Appellate review of multi-claim general verdicts: the life and premature death of the Baldwin principle.

I. INTRODUCTION

Consider the following scenario. A law student slips and falls in a dark hallway after hours. He brings a tort action against the law school alleging that it was negligent because (1) it failed to provide adequate lighting in the hallways at night and (2) it failed to provide sufficient notice that the floors had just been waxed. The judge instructs the jury to return a general verdict. After deliberating, the jury finds the law school liable. The law school feels wronged and appeals the case. On appeal, the circuit court finds that there was insufficient evidence to find the law school liable on the second theory of liability. Indeed, a sign had been posted in the hallway to the effect that the floors had just been waxed, but the law student was too tired and dazed to notice. The law school argues that the general verdict must be vacated and the case remanded to the trial court for a new trial because, they argue, there is no way to know if the jury based its general verdict on the improper theory of liability. In response, the law student argues that the focus of the entire case had been on the failure to provide lighting. It was emphasized at closing argument, and almost all of the evidence related to this point. He thus urges the circuit court to exercise its discretion and find that the error was harmless. How should the circuit court dispose of this case?

In the early part of this century, the Supreme Court developed a general principle--the Baldwin (1) principle--that mandated the result urged by the law school. Essentially, the court vacated the verdict and granted a new trial due to the ambiguity in the jury's verdict. However, in the 1970s and 1980s, the federal circuit courts began to move towards the law student's position by distinguishing the Baldwin principle in myriad ways. In response, however, a handful of dissenting judges sharply criticized the movement away from the Baldwin principle as a direct contravention of binding United States Supreme Court authority. The appellate court in the law school case is therefore faced with a difficult situation. Should it adhere to possibly outdated Supreme Court precedent? Or should it embrace the modern analysis and its controversial methods of distinguishing the Court's early precedents? What happens if the federal rule, whatever it may be, conflicts with the rule of the state where the law student brought suit? This article endeavors to answer those questions.

Part II describes the genesis of the Baldwin principle and the early Supreme Court precedents that nurtured it. It also examines the circuit courts' initial endorsement of the Baldwin principle. Part III moves on to the modern era. It traces the development of discretionary rules, such as the Ninth Circuit's Traver (2) rule and the harmless error exception embraced in other circuits, from the 1970s and 1980s to the present day, and notes the circuit courts' increased receptiveness to discretionary arguments in the past six years. Part III also delineates the arguments advanced by the circuit courts to rationalize their departure from the Baldwin principle, as well as the objections of those old-fashioned judges who refused to ride the wave. Next, in Part IV, a detailed analysis is presented of the various arguments made by the modern circuit courts. After rebutting each of the arguments advanced, Part IV concludes that, contrary to the picture painted by the modern circuit courts, the Baldwin principle is alive and well. Finally, in Part V, a policy-based defense of the Baldwin principle is presented. After examining the requirement of a unanimous jury verdict in federal civil cases, Part V asserts that the appellate courts advocating harmless error analysis greatly overestimate their ability to ascertain when an error is in fact harmless. In addition, because a harmless error rule presents a severe danger of abrogating the unanimity requirement ex post, Part V maintains that it can not be justified when there are less blunt methods, such as a waiver rule, that achieve the same result at less potential cost.

II. AN OVERVIEW OF THE BALDWIN PRINCIPLE IN THE EARLY YEARS

A. The Development of the Baldwin Principle in the Supreme Court

While there has been a long debate over the proper appellate disposition of multi-faceted general verdicts where an error is found to lie in one facet, the issue has become highly controversial in recent years. Yet, up until the mid-1970s, the Supreme Court had enunciated and adhered to a general principle that required the appellate court to vacate the verdict and remand the case to the district court for a new trial. In response, the federal circuit courts respected the Court's decision and routinely applied the general principle to vacate and remand such cases.

The genesis of the principle can be traced all the way back to a Supreme Court case decided by Justice Field in 1884. In Maryland v. Baldwin, (3) a young man named Markley sued the administrator of what he alleged to be his father's estate for the one-fourth inheritance to which he claimed to be entitled under law. As part of their defense, the defendants--the administrators of the deceased's estate--asserted claims that included the following: (1) Markley was not one of the heirs of the deceased because his mother was not married to the deceased; and (2) they had executed a $3,500 accord with Markley. At trial, Markley asserted that his mother was married to the deceased, even though he could not produce a marriage certificate or a witness who attended the wedding. The star witness was a good friend of the deceased named Cross. Cross testified that the deceased had admitted to him that he was legally married to Markley's mother and explained his reasons for concealing the marriage. The defendants objected to this testimony as hearsay. The thai court overruled the objection, but permitted one of the defendants to testify as to conversations he had with the deceased about Cross's dishonesty.

At the close of evidence, the trial court submitted the case to the jury on a general verdict, (4) and the jury found in favor of the defendants. On appeal, the Supreme Court held that it was error to admit the defendant's testimony about what the deceased had told him about Cross as it was mere hearsay. (5) The question then was the proper disposition of the case. The Court noted that there was evidence introduced at trial bearing on all of the issues that the defendants asserted, and that "if any one of the pleas was, in the opinion of the jury, sustained, their verdict was properly rendered." (6) Yet, the Court went on to formulate what has become known as the Baldwin principle:

   [The] generality [of the verdict] prevents us from perceiving upon which
   [issue] [the jury] found. If, therefore, upon any one issue error was
   committed, either in the admission of evidence, or in the charge of the
   court, the verdict cannot be upheld, for it may be that by that evidence
   the jury were controlled under the instructions given. (7)

As a result of this principle, the Court vacated the general verdict and ordered a new trial for Markley. (8)

The Baldwin decision arose in the context of a single theory of liability to which there were multiple defenses, one of which was in error. It thus applied when a plaintiff sought to attack on appeal a general verdict for the defendant.

The courts soon extended the logic of the Baldwin principle to the realm of multiple theories of liability. In the 1907 case of Wilmington Star Mining Co. v. Fulton, (9) Samuel Fulton, a trackman and mine laborer in a mine operated by the Wilmington Star Mining Company, was killed by an explosion of mine gas. Fulton's widow subsequently brought a wrongful death claim against Wilmington and alleged eight specific theories of negligence. The Court, however, found there to be insufficient evidence to support one of the theories, and since "it [was] impossible ... to say upon which of the counts ... the [general] verdict was based," the judgment was vacated. (10) The conclusion, the Court believed, flowed naturally from the Baldwin decision. (11)

Forty years later, in the 1959 case of United New York & New Jersey Sandy Hook Pilots Association v. Halecki, (12) the Court once again affirmed the applicability of the Baldwin principle in this context. An electrician named Halecki died of carbon tetrachloride fume inhalation after spraying the pilot boat New Jersey with the chemical as part of a generator repair job that the ship's owners had sub-contracted to Halecki's employer. The administratrix of Halecki's estate brought a wrongful death action, and the court issued jury instructions that allowed Halecki's estate to recover on two separate theories of liability--negligence or the unseaworthiness doctrine. The jury returned a general verdict in favor of Halecki's estate. (13) On appeal, the Supreme Court held that it was error to submit the case to the jury on the unseaworthiness claim because Halecki was not a member of the ship's crew, and so the ship did not owe him a special duty. (14) The negligence claim, however, was proper. (15) Echoing (but not citing) the language of Baldwin, the Court then ordered a new trial because there was "no way to know that the invalid claim of unseaworthiness was not the sole basis for the verdict." (16) Three years later, in Sunkist Growers, Inc. v. Winckler & Smith Citrus Products Co., (17) the Court yet again affirmed the rationale of the Baldwin principle in an antitrust case in which the court submitted multiple theories of liability to the jury and a general verdict was returned. (18)

The Supreme Court was thus quite clear that the Baldwin principle forced a federal appellate court to vacate a general verdict based on multiple defenses, as in Baldwin, or multiple theories of liability, as in Halecki and Sunkist Growers, wherever one of the defenses or theories, respectively, was in error as a matter of law. (19) The proper disposition then was to vacate the judgment, remand the case, and order a new trial. It was simply improper for an appellate body to attempt to divine the defense or theory upon which the jury had based its decision.

B. The Baldwin Principle in the Lower Courts

The Court had spoken and, for the most part, the federal bench listened. Early cases in the federal appellate courts steadfastly adhered to the Baldwin principle. The Eighth Circuit, in particular, promulgated a number of decisions between 1893 and 1903 that seemed to entrench the Baldwin principle deeply in the case law. (20) Surprisingly, the Eighth Circuit did not publish another opinion referring to the Baldwin principle until 1978. Nevertheless, the Fifth Circuit picked up the slack in the first half of the century, (21) and with the Eighth and Fifth Circuits' bodies of precedent firmly established, the First, (22) Third, (23) Fourth, (24) Tenth, (25) and D.C. (26) Circuits each followed suit, often citing both to Baldwin and either the Fifth or Eighth Circuit's decisions.

Some courts extended the Baldwin principle further, to a point beyond the specific constraints of the Supreme Court's decisions. For example, in Laguna Royalty Co. v. Marsh, (27) the Fifth Circuit extended the logic of the Baldwin principle to a judgment in which a specific interrogatory was so open-ended that it was the functional equivalent of a general verdict: "The jury's answer [to the interrogatory] is, for all practical purposes, obscured in the impenetrable mystery of a general verdict.... That the mystery is less perhaps than had there been nothing but a general charge and verdict is beside the point." (28) Moreover, in New York & Porto Rico Steamship Co. v. Garcia, (29) the First Circuit found that the Baldwin principle applied to a judgment in which a jury returned a general verdict that encompassed two separate causes of action, although it held that the very submission of a general verdict on two separate causes of action is error in and of itself. (30)

In the early years, then, the Baldwin principle became firmly established in the case law and the courts extended its underlying logic to scenarios outside of the Court's precedent. But that is not to say that there were no rogues or dissenters. Indeed there were. Perhaps the most notorious of all these courts was the Seventh Circuit, although the Second and Sixth Circuits were to a lesser extent complicit in this endeavor. The point of departure for these circuits was the question of whether federal or state law governed the scenarios covered by the Baldwin principle.

In Firemen's Insurance Co. v. Follett, (31) the Seventh Circuit addressed the Wilmington variation (32) of the Baldwin principle and found state law to govern. The court noted that the Illinois "two issue" rule, as developed by an Illinois statute and a long line of state case law dating back to 1845, was that "where there are several counts in the declaration, if there is one good count sustained by the evidence, where a general verdict is returned by the jury, the verdict will not be set aside by reason of the unsustained counts being permitted to go to the jury." (33) This rule, of course, runs counter to the Baldwin principle. It essentially assumes that a jury's general verdict was based on a sufficient theory of liability, even though there was insufficient evidence to support at least one of the other theories of liability. Why, however, did Wilmington not cover this case? In Wilmington, the Court had suggested that the Illinois rule was not premised on the inability of the appellate court to ascertain the basis for the jury's decision, but rather on a waiver principle. The Court noted that the Illinois courts had construed the rule so as to apply only to situations in which the appellant had not objected to the return of a general verdict on multiple theories of liability, (34) and in Wilmington, the appellant had done so. (35) In contrast, in Firemen's Insurance, the Seventh Circuit found that the appellant had not objected and distinguished Wilmington on this basis. (36) The Firemen's Insurance decision thus exposed the Supreme Court's willingness to be somewhat receptive to the idea that a state rule could trump the Baldwin principle, and upon this basis, several courts began to explore the idea.

Although the Firemen's Insurance decision seems to have plausibly distinguished the Wilmington case, the Seventh Circuit dispensed with the niceties seven years later. In Cross v. Ryan, (37) the Seventh Circuit found the "two issue" rule to be operative without referencing the Wilmington or Firemen's Insurance decisions, Illinois law, or whether the appellant had objected prior to the close of evidence. (38) Because there was sufficient evidence supporting each theory of liability, however, the disposition of the case was not affected by this decision (or at least not in any manifest way). No harm, no foul? Perhaps. Of course, in the Cross decision itself, the failure to recognize the distinction of Wilmington offered in Firemen's Insurance was troublesome, (39) but a greater harm could have arisen by virtue of the distinction being lost in the case law. Fortunately, in the early years, it did not.

