JUSTICE STEVENS, dissenting.

While I agree with most of the reasoning in the Court's opinion, I disagree with its disposition of the case. I would affirm the judgment of the Court of Appeals. I would also reject the suggestion that the Seventh Amendment limits the power of a federal appellate court sitting in diversity to decide whether a jury's award of damages exceeds a limit established by state law.

I

The Court correctly explains why the 1986 enactment of § 5501(c) of the N.Y. Civ. Prac. Law and Rules (McKinney 1995) changed the substantive law of the State. A state-law ceiling on allowable damages, whether fixed by a dollar limit or by a standard that forbids any award that "deviates materially from what would be reasonable compensation," ibid., is a substantive rule of decision that federal courts must apply in diversity cases governed by New York law.

I recognize that state rules of appellate procedure do not necessarily bind federal appellate courts. The majority persuasively shows, however, that New York has not merely adopted a new procedure for allocating the decisionmaking function between trial and appellate courts. Ante, at 422-425. Instead, New York courts have held that all jury awards, not only those reviewed on appeal, must conform to the requirement that they not "deviat[e] materially" from amounts awarded in like cases. Ante, at 425. That New York has chosen to tie its damages ceiling to awards traditionally recovered in similar cases, rather than to a legislatively determined but inflexible monetary sum, is none of our concern.

Given the nature of the state-law command, the Court of Appeals for the Second Circuit correctly concluded in Consorti v. Armstrong World Industries, Inc., 64 F.3d 781, superseded, 72 F.3d 1003 (1995), that New York's excessiveness standard applies in federal court in diversity cases controlled by New York law. Consorti erred in basing that conclusion in part on the fact that a New York statute requires that State's appellate division to apply the standard, but it was nevertheless faithful to the Rules of Decision Act, as construed in Erie R. Co. v. Tompkins, 304 U.S. 64 (1938), in holding that a state-law limitation on the size of a judgment could not be ignored. Similarly, the Court of Appeals correctly followed Consorti in this case and considered whether the damages awarded materially deviated from damages awarded in similar cases. 66 F.3d 427, 431 (CA2 1995). I endorse both opinions in these respects.

Although the majority agrees with the Court of Appeals that New York law establishes the size of the damages that may be awarded, it chooses to vacate and remand. The majority holds that a federal court of appeals should review for abuse of discretion a district court's decision to deny a motion for new trial based on a jury's excessive award. As a result, it concludes that the District Court should be given the opportunity to apply in the first instance the "deviates materially" standard that New York law imposes. Ante, at 439.

The District Court had its opportunity to consider the propriety of the jury's award, and it erred. The Court of Appeals has now corrected that error after "drawing all reasonable inferences in favor of" petitioner. 66 F.3d, at 431. As there is no reason to suppose that the Court of Appeals has reached a conclusion with which the District Court could permissibly disagree on remand, I would not require the District Court to repeat a task that has already been well performed by the reviewing court. I therefore would affirm the judgment of the Court of Appeals.

II

Although I have addressed the question presented as if our decision in Erie alone controlled its outcome, petitioner argues that the second clause of the Seventh Amendment, which states that "no fact tried by jury, shall be otherwise re-examined in any Court of the United States, than according to the rules of the common law," U.S. Const., Amdt. 7, bars the procedure followed by the Court of Appeals. There is no merit to that position.

Early cases do state that the Reexamination Clause prohibits appellate review of excessive jury awards, but they do not foreclose the practice altogether. See, e.g., Southern Railway-Carolina Div. v. Bennett, 233 U.S. 80, 87 (1914) ("It may be admitted that if it were true that the excess appeared as [a] matter of law; that if, for instance, the statute fixed a maximum and the verdict exceeded it, a question might arise for this court"); 11 C. Wright, A. Miller, M. Kane, Federal Practice and Procedure § 2820, pp. 207-209 (2d ed. 1995). Indeed, for the last 30 years, we have consistently reserved the question whether the Constitution permits such review, ante, at 434-435, and, in the meantime, every Court of Appeals has agreed that the Seventh Amendment establishes no bar. 11 Wright Miller § 2820, at 209.

Taking the question to be an open one, I start with certain basic principles. It is well settled that jury verdicts are not binding on either trial judges or appellate courts if they are unauthorized by law. A verdict may be insupportable as a matter of law either because of deficiencies in the evidence or because an award of damages is larger than permitted by law. If an award is excessive as a matter of law — in a diversity case if it is larger than applicable state law permits — a trial judge has a duty to set it aside. A failure to do so is an error of law that the court of appeals has a duty to correct on appeal.

These principles are sufficiently well established that no Seventh Amendment issue would arise if an appellate court ordered a new trial because a jury award exceeded a monetary cap on allowable damages. That New York has chosen to define its legal limit in less mathematical terms does not require a different constitutional conclusion.

