has adopted the second approach instead. Ebanks v. Great Lakes Dredge Dock Co., 688 F.2d 716 (11th Cir. 1982), cert. denied, 460 U.S. 1083, 103 S.Ct. 1774, 76 L.Ed.2d 346 (1983), on remand sub nom. Self, 832 F.2d at 1546. The First Circuit has also adopted the second approach. Joia v. Jo-Ja Serv. Corp., 817 F.2d 908, 915-17 (1st Cir. 1987) (relying on Edmonds and agreeing with Ebanks), cert. denied, 484 U.S. 1008, 108 S.Ct. 703, 98 L.Ed.2d 654 (1988). The allegiance of the Fifth Circuit itself to the Leger decision has shown some signs of erosion. See Simeon v. T. Smith Son, Inc., 852 F.2d 1421, 1430 n. 11, 1444-45 n. 9, 1454-55 (5th Cir. 1988); cert. denied, ___ U.S. ___, 109 S.Ct. 3156, 104 L.Ed.2d 1019 (1989); Hernandez v. M/V RAJAAN, 841 F.2d 582, 591 (5th Cir.), cert. denied, ___ U.S. ___, 109 S.Ct. 530, 102 L.Ed.2d 562 (1988).

However, other post- Edmonds cases have expressed faith in the continuing validity of Leger and its approach. In a leading district court decision, Kizer v. Peter Kiewit Sons' Co., 489 F. Supp. 835, 838-40 (N.D.Cal. 1980), Judge Poole followed the third approach, drawing a distinction between the statutory context of Edmonds and the common law approach of the 1977 Comparative Fault Act, and choosing to follow the latter. See also Bordelon v. Consolidated Georex Geophysics, 628 F. Supp. 810, 812-13 (W.D.La. 1986) (applying Leger); but see Sherrill v. Brinkerhoff Maritime Drilling, 615 F. Supp. 1021, 1037 (N.D.Cal. 1985) (following the second approach and applying Cal.Civ.Proc.Code § 877.6), aff'd in part, reversed in part on other grounds sub nom. Zipfel v. Halliburton Co., 832 F.2d 1477 (9th Cir. 1987), cert. denied, ___ U.S. ___, 108 S.Ct. 2819, 100 L.Ed.2d 921 (1988). Commentators also have been unable to come to any definite consensus on the proper approach to follow.

We have found no federal admiralty decision that adopts the first approach, permitting contribution even in the face of a good faith settlement. Although Sea-Land Serv. v. American Logging Tool Corp., 637 F. Supp. 240 (W.D.Wash. 1985), has been cited to us in support of that approach, that case deals with a different problem. That case concerns whether or not a settling tortfeasor may sue another joint tortfeasor for contribution. The extremely short opinion notes that both tortfeasors had settled with the plaintiff, id. at 240, but merely goes on to state that "admiralty law recognizes that even where a party to a lawsuit settles, it may still bring an indemnity action against a joint tort feasor." Id. at 241 (citing cases). The issue in this case, whether a settling joint tortfeasor may use a settlement as a defense against a suit for contribution, was apparently never raised there.

In this case we need not decide whether the good faith settlement bar of the second Restatement approach is better or more consistent with Supreme Court precedent than the third approach's pro rata reduction of plaintiff's recovery. Compare Kizer, 489 F. Supp. at 840-41, with Sherrill, 615 F. Supp. at 1037. This court recently grappled with similar issues in the context of a securities class action settlement. Franklin v. Kaypro Corp. 884 F.2d 1222 (9th Cir. 1989) (pet. for reh. pending). The only options presented to the district court below were to permit post-settlement contribution, which would have the effect of adopting the first approach, or else to deny contribution upon a finding of good faith settlement, in effect following the second approach, but not necessarily precluding the third approach. Given those alternatives, we conclude the district court chose the wiser course.

