dissenting.
This is a hard case making, I think, proverbially bad law. On the surface what seems to be decided is simply a question of res judicata. Actually the decision rests on an “and/or” hodgepodge of res judicata and Erie doctrines.1 In my judgment the admixture not only is unnecessary but distorts and misapplies both doctrines. If res judicata properly applies and is adequate to dispose of the cause, there is no occasion for the sidewise introduction of Erie ideas. Likewise, if Erie appropriately governs the case, the Court’s elaborate and altogether novel discussion of res judicata is superfluous.
The Court has not decided this case on any basis of full faith and credit.2 Accordingly res judicata as it is applied *202has neither constitutional nor statutory status. For present purposes it is therefore purely a rule of judicial administration to be applied, like all such rules, as considerations of justice and right application of the policy require, not omitting due regard for its appropriate limits.
Res judicata is a generally sound but by no means unlimited policy of judicial action. The doctrine is grounded in the need for putting an end to litigation.3 It does this by precluding the parties from showing what is or may be the truth.4 The sound core of the policy is that ordinarily one suit which determines or gives a full and fair chance *203for determining causes of action and issues5 between litigants should be enough and when this much has been given further opportunity should be denied.
Stated so simply, however, the doctrine would be as much trap for the unwary as boon for the wise or lucky litigant. Exceptions and qualifications are so numerous as to make the field not only technical but treacherous, this case being a nice illustration. Qualification may itself lose sight of basic policy and become sheer technicality.6 But general rules are not qualified so extensively as this one has been without reason. There is good reason for much of what has been done in this respect with res judicata.
The effect of the rule qualifies its scope. It is not every case in which a litigant has had “one bite at the cherry” that the law forbids another. In other words, it is not every such case in which the policy of stopping litigation outweighs that of showing the truth. This is so not only where the first suit actually gives no real chance to secure a substantial determination,7 but also though less generally of others in which the litigant has such a chance and foregoes or misses it.8 It is so too whether the claimed *204estoppel by prior suit is “direct” or “indirect,” that is, on the same or a different cause of action.9
Upon the law as well as the policy, the question has been one of balancing considerations of justice and convenience between stopping litigation and stopping the showing of the truth.10 That balance has never been so one-sided in favor of the former that the matter is ended simply by showing that a party has had some chance, however slight, in a previous litigation to secure a favorable decision.
If this were the law every case where a party takes a nonsuit or a dismissal expressly for the purpose of starting over again would be a final and conclusive determination against him. I know of no jurisdiction where the law has been so harsh. Nor do I think it should be in this one.
There are too many good reasons why persons starting out in litigation should not be barred of their rights by the fact alone that they withdraw in order to start again, even though by going on to the end they might pull through successfully against great odds. Crucial witnesses may disappear or die and time be required for finding them or others. Surprise in the course of trial may occur justifying withdrawal without fatal loss of rights. Even as in *205this case, jurisdictional and other uncertainties may arise putting in jeopardy or making comparatively or completely futile further pursuit of the pending litigation when another suit in the same or a different court might provide a more certain and less expensive mode of disposing of the controversy for all the parties.
These and other reasons have qualified flat application of res judicata too long and too universally for their qualifying effects to be thrown overboard now simply because a withdrawing litigant might conceivably have come out victorious had he gone on to the very farthest end.11 Such a criterion would turn res judicata into a rigid rule requiring exhaustion of judicial remedies, a notion heretofore wholly alien to the doctrine.12
This course, moreover, seems to be justified on the basis that the grounds of an adjudication have nothing to do with the adjudication or its effects, for purposes of applying res judicata. That is true, apparently, for applying res judicata to Buffington’s failure to take his appeal here in the North Carolina state court suit, so as to cut off his right ever to secure a decision on the "'merits’ in the sense of the ultimate substantive issues of a litigation.” But it is not true, apparently, for application of the doctrine to different jurisdictional rulings. For "an adjudication declining to reach such ultimate substantive issues may bar a second attempt to reach them in another court of the *206State. Such a situation is presented when the first decision is based not on the ground that the distribution of judicial power among the various courts of the State requires the suit to be brought in another court in the State, hut on the inaccessibility of all the courts of the State to such litigation. And that is the essence of the present case.” (Emphasis added.)