New York's limitation requires a legal inquiry that cannot be wholly divorced from the facts, but that quality does not necessarily make the question one for the factfinder rather than the reviewing court. Three times this Term we have assigned appellate courts the task of independently reviewing similarly mixed questions of law and fact. See Ornelas v. United States, 517 U.S. 690, 696-697 (1996); Markman v. Westview Instruments, Inc.,, 517 U.S. 370, 388-390 (1996); Thompson v. Keohane, 516 U.S. 90, 112-116 (1995). Such appellate review is proper because mixed questions require courts to construe all record inferences in favor of the factfinder's decision and then to determine whether, on the facts as found below, the legal standard has been met. See Ornelas, 517 U.S., at 696-697 (quoting Pullman-Standard v. Swint, 456 U.S. 273, 289, n. 19 (1982). In following that procedure here, the Court of Appeals did not reexamine any fact determined by a jury. 66 F.3d, at 431. It merely identified that portion of the judgment that constitutes "unlawful excess." See Dimick v. Schiedt, 293 U.S. 474, 486 (1935).

Even if review by the Court of Appeals implicates the Reexamination Clause, it was "according to the rules of the common law." U.S. Const., Amdt. 7. At common law, the trial judge sitting nisi prius recommended whether a judicial panel sitting en banc at Westminster should accept the jury's award. The en banc court then ruled on the motion for new trial and entered judgment. 11 Wright Miller § 2819, at 203.

Petitioner correctly points out that under this procedure motions for new trial based on excessiveness were not technically subject to appellate review. Riddell, New Trial at the Common Law, 26 Yale L. J. 49, 57 (1916) ("It seems clear that in criminal as in civil cases, the trial Judge had not the power to grant a new trial, but that recourse must be had to 'the Court above'"); id., at 60. However, because the nisi prius judge often did not serve on the en banc court, the "court above" was in essentially the same position as a modern court of appeals. It considered the legality of the jury's award in light of the trial judge's opinion, but without any firsthand knowledge of what had transpired below. See Blume, Review of Facts in Jury Cases — The Seventh Amendment, 20 J. Am. Jud. Soc. 130, 131 (1936).

Petitioner also contends that at common law the en banc court could only grant a new trial if the trial judge so recommended. That contention is undermined by numerous cases in which the "court above" granted new trials without making any reference to the trial judge's view of the damages. See, e.g., Honda Motor Co. v. Oberg, 512 U.S. 415, 422-425 (1994) (citing cases). Moreover, early English cases repeatedly state that the power to order a new trial when the jury returned an excessive award rested with "the Court," rather than the judge below, and Blackstone identifies excessive damages as an independent basis on which the "court above" may grant a new trial but makes no mention of a requirement that the trial judge must so recommend. 3 W. Blackstone, Commentaries *387.

Even when read most favorably to petitioner, therefore, no meaningful distinction exists between the common-law practice by which the "court above" considered a new trial motion in the first instance, and the practice challenged here, by which an appellate court reviews a district court's ruling on a new trial motion. See Riddell, 26 Yale L. J., at 57. As Justice Stone explained, in a dissenting opinion joined by Chief Justice Hughes, Justice Brandeis, and Justice Cardozo:

"[The Seventh Amendment], intended to endure for unnumbered generations, is concerned with substance and not with form. There is nothing in its history or language to suggest that the Amendment had any purpose but to preserve the essentials of the jury trial as it was known to the common law before the adoption of the Constitution. For that reason this Court has often refused to construe it as intended to perpetuate in changeless form the minutiae of trial practice as it existed in the English courts in 1791. From the beginning, its language has been regarded as but subservient to the single purpose of the Amendment, to preserve the essentials of the jury trial in actions at law, serving to distinguish them from suits in equity and admiralty, see Parsons v. Bedford, 3 Pet. 433, 446, and to safeguard the jury's function from any encroachment which the common law did not permit.

"Thus interpreted, the Seventh Amendment guarantees that suitors in actions at law shall have the benefits of trial of issues of fact by a jury, but it does not prescribe any particular procedure by which these benefits shall be obtained, or forbid any which does not curtail the function of the jury to decide questions of fact as it did before the adoption of the Amendment. It does not restrict the court's control of the jury's verdict, as it had previously been exercised, and it does not confine the trial judge, in determining what issues are for the jury and what for the court, to the particular forms of trial practice in vogue in 1791." Dimick v. Schiedt, 293 U.S., at 490-491.

Because the Framers of the Seventh Amendment evinced no interest in subscribing to every procedural nicety of the notoriously complicated English system, see Henderson, The Background of the Seventh Amendment, 80 Harv. L. Rev. 289, 290 (1966), the common-law practice certainly does not demonstrate that the Reexamination Clause prohibits federal appellate courts from ensuring compliance with state-law limits on jury awards.

Nor does early and intricate English history justify the more limited assertion that federal appellate courts must be limited to a particular, highly deferential standard of excessiveness review. Common-law courts were hesitant to disturb jury awards, but less so in cases in which "a reasonably certain measure of damages is afforded." 1 D. Graham, Law of New Trials in Cases Civil and Criminal 452 (2d ed. 1855); Washington, Damages in Contract at Common Law, 47 L. Q. Rev. 345, 363-364 (1931).