Denial of a settlement bar would interfere with policies favoring settlement. The Commissioners' Comment to the 1955 UCATA, which imposed the settlement bar of the second approach and replaced the first approach as implemented in the 1939 UCATA, bears ample testimony to the disadvantages of the first approach:

The idea underlying the 1939 provision was that the plaintiff should not be permitted to release one tortfeasor from his fair share of liability and mulct another instead, from motives of sympathy or spite, or because it might be easier to collect from one than from the other; and that the release from contribution affords too much opportunity for collusion between the plaintiff and the released tortfeasor against the one not released. Reports from the state where the Act is adopted appear to agree that it has accomplished nothing in preventing collusion. In most three-party cases two parties join hands against the third, and this occurs even when the case goes to trial against both defendants. "Gentlemen's agreements" are still made among lawyers, and the formal release is not at all essential to them. If the plaintiff wishes to discriminate as to the defendants, the 1939 provision does not prevent him from doing so.

The effect of [the first approach] has been to discourage settlements in joint tort cases, by making it impossible for one tortfeasor alone to take a release and close the file. . . . In many cases [plaintiff's] chief reason for settling with one [defendant] rather than another is that they hope to get more from the party with whom they do not settle. . . . No defendant wants to settle when he remains open to contribution in an uncertain amount, to be determined on the basis of a judgment against another in a suit to which he will not be a party. Some reports go so far as to say that the [first approach] has made independent settlements impossible. Many of the complaints come from plaintiff's attorneys, who say that they can no longer settle cases with one tortfeasor. . . .

It seems more important not to discourage settlements than to make an attempt of doubtful effectiveness to prevent discrimination by plaintiffs, or collusion in the suit.

12 U.L.A. 99-100 (1975). Accordingly, we uphold the district court's holding that a good faith settlement bars the action for contribution.

Good Faith Ruling

The district court concluded that the settlement between the plaintiff and Christopher was in good faith because it was not disproportionate to Christopher's liability. The court applied the standard enunciated in the leading California case of Tech-Bilt, Inc. v. Woodward-Clyde Assocs., 38 Cal.3d 488, 213 Cal.Rptr. 256, 698 P.2d 159 (1985). In this appeal, the appellant does not challenge the use of that standard. Rather, the appellant keys on the district court's methodology. The district court, after determining the value of the plaintiff's total claim and the settlement, estimated a range of potential liability. The appellant would have us hold that the district court was required to pick a particular figure, rather than estimate a range. The appellant admits he can find no authority for this proposition.

Because both defendants settled with the plaintiff, the district court did not have a full trial court record or damage judgment with which to work. We cannot say that there was error in the district court's methodology. Since the point of settlement is to avoid a trial, the district court in such circumstances must exercise a good deal of discretion in attempting to gauge exposure. We decline to require the court to commit to one precise but necessarily arbitrary figure.

The appellant also contends that even if the use of a range for percentage of fault is permissible, the district court erred in finding that this particular settlement was in good faith. The district court estimated the plaintiff's total claim at $500,000 and the settling defendant's liability as between thirty-three and sixty-seven percent. The court, however, estimated the value of the settlement at $100,000, or just twenty percent of the value of the plaintiff's total claim. The appellant contends that the district court could not have found the settlement to be in good faith when the amount of the settlement represented a value discounted below the estimated range of liability. In other words, the appellant contends that the district court was required to hold that the settlement was not in good faith if the settlement fell below the range of estimated actual damages assessible after trial.

Settlement amounts, however, are often discounted to reflect the cost of trial to the plaintiff and the uncertainties of the trial's outcome. Further, requiring perfect foresight on the part of the settling parties would not be in accordance with the agreed-upon California "grossly disproportionate" standard which finds good faith in a settlement which is in the "ballpark." Id. at 501 n. 9, 213 Cal.Rptr. at 264 n. 9, 698 P.2d at 168 n. 9. The district court committed no error in finding the settlement in good faith even though the amount settled on represented a discounted liability outside the range hypothesized by the court.

The judgment of the district court is AFFIRMED.