I can understand the distinction drawn. But I find difficulty in understanding why res judicata turns for application in this case to cut off determination of substantive issues not at all upon the grounds of decision but only on the fact of adverse decision; but, for application to such issues when a jurisdictional question is also involved, it turns not simply upon the adjudication, but upon the grounds for the jurisdictional determination. If res judicata is governed solely by the adjudication without reference to what is adjudicated, that is, merely by the fact of adverse decision, I should think that rule would apply in all cases. If, on the contrary, the grounds of adjudication are relevant and controlling for the one class of questions, I should think they would be for the other.
The fallacy lies in the novel and unprecedented idea that the groundings of a court’s decision have nothing to do with whether res judicata applies, except when they relate to one kind of jurisdictional determination rather than another. Apart from the exception, the idea ignores the vast body of law which has grown up on the basis that the grounding of the decision is the criterion for applying the doctrine.13 And much of that case law has been that if the “ ‘merits’ in the sense of the ultimate substantive issues of a litigation” are not reached, their later determination is not foreclosed.14 This is true whether or not the jurisdic*207tional ruling is erroneous or valid and whether or not if erroneous it might have been corrected on appeal.
That law I think is sound, and I think it is just as sound when the jurisdictional decision goes off erroneously on a federal ground or erroneously ignoring one as when it rests on a valid basis. It is grounded in the policy that unless a litigant gets a real bite at the apple of discord he should not be foreclosed from another attempt. Its basis is that in such a case it is better and more just not to stop litigation than it is to stop the showing of the truth and thereby bring about a forfeiture of valuable substantive rights without giving at least one full and fair, which means fairly certain, opportunity for securing decision upon them.
Bullington has not had such an opportunity. He has never received, and now never can receive a decision on the substantive merits of his claim, unless possibly he can catch and serve Angel in another state and after prolonged further litigation succeed in inducing this Court to hold the North Carolina bar and res judicata not operative there. See Riley v. New York Trust Co., 315 U. S. 343, 349: “By the Constitutional provision for full faith and credit, the local doctrines of res judicata, speaking generally, become a part of national jurisprudence, and therefore federal questions cognizable here.”
Conceivably Bullington by coming here in the North Carolina suit might have secured a decision that the North Carolina statute and decision were invalid constitutionally in excluding him from all the state’s courts and that the state must afford him a remedy on proof of his substantive claim. But the very multiplicity of the constitutional questions enumerated in the Court’s opinion which were or might have been pertinent made that chance slim indeed. What is more important is that if the j udgment had been thus reversed and remanded, it would have been *208wiped out and he then, would have been free to dismiss the suit and start over again in the federal courts sitting in the state or in the state courts. Bucher v. Cheshire R. R. Co., 125 U. S. 555, 578-579.15
On the other hand, if the case had gone against him here, then his right to start over in the federal court in North Carolina would depend upon whether one of two grounds were accepted for this Court’s decision, namely, on the one hand, that North Carolina had power to bar his substantive claim and had done so in effect though not in words, thus closing the doors of the federal court to it under the Erie rule; on the other, that the state had power to close its courts against his claim without adjudicating its substantive validity, thus leaving him free to go to the federal court under the Erie rule. Either result was a conceivable one, depending on whether the Court should conclude that a "right without a remedy” remains a right, for this purpose, or becomes none at all. But the only chance for Bullington’s ultimate success, in the event of adverse deci*209sion here, would have been for the decision to have turned out on the latter ground.
That chance was hardly worth the gamble. For this Court has declared in Guaranty Trust Co. v. York, 326 U. S. 99, 108-109, that a right without a remedy is no right at all for purposes of enforcement by a diversity suit in a federal court sitting in the state.16 And the nature of the North Carolina statute as construed by the state court reaches exactly the result which the York case says precludes resort to the federal court on the same cause in a diversity suit.17 Indeed this seems to be an alternative basis for the present decision.18 Bullington’s chance to get to the federal court on such a basis was therefore practically nil.