Here, New York has prescribed an objective, legal limitation on damages. If an appellate court may reverse a jury's damages award when its own conscience has been shocked, 66 F.3d, at 430, or its sense of justice outraged, Dagnello v. Long Island R. Co., 289 F.2d 797, 802 (CA2 1961); cf. Honda Motor Co. v. Oberg, 512 U.S., at 422-424 (citing English cases), it may surely follow a sovereign's command that it do so when a jury has materially deviated from awards granted by other juries. If anything, the New York standard, though less deferential, is more certain.

III

For the reasons set forth above, I agree with the majority that the Reexamination Clause does not bar federal appellate courts from reviewing jury awards for excessiveness. I confess to some surprise, however, at its conclusion that "'the influence — if not the command — of the Seventh Amendment,'" ante, at 432 (quoting Byrd v. Blue Ridge Rural Elec. Cooperative, Inc., 356 U.S. 525, 537 (1958), requires federal courts of appeals to review district court applications of state-law excessiveness standards for an "abuse of discretion." Ante, at 438.

The majority's persuasive demonstration that New York law sets forth a substantive limitation on the size of jury awards seems to refute the contention that New York has merely asked appellate courts to reexamine facts. The majority's analysis would thus seem to undermine the conclusion that the Reexamination Clause is relevant to this case.

Certainly, our decision in Byrd does not make the Clause relevant. There, we considered only whether the Seventh Amendment's first clause should influence our decision to give effect to a state-law rule denying the right to a jury altogether. 356 U.S., at 537. That holding in no way requires us to consult the Amendment's second clause to determine the standard of review for a district court's application of state substantive law.

My disagreement is tempered, however, because the majority carefully avoids defining too strictly the abuse-of-discretion standard it announces. To the extent that the majority relies only on "practical reasons" for its conclusion that the Court of Appeals should give some weight to the District Court's assessment in determining whether state substantive law has been properly applied, ante, at 438, I do not disagree with its analysis.

As a matter of federal-court administration, we have recognized in other contexts the need for according some deference to the lower court's resolution of legal, yet fact-intensive, questions. See Ornelas v. United States, 517 U.S., at 699; Pierce v. Underwood, 487 U.S. 552, 558, n. 1 (1988). Indeed, it is a familiar, if somewhat circular, maxim that deems an error of law an abuse of discretion.

In the end, therefore, my disagreement with the label that the majority attaches to the standard of appellate review should not obscure the far more fundamental point on which we agree. Whatever influence the Seventh Amendment may be said to exert, Erie requires federal appellate courts sitting in diversity to apply "the damage-control standard state law supplies." Ante, at 438.

IV

Because I would affirm the judgment of the Court of Appeals, and because I do not agree that the Seventh Amendment in any respect influences the proper analysis of the question presented, I respectfully dissent.

JUSTICE SCALIA, with whom the CHIEF JUSTICE and JUSTICE THOMAS join, dissenting.

Today the Court overrules a longstanding and well-reasoned line of precedent that has for years prohibited federal appellate courts from reviewing refusals by district courts to set aside civil jury awards as contrary to the weight of the evidence. One reason is given for overruling these cases: that the Courts of Appeals have, for some time now, decided to ignore them. Such unreasoned capitulation to the nullification of what was long regarded as a core component of the Bill of Rights — the Seventh Amendment's prohibition on appellate reexamination of civil jury awards — is wrong. It is not for us, much less for the Courts of Appeals, to decide that the Seventh Amendment's restriction on federal-court review of jury findings has outlived its usefulness.

The Court also holds today that a state practice that relates to the division of duties between state judges and juries must be followed by federal courts in diversity cases. On this issue, too, our prior cases are directly to the contrary.

As I would reverse the judgment of the Court of Appeals, I respectfully dissent.

I

Because the Court and I disagree as to the character of the review that is before us, I recount briefly the nature of the New York practice rule at issue. Section 5501(c) of the N.Y. Civ. Prac. Law and Rules (CPLR) (McKinney 1995) directs New York intermediate appellate courts faced with a claim "that the award is excessive or inadequate and that a new trial should have been granted" to determine whether the jury's award "deviates materially from what would be reasonable compensation." In granting respondent a new trial under this standard, the Court of Appeals necessarily engaged in a two-step process. As it has explained the application of § 5501(c), that provision "requires the reviewing court to determine the range it regards as reasonable, and to determine whether the particular jury award deviates materially from that range." Consorti v. Armstrong World Industries, Inc., 72 F.3d 1003, 1013 (CA2 1995) (amended). The first of these two steps — the determination as to "reasonable" damages plainly requires the reviewing court to reexamine a factual matter tried by the jury: the appropriate measure of damages, on the evidence presented, under New York law. The second step — the determination as to the degree of difference between "reasonable" damages and the damages found by the jury (whether the latter "deviates materially" from the former) — establishes the degree of judicial tolerance for awards found not to be reasonable, whether at the trial level or by the appellate court. No part of this exercise is appropriate for a federal court of appeals, whether or not it is sitting in a diversity case.