Instead, the more technically correct analysis in the Firemen's Insurance decision began to seep into the case law of the other circuits, most notably the Second Circuit. In the early years, the Second Circuit had adhered to the Baldwin principle. (40) However, from the mid-1940s until the 1960s, the Second Circuit began to come under the influence of the Seventh Circuit's analysis, often citing its decisions to support the application of a "two issue" rule. In Lee v. Pennsylvania Railroad Co., (41) the Second Circuit appeared to adopt Firemen's Insurance's distinguishing of Wilmington:

   A general verdict is upheld where there is substantial evidence supporting
   any ground of recovery in favor of an appellee. The situation is otherwise
   where timely objection has been made to the submission of a ground
   inadequately supported along with one duly supported by the evidence. But
   to preserve such a contention for the consideration of the appellate
   tribunal, the matter must be specifically called to the attention of the
   trial judge in order that he may have the opportunity to consider the
   asserted insufficiency as to one [theory of liability] and correct himself,
   if necessary, by removing it from the jury's consideration. (42)

Oddly enough, the Second Circuit cited the Cross decision for this proposition; it made no mention of Firemen's Ins. Co. itself. (43) Regardless, the "two issue" rule in the Second Circuit was short-lived. In the 1960 decision of Fatovic v. Holland America Line, (44) the Second Circuit, with little discussion, retreated from this position to once again embrace the Baldwin principle. (45)

The Sixth Circuit, on the other hand, had a different analysis of the question of whether federal or state law applied. In the earliest case, McCrate v. Morgan Packing Co., (46) the Sixth Circuit, citing Erie Railroad Co. v. Tompkins, (47) first held that Ohio's "two issue" rule did not apply. (48) Subsequently, in Volasco Products Co. v. Lloyd A. Fry Roofing Co., (49) the court suggested that under the right circumstances, the state's "two issue" rule might apply. The court explained that Erie required a federal appellate court to follow the state "two issue" rule only when jurisdiction was founded on diversity and state law governed the substantive issues in the case; however, since Volasco arose under the federal antitrust laws, the Baldwin principle applied. (50) In subsequent cases, the Sixth Circuit maintained this distinction between cases based on diversity or federal question jurisdiction. (51)

In summary, in the early part of the twentieth century, the Supreme Court articulated and gradually extended the Baldwin principle, and the lower federal appellate courts, especially the Fifth and Eighth Circuits, began to entrench it into the case law. A question arose, however, as to whether (and when) a federal appellate court should apply a state's "two issue" rule or the federal Baldwin principle, an issue complicated by the emergence of a nascent Erie doctrine. Two approaches developed. The Seventh Circuit, relying on its interpretation of Wilmington, held that state law applied whenever the appellant had failed to object to the alleged error at the trial court level, and the Second Circuit, for a time, agreed with this position. In contrast, the Sixth Circuit held that state law applied whenever jurisdiction was premised on diversity and the action was governed by state law. Otherwise, in cases governed by federal law, the Baldwin principle applied.

III. THE EFFICACY OF THE BALDWIN PRINCIPLE IN THE MODERN ERA

In the late 1970s and early 1980s, the federal appellate courts began to develop a more discretionary standard in considering the disposition of multi-faceted general verdicts in which an error lay in one facet. Two prominent approaches developed--the Ninth Circuit's Traver rule and the harmless error doctrine--although in practice they seemed to be functionally equivalent. In addition, the courts recognized three new arguments to reconcile discretion with the Baldwin principle. First, many circuit courts reinterpreted the Baldwin line of cases as holding only that the vacate-and-remand technique was appropriate if it was not possible to ascertain whether the jury based its general verdict on an invalid claim. The Baldwin line of cases was conditional, and as a result, it left open the possibility of a discretionary harmless error rule. A second line of reasoning used by the circuit courts posited that the federal appellate courts were bound by federal harmless error provisions enacted after the development of the Baldwin principle. These provisions trumped, according to some circuits. Third, courts suggested that the Supreme Court's decisions in the criminal context applying a discretionary harmless error analysis justified departure from the absolute civil rule coming from Baldwin. During this time period, the reconciliation of these competing lines of reasoning often occurred in dicta. More and more courts began to discuss the possibility of a exercised it. Finally, the old questions of conflicts of law and waiver remained, and the federal courts continued to struggle with the issue.

A. The Emergence of a Discretionary Rule.

During the modern era, the circuit courts developed two types of discretionary rules: the Ninth Circuit's Traver rule and the harmless error rule developed by the other circuits. Both involved an element of discretion, and both could be seen as efforts to conserve scarce judicial resources.

1. The Ninth Circuit's Traver Rule.

In Traver v. Meshriy, (52) the Ninth Circuit, speaking through then-Judge Anthony Kennedy, articulated what has become known as the Traver rule. William Traver, a collector of antique American clocks, alleged a variety of causes of action (including false imprisonment, assault, slander, intentional infliction of emotional distress, and a Section 1983 claim) against myriad Bank of America officers who had initially refused to let him withdraw $1,000 from his account to purchase a valuable clock and subsequently detained him. A general verdict was returned in favor of Traver. On appeal, the Ninth Circuit noted that the Section 1983 claim against the bank itself, based on a theory of respondeat superior, was likely insufficient. (53) Judge Kennedy, however, found it unnecessary to pass on this question because he held that:

   Where more than one theory of recovery has been submitted to the jury in a
   civil case, and where on appeal it is claimed that as to one of the
   theories there was a lack of evidential support or an error of law in
   submitting the theory to the jury, the reviewing court has discretion to
   construe a general verdict as attributable to another theory if it was
   supported by substantial evidence and was submitted to the jury free from
   error. (54)

Judge Kennedy relied on the Seventh Circuit's Cross decision and the Sixth Circuit's Adkins decision as the basis for this holding. However, neither Cross nor Adkins supports this decision. First, the Cross decision ignored the distinction of Wilmington explained in the Firemen's Insurance decision and, indeed, even if it had recognized the distinction, the fact that the appellant in Cross had not objected made Wilmington binding on it. This is, therefore, an instance of harm caused by the Cross decision dicta as discussed above. (55) Perhaps we could excuse Judge Kennedy for this one mis-citation, but it should be clear that, like Cross, the valid distinction of Wilmington described in Firemen's Insurance does not apply here. Judge Kennedy makes no mention of whether the appellants had objected at trial to the use of a general verdict. If they had, under the Firemen's Insurance decision the Baldwin principle would apply. If they had not, it would be a question of state law, and Judge Kennedy makes no mention of what California law is on this subject. Second, even if Cross was adequately supported in law itself, it does not support the proposition for which Judge Kennedy cites it. The Cross decision holds that a court has an obligation, under the "two issue" rule, to affirm the generalized verdict if any of the theories of liability is supported by evidence. It speaks of a duty, not discretion. Third, Judge Kennedy's cite to the Sixth Circuit's Adkins decision also is misguided. The Adkins decision holds that the Baldwin rule applies whenever jurisdiction is not premised solely on diversity jurisdiction. In this case, however, Traver based jurisdiction on 28 U.S.C. [section] 1331 and [section] 1343, not 28 U.S.C. [section] 1332. (56) Thus, Judge Kennedy's decision (and the Traver rule) should be considered controversial from the start.

Nevertheless, Judge Kennedy promulgated a number of factors to help inform future appellate discretion:

   [1] the potential for confusion of the jury which may have resulted from an
   erroneous submission of a particular claim or cause of action, (57) [2]
   whether privileges or defenses of the losing party apply to the count upon
   which the verdict is being sustained so that they would have been
   considered by the jury with reference to the count, [3] the strength of the
   evidence supporting the count being relied upon to sustain the verdict, and
   [4] the extent to which the same disputed issues of fact apply to one or
   more of the theories in question (58)

Based on these factors, Judge Kennedy found that Traver's case was "an appropriate case to exercise our discretion to construe the general verdict as attributable to the state tort law theories of recovery." (59) The court failed to mention the first factor specifically. (60) It did note, however, with regard to the second factor, that the "defenses of reasonable conduct and action taken in good faith were fully applicable to the state claims [as well as the [section] 1983 claim against the bank]." (61) The third factor also was satisfied because there was "ample evidence to support an award to the plaintiff based upon [the state claims]." (62) Finally, under the fourth factor, the court noted that

   [T]he section 1983 claim is all but derivative of the state torts alleged
   in the claim and the facts bearing upon it are substantially the same. A
   jury finding of liability on a section 1983 claim would necessarily
   encompass a finding of liability on one or more of the state law claims,
   and state law provides for the employer's liability in this instance for
   the state torts committed by the employees in the scope of their
   employment. (63)

Therefore, Judge Kennedy held that the Ninth Circuit would exercise its discretion to construe the general verdict as applicable to the state claims against the individuals and the bank itself, notwithstanding that it doubted the viability of a respondeat superior theory against the bank under Section 1983. (64)

In subsequent cases, the Ninth Circuit continued to pay lip service to the Baldwin rule, but the Traver rule has been so consistently recognized and applied that it is safe to say that the Ninth Circuit has entrenched the discretionary principle in its case law. With respect to the Traver factors, in Hendry v. Exide Electronics Corp., the court, summarizing the Ninth Circuit case law on this issue, noted that it balances these factors "in an ad hoc manner, without any one factor being accorded particular weight and without necessarily considering all four factors in each case." (65) The court has on occasion applied the Traver rule to affirm a general verdict without any mention of the factors, (66) and yet, in some cases, a problem with one of the factors is enough for the court to refuse to exercise discretion. (67) The Traver rule is thus a free-form inquiry giving the appellate court carte blanche to affirm or reverse a general verdict without any coherent unifying principle.

Not every Ninth Circuit judge, however, embraces the Traver rule. Judge Kozinski, in particular, has launched a scathing attack on it, arguing that it blatantly flouts binding Supreme Court precedent:

   More than a century ago, the Supreme Court explained that, because we
   cannot read the minds of jurors, a general verdict that may have been based
   on an improper theory of liability must be reversed. "The verdict's
   generality prevents us from perceiving upon which plea they found. If,
   therefore, upon any one issue error was committed, either in the admission
   of evidence, or in the charge of the court, the verdict cannot be upheld"
   [citing Baldwin]. The Court has three times reiterated the rule in
   unequivocal terms [citing Wilmington, Halecki, and Sunkist Growers]. The
   wisdom of this rule is beyond cavil. As one commentator said of the general
   verdict: "No one but the jurors can tell what was put into it and the
   jurors will not be heard to say. The general verdict is as inscrutable and
   essentially mysterious as the judgment which issued from the ancient oracle
   of Delphi." When the Supreme Court speaks consistently and repeatedly on an
   issue, the inferior federal courts are bound to follow suit.... The Baldwin
   rule exists precisely to deal with the kind of uncertainties this case
   presents; fine tuning is unnecessary when the facts and law coincide. (68)

In addition, Judge Kozinski chastised the Ninth Circuit's "erosion of a perfectly good rule." (69) Although Traver may be read as a narrow exception to the Baldwin principle, in subsequent cases, Judge Kozinski complains, the Ninth Circuit has neglected to even acknowledge the Supreme Court decisions and the Baldwin principle. (70) Judge Kozinski did not find this surprising: "As is often the case when courts allow themselves discretion to avoid absolute principles, the exceptions have now submerged the rule." (71) Nonetheless, the Ninth Circuit as a whole has so far resisted Judge Kozinski's challenge and consistently applied the Traver rule. (72)

2. The Harmless Error Exception.

The Baldwin principle has also been somewhat eroded, albeit a bit more gradually, in several other circuit courts. Although these courts have styled their discretionary analysis a "harmless error" exception to the Baldwin principle, it differs little from the Traver rule. However, courts using the harmless error rule have not applied it as often as the Ninth Circuit has applied the Traver rule. Instead, most circuit courts attempt to keep it cabined within rather strict bounds. It thus clings a little more tightly to the Baldwin principle than the Traver rule does.

a. Genesis.