Should he now be barred because he did not take the extremely remote chance of securing a favorable decision, reversing the state court’s judgment and forcing the state to hear his case on the merits ? Not, I think, unless we can say he then would have been forced, if successful, to continue the litigation in the state courts and could not withdraw to start over in the North Carolina federal court. This we could not say unless we were to overrule the Bucher case, which Bullington had a right to assume we would not do. Why he should be barred from doing now, because he did not take his almost hopeless appeal, what he would have been at liberty to do if he had taken it successfully, I am not able to understand. No sound policy of ending litigation, conserving judicial time or litigants’ rights or in any other respect can possibly be served by *210such a ruling.19 Moreover, the very difficulties in his way for securing a successful determination here, which would give him some certain remedy either in the state or in the federal courts, were sufficient reason, in my opinion, to justify his foregoing that dubious procedure and starting over again in the federal court.
The real trouble here is not with the law of res judicata, for that law has no valid application to these facts. It is that the doctrine is used as an escape from facing squarely the real question presented. This is whether North Carolina’s decision made the Erie doctrine applicable. The Court’s opinion does state expressly that the effect of the North Carolina decision was to create a policy of the state against the validity of all claims for deficiency judgments, and comes almost but not quite to saying this requires the case to go off on application of the York rule.
That issue is inescapable here. The Erie rule did not purport to change the law of federal jurisdiction in diversity cases, taking it out of the hands of Congress and the federal courts and putting it within the states’ power to determine. It purported only to prescribe the rule federal courts should follow in applying the substantive law. If the North Carolina decision was exclusively a jurisdic*211tiónal one, it had no effect on the power of the federal courts in that state to hear controversies excluded by it from the state courts, and the decision neither reached the merits of the controversy "in the sense of the ultimate substantive issues of a litigation” nor barred Bullington from going to the federal court. See Lupton’s Sons Co. v. Automobile Club, 225 U. S. 489. If on the other hand the decision was in effect, although not in words, a determination of the merits in that sense, it both adjudicated Bullington’s substantive rights and barred him from maintaining the later suit successfully in the federal court. That question is here and until it is resolved he is deprived of any day in court except to go from one to another without securing decision either on the merits substantially or “on the merits” jurisdictionally.
From the Court’s opinion I cannot say whether the question has been resolved. Its discussion of North Carolina’s “policy” and its overruling of the Lupton’s Sons case, supra, would seem to indicate that it is applying York, though without saying so frankly. But, if so, why speak also of res judicata? The law should not be made into such a merry-go-round. Bullington is entitled to one full day in court on the substance of his claim. This he has not had.
I hardly need add that I agree with the views expressed by Mr. Justice Reed.
Mr. Justice Jackson joins in this opinion.Erie R.R. Co. v. Tompkins, 304 U. S. 64.
The Court does not hold that the full faith and credit clause, Const., Art. IV, § 1, binds the federal courts to give the North Carolina judgment the effect of precluding a further suit in the federal *202courts on the substantive cause of action. Two difficulties would arise. (1) If, as the Court asserts, the federal court in diversity cases were only “another North Carolina court,” the full faith and credit clause would have no application; but, that it may, see Cooper v. Newell, 173 U. S. 555, 567: “. . . the courts of the United States are tribunals of a different sovereignty, and exercise a distinct and independent jurisdiction from that exercised by the state courts, and this is true in respect of the courts of the several States as between each other. And the courts of the United States are bound to give to the judgments of the state courts the same faith and credit that the courts of one State are bound to give to the judgments of the courts of her sister States.” See also American Surety Co. v. Baldwin, 287 U. S. 156, where Mr. Justice Brandeis spoke of the full faith and credit clause in connection with the obligation of a federal court sitting in Idaho to follow, as res judicata, a previous Idaho decision. (2) The decision would contradict, not confirm, as full faith and credit require, the basis for the North Carolina court’s judgment, namely, that the statute does not outlaw substantive claims but only deprives the state courts of power to entertain them.
See Moschzisker, Res Judicata (1929) 38 Yale L. J. 299, 300. In this respect, of course, res judicata resembles both statutes of limitations and the doctrine of laches in equity, as well as full faith and credit when applicable to judgments.
Scott, Collateral Estoppel by Judgment (1942) 56 Harv. L. Rev. 1. So do statutes of limitations, laches and full faith and credit re judgments, when applicable.
Cf. note 9 infra and text.
A sign generally that something is radically wrong with the rule or with it and the exceptions together. Cf. Georgetown College v. Hughes, 76 U. S. App. D. C. 123, 130 F. 2d 810.