A

Granting appellate courts authority to decide whether an award is "excessive or inadequate" in the manner of CPLR § 5501(c) may reflect a sound understanding of the capacities of modern juries and trial judges. That is to say, the people of the State of New York may well be correct that such a rule contributes to a more just legal system. But the practice of federal appellate reexamination of facts found by a jury is precisely what the People of the several States considered not to be good legal policy in 1791. Indeed, so fearful were they of such a practice that they constitutionally prohibited it by means of the Seventh Amendment.

That Amendment was Congress's response to one of the principal objections to the proposed Constitution raised by the Anti-Federalists during the ratification debates: its failure to ensure the right to trial by jury in civil actions in federal court. The desire for an explicit constitutional guarantee against reexamination of jury findings was explained by Justice Story, sitting as Circuit Justice in 1812, as having been specifically prompted by Article III's conferral of "appellate Jurisdiction, both as to Law and Fact" upon the Supreme Court. "[O]ne of the most powerful objections urged against [the Constitution]," he recounted, was that this authority "would enable that court, with or without a new jury, to re-examine the whole facts, which had been settled by a previous jury." United States v. Wonson, 28 F. Cas. 745, 750 (No. 16,750) (CC Mass.).

The second clause of the Amendment responded to that concern by providing that "[i]n [s]uits at common law . . . no fact tried by a jury, shall be otherwise re-examined in any Court of the United States, than according to the rules of the common law." U.S. Const., Amdt. 7. The Reexamination Clause put to rest "apprehensions" of "new trials by the appellate courts," Wonson, 28 F. Cas., at 750, by adopting, in broad fashion, "the rules of the common law" to govern federal-court interference with jury determinations. The content of that law was familiar and fixed. See, e.g., ibid. ("[T]he common law here alluded to is not the common law of any individual state, (for it probably differs in all), but it is the common law of England, the grand reservoir of all our jurisprudence"); Dimick v. Schiedt, 293 U.S. 474, 487 (1935) ( Seventh Amendment "in effect adopted the rules of the common law, in respect of trial by jury, as these rules existed in 1791"). It quite plainly barred reviewing courts from entertaining claims that the jury's verdict was contrary to the evidence.

At common law, review of judgments was had only on writ of error, limited to questions of law. See, e.g., Wonson, supra, at 748; 3 W. Blackstone, Commentaries on the Laws of England 405 (1768) ("The writ of error only lies upon matter of law arising upon the face of the proceedings; so that no evidence is required to substantiate or support it"); 1 W. Holdsworth, History of English Law 213-214 (7th ed. 1956); cf. Ross v. Rittenhouse, 2 Dall. 160, 163 (Pa. 1792) (McKean, C. J.). That principle was expressly acknowledged by this Court as governing federal practice in Parsons v. Bedford, 3 Pet. 433 (1830) (Story, J.). There, the Court held that no error could be assigned to a district court's refusal to allow transcription of witness testimony "to serve as a statement of facts in case of appeal," notwithstanding the right to such transcription under state practices made applicable to federal courts by Congress. Id., at 443 (emphasis deleted.) This was so, the Court explained, because "[t]he whole object" of the transcription was "to present the evidence here in order to establish the error of the verdict in matters of fact," id., at 445 — a mode of review simply unavailable on writ of error, see id., at 446, 448. The Court concluded that Congress had not directed federal courts to follow state practices that would change "the effect or conclusiveness of the verdict of the jury upon the facts litigated at the trial," id., at 449, because it had "the most serious doubts whether [that] would not be unconstitutional" under the Seventh Amendment, id., at 448.

"This is a prohibition to the courts of the United States to re-examine any facts tried by a jury in any other manner. The only modes known to the common law to re-examine such facts, are the granting of a new trial by the court where the issue was tried, or to which the record was properly returnable; or the award of a venire facias de novo, by an appellate court, for some error of law which intervened in the proceedings.

. . . . .

"[I]f the evidence were now before us, it would not be competent for this court to reverse the judgment for any error in the verdict of the jury at the trial. . . ." Id., at 447-449.

Nor was the common-law proscription on reexamination limited to review of the correctness of the jury's determination of liability on the facts. No less than the existence of liability, the proper measure of damages "involves only a question of fact," St. Louis, I. M. S. R. Co. v. Craft, 237 U.S. 648, 661 (1915), as does a "motio[n] for a new trial based on the ground that the damages . . . are excessive," Metropolitan R. Co. v. Moore, 121 U.S. 558, 574 (1887). As appeals from denial of such motions necessarily pose a factual question, courts of the United States are constitutionally forbidden to entertain them.

"No error of law appearing upon the record, this court cannot reverse the judgment because, upon examination of the evidence, we may be of the opinion that the jury should have returned a verdict for a less amount. If the jury acted upon a gross mistake of facts, or were governed by some improper influence or bias, the remedy therefore rested with the court below, under its general power to set aside the verdict. . . . Whether [the refusal to exercise that power] was erroneous or not, our power is restricted by the Constitution to the determination of the questions of law arising upon the record. Our authority does not extend to a re-examination of facts which have been tried by the jury under instructions correctly defining the legal rights of parties. Parsons v. Bedford, [ supra]. . . ." Railroad Co. v. Fraloff, 100 U.S. 24, 31-32 (1879).