The genesis of a harmless error exception to the Baldwin principle in the circuit courts occurred in the 1970s. In Amer. Airlines, Inc. v. U.S., (73) the Fifth Circuit became the first circuit to delineate a harmless error exception not based on a state "two issue" rule, as in the Sixth and (arguably) the Seventh Circuit. (74) An American Airlines jet had crashed on its approach to the Greater Cincinnati Airport, killing fifty-eight out of sixty-two people on board, and the families of the victims brought a suit for wrongful death. The plaintiffs offered thirty-two theories of negligence, and the jury returned a general verdict in favor of the plaintiffs. On appeal, the Fifth Circuit found that "each and every one of the elements was supported by substantial evidence in the record, except one." (75) Yet, the court felt that it was "inconceivable that in the mass of testimony so clearly establishing negligence in thirty other particulars this issue could have influenced the verdict against American," and thus, the court found "no justification for reversal and a new trial on such insubstantial grounds." (76) Surprisingly, the Fifth Circuit cited Wilmington for this proposition. (77) According to its unique reading of Wilmington, the Baldwin principle only applies "if it is impossible to say upon which counts the verdict was based." (78) Although it is plausible to read Wilmington in this way, (79) it is a bit of a stretch to do so based on Baldwin as well as on both Halecki and Sunkist Growers, which came after Wilmington but before the instant case and did not speak in such conditional terms.

Although American Airlines offered at least a plausible attempt to distinguish Supreme Court precedent, it was neglected outside of the Fifth Circuit. A trio of cases, however, arising in the Second, Seventh, and Tenth Circuits became the force that infused the harmless error exception into the case law of the circuit courts across the nation. (80) Ironically, the trio of cases was less analytically sound than American Airlines. Indeed, they dispensed with the obligation to conform to Supreme Court precedent altogether.

With these four precedents on the books, the issue came before the Second Circuit. In Morrissey v. National Maritime Union of America, (81) Judge Friendly, writing for the Second Circuit panel, began his opinion by citing Halecki and the early cases adhering to the Baldwin principle and noting that "the language used is generally quite absolute." (82) After recognizing the existence of the harmless error exception in Collum, Gardner, and Roginsky, Judge Friendly, without passing on the merits of these cases, suggested that "the qualification on the general rule of Halecki and other cases must be kept within rather strict bounds." (83) He then promptly held that the instant case was not a proper case in which to exercise discretion given the high burden he felt necessary to invoke it. (84)

Two years later, the Eighth Circuit, the court most responsible for entrenching the Baldwin principle in the early years, also came to embrace the harmless error exception, albeit cautiously and without applying it. In Mueller v. Hubbard Milling C0., (85) the Court was faced once again with a problematic general verdict: "the insurmountable difficulty is that the general verdicts returned by the jury ... may rest solely upon the contract claims, which are tainted with the trial court's erroneous and prejudicial admission of parol evidence." (86) After quoting Judge Friendly's opinion in Morrissey at length, (87) the court held:

   We are in agreement with [Judge Friendly's] analysis, which is decisive
   here. First, there is no material distinction between a situation, like
   that in Morrissey, in which one of several theories of liability should not
   have been submitted to the jury at all, and a situation, like that here, in
   which one of several theories of liability is not sustainable because of an
   erroneous and prejudicial admission of evidence. The essential inquiry in
   either case is whether the appellate court is fairly convinced that the
   jury proceeded on a sound basis. Second, we are unable to say with any
   confidence that the verdicts returned below rested on anything other than
   the contract claims. Plaintiff's counsel stated in closing argument: "Our
   principal focus has been on the contract claims, the promises that John
   McNeal made...." A careful review of the record convinces us that all of
   the plaintiff's theories were of marginal validity at best and that the
   contract theory might well have provided the sole basis for the verdicts.

The Eighth Circuit thus agreed with Judge Friendly's warning that the harmless error exception, if it was consistent with the Baldwin line of cases, should be kept within rather strict bounds, and it believed that a "fairly convinced" standard would do that.

b. The Modern Harmless Error Exception.

The harmless error exception arose in the Second, Fifth, Eighth, and Tenth Circuits. Going forward, however, the application of the harmless error rule oscillated in these circuits--a result that is perhaps unavoidable given the cycling on multi-member courts. To be sure, the harmless error rule very rarely has been applied to uphold a general verdict. (89) On the theoretical issue, however, the circuit courts have been highly inconsistent in their reasoning. After Mueller's articulation of a narrow harmless error exception, the Eighth Circuit not only never applied that exception, it did not even mention the possibility of it again. (90) In the Second Circuit, (91) the court, at least until the 1998 opinion in Bruneau, (92) consistently applied the general rule, (93) except that it occasionally noted that it did possess the power to apply a harmless error exception. (94) Likewise, in the Fifth Circuit, the court has also seemed to disfavor the harmless error rule. After the American Airlines decision in 1969, it issued a series of decisions applying the general rule without so much as mentioning harmless error. (95) In the 1984 case of Braun v. Flynt, however, the Fifth Circuit recognized the emergence of a harmless error rule in other circuits, but ironically not the American Airlines decision. It applied the rule to a general verdict encompassing a valid "false light" claim and an invalid appropriation claim because the focus of the case was on the "false light" claim. (96) Yet, after Braun became one of the rare cases in which harmless error was found, the Fifth Circuit only mentioned the possibility of harmless error in two out of the next seven cases and applied it in neither case. (97) The Tenth Circuit has done the same. (98) Thus, in the circuits that originated the harmless error exception, there has been little enthusiasm for it in subsequent cases.

It should be noted that three out of the four cases in which the court applied the harmless error exception were handed down in the past six years, perhaps indicating an increased willingness to use it. (99) In addition to the Second Circuit's Bruneau decision in 1998, two of these cases arose in circuits that had been much stronger adherents (as compared to the four circuits that originated the harmless error exception) to the Baldwin principle during the debate that arose in the 1970s and 1980s. The Fourth Circuit, for example, had adhered to the Baldwin principle throughout this time. (100) But in the 1995 case of Henderson v. Winston, (101) the Fourth Circuit issued a per curiam opinion applying a harmless error exception in the context of a verdict for the defendant. (102) It distinguished Sunkist Growers:

   [T]he rationale for the Sunkist rule ... is that a reviewing court cannot
   be certain that the legal basis for the verdict was valid, meaning that
   prejudice is necessarily established. Here, by contrast, there was only one
   theory of liability pressed, and there is no doubt that [a critical element
   of the theory of liability] went against the plaintiff. Hence, even if the
   jury did err on one of the other elements, the verdict for defendant was
   still unquestionably valid. (103)

The Third Circuit likewise consistently had applied the Baldwin principle up until 1999; (104) however, in Hurley v. Atlantic City Police Department, (105) Chief Judge Becker, relying both on Federal Rule of Civil Procedure 61 and the proposition that the Baldwin principle is narrow and does not apply where the record made it possible to determine the basis for the jury's decision, held that the harmless error exception applied where a jury returned a general verdict that encompassed an improper quid pro quo theory of sexual harassment. (106) In a dissent bordering on disgust, Judge Cowen, echoing but not citing Judge Kozinski's Kern dissent, sharply criticized the majority's departure from both circuit and Supreme Court precedents:

   [O]ur precedents could not be more clear [in applying the Baldwin
   principle].... The Supreme Court has been equally consistent in holding
   that a general verdict in a civil case must be reversed if any of the
   claims submitted to the jury are found to be unsound.... Since Baldwin, the
   Supreme Court has reaffirmed this rule, without exception, on at least
   three separate occasions.... [T]he fact of the matter is that prior to
   today's decision, this court has never relied on a harmless error analysis
   to affirm a general verdict that may have rested on an improper ground. It
   is remarkable that after more than one hundred years of reviewing civil
   judgments founded on general verdicts, the majority believes that we have
   only now come upon the case that calls for such drastic action, I do not
   believe that that day has arrived. (107)

But Judge Cowen did not stop there. As a policy matter, he argued that the Baldwin rule made perfect sense:

   [C]an any of us really purport to know how the jury viewed the evidence in
   this case? Any experienced trial lawyer can attest to the fact that juries
   sometimes view cases in surprising ways.... I am very concerned that the
   majority's newly-minted harmless error analysis will invite further efforts
   by appellate judges, in even more difficult cases, to divine what a jury
   may have been thinking when it rendered a general verdict. Although the
   majority emphasizes that their decision to employ a harmless-error doctrine
   is founded upon the "extreme" facts of this case, that is simply to say
   that we will only affirm a tainted general verdict in the future when we
   believe that the jury has reached an obviously correct result. Because I
   believe that we have no role in speculating how a jury might have viewed
   the evidence presented at trial, and that such attempts at judicial
   telepathy are unwise and contrary to our precedents, I respectfully
   dissent. (108)

It thus seems that the harmless error exception has become more prevalent in the last six years. (109) Indeed, in recent years, the question has now turned to what level of confidence is required that an error was harmless. Some circuits have set a standard of "reasonable certainty," (110) others have cast the standard in terms of "sufficient confidence." (111) The Tenth Circuit, in particular, has struggled with this standard. Initially, the standard was reasonable certainty. (112) In subsequent cases, however, the Tenth Circuit beefed up the standard to require absolute certainty. (113) The absolute certainty standard is, of course, extraordinarily narrow, and it is perhaps functionally equivalent to the Baldwin principle. (114) Although these decisions recognize the possibility that the Supreme Court's precedents could be read to embrace a harmless error exception, as described below, they jack the standard up so high that they basically restore the Baldwin principle without overruling the earlier decision recognizing the harmless error doctrine.

Coupled with the widespread emergence of the harmless error exception in recent years are the modern circuit courts' attempts to assert new justifications for departing from the Baldwin principle and for downplaying the force of Supreme Court precedent. In Asbill v. Housing Authority of the Choctaw Nation of Oklahoma, (115) a decision heavily relied upon in Chief Judge Becker's Hurley opinion, (116) the Tenth Circuit noted that the Court's precedents "[do] not paint with as broad a brush as appears." (117) Instead, it believed that there was a statutory basis for the harmless error exception: "As with all errors committed at trial, a litmus test for reversal is whether the appellant was thereby unjustly prejudiced." (118)

Judge Barrett's opinion apparently sought to escape the vice of the Baldwin line of precedent by asserting that the promulgation of the Federal Rules of Civil Procedure--and in particular Rule 61--after Baldwin and Wilmington somehow abrogated the decision, and similarly, that the enactment of 28 U.S.C. [section] 2111 after Halecki and Sunkist Growers somehow abrogated them as well. In addition, Judge Barrett reread Wilmington to apply to only "vital" errors that are prejudicial to the "substantial rights" of the objecting party--terms that are mirrored in Rule 61 and 28 U.S.C. [section] 2111. Yet, it is unclear what precisely Judge Barrett meant. Was he suggesting that the Baldwin line of cases was still authoritative, but included a never before discovered harmless error exception pregnant in its language? Of course, Wilmington could not be read as referring to Rule 61 or Section 2111 because it came before their enactment. But maybe Judge Barrett was suggesting the opposite--that enactment of Rule 61 and Section 2111 was premised on a reading of Wilmington and, because of that, the application of the statutory rule was fully consistent with Supreme Court precedent. The opinion is a bit confusing, but as will be discussed in Part IV, Judge Barrett may have been on to something.

The courts could also escape the Baldwin line of cases by referring to Supreme Court precedent in the criminal context that applied the harmless error exception. However, the only circuit to attempt to do so was the Second. In Bruneau, Judge Cardamone's opinion applying a harmless error exception asserted that the Supreme Court's criminal rule, as discussed in Kotteakos v. U.S. (119) was authoritative. (120) In Kotteakos, the Court had indeed applied a harmless error analysis to affirm a criminal conviction tainted by an improper theory of liability. The problem with Judge Cardamone's importation of this rule into the civil context was that the Wilmington decision strongly suggested that the criminal rule is inapplicable, although it failed to elaborate on this point. (121) At the very least, then, Judge Cardamone should have resolved this embarrassing conflict.

Ironically, in the Fifth Circuit, the court invoked two Supreme Court criminal cases for precisely the opposite reason--to support the Baldwin principle. In Neubauer, the Fifth Circuit relied heavily on the Supreme Court's criminal decisions in Stromberg v. California (122) and Zant v. Stephens (123) to justify a harmless error exception. (124) Thus, in both the Second and the Fifth Circuits, the courts transposed the Court's criminal decisions, albeit to reach opposite results. As will be discussed in Part IV, however, both of these decisions are flawed in that they fail to take into account crucial distinctions in the fabric of the Court's criminal and civil jurisprudence on this issue.