See Walden v. Bodley, 14 Pet. 156, 161; Hughes v. United States, 4 Wall. 232, 237; Restatement, Judgments (1942) § 49.
“Judgments of nonsuit, of non prosequitur, of nolle prosequi, of discontinuance and of dismissal generally, are exceptions to the general rule that when the pleadings, the court, and the parties are such as to permit of a trial on the merits, the judgment will be considered as final and conclusive of all matters which could have been so tried.” 2 Freeman, Judgments (5th ed.) 1579-1580. And “generally speaking, judgments merely of dismissal, whether voluntary or involuntary, in actions at law are not on the merits and do not operate *204as a bar or estoppel in subsequent proceedings involving the same matters.” Id. at 1582. See Haldeman v. United States, 91 U. S. 584; Jacobs v. Marks, 182 U. S. 583; Manhattan Life Ins. Co. v. Broughton, 109 U. S. 121; Restatement, Judgments (1942) §§ 53, 54.
Scott, op. cit. supra note 4, p. 2; Restatement, Judgments (1942) 175-176. Usually “direct” estoppel is said to preclude redetermination of issues actually determined or which might have been determined “on the merits.” Ibid. But “indirect” estoppel precludes relitigating only issues actually decided. Cromwell v. County of Sac, 94 U. S. 351, 352-353.
The process of course crystallizes in definite rules for types of cases, but the important fact is that these rules do not all come out the same way for application of the rule of preclusion and that in the process of crystallization the weighing of the opposing considerations forms the rule for or against that policy.
The Court clearly implies that Bullington would not be foreclosed on res judicata by the North Carolina decision if his route to this Court for review had been by certiorari rather than by appeal. But the ruling as made, in so far as it rests on the failure to appeal, ignores the settled law that for purposes of applying res judicata failure to take appeal has no bearing once the judgment becomes final. See note 19. It also defeats the policy of res judicata; for a party, instead of being allowed to accept the jurisdictional ruling, is forced to appeal to the highest court in order to save his rights no matter how meritless the appeal, thus prolonging rather than shortening, litigation.
See note 19.
See Scott, op. cit., supra note 4; Restatement, Judgments (1942) §§ 49, 50, 53, 54.
See authorities cited in note 7 supra.
“If ... a judgment has been vacated by the trial court or reversed by an appellate court, it is no longer conclusive between the parties, either as a merger of the original cause of action or as a bar to an action upon the original cause of action . . . .” Restatement, Judgments (1942) 163.
“Ordinarily, after a judgment has been reversed on appeal and the cause remanded, the case stands for trial de novo on the issues properly joined. . . . With respect to the right of plaintiff to take a voluntary nonsuit, it stands in the same relative position which it occupied before the trial in the first instance.” Shell Petroleum Corp. v. Shore, 80 F. 2d 785, 786.
This is the general rule. 89 A. L. R. 109; 126 A. L. R. 305. It would seem to apply in North Carolina. North Carolina follows both the doctrine that the trial court, upon remand by an appellate court, is to proceed as if there had been no previous trial and the doctrine that judgments not on the merits do not constitute an estoppel to subsequent actions. Hickory v. Railroad, 138 N. C. 311, 318; Grimes v. Andrews, 170 N. C. 515; cf. Gen. Stat. N. C. (1943) § 1-25, as interpreted in Grimes v. Andrews, 170 N. C. at 522.
The York case however did not purport to apply or extend the rule to a cause of action arising under and governed by the laws of another state than that in which the federal court was sitting. But cf. Klaxon Co. v. Stentor Elec. Mfg. Co., 313 U. S. 487.
Except for the factor noted in note 16.
See text infra.
“The application of the principle of res judicata has not in any way been made to depend upon whether the judgment in question is subject to review in another tribunal. Except in so far as it may affect the question of its finality, as in the case of orders on motions, the fact that a judgment may or may not be appealable should have no bearing upon its effect as res judicata.” 2 Freeman, Judgments (5th ed.) 1339.
Thus, there is no doctrine of exhaustion of judicial remedies. If the judgment of a court goes on jurisdictional grounds, the party may accept it and, instead of appealing, may institute another action where he will not be met by the jurisdictional bar. Cf. Restatement, Judgments (1942) 194-195; Cook, The Logical and Legal Bases of the Conflict of Laws (1942) 133-135.