This view was for long years not only unquestioned in our cases, but repeatedly affirmed.

B

Respondent's principal response to these cases, which is endorsed by JUSTICE STEVENS, see ante, at 443-445, is that our forebears were simply wrong about the English common law. The rules of the common-law practice incorporated in the Seventh Amendment, it is claimed, did not prevent judges sitting in an appellate capacity from granting a new trial on the ground that an award was contrary to the weight of the evidence. This claim simply does not withstand examination of the actual practices of the courts at common law. The weight of the historical record strongly supports the view of the common law taken in our early cases.

At common law, all major civil actions were initiated before panels of judges sitting at the courts of Westminster. Trial was not always held at the bar of the court, however. The inconvenience of having jurors and witnesses travel to Westminster had given rise to the practice of allowing trials to be held in the countryside, before a single itinerant judge. This nisi prius trial, as it was called, was limited to the jury's deciding a matter of fact in dispute; once that was accomplished, the verdict was entered on the record which — along with any exceptions to the instructions or rulings of the nisi prius judge — was then returned to the en banc court at Westminster. See generally 1 Holdsworth, History of English Law, at 223-224, 278-282; G. Radcliffe G. Cross, The English Legal System 90-91, 183-186 (3d ed. 1954). Requests for new trials were made not to the nisi prius judge, but to the en banc court, prior to further proceedings and entry of judgment. See 1 Holdsworth, supra, at 282; Riddell, New Trial at the Common Law, 26 Yale L. J. 49, 53, 57 (1916). Such motions were altogether separate from appeal on writ of error, which followed the entry of judgment. 1 Holdsworth, supra, at 213-214; Radcliffe Cross, supra, at 210-212.

Nonetheless, respondent argues, the role of the en banc court at Westminster was essentially that of an appellate body, reviewing the proceedings below; and those appellate judges were capable of examining the evidence, and of granting a new trial when, in their view, the verdict was contrary to the weight of the evidence. See Blume, Review of Facts in Jury Cases — The Seventh Amendment, 20 J. Am. Jud. Soc. 130, 131 (1936); Riddell, supra, at 55-57, 60. There are two difficulties with this argument. The first is the characterization of the court at Westminster as an appellate body. The court's role with respect to the initiation of the action, the entertaining of motions for new trial, and the entry of judgment was the same in all cases — whether the cause was tried at the bar or at nisi prius. To regard its actions in deciding a motion for a new trial as "appellate" in the latter instance supposes a functional distinction where none existed. The second difficulty is that when the trial had been held at nisi prius, the judges of the en banc court apparently would order a new trial only if the nisi prius judge certified that he was dissatisfied with the verdict. To be sure, there are many cases where no mention is made of the judge's certificate, but there are many indications that it was a required predicate to setting aside a verdict rendered at nisi prius, and respondent has been unable to identify a single case where a new trial was granted in the absence of such certification. In short, it would seem that a new trial could not be had except upon the approval of the judge who presided over the trial and heard the evidence.

I am persuaded that our prior cases were correct that, at common law, "reexamination" of the facts found by a jury could be undertaken only by the trial court, and that appellate review was restricted to writ of error which could challenge the judgment only upon matters of law. Even if there were some doubt on the point, we should be hesitant to advance our view of the common law over that of our forbears, who were far better acquainted with the subject than we are. But in any event, the question of how to apply the "rules of the common law" to federal appellate consideration of motions for new trials is one that has already been clearly and categorically answered, by our precedents. As we said in Dimick v. Schiedt, 293 U.S. 474 (1935), in discussing the status of remittitur under "the rules of the common law," a doctrine that "has been accepted as the law for more than a hundred years and uniformly applied in the federal courts during that time" and "finds some support in the practice of the English courts prior to the adoption of the Constitution" will not lightly "be reconsidered or disturbed," id., at 484-485. The time to question whether orders on motions for a new trial were in fact reviewable at common law has long since passed. Cases of this Court reaching back into the early 19th century establish that the Constitution forbids federal appellate courts to "reexamine" a fact found by the jury at trial; and that this prohibition encompasses review of a district court's refusal to set aside a verdict as contrary to the weight of the evidence.

C

The Court, as is its wont of late, all but ignores the relevant history. It acknowledges that federal appellate review of district-court refusals to set aside jury awards as against the weight of the evidence was "once deemed inconsonant with the Seventh Amendment's Reexamination Clause," ante, at 434, but gives no indication of why ever we held that view; and its citation of only one of our cases subscribing to that proposition fails to convey how long and how clearly it was a fixture of federal practice, see ibid. (citing only Lincoln v. Power, 151 U.S. 436 (1894)). That our earlier cases are so poorly recounted is not surprising, however, given the scant analysis devoted to the conclusion that "appellate review for abuse of discretion is reconcilable with the Seventh Amendment," ante, at 435.