Finally, it should be noted that the harmless error exception is inherently discretionary, and in many cases, the analysis of whether harmless error lies often tracks the Traver factors delineated by then-Judge Kennedy: potential for jury confusion; joint applicability of defenses; strength of evidence of proper claims; and factual similarities. For example, like the Traver factor involving potential jury confusion, the harmless error analysis has often turned on whether parties stressed the proper claim at trial and in closing arguments. A second example is the comparability between the "factual similarities" factor of the Traver rule and the First Circuit's subset theory. In Brochu v. Ortho Pharmaceutical Corp., (125) the First Circuit, despite professing its adherence to the Baldwin principle, noted:

   The case at bar, however, is unusual in that the elements needed for fraud
   and for strict liability corresponded, except that [the potentially invalid
   fraud claim] required an additional element. If the jury erroneously
   thought this element was established, and wrongly concluded that there was
   fraud, it necessarily and warrantably must have found all the elements
   establishing strict liability. Since the damages were the same in either
   event ... defendant was not harmed. Rather, it was a case where all roads
   led to Rome. (126)

The subset theory of harmless error, as this line of reasoning has become known, was also applied by the First Circuit in Shepp v. Uehlinger. (127) To be sure, the subset theory is a bit narrower than the Traver rule's factual similarities factor, (128) but the underlying principle is the same--harmless error will lie where the jury must consider the same evidence on two different theories of liability encompassed in a general verdict. (129) In this sense, it seems for the most part as if the Traver rule is really just the harmless error exception by another name.

c. The Seventh Circuit's McGrath Rule

In Firemen's Insurance and Cross, the Seventh Circuit had created a rule whereby the court could uphold a multi-claim general verdict if there was a valid basis for one of the claims based on its reading of the Wilmington decision. (130) In a subsequent case, the circuit seemed to back off of this position in favor of the Baldwin principle. (131) Yet, in 1981, the Seventh Circuit decided to go in a completely different direction. In McGrath v. Zenith Radio Corp., (132) the court devised a rule whereby a defendant challenging a multi-claim general verdict on appeal could triumph only if he could show that the plaintiff could not prevail under any of the theories of liability. (133) The purported rationale offered by the court was that it would be mere "speculation" to assume that the invalid theory was the basis of the jury's decision. (134) The McGrath decision thus turned the Baldwin principle on its head. Baldwin held that where one theory of liability is improper, but is cloaked in the guise of a general verdict, the appellate court must reverse because it cannot know that the jury did not base its decision on the invalid ground. The McGrath rule, in contrast, states that an appellate court must uphold the verdict because there is no way to know that the verdict was not based on the valid ground. For this reason, Judge Kozinski is sharply critical of the McGrath role: "The Seventh Circuit, for reasons of its own, has adopted a maverick rule precisely the opposite of that repeatedly announced by the Supreme Court.... For reasons explained in Baldwin, this rules makes no sense at all, never mind that it contravenes Supreme Court authority." (135) Indeed, in McGrath, the Seventh Circuit never even mentions any of the Supreme Court's decisions on point nor, for that matter, any of its own previous decisions waffling between the Baldwin rule and the harmless error exception. It is truly a maverick rule.

B. The Choice-of-Law Issue

In the modern era, the conflict between state "two issue" rules and the Baldwin principle in federal cases based on diversity jurisdiction has created a split in the circuits. The Sixth Circuit has stuck with the rule it developed in the early years whereby state law governs when the case is tried under state law and the Baldwin principle governs when a case is tried under federal law, such as a civil rights or antitrust case. (136) The Eleventh Circuit has jumped into the fray as well only to issue inconsistent decisions. (137) Yet, in perhaps a more surprising development, the Ninth Circuit's Kern decision intimated in a footnote that perhaps state law controlled its adoption of the Traver rule, (138) On the other hand, the Fourth Circuit and the Fifth Circuit (to some degree) have both come down squarely on the side of applying federal law, reasoning that appellate review of general verdicts was a matter of procedure and not substance. (139) Noticeably absent in this debate is any direct discussion of the Erie doctrine.

C. Waiver

A number of modern appellate courts have held that an appellant is required to object to a general verdict before he can invoke the Baldwin principle on appeal. The opinions follow a predictable path. Typically, they will express some frustration at the district court's failure to employ a specialized verdict, (140) recognize that Rule 49(a) commits the decision on verdict form to the district court's discretion, and finally suggest greater use of the specialized verdict, often citing to an article by former Chief Judge Brown of the Fifth Circuit. (141) Since they are powerless to force the district court to adopt a specialized verdict, they shift the burden to the appellant to properly object to the use of the general verdict so as to put the district court on notice and thereby preserve the applicability of the Baldwin principle on appeal.

The Ninth Circuit seems to have the most established waiver rule. In McCord v. McGuire, (142) Judge Kozinski held that a request for a specialized verdict and a proper objection, if refused, is necessary to preserve the Baldwin principle on appeal:

   Maguire's failure to request a special verdict as to each factual theory in
   the case prevents him from pressing [the Baldwin] argument on appeal.
   [While] Federal Rule of Civil Procedure 49(a) gives district courts wide
   discretion in the use of special verdicts ... litigants have the
   responsibility to request or submit special verdict forms. Litigants like
   Maguire who wish to challenge the sufficiency of the evidence as to some,
   but not all, specifications of negligence must present an appropriate
   record for review by asking the jury to make separate factual
   determinations as to each specification. Any other rule would unnecessarily
   jeopardize jury verdicts that are otherwise fully supported by the record
   on the mere theoretical possibility that the jury based its decision on
   unsupported specifications. We will not allow litigants to play procedural
   brinkmanship with the jury system and take advantage of uncertainties they
   could well have avoided. (143)

Judge Kozinski's move may be quite clever. All can agree that judicial economy is harmed by the Baldwin role, and that the Traver rule and the harmless error doctrine are attempts to mitigate the harsh effect of the Baldwin principle. Yet, the same result may be obtained if the appellate court is never placed in the position of having to apply the Baldwin principle in the first place. Although an appellate court cannot force a district court to employ a specialized verdict, it can force the litigant to bring it to the court's attention, which may place the necessary pressure on the court to craft the verdicts specially. Judge Kozinski's waiver rule thus remains faithful to the Baldwin principle as precedent while at the same time doing the work of the harmless error doctrine with respect to economizing on scarce judicial resources without engaging in judicial telepathy. It should therefore not be surprising that other circuits have toyed with this approach. (144)

IV. THE VITALITY OF THE BALDWIN PRINCIPLE IN THE MODERN ERA

At its most abstract level, the modern debate in the circuit courts between the Baldwin principle and a discretionary harmless error rule is focused on the weight and scope of Supreme Court precedent versus a perceived statutory command. In recent years, the debate has become a prize fight, with Judges Kozinski and Cowen in one comer arguing for adherence to Supreme Court precedents; and Judges Barrett and Cardamone, Chief Judge Becker and then-Judge Kennedy in the other distinguishing the precedents and pointing to the statutory provisions to buttress their position. No one knows who the ultimate winner in this battle will be. Indeed, the Supreme Court likely will have to step in at some point and resolve the issue. For present purposes, however, the purpose of this article is to ascertain who has a better case on the merits as the law stands right now. (145)

The first line of analysis is the authority, strength, and scope of the Supreme Court's Baldwin line of cases. There are two possible interpretations. One interpretation, the "absolute" interpretation endorsed by Judges Kozinski and Cowen, is that the Supreme Court has affirmatively held that because it is impossible to ascertain the basis for a jury's general verdict when one of the claims is invalid, the case must be remanded for retrial. In Sunkist Growers, for example, the Court seemed to embrace this absolute view; (146) it did not intimate (or even inquire into) whether it was possible that the jury verdict could have rested on the valid claim, but rather merely cited Baldwin to suggest that it was "unnecessary for us to explore the legality of the other theories." (147) Nowhere did the court inquire into whether the jury's general verdict was premised .on the valid claim. It only cited Baldwin. The second interpretation, which we might call the "conditional" interpretation, is that the Baldwin line of cases applies only if it is impossible to say upon which claim the general verdict rested. Recall Judge Barrett's assertion in Asbill that the Baldwin line of cases does "not paint with as broad a brush as appears" (148) and the Fifth Circuit's citation to Wilmington for this proposition in the American Airlines case. (149) In Wilmington, the Court's position may perhaps have suggested a conditional approach in the following passage:

   [W]here a jury is wrongfully permitted ... to take into consideration in
   reaching a verdict counts of a declaration which have not been supported by
   any evidence, and where it is impossible from the record to say upon which
   of the counts of the declaration the verdict was based, the judgment
   entered under such circumstances can [not] be sustained upon the theory
   that substantial rights of the objecting party had not been invaded. (150)

The conditional inference is a negative one deriving from the highlighted language. The Court's use of the "where it is impossible" language suggests that it may not be possible in certain cases to ascertain the basis for a general verdict. It is also at least plausible to infer from Baldwin and Halecki a similar conditional rule. (151) But to say a particular interpretation is plausible is not to say that it is persuasive. It is true that the Baldwin line of cases could be read either way. But the question is which interpretation conforms more closely to the feel of these cases and the Supreme Court's intent. The answer, I believe, is that the Baldwin principle is absolute. In each case in which the Court has discussed the issue, the Baldwin rule has appeared to be automatic. The Court has never even hinted at the possibility of harmless error nor has it ever found that it could somehow divine the basis for a multi-faceted general verdict. (152) Neither Judge Barrett nor the Fifth Circuit can point to any case to this effect. We should not, then, read too much into negative inferences, especially when they are such a sharp departure from the traditional rule. If the Court intended such a drastically different rule than the one presumed in the early years, one would naturally expect that the Court would have been more explicit about it. Pregnant inferences from casual language are not a sound analysis of precedent. Thus, in the absence of a clearer statement from the Court, the absolute interpretation of the Baldwin principle should control. It should thus have a binding effect on the lower federal courts--an effect that those circuits employing discretionary rules, and most certainly the Seventh Circuit's McGrath rule, are refusing to recognize.

The modern circuit courts have attempted to circumvent the absolute interpretation of the Baldwin line of cases by pointing to various external sources. Recall Judge Barrett's citation to Rule 61 of the Federal Rules of Civil Procedure in Asbill. (153) The caption above Rule 61 is entitled "harmless error," and it reads as follows:

   No error in either the admission or the exclusion of evidence and no error
   or defect in any ruling or order or in anything done or omitted by the
   court or by any of the parties is ground for granting a new trial or for
   setting aside a verdict or for vacating, modifying, or otherwise disturbing
   a judgment or order, unless refusal to take such action appears to the
   court inconsistent with substantial justice. The court at every stage of
   the proceeding must disregard any error or defect in the proceeding which
   does not affect the substantial rights of the parties. (154)

Rule 61 thus raises a difficult interpretive question. Although the drafters of Rule 61 clearly set out to economize judicial resources by instructing courts to ignore harmless errors, it is entirely possible to read the Baldwin line of cases as a considered judgment that when a jury returns a general verdict premised on multiple claims, one of which is invalid, the "substantial rights of the parties" would be affected per se if the court ascribed the verdict to the valid claim. If so, then there is no conflict between the Baldwin line of cases and Rule 61's procedural mandate. The rule simply speaks to other more technical errors that do not invade the "substantial rights" of the parties.

Alternatively, Rule 61 could be read as conflicting directly with the absolute interpretation of the Baldwin line of cases if one were to interpret the rule as contemplating a case-by-case inquiry into harmless error. Because Congress has ratified Rule 61 it should trump. Rule 61 was enacted in 1937, well after Baldwin and Wilmington, but earlier than Halecki and Sunkist Growers. On this theory, Rule 61 abrogated the former decisions; thus, the Court simply made a mistake in Halecki and Sunkist Growers by blindly adhering to an absolute interpretation of the earlier precedents without taking into account a rule that was enacted after they were decided. Yet, it was the Supreme Court that promulgated the rules of civil procedure, and it seems odd to suggest that it was unaware of its own rules. Moreover, it is important to recognize that the language of Rule 61 mirrors the language of the Wilmington decision, (155) and if I am correct in reading Wilmington as absolute in light of Halecki and Sunkist Growers's treatment of it as such, then it also must be the case that the absolute interpretation of the Baldwin principle fits squarely within Rule 61.