No precedent of this Court affirmatively supports that proposition. The cases upon which the Court relies neither affirmed nor rejected the practice of appellate weight-of-the-evidence review that has been adopted by the courts of appeals — a development that, in light of our past cases, amounts to studied waywardness by the intermediate appellate bench. Our unaccountable reluctance, in Grunenthal v. Long Island R. Co., 393 U.S. 156, 158 (1968), and Neese v. Southern R. Co., 350 U.S. 77 (1955), to stand by our precedents, and the undeniable illogic of our disposition of those two cases — approving ourselves a district-court denial of a new trial motion, so as not to have to confront the lawfulness of reversal by the court of appeals — is authority of only the weakest and most negative sort. Nor can any weight be assigned to our statement in Browning-Ferris Industries of Vt., Inc. v. Kelco Disposal, Inc., 492 U.S. 257, 279 (1989), seemingly approving appellate abuse-of-discretion review of denials of new trials where punitive damages are claimed to be excessive. Browning-Ferris, like Grunenthal and Neese, explicitly avoided the question that is before us today, see 492 U.S., at 279, n. 25. Even more significantly, Browning-Ferris involved review of a jury's punitive damages award. Unlike the measure of actual damages suffered, which presents a question of historical or predictive fact, see, e.g., Craft, 237 U.S., at 661, the level of punitive damages is not really a "fact" "tried" by the jury. In none of our cases holding that the Reexamination Clause prevents federal appellate review of claims of excessive damages does it appear that the damages had a truly "punitive" component.

In any event, it is not this Court's statements that the Court puts forward as the basis for dispensing with our prior cases. Rather, it is the Courts of Appeals' unanimous "agree[ment]" that they may review trial-court refusals to set aside jury awards claimed to be against the weight of the evidence. Ante, at 435. This current unanimity is deemed controlling, notwithstanding the "relatively late" origin of the practice, ante, at 434, and without any inquiry into the reasoning set forth in those Court of Appeals decisions. The Court contents itself with citations of two federal appellate cases and the assurances of two leading treatises that the view (however meager its intellectual provenance might be) is universally held. See ante, at 435-436. To its credit, one of those treatises describes the "dramatic change in doctrine" represented by appellate abuse-of-discretion review of denials of new trial orders generally as having been "accomplished by a blizzard of dicta" that, through repetition alone, has "given legitimacy to a doctrine of doubtful constitutionality." 11 C. Wright, A. Miller, M. Kane, Federal Practice and Procedure § 2819, pp. 200, 204 (2d ed. 1995).

The Court's only suggestion as to what rationale might underlie approval of abuse-of-discretion review is to be found in a quotation from Dagnello v. Long Island R. Co., 289 F.2d 797 (CA2 1961), to the effect that review of denial of a new trial motion, if conducted under a sufficiently deferential standard, poses only "'a question of law.'" Ante, at 435 (quoting Dagnello, supra, at 806). But that is not the test that the Seventh Amendment sets forth. Whether or not it is possible to characterize an appeal of a denial of new trial as raising a "legal question," it is not possible to review such a claim without engaging in a "reexamin[ation]" of the "facts tried by the jury" in a manner "otherwise" than allowed at common law. Determining whether a particular award is excessive requires that one first determine the nature and extent of the harm — which undeniably requires reviewing the facts of the case. That the court's review also entails application of a legal standard (whether "shocks the conscience," "deviates materially," or some other) makes no difference, for what is necessarily also required is reexamination of facts found by the jury.

In the last analysis, the Court frankly abandons any pretense at faithfulness to the common law, suggesting that "the meaning" of the Reexamination Clause was not "fixed at 1791," ante, at 436, n. 20, contrary to the view that all our prior discussions of the Reexamination Clause have adopted, see supra, at 451-454. The Court believes we can ignore the very explicit command that "no fact tried by a jury shall be otherwise reexamined in any Court of the United States, than according to the rules of the common law" because, after all, we have not insisted that juries be all male, or consist of 12 jurors, as they were at common law. Ante, at 436, n. 20. This is a desperate analogy, since there is of course no comparison between the specificity of the command of the Reexamination Clause and the specificity of the command that there be a "jury." The footnote abandonment of our traditional view of the Reexamination Clause is a major step indeed.

II

The Court's holding that federal courts of appeals may review district-court denials of motions for new trials for error of fact is not the only novel aspect of today's decision. The Court also directs that the case be remanded to the District Court, so that it may "test the jury's verdict against CPLR § 5501(c)'s 'deviates materially' standard." Ante, at 439. This disposition contradicts the principle that "[t]he proper role of the trial and appellate courts in the federal system in reviewing the size of jury verdicts is . . . a matter of federal law." Donovan v. Penn Shipping Co., 429 U.S. 648, 649 (1977) (per curiam).