An argument could be made that reliance on Rule 61 to sidestep the Baldwin line of cases is misplaced because Rule 61 does not apply to appellate courts. As Rule 1 makes clear, the Federal Rules of Civil Procedure apply only to the district courts. (156) In McDonough Power Equipment, Inc. v. Greenwood, (157) however, the Court noted that "while in a narrow sense Rule 61 applies only to the district courts, see Fed. Rule Civ. Proc. 1, it is well-settled that the appellate courts should act in accordance with the salutary policy embodied in Rule 61." (158) The Court itself thus suggested that the policies underlying Rule 61 should be applicable to appellate review. In McDonough, the Court went on to note that "Congress has further reinforced the application of Rule 61 by enacting the harmless error statute, 28 U.S.C. [section] 2111, which applies directly to appellate courts and which incorporates the same principle as that found in Rule 61." (159) In Asbill, Judge Barrett had cited this provision in conjunction with Rule 61 to justify using harmless error analysis. (160) Section 2111 is also entitled "harmless error" and provides:

   On the hearing of any appeal or writ of certiorari in any case, the court
   shall give judgment after an examination of the record without regard to
   errors or defects which do not affect the substantial rights of the
   parties. (161)

Even if, then, Rule 61 did not apply to appellate courts, Section 2111 would. Yet, Section 2111 uses the same "substantial rights" language that both Wilmington and Rule 61 employed, and if the analysis above is correct, the same result should obtain here. The Baldwin principle, read in light of Halecki and Sunkist Growers, would suggest that if an appellate court were to ascribe a jury's general verdict to a valid claim where another was improper, the substantial rights of the parties would be affected per se. Thus, external sources do not justify a departure from the Baldwin principle.

An alternative avenue of escaping the Baldwin principle is based on the Court's criminal rule. The circuit courts, however, have read the criminal rule in conflicting ways. In Bruneau, Judge Cardamone's opinion for the Second Circuit relied on Kotteakos v. U.S. as authority for harmless error analysis; (162) however, the Fifth Circuit's opinion in Neubauer relied on the Court's decision in Stromberg v. California to affirm the vitality of the Baldwin principle. (163) Who is right? The disentanglement of the Court's criminal harmless error rule is not easy, and a thorough analysis of it is beyond the scope of this article. But as a precedential matter, we can analyze the viability of relying on these decisions to justify either harmless error or the Baldwin principle.

In Kotteakos, (164) a large group of people was charged with a single conspiracy to defraud various credit institutions and the Federal Housing Administration. A jury returned a verdict against all the defendants, but on appeal, the circuit court held that there was not a single conspiracy engaged in by all the defendants, but rather separate and distinct conspiracies as a matter of law. (165) The basic defect in the single conspiracy charge was that one defendant ran the operation. The other defendants, without knowledge of each other, each individually acted through him; therefore, a single conspiracy could not be proven. The circuit court, however, relying on a predecessor to 28 U.S.C. [section] 2111, had refused to overturn the convictions on the ground that the error was not prejudicial and thus harmless. The Court reversed. The substantial right of the parties that was invaded, according to the Court, was "the right not to be tried en masse for the conglomeration of distinct and separate offenses committed by others as shown by this record." (166) It is curious, then, that Judge Cardamone relied on a case that actually reversed and refused to find harmless error to justify such a sharp departure from otherwise controlling Supreme Court precedent.

But there are deeper problems with Judge Cardamone's reliance on Kotteakos. First, the scenario is analytically different from the Baldwin line of cases. It does not involve a general verdict encompassing multiple theories of liability, one of which is invalid; rather, it involves a situation in which one theory of liability is promulgated, and is itself invalid. In Kotteakos, an appellate court could not construe the verdict to be anything else. Thus, it simply does not speak to the Baldwin principle. Second, Judge Cardamone's opinion fails to recognize that the Kotteakos Court had noted that the predecessor statute to 28 U.S.C. [section] 2111 was primarily concerned with criminal cases:

   [A]nyone familiar with it knows that [the predecessor statute] grew out of
   widespread and deep conviction over the general course of appellate review
   in American criminal causes. This was shortly, as one trial judge put it
   after [the predecessor statute] had become law, that courts of review,
   "tower above the trials of criminal cases as impregnable citadels of
   technicality." So great was the threat of reversal, in many jurisdictions,
   that criminal trial became a game for sowing reversible error in the
   record, only to have repeated the same matching of wits when a new trial
   had been thus obtained. (167)

The Wilmington Court's recognition that criminal cases are somehow different taps into this history of concern with technicalities in the criminal context. (168) The legal system has historically viewed a technicality that allows a murderer or rapist to walk out of the courtroom much differently than a technicality that allows a negligent driver to do the same, and consequently, the federal harmless error statute is primarily intended to address the former situation. Moreover, even if there were a similar concern with technicalities in the civil context, it does not follow that the Baldwin principle needs to be abrogated as a result. The Baldwin principle is not based on a technicality, but instead on the inability of an appellate court to ascertain whether a jury based its general verdict on a proper or improper claim. It is one thing to say that an evidentiary mistake is harmless; it is quite another to say that a general verdict containing an invalid theory of liability is harmless where the court has no way of knowing on what the jury based its verdict. There is a difference between judicial prognostication as to what effect an error may have had on a jury and judicial telepathy as to what the jury actually decided.

Judge Cardamone's attempt to tap into Kotteakos and the criminal cases to support a harmless error rule was unwise. But what about the Fifth Circuit's attempt to tap into the same genus of cases to justify the Baldwin principle? The Fifth Circuit, as noted above, relied on Stromberg to buttress its application of the Baldwin principle. In Stromberg v. California, (169) Yetta Stromberg had been convicted of violating a California criminal syndicalism statute prohibiting the display of a red flag (1) as an emblem of opposition to the organized government; or (2) as an invitation to anarchistic action; or (3) as an aid to propaganda of a seditious nature. On appeal, the Court found the first clause of the statute to be repugnant to the Constitution on its face. For our purposes, though, the crucial passage related to the disposition of the case:

   The verdict against the appellant was a general one. It did not specify the
   ground upon which it rested. As there were three purposes set forth in the
   statute, and the jury was instructed that their verdict might be given with
   respect to any one of them, independently considered, it is impossible to
   say under which clause of the statute the conviction was obtained. If any
   one of these clauses ... was invalid, it cannot be determined upon this
   record that the appellant was not convicted under that clause. It may be
   added that this is far from being a merely academic proposition, as it
   appears, upon an examination of the original record filed with this Court,
   that the State's attorney upon the trial emphatically urged upon the jury
   that they could convict the appellant under the first clause alone, without
   regard to the other clauses. It follows that ... the necessary conclusion
   from the manner in which the case was sent to the jury is that, if any of
   the clauses in question is invalid under the Federal Constitution, the
   conviction cannot be upheld. (170)

On its face, it would seem that the disposition of the Stromberg case is fully in accord with the Baldwin principle, and the Fifth Circuit was thereby correct to rely upon it. It is important to note, however, that Stromberg, like the Baldwin line of cases, may be susceptible to an absolute interpretation or a conditional interpretation focusing on the "if any of the clauses in question is invalid under the Federal Constitution" language. In subsequent cases, the Court, unlike in the Baldwin context, made clear that the conditional interpretation is actually the correct one. In the most recent analysis of the Stromberg line of cases, Griffin v. U.S., (171) the Court stated:

   [T]he holding of Stromberg do[es] not necessarily stand for anything more
   than the principle that, where a provision of the Constitution forbids
   conviction on a particular ground, the constitutional guarantee is violated
   by a general verdict that may have rested on that ground.

   Petitioner cites no case, and we are aware of none, in which we have set
   aside a general verdict because one of the possible bases of conviction was
   neither unconstitutional as in Stromberg, nor even illegal as in Yates, but
   merely unsupported by sufficient evidence. (172)

The Stromberg line of criminal cases, therefore, only applies where the invalid ground is constitutionally defective and it is possible that the jury rested its general verdict on that ground. Although the essence of the Baldwin principle is somewhat manifest in these circumstances, it is clear that the Stromberg line of cases is much narrower. (173) Thus, the Fifth Circuit's reliance on these cases in Neubauer is misplaced.

A final point of discussion is the choice-of-law question. I have argued above that 28 U.S.C. [section] 2111, read in light of the Court's precedents, incorporates the Baldwin principle. It is (or at least should be) the law of the federal appellate courts. Almost all of the states, however, have enacted "two issue" rules that essentially make the Baldwin scenario subject to a discretionary harmless error analysis. There is often a conflict between these two rules in appellate cases, and there is a split in resolving the conflict among the circuit courts. As noted in Parts III and IV, the Sixth Circuit has consistently applied a rule whereby a state "two issue" rule governs when a case is tried under state law and the Baldwin principle applies when the case is tried under federal law. On the other hand, the Fourth Circuit has come down squarely on the side of federal law, determining that appellate review of general verdicts is a matter of procedure not substance. (174) How should this conflict be resolved?

In Wilmington, the Court had suggested that state law might be applicable by distinguishing an Illinois statute on point. (175) Yet, it is hard to know what to make of this suggestion. The Wilmington decision came down in 1907, twenty-one years prior to the creation of the Erie doctrine and far removed from its modern development. We should be careful then not to place too much emphasis on Wilmington's discussion of state law. It may be a product of the era when the application of state and federal law was not as clearly demarcated.

The Erie doctrine is thus the proper place to begin the inquiry. As an initial matter, it should be pointed out that there is now no question that the Baldwin principle applies in cases based on federal law. Federal courts apply federal procedural law. On this point, the Sixth and Fourth Circuits seem to agree. (176) The real source of contention between these two circuits is the rule to be applied in diversity cases. In the early years, the conflict between federal and state law was a conflict between a state statute--such as the Illinois "two issue" statute--and a federal common law rule embodied in the Baldwin principle. However, with the enactment of 28 U.S.C. [section] 2111 in 1949, the conflict has changed to a conflict between a state statute and a federal statute with a Baldwin gloss. The difference is crucial because federal statutes bring the Supremacy Clause into play. In Stewart Organization, Inc. v. Ricoh Corp., (177) the Court held that a federal statute that is "sufficiently broad" to control the conflict should control by virtue of the Supremacy Clause. Section 2111 is in direct conflict with the state "two issue" rules. At some level, of course, the state "two issue" rules and 28 U.S.C. [section] 2111 are essentially similar efforts to proscribe harmless error analysis. But 28 U.S.C. [section] 2111 exempts errors that affect the "substantial rights" of the parties, and the Baldwin line of cases has interpreted this language where a general verdict encompassing multiple theories of liability is returned. The interpretation is based on the statute itself, and so does not mitigate the Supremacy Clause's force in displacing state law. The Sixth Circuit's rule, therefore, must be rejected. In addition, because Section 2111 is a procedural rule, the second prong of the Stewart test is satisfied:" `Erie and its offspring cast no doubt on the long-recognized power of Congress to prescribe housekeeping rules for federal courts.'" (178)

V. A POLICY-BASED DEFENSE OF THE BALDWIN PRINCIPLE

In the preceding section, I argued, as a matter of law, that the Baldwin principle remains alive and well and that courts invoking harmless error analysis flout standing Supreme Court precedent. In contrast, this section endeavors to mount a policy-based defense of the Baldwin principle.

As a preliminary matter, it is important to recognize that the federal courts have a longstanding policy of requiring unanimous jury verdicts. (179) Indeed, in Apodaca v. Oregon, (180) Justice White wrote that the requirement arose during the Middle Ages. (181) Moreover, in Johnson v. Louisiana, (182) Justice Powell elaborated on its common law pedigree:

   In an unbroken line of cases reaching back into the late 1800's, the
   Justices of this Court have recognized, virtually without dissent, that
   unanimity is one of the indispensable features of federal jury trial. In
   these cases, the Court has presumed that unanimous verdicts are essential
   in federal jury trials, not because unanimity is necessarily fundamental to
   the function performed by the jury, but because that result is mandated by
   history.... [T]he framers desired to preserve the jury safeguard as it was
   known to them at common law. At the time the Bill of Rights was adopted,
   unanimity had long been established as one of the attributes of a jury
   conviction at common law. (183)

It is thought that a unanimous verdict is desirable for several reasons, but most prominently because it is believed to encourage robust jury deliberation and ensure a more accurate outcome. (184) Although the Court has flirted with non-unanimity in criminal cases, (185) and legal scholars have occasionally argued for a non-unanimity rule, (186) the rule that jury verdicts must be unanimous in civil cases is still the prevailing federal law today. (187) The 1991 amendments to Rule 48 expressly provided that a verdict "shall be unanimous" unless the parties otherwise stipulate, (188) and modern court cases have asserted that a unanimity role is pregnant in the language of the Seventh Amendment. (189) It is therefore now clear that unanimity is required in federal civil cases and is probably constitutionally mandated.

The unanimity requirement is relevant to the debate regarding whether the Baldwin principle or the harmless error exception applies. The cases invoking a discretionary harmless error exception quixotically attempt to ascertain the basis of the jury's general verdict. If the court somehow believes that the jury based its verdict on one of the valid remaining theories of liability, it has shown some willingness to hold the error harmless. The problem with this approach, however, is that it often treats the jury as a monolithic body. It is not. It is a body consisting of several members, and if only one member based her decision on the invalid theory of liability, and the court persists in viewing the error as harmless, the general verdict could no longer be considered unanimous. It may have been harmless error for eleven out of twelve jurors. However, if one person based her decision on the invalid theory of liability, that means that the error is not truly harmless because the jury would not have returned the general verdict absent that one person. A court, applying a discretionary harmless error exception, would have thus allowed a general verdict that was not unanimous to stand--a result that contravenes the unanimity requirement and therefore is not harmless. Unless we are willing to tolerate an ex post abrogation of a strong federal policy, probably constitutionally mandated, (190) that has been around since the Middle Ages, this error could not be characterized as harmless. Once this is understood, it becomes clear that those courts professing adherence to a harmless error exception grossly overstate their ability to ascertain the basis for a jury's general verdict. A court would have to be confident that not even one juror based her decision on the invalid theory of liability. Apart from the narrow case of subsets discussed in Part IV, it is simply impossible to imagine a situation where an appellate court could have such confidence, and it is dangerous for it to even think it does. On what basis could it possibly make such a judgment? Could it be anything other than judicial telepathy?

Presumably, those circuit courts advocating harmless error analysis are attempting to mitigate the prospect of litigants getting off on technicalities. This was the impetus for the harmless error statute, and there are indications in a handful of opinions that this is indeed the case. The unanimity rule, however, has a built-in preference for false non-liability over false liability in civil cases just as it has a built-in preference for false acquittal over false conviction in criminal cases. (191) False liability is generally regarded as a horrific result, and a unanimous decision rule for juries endeavors to minimize its occurrence, even if doing so raises the incidence of false nonliability. As Rae argues, the federal court system exhibits a positional preference regarding outcomes, and the decision rule (unanimity) is thus weighted to reflect that positional preference. (192) Yet, if an appellate court were to affirm a general verdict where only a single juror based her decision on an invalid theory of liability, the positional preference for nonliability embodied in the unanimity rule would be turned on its head. The appellate courts would be favoring false liability over false non-liability. They would, in effect, be reversing the most basic principle of our legal system.

The dangers mentioned above are real, and an appellate court should not underestimate them. Let us suppose, however, that there is a legitimate concern with technicalities and that the legal system is willing (and properly able) to tolerate some level of error to compensate for it. If so, would an appellate court be justified in applying harmless error analysis? Again, I think the answer is no, at least not when there are less dangerous means of obtaining the same result. The discretionary harmless error exception is a blunt instrument, but alternative methods of achieving the same result, such as Judge Kozinski's waiver rule, (193) or a rule that makes the use of general verdicts in multi-claim scenarios reversible error, (194) could mitigate the potential danger associated with harmless error analysis while addressing the concern with technicalities. Both of these rules, I believe, would be excellent salves. In addition, the appellate courts routinely apply similar rules in a wide variety of contexts, so it is not as if they would be breaking any new ground by instituting such rules. Indeed, given the prevalence of per se abuse of discretion and waiver rules in modern appellate practice, it seems odd that the appellate courts have not embraced such rules already. Regardless, the point here is that these rules would not present the same dangers as a discretionary harmless error exception, and as a result, a more dangerous harmless error rule cannot be justified as a policy matter.

VI. CONCLUSION.

The Baldwin principle enjoyed a brief period of dominance in the early 1900s. In the last thirty years, however, the circuit courts have attempted to conserve on judicial resources, and as a result, they have endeavored to find a plausible reason for distinguishing the Baldwin line of cases. In this article, I suggest that the circuit courts have been acting lawlessly. The reasons they offer for distinguishing the Baldwin line of cases do not stand up upon close scrutiny. The Baldwin line of cases is an automatic rule requiring vacatur of the judgment, a remand of the case, and a new trial order. The Court has never backed away from this principle. The passage of 28 U.S.C. [section] 2111 did not change the analysis; instead, it simply folded the Wilmington decision within it. If a circuit court were to uphold a general verdict resting upon an invalid claim, the Wilmington decision suggests that it would affect the substantial rights of the parties per se. The Court's criminal rule is not to the contrary. It is based on a predecessor statute to 28 U.S.C. [section] 2111 that was designed primarily to curb abuses in the criminal appeals process. It not only does not apply, but it suggests that 28 U.S.C. [section] 2111 is aimed at the same type of abuse. Finally, after the Court's decision in Stewart Org., Inc. v. Ricoh Corp., (195) a circuit court may not rely on a state harmless error rule to escape the Baldwin line of cases, even in diversity cases, because 28 U.S.C. [section] 2111 incorporates the Baldwin principle, and the Supremacy Clause ensures that it trumps.

The life of the Baldwin principle has been highly controversial. The Supreme Court may eventually have to revisit it. Until it does so, I hope to have shown, as the cliche goes, that the reports of the Baldwin principle's death have been greatly exaggerated.

(1.) See Maryland v. Baldwin, 112 U.S. 490 (1884).

(2.) See Traver v. Meshriy, 627 F.2d 934 (9th Cir. 1980).

(3.) 112 U.S. 490 (1884).

(4.) A general verdict is one in which the jury finds for one or the other party, without making a specific finding of fact, such as on which theory of liability it is basing its verdict.

(5.) Baldwin, 112 U.S. at 494 ("This testimony was clearly inadmissible; it was mere hearsay.").

(6.) Id. at 493.

(7.) Id.

(8.) Id. at 495.

(9.) 205 U.S. 60 (1907).

(10.) Id. at 79. ("[W]here a jury is wrongfully permitted, over the objection of the opposing party, to take into consideration in reaching a verdict counts of a declaration which have not been supported by any evidence, and where it is impossible from the record to say upon which of the counts of the declaration the verdict was based, [it cannot be said] that the judgment entered under such circumstances can be sustained upon the theory that substantial rights of the objecting party had not been invaded.").

(11.) Id. at 78 (citing Baldwin and noting that "[u]nder this [scenario] we find it impossible to say that prejudicial error did not result"). It should be noted, however, that the Wilmington decision foreshadowed later developments in this area. First, it noted that the rule is different in the criminal context. Id. ("And, of course, in a case like the one we are considering we cannot maintain the verdict, as might be done in a criminal case upon a general verdict of guilty upon all the counts of an indictment."). Second, it specifically distinguished the applicability of Illinois's "two issue" rule. Id. at 78-79. Note: In a nutshell, the two-issue rule is simply that if a general verdict is rendered upon multiple theories of liability (issues), and at least one of those theories is supported by sufficient evidence, the general verdict will be allowed to stand.

(12.) 358 U.S. 613 (1959).

(13.) Id. at 614.

(14.) Id. at 617-618.

(15.) Id. at 618.

(16.) Id.

(17.) 370 U.S. 19 (1962).

(18.) Id. at 29-30. ("Since we hold erroneous one theory of liability upon which the general verdict may have rested [...] it is unnecessary for us to explore the legality of the other theories.") (citing Baldwin).

(19.) It is important to note that a wide range of errors is possible, ranging from evidentiary problems to insufficient evidence to bad law.

(20.) See e.g Patton v. Wells, 121 F. 337, 340 (8th Cir. 1903); Fireman's Fund Ins. Co. v. McGreevy, 118 F. 415, 419 (8th Cir. 1902) (citing Baldwin); Durant Min. Co. v. Percy Consol. Min. Co., 93 F. 166, 169-70 (8th Cir. 1899); Lyon, Potter & Co. v. First Nat. Bank of Sioux City, Iowa, 85 F. 120, 125 (8th Cir. 1898); Cresswell Ranch & Cattle Co. v. Martindale, 63 F. 84, 90 (8th Cir. 1894) (citing Baldwin); St. Louis, I.M. & S. Ry. Co. v. Needham, 63 F. 107, 114 (8th Cir. 1894) (citing Baldwin); What Cheer Coal Co. v. Johnson, 56 F. 810, 813 (8th Cir. 1893) (citing Baldwin).

(21.) See Lyle v. Bentley, 406 F.2d 325, 327-28 (5th Cir. 1969) ("If the jury's verdict was in fact based on [a proper claim of] undue influence, which it could have been, and if the evidence to sustain such a finding was insufficient, then the verdict must not be allowed to stand. Under such circumstances it matters not how much evidence there is in the record of [the valid] mental incapacity [claim]. Since the general verdict speaks affirmatively with respect to either or both of the issues submitted to the jury, we have no way of knowing upon what basis the jury's verdict was founded. Therefore, while the subjectives of the jury's deliberations are not for us to examine, if those deliberations as directed by the court's instructions permit a verdict to be based on an issue not supported by sufficient evidence, the jury verdict must be set aside and a new trial held."); E.L Cheeney Co. v. Gates, 346 F.2d 197, 200 & n. 4 (5th Cir. 1965); Vandercook & Son, Inc. v. Thorpe, 344 F.2d 930 (5th Cir. 1965) (same); Travelers Ins. Co. v. Wilkes, 76 F.2d 701 (5th Cir. 1935) (same); Am. Sugar Refining Co. v. J.E, Jones & Co., 293 F. 560, 562-63 (5th Cir. 1923) (same) (Hugo Black arguing for defendants in error). See also Mixon v. A. Coast Line R.R. Co., 370 F.2d 852, 860-62 (5th Cir. 1966) (Brown, J., concurring) (concurring to emphasize the "woeful and unfortunate loss of precious judicial time" resulting from the trial court's failure to use a specialized verdict). Chief Judge Brown later felt so strongly about the issue that he wrote a tract on it. See John R. Brown, Federal Special Verdicts: The Doubt Eliminator, 44 F.R.D. 245, 338 (1968).

(22.) See N.Y. & Porto Rico S.S. Co. v. Garcia, 16 F.2d 734, 737 (1st Cir. 1926) ("To add to the difficulties, the jury returned a general verdict, and, such being the case, it cannot be said whether the verdict was based upon the action of contract or the action for slander.") (citing Baldwin).

(23.) See Albergo v. Reading Co., 372 F.2d 83, 85-86 (3rd Cir. 1966) (In a Wilmington scenario, "since the evidence was insufficient to warrant submission of the first [FELA] claim to the jury, the judgment of the court below would be sustainable only if the verdict rested solely on the second [separate FELA] claim. The form of the verdict makes it impossible for us to determine whether it rested on the first claim, the second, or both. Where, as here, a general verdict may rest on either of two claims--one supported by the evidence and the other not--a judgment thereon must be reversed."). After the Third Circuit ordered a new trial, it pointed out that it would not have been necessary had the trial court employed either a specialized verdict or a general verdict accompanied by answers to specific interrogatories, both of which Rule 49 was designed to encourage. Id. at 86.