The Court acknowledges that state procedural rules cannot, as a general matter, be permitted to interfere with the allocation of functions in the federal court system, see ante, at 436-437. Indeed, it is at least partly for this reason that the Court rejects direct application of § 5501(c) at the appellate level as inconsistent with an "'essential characteristic'" of the federal court system — by which the Court presumably means abuse-of-discretion review of denials of motions for new trials. See ante, at 431, 437-438. But the scope of the Court's concern is oddly circumscribed. The "essential characteristic" of the federal jury, and, more specifically, the role of the federal trial court in reviewing jury judgments, apparently counts for little. The Court approves the "accommodat[ion]" achieved by having district courts review jury verdicts under the "deviates materially" standard, because it regards that as a means of giving effect to the State's purposes "without disrupting the federal system," ante, at 437. But changing the standard by which trial judges review jury verdicts does disrupt the federal system, and is plainly inconsistent with the "strong federal policy against allowing state rules to disrupt the judge-jury relationship in the federal court." Byrd v. Blue Ridge Rural Elec. Cooperative, Inc., 356 U.S. 525, 538 (1958). The Court's opinion does not even acknowledge, let alone address, this dislocation.

We discussed precisely the point at issue here in Browning-Ferris Industries of Vt., Inc. v. Kelco Disposal, Inc., 492 U.S. 257 (1989), and gave an answer altogether contrary to the one provided today. Browning-Ferris rejected a request to fashion a federal common-law rule limiting the size of punitive-damages awards in federal courts, reaffirming the principle of Erie R. Co. v. Tompkins, 304 U.S. 64 (1938), that "[i]n a diversity action, or in any other lawsuit where state law provides the basis of decision, the propriety of an award of punitive damages . . ., and the factors the jury may consider in determining their amount, are questions of state law." 492 U.S., at 278. But the opinion expressly stated that "[f]ederal law . . . will control on those issues involving the proper review of the jury award by a federal district court and court of appeals." Id., at 278-279. "In reviewing an award of punitive damages," it said, "the role of the district court is to determine whether the jury's verdict is within the confines set by state law, and to determine, by reference to federal standards developed under Rule 59, whether a new trial or remittitur should be ordered." Id., at 279. The same distinction necessarily applies where the judgment under review is for compensatory damages: State substantive law controls what injuries are compensable and in what amount; but federal standards determine whether the award exceeds what is lawful to such degree that it may be set aside by order for new trial or remittitur.

The Court does not disavow those statements in Browning-Ferris (indeed, it does not even discuss them), but it presumably overrules them, at least where the state rule that governs "whether a new trial or remittitur should be ordered" is characterized as "substantive" in nature. That, at any rate, is the reason the Court asserts for giving § 5501(c) dispositive effect. The objective of that provision, the Court states, "is manifestly substantive," ante, at 429, since it operates to "contro[l] how much a plaintiff can be awarded" by "tightening the range of tolerable awards," ante, at 425, 426. Although "less readily classified" as substantive than "a statutory cap on damages," it nonetheless "was designed to provide an analogous control," ante, at 428, 429, by making a new trial mandatory when the award "deviat[es] materially" from what is reasonable, see ante, at 428-429.

I do not see how this can be so. It seems to me quite wrong to regard this provision as a "substantive" rule for Erie purposes. The "analog[y]" to "a statutory cap on damages," ante, at 428, 429, fails utterly. There is an absolutely fundamental distinction between a rule of law such as that, which would ordinarily be imposed upon the jury in the trial court's instructions, and a rule of review, which simply determines how closely the jury verdict will be scrutinized for compliance with the instructions. A tighter standard for reviewing jury determinations can no more plausibly be called a "substantive" disposition than can a tighter appellate standard for reviewing trial-court determinations. The one, like the other, provides additional assurance that the law has been complied with; but the other, like the one, leaves the law unchanged.

The Court commits the classic Erie mistake of regarding whatever changes the outcome as substantive, see ante, at 428-431. That is not the only factor to be considered. See Byrd, supra, at 537 ("[W]ere 'outcome' the only consideration, a strong case might appear for saying that the federal court should follow the state practice. But there are affirmative countervailing considerations at work here"). Outcome determination "was never intended to serve as a talisman," Hanna v. Plumer, 380 U.S. 460, 466-467 (1965), and does not have the power to convert the most classic elements of the process of assuring that the law is observed into the substantive law itself. The right to have a jury make the findings of fact, for example, is generally thought to favor plaintiffs, and that advantage is often thought significant enough to be the basis for forum selection. But no one would argue that Erie confers a right to a jury in federal court wherever state courts would provide it; or that, were it not for the Seventh Amendment, Erie would require federal courts to dispense with the jury whenever state courts do so.