(24.) See A. Coast LineR. Co. v. Tiller, 142 F.2d 718, 722 (4th Cir. 1944) ("[S]ince the verdict was general, it is impossible to say whether it was based upon the [general negligence] issue that was properly submitted to the jury or upon the [special negligence arising out of the alleged violation of the Federal Boilers Inspection Act] issue that should have been withdrawn.") (citing Baldwin and Wilmington).

(25.) See Murphy v. Dyer, 409 F.2d 747, 748 (10th Cir. 1969) (citing Baldwin and Sunkist Growers).

(26.) See N. Am. Graphite Corp. v. Allan, 184 F.2d 387, 389 (D.C. Cir. 1950).

(27.) 350 F.2d 817 (5th Cir. 1965).

(28.) Id. at 823-24. See also Prudential Ins. Co. v. Morrow, 339 F.2d 411, 412-13 (5th Cir. 1964) (same).

(29.) 16 F.2d at 736-37.

(30.) In this case, Dolores Carmoega Garcia alleged that a steamship's refusal to transport her to New York constituted a breach of contract and that the steamship's pretextual justification, stated openly and in public, that she suffered from a contagious venereal disease constituted slander. The jury returned a general verdict for Garcia and assessed damages in the sum of $4,900, costs, and attorney's fees. On appeal, the First Circuit vacated the judgment. As a preliminary matter, the First Circuit noted that the "causes of action set out in the complaint are distinct and independent, and should not have been joined. They cannot be regarded as counts in a declaration, each stating the same cause of action, but with slight modifications, for they are not such." Id. at 736-37. In effect, the First Circuit held that it was error to even allow the jury to return a general verdict when there are multiple causes of action (as opposed to multiple theories of liability). Nevertheless, even putting this aside, the court held that, assuming there was an error in one theory of liability (as there was with respect to the slander cause of action), it could not have been corrected: "To add to the difficulties, the jury returned a general verdict, and, such being the case, it cannot be said whether the verdict was based upon the action of contract or the action for slander." Id. at 737 (citing Baldwin).

(31.) 72 F.2d 49 (1934).

(32.) That is, the scenario where there is insufficient evidence to sustain one of several theories of liability and a general verdict is returned.

(33.) Firemen's Ins. Co., 72 F.2d at 51 (collecting case law and summarizing decisions).

(34.) Wilmington, 205 U.S. at 79 ("[A] party cannot wait until after the close of the evidence at the trial, and a fortiori, after verdict, and then for the first time question the sufficiency of the counts.") (citing Chicago v. Lonergan, 63 N.E. 1018 (Ill. 1902) and Consol. Coal Co. v. Scheiber, 47 N.E. 1052 (Ill. 1897)).

(35.) Id. at 78-79 ("Nor does section 57 of the Illinois practice act ... support the contention that errors of the character of those we have just been considering must be treated as not prejudicial" because the jury was permitted to return a general verdict "over the objection of the opposing party").

(36.) 72 F.2d at 51-52 ("No such motion was made in the present case.... So there really was a different question in the Wilmington case from the one here presented.... The Supreme Court of the United States, in the Wilmington case, was not attempting to lay down a new rule to supersede the well-established rule in Illinois ... but it held that in that case the failure of the court to allow a defendant's motion before verdict to strike from the record certain counts, to establish which no proof had been offered, was prejudicial error. Had a similar motion been made in this case ... and no evidence offered to sustain it, and an adverse ruling upon the motion, then a [different] situation would have been presented.").

(37.) 124 F.2d 883 (1941).

(38.) Id. at 887 ("We have held each count good, and therefore if there is substantial evidence to sustain any one count in favor of each plaintiff, the general verdict must be upheld.").

(39.) It is arguably less troublesome than one might initially think considering that the Supreme Court denied certiorari in the case. See Ryan v. Cross, 316 U.S. 682 (1942). Of course, the refusal of the Supreme Court to grant certiorari does not necessarily mean that it approved of the Cross decision. Indeed, it is wise not to read too much into denials of certiorari.

(40.) See Rashaw v. C. Vt. Ry., Inc., 133 F.2d 253, 256 (2d Cir. 1943); Christian v. Boston & M.R.R., 109 F.2d 103, 105 (2d Cir. 1940) (citing Wilmington).

(41.) 192 F.2d 226 (2d Cir. 1951). This case contains the most thorough analysis of the issue. For other Second Circuit cases during this era invoking the distinction, see Ortiz v. Grace Line, Inc., 250 F.2d 124 (2d Cir. 1957) and Vareltzis v. Luckenbach Steamship Co., 258 F.2d 78, 80 (2d Cir. 1958).

(42.) Lee, 192 F.2d at 229 (citations omitted).

(43.) Id. (citing Cross mistakenly). The Second Circuit repeated this mistake in Vareltzis. See 258 F.2d at 80.

(44.) 275 F.2d 188 (2d Cir. 1960).

(45.) Id. at 190-91 (citing D.C. and Fifth Circuit cases adhering to the Baldwin principle and neglecting to mention any of the previous Second Circuit cases applying a "two issue" rule).

(46.) 117 F.2d 702 (6th Cir. 1941).

(47.) 304 U.S. 64 (1938).

(48.) McCrate, 117 F.2d at 704-05 (citing Erie and noting that Ohio's "two issue" rule did not apply). See also Louisville & Nashville R.R. Co. v. Rochelle, 252 F.2d 730 (6th Cir. 1958) (Tennessee's two-issue rule applied); A. Coastline R.R. Co. v. Smith, 264 F.2d 428 (6th Cir. 1959) (same).

(49.) 308 F.2d 383 (6th Cir. 1962).

(50.) Id. at 390 ("This case is tried under federal law where the `two issue' rule is not applicable.") (citing Wilmington).

(51.) In Elder-Beerman Stores Corp. v. Federated Department Stores, Inc., 459 F.2d 138 (6th Cir. 1972), the Sixth Circuit, as in Valasco, also upheld the applicability of the Baldwin principle in federal antitrust cases. Id. at 147-148 (noting that the conclusion is in accord with Sunkist Growers). In other cases based on diversity jurisdiction and state law, the court found the state "two issue" rule to govern. See Keet v. Serv. Mach. Co., Inc., 472 F.2d 138, 140 (6th Cir. 1972) (applying Ohio's "two issue" rule); Adkins v. Ford Motor Co., 446 F.2d 1105, 1108 (6th Cir. 1971) (applying Tennessee's "two issue" rule and stating that "[w]here jurisdiction is founded upon diversity of citizenship and state-created rights are sought to be enforced, this Court has consistently construed a general verdict as the state courts would.").

(52.) 627 F.2d 934 (9th Cir. 1980).

(53.) Id. at 938.

(54.) Id. (emphasis added).

(55.) See supra n. 39 and accompanying text.

(56.) Traver, 627 F.2d at 936.

(57.) In subsequent cases, the Ninth Circuit has found a strong likelihood of jury confusion where the case is complex, such as an antitrust case, or where the factual and legal issues are intermingled. See infra n. 60.

(58.) Traver, 627 F.2d at 938-39. For an excellent overview of how these factors have played out in the case law, see Hendry v. Exide Elecs. Corp., No. C-89-0077 SAW, 1992 WL 217785, at *2-5 (9th Cir. 1992).

(59.) Traver, 627 F.2d at 939.

(60.) Ironically, in other cases, the court's discretionary inquiry has often turned on this factor. The court has found a strong likelihood of confusion where the case is complex, such as an antitrust case, where the legal issue is unsettled, or where the factual and legal issues are intermingled. See Syufy Enters. v. Am. Multicinema, Inc., 793 F.2d 990, 10011003 (9th Cir. 1986) (refusing to exercise discretion under first and third factors where "the jury was confronted with a complex case involving multiple antitrust theories" and "each claim also involved distinctive legal and factual components"); Malone v. Cal. St. College, Stanislaus, No. CV-84-00601-EDP, 1993 WL 51303, at **3-6 (9th Cir. 1993) (refusing to exercise discretion to uphold verdict, after disparate impact claim rejected on appeal, under all four factors, but most prominently because "the potential for confusion of the jury was substantial due to the erroneous submission of the retaliatory discharge claim to the jury without any admonition regarding the elements of this theory of liability"); Hesse v. Air France, 1990 WL 58237, at **2-4 (9th Cir. 1990) (refusing to exercise discretion due to "the intermingled nature of [plaintiff's] claims, [and] the confusion in the law [of wrongful discharge] at the time of trial"); Kern v. Levolor Lorentzen, Inc., 899 F.2d 772, 791-92 (9th Cir. 1990) (Kozinski, J., dissenting) (arguing that interrelation of age discrimination and breach of contract claims confused jury). It should be noted, however, that the court has occasionally found that the jury is unlikely to be confused. See Knapp v. Ernst & Whinney, 90 F.3d 1431, 1440 (9th Cir. 1996) (noting that separation of issues rendered jury unlikely to be confused). In addition, on at least one occasion, the court emphasized that there is little likelihood of jury confusion where the appellant's lawyer stressed the valid claim during the trial's closing arguments. See Kern, 899 F.2d at 777 (exercising discretion where no possibility of jury confusion existed because appellee's counsel stressed proper claim during closing argument and where "[a]ge discrimination was offered not as an independent ground for recovery, but only as a possible explanation for ... disparate treatment").

(61.) Traver, 627 F.2d at 939. In subsequent cases, the Ninth Circuit has emphasized that the defenses between the claims should be nearly identical. Otherwise, it has reasoned that the jury could have found for the appellee on any one claim without necessarily being compelled to reject a valid defense that was applicable only to that separate claim. See Knapp, 90 F.3d at 1440 (defenses applied equally to both counts); Kern, 899 F.2d at 777 (same). But see Oglesby v. S.P. Trans. Co., 6 F.3d 603, 610 (9th Cir. 1993) (refusing to exercise discretion under second and fourth factors where defense of comparative negligence is available under negligence claim but not under Boiler Inspection Act claim); Hendry, 1992 WL 217785 at **4 (refusing to exercise discretion where there were different defenses to each of five theories of liability).

(62.) Traver, 627 F.2d at 939. The Ninth Circuit has really provided little meat on the bones of this factor. It is inherently a subjective assessment of the merits of the appellant's case. It is important to note, however, that the Ninth Circuit does require more than marginal evidence. See Syufy, 793 F.2d at 1001-1003 (refusing to exercise discretion under first and third factors where the evidence of the surviving theories was "marginal").

(63.) Traver, 627 F.2d at 939 (citing California cases). In many subsequent cases, the factual similarities analysis is subsumed within the first factor, but occasionally it is analyzed separately. The court seems to require virtually identical disputed issues of fact; however, on occasion, the court may exercise discretion where the facts and evidence regarding the valid claim are a subset of the facts and evidence regarding the improper claim. See Lancer Ins. Co. v. D.W. Ferguson & Assoc., No. CIV 91-3314 ER, 1995 WL 21705, **3 (9th Cir. 1995) (refusing to exercise discretion on ground that "disputed issues of fact were different as between the contract and tort theories" and only one of five theories of recovery survived appellate review); Counts v. Burlington N.R.R. Co., 952 F.2d 1136, 1140 (9th Cir. 1991) (refusing to exercise discretion where facts used for several theories, such as lack of consideration, mutual mistake, economic duress, fraud in inducement, "were not largely identical"). But see Knapp, 90 F.3d at 1441-42 (exercising discretion where facts regarding deceptive statement and scienter were elements of both primary and aiding and abetting liability under SEC Rule 10b-5); Kern, 889 F.2d at 778 (exercising discretion where facts were "largely identical," except that invalid claim required additional evidence).

(64.) Traver, 627 F.2d at 938-39.