In any event, the Court exaggerates the difference that the state standard will make. It concludes that different outcomes are likely to ensue depending on whether the law being applied is the state "deviates materially" standard of § 5501(c) or the "shocks the conscience" standard. See ante, at 429-430. Of course it is not the federal appellate standard but the federal district-court standard for granting new trials that must be compared with the New York standard to determine whether substantially different results will obtain — and it is far from clear that the district-court standard [ ought] to be "shocks the conscience." Indeed, it is not even clear (as the Court asserts) that "shocks the conscience" is the standard (erroneous or not) actually applied by the district courts of the Second Circuit. The Second Circuit's test for reversing a grant of a new trial for an excessive verdict is whether the award was " clearly within the maximum limit of a reasonable range," Ismail v. Cohen, 899 F.2d 183, 186 (CA2 1990) (internal quotation marks omitted), so any district court that uses that standard will be affirmed. And while many district-court decisions express the "shocks the conscience" criterion, see, e.g., Koerner v. Club Mediterranee, S. A., 833 F. Supp. 327, 333 (SDNY 1993), some have used a standard of "indisputably egregious," Banff v. Express, Inc., 921 F. Supp. 1065, 1069 (SDNY 1995), or have adopted the inverse of the Second Circuit's test for reversing a grant of new trial, namely, " clearly outside the maximum limit of a reasonable range," Paper Corp. v. Schoeller Technical Papers, Inc., 807 F. Supp. 337, 350-351 (SDNY 1992). Moreover, some decisions that say "shocks the conscience" in fact apply a rule much less stringent. One case, for example, says that any award that would not be sustained under the New York "deviates materially" rule "shocks the conscience." See In re Joint Eastern S. Dist. Asbestos Litigation, 798 F. Supp. 925, 937 (E—SDNY 1992), rev'd on other grounds, 995 F.2d 343, 346 (CA2 1993). In sum, it is at least highly questionable whether the consistent outcome differential claimed by the Court even exists. What seems to me far more likely to produce forum shopping is the consistent difference between the state and federal appellate standards, which the Court leaves untouched. Under the Court's disposition, the Second Circuit reviews only for abuse of discretion, whereas New York's appellate courts engage in a de novo review for material deviation, giving the defendant a double shot at getting the damages award set aside. The only result that would produce the conformity the Court erroneously believes Erie requires is the one adopted by the Second Circuit and rejected by the Court: de novo federal appellate review under the § 5501(c) standard.

To say that application of § 5501(c) in place of the federal standard will not consistently produce disparate results is not to suggest that the decision the Court has made today is not a momentous one. The principle that the state standard governs is of great importance, since it bears the potential to destroy the uniformity of federal practice and the integrity of the federal court system. Under the Court's view, a state rule that directed courts "to determine that an award is excessive or inadequate if it deviates in any degree from the proper measure of compensation" would have to be applied in federal courts, effectively requiring federal judges to determine the amount of damages de novo, and effectively taking the matter away from the jury entirely. Cf. Byrd, 356 U.S., at 537-538. Or consider a state rule that allowed the defendant a second trial on damages, with judgment ultimately in the amount of the lesser of two jury awards. Cf. United States v. Wonson, 28 F. Cas., at 747-748 (describing Massachusetts practice by which a second jury trial could be had on appeal). Under the reasoning of the Court's opinion, even such a rule as that would have to be applied in the federal courts.

The foregoing describes why I think the Court's Erie analysis is flawed. But in my view, one does not even reach the Erie question in this case. The standard to be applied by a district court in ruling on a motion for a new trial is set forth in Rule 59 of the Federal Rules of Civil Procedure, which provides that "[a] new trial may be granted . . . for any of the reasons for which new trials have heretofore been granted in actions at law in the courts of the United States." (Emphasis added.) That is undeniably a federal standard. Federal District Courts in the Second Circuit have interpreted that standard to permit the granting of new trials where "'it is quite clear that the jury has reached a seriously erroneous result'" and letting the verdict stand would result in a "'miscarriage of justice.'" Koerner v. Club Mediterranee, S. A., supra, at 331 (quoting Bevevino v. Saydjari, 574 F.2d 676, 684 (CA2 1978)). Assuming (as we have no reason to question) that this is a correct interpretation of what Rule 59 requires, it is undeniable that the Federal Rule is "'sufficiently broad' to cause a 'direct collision' with the state law or, implicitly, to 'control the issue' before the court, thereby leaving no room for the operation of that law." Burlington Northern R. Co. v. Woods, 480 U.S. 1, 4-5 (1987). It is simply not possible to give controlling effect both to the federal standard and the state standard in reviewing the jury's award. That being so, the court has no choice but to apply the Federal Rule, which is an exercise of what we have called Congress's "power to regulate matters which, though falling within the uncertain area between substance and procedure, are rationally capable of classification as either," Hanna, 380 U.S., at 472.

* * *

There is no small irony in the Court's declaration today that appellate review of refusals to grant new trials for error of fact is "a control necessary and proper to the fair administration of justice," ante, at 435. is objection to precisely that sort of "control" by federal appellate judges that gave birth to the Reexamination Clause of the Seventh Amendment. Alas, those who drew the Amendment, and the citizens who approved it, did not envision an age in which the Constitution means whatever this Court thinks it ought to mean — or indeed, whatever the courts of appeals have recently thought it ought to mean.

When there is added to the revision of the Seventh Amendment the Court's precedent-setting disregard of Congress's instructions in Rule 59, one must conclude that this is a bad day for the Constitution's distinctive Article III courts in general, and for the role of the jury in those courts in particular. I respectfully dissent.