(65.) Hendry, 1992 WL 217785 at **3 (analyzing three factors). See also Kelly v. City of Oakland, 198 F.3d 779, 785 (9th Cir. 1999) (exercising discretion upon analysis of all four factors); Knapp, 90 F.3d at 1439-42 (affirming district court's invocation of Traver discretion based on analysis of all four factors to sustain a general verdict against a subsequent motion for a new trial where one of the claims was mooted by a Supreme Court decision); Kern, 899 F.2d at 777-79 (exercising discretion upon analysis of all four factors); Nutri-metics Intl., Inc. v. Carrington Labs, Inc., No. CV-87-0935-RMT, 1992 WL 389246 (9th Cir. 1992) (exercising discretion upon informal analysis of last three factors); Benigni v. City of Hemet, 879 F.2d 473, 480 (9th Cir. 1988) (exercising discretion on the basis of two factors). But see Malone, 1993 WL 51303 at **3-6 (refusing to exercise discretion on poor showing of all four factors); Portland Feminist Women's Health Ctr. v. Advocates for Life, Inc., 62 F.3d 280, 285-86 (9th Cir. 1994) (applying Traver rule where one claim was rendered moot by a Supreme Court decision, but refusing to exercise discretion based on third and fourth factors); Syufy, 793 F.2d at 1001-1003 (refusing to exercise discretion on poor showing of only two factors); Oglesby, 6 F.3d at 610 (two factors).

(66.) See Roberts v. College of the Desert, 870 F.2d 1411 (9th Cir. 1988) (applying Traver rule with no discussion of factors to affirm general verdict where it could be attributed to either due process or equal protection claim).

(67.) See e.g. Lancer Ins., 1995 WL 21705 at **3 (refusing to exercise discretion on the basis of poor showing of one factor); Hesse, 1990 WL 58237 at **2-3 (same); Counts, 952 F.2d at 1140-41 (same).

(68.) Kern, 899 F.2d at 790-792 (Kozinski, J., dissenting). See also Lucas v. J.C. Penney Co., Inc., No. CV-93-01804-JCC, 1997 WL 730328 **2 (9th Cir. 1997). Ironically, Judge Kozinski did not always feel so strongly about the precedential force of the Baldwin principle. Indeed, in 1989, just a year before his Kern dissent, Judge Kozinski invoked the Traver rule to uphold a general verdict where factual evidence was sufficient to support a medical negligence verdict, despite the inadequacy of four of the eight factual theories supporting that claim. See McCord v. Maguire, 873 F.2d 1271, 1273-74 (9th Cir. 1989) (citing Landes, Syufy, and Traver).

(69.) Kern, 899 F.2d at 791.

(70.) Id.

(71.) Id.

(72.) It would be interesting to see what Judge Kennedy would do if the Traver rule ever got before the Supreme Court because he would be in the odd position of potentially having to admit that the Traver rule--a rule he crafted--flouted the precedents of the Court on which he now sits.

(73.) 418 F.2d 180 (5th Cir. 1969).

(74.) For Sixth Circuit cases, see supra nn. 48-51 and accompanying text; for Seventh Circuit cases, see supra nn. 31-39 and accompanying text.

(75.) Am. Airlines, 418 F.2d at 195.

(76.) Id.

(77.) Id. The Fifth Circuit also cited to its earlier decision in Lyle v. Bentley, 406 F.2d 325, 327-28 (5th Cir. 1969), but apparently did not find it troubling that Lyle had actually gone the other way. It had applied the general rule using language that suggested that it was mandatory.

(78.) Am. Airlines, 418 F.2d at 195 (emphasis added).

(79.) Wilmington, 205 U.S. at 79 (stating that "and where it is impossible from the record to say upon which of the counts of the declaration the verdict was based ..") (emphasis added).

(80.) See Gardner v. Gen. Motors Corp., 507 F.2d 525, 529 (10th Cir. 1974) ("Although we might well hesitate to affirm the judgments if duty to warn were the single theory of liability we consider plaintiff's other theories of recovery, none of which depend on jury conclusions regarding duty to warn, to overwhelm any potential error in this regard."); Collum v. Butler, 421 F.2d 1257, 1260 (7th Cir. 1970); Roginsky v. Richardson-Merrell, Inc., 378 F.2d 832 (2d Cir. 1967).

(81.) 544 F.2d 19 (2d Cir. 1976).

(82.) Id. at 26-27.

(83.) Id.

(84.) Id. ("The jury was thus repeatedly invited to find for the plaintiff under either section of the Landrum-Griffin Act. While it may be tempting to say that the jury could not really have thought that Morrissey was entitled to a heating before the police were called, they had been told they could base their verdict on just that. As a practical matter, judging from the large amount of punitive damages awarded, we would guess that the jury sustained both theories of recovery.").

(85.) 573 F.2d 1029 (8th Cir. 1978).

(86.) Id. at 1038.

(87.) Id. at 1038-39 (quoting Morrissey, 544 F.2d at 26-27).

(88.) Id. at 1039 (emphasis added).

(89.) There have only been four cases outside of the Ninth Circuit that have exercised a harmless error exception to the Baldwin line of cases. See Hurley v. Atlantic City Police Dept., 174 F.3d 95, 120-22 (3d Cir. 1999) (finding harmless error to lie in a sexual harassment case where a general verdict encompassing an improper quid pro quo theory of liability was returned); Bruneau v. South Kortright C. Sch. Dist., 163 F.3d 749, 759-61 (2d Cir. 1998) (finding harmless error to lie in a Title IX case); Henderson v. Winston, 59 F.3d 166 (4th Cir. 1995) (per curiam) (finding harmless error to lie in a [section] 1983 action where one element of a denial of due process theory of liability was improper); Braun v. Flynt, 731 F.2d 1205, 1206 (5th Cir. 1984) (finding harmless error to lie in an invasion of privacy case where general verdict encompassed an invalid appropriation theory and a valid "false light" theory because the focus of the case had been on the false light claim).

(90.) See E.I. DuPont de Nemours & Co. v. Berkeley & Co., 620 F.2d 1247, 1257-58 (8th Cir. 1980); Bone v. Refco, Inc., 774 F.2d 235, 242-44 (8th Cir. 1985); Dudley v. Dittmer, 795 F.2d 669, 673-75 (8th Cir. 1986).

(91.) As noted above, the Second Circuit initially oscillated between adherence to the "two issue" rule and the Baldwin principle in the early years. After Morrissey, however, things evened out a bit during the 1980s and early 1990s.

(92.) 163 F.3d 749 (2d Cir. 1998). In this case, the Second Circuit broke from the past two decades by invoking and applying a harmless error exception. Id. at 759-61. Judge Cardamone's opinion noted that Morrissey "specifically recognized the possibility of applying harmless error analysis where appropriate to avoid the usual course of reversal and new trial," and as a result, he felt that this was the appropriate case in which to do so. Id. at 759 (citing Morrissey).

(93.) See Yentsch v. Texaco, Inc., 630 F.2d 46, 59 (2d Cir. 1980) (applying Baldwin principle); Katara v. D.E. Jones Commodities, Inc., 835 F.2d 966, 971 (2d Cir. 1987) (same); O'Neill v. Krzeminski, 839 F.2d 9, 12 (2d Cir. 1988) (same but extending it to a jury answer to a single interrogatory that comprehends two distinct bases of liability); BAII Banking Corp. v. UPG, Inc., 985 F.2d 685, 704 (2d Cir. 1993).

(94.) See Yentsch, 630 F.2d at 59 ("Yenstch urges that we can nevertheless affirm the judgment below if we are `fairly convinced that the jury proceeded only on the sound ground.' We have no confidence, however, that the jury based its decision exclusively on the price-fixing claim, the only one sufficiently supported by the evidence to warrant submission.").

(95.) See Jones v. Miles, 656 F.2d 103, 106-08 (5th Cir. 1981); Kicklighter v. Nails by Jannee, Inc., 616 F.2d 734, 742 (5th Cir. 1980); King v. Ford Motor Co., 597 F.2d 436, 439 (5th Cir. 1979) (distinguishing between a harmless error rule for reviewing JNOV motions and the Baldwin principle for reviewing motions for a new trial and general appellate review); Dougherty v. Continental Oil Co., 579 F.2d 954, 960 & n. 2 (5th Cir. 1978); Smith v. S. Airways, Inc., 556 F.2d 1347, 1347-48 (5th Cir. 1977); Yoder Brothers, Inc. v. Cal.-Fla. Plant Corp., 537 F.2d 1347 n. 48 (5th Cir. 1976) (Brown, C.J., dissenting). In addition, it should be noted that the Eleventh Circuit adopted all of the Fifth Circuit's case law prior to Oct. 1, 1981, as binding precedent. See Bonner v. City of Prichard, 661 F.2d 1206, 1209 (11th Cir. 1981) (en banc). There is in consequence a certain amount of path dependence in its case law, and many of its opinions rely heavily on the King decision. See Carroll Kenworth Truck Sales, Inc. v. Kenworth Truck Co., 781 F.2d 1520, 1529 (11th Cir. 1986); Royal Typewriter Co. v. Xerographic Supplies Corp., 719 F.2d 1092, 1098-99 (11th Cir. 1983); Hunter v. Reardon Smith Lines, Ltd., 719 F.2d 1108, 1113 & n.15 (11th Cir. 1983) (applying Baldwin principle, but noting in a footnote that if assuming arguendo the harmless error exception applied, reversal was nonetheless required); Walden v. U.S. Steel Corp., 759 F.2d 834, 838 (11th Cir. 1985); Michigan Abrasive Co., v. Poole, 805 F.2d 1001, 1005 (11th Cir. 1986) (relying on mistaken reading of Walden as relying on state law); Maccabees Mut. Lif. Ins. Co. v. Morton, 941 F.2d 1181, 1184 (11th Cir. 1991); Cronin v. Washington Natl. Ins. Co., 980 F.2d 663, 669 n .7 (11th Cir. 1993); Richards v. Michelin Tire Corp., 21 F.3d 1048, 1055-56 (11th Cir. 1994) (relying on King's (as cited in Carroll Kenworth and Cronin) distinction between the rule applicable to JNOV and new trial motions). But see O'Donnell v. Ga. Osteopathic Hosp., Inc., 748 F.2d 1543, 1549 (11th Cir. 1984) (applying "two issue" rule and failing to distinguish between JNOV and new trial motions). Oddly, the Eleventh Circuit seems to place much more faith in the soundness of the King distinction than does the Fifth Circuit itself, or any other circuit for that matter. It appears to be the only circuit to attach any significance to the JNOV/new-trial motion distinction.

(96.) Braun, 731 F.2d at 1206.

(97.) The cases not mentioning the possibility of harmless error and applying the Baldwin principle include Ratner v. Sioux Natural Gas Corp., 770 F.2d 512, 518 (5th Cir. 1985); Neubauer v. City of McAllen, 766 F.2d 1567, 1575-78 (5th Cir. 1985); Nowell v. Universal Electric Co., 792 F.2d 1310, 1312 (5th Cir. 1986); Pan Eastern Exploration Co. v. Hufo Oils, 855 F.2d 1106, 1123-25 (5th Cir. 1988) (noting general rule, but holding that appellant had waived the right by arguing only that none of the possible theories of recovery were supported by the evidence, not that the cause should be reversed and remanded if any one of the theories was invalid) and Crist v. Dickson Welding, Inc., 957 F.2d 1281, 1286 (5th Cir. 1992). The cases briefly mentioning the possibility of harmless error but adhering to the Baldwin principle include Olney Savings & Loan Association v. Trinity Banc Savings Association, 885 F.2d 266, 271 (5th Cir. 1989) (recognizing Baldwin principle and Braun's harmless error exception but not applying because there was sufficient evidence to support all the claims) and Ward v. Freeman, 854 F.2d 780, 790 (5th Cir. 1988) (noting both but applying the Baldwin principle).

(98.) See Collis v. Ashland Oil & Refining Co., 722 F.2d 625, 627 (10th Cir. 1983) (applying general rule without discussion of harmless error exception); Asbill v. Hous. Auth. of Choctaw Nation of Ok., 726 F.2d 1499, 1504 (10th Cir. 1984) (applying general rule although discussing and justifying harmless error exception); Smith v. FMC Corp., 754 F.2d 873, 877 (10th Cir. 1985) (applying general rule without discussion of harmless error exception); McMurray v. Deere & Co., 858 F.2d 1436, 1445-46 (10th Cir. 1988) (same). In Farrell v. Klein Tools, Inc., 866 F.2d 1294, 1298-1300 ( 10th Cir. 1989), the court discussed the issue at length and acknowledged that the "law on this subject is not characterized by any great clarity." Id. at 1299. After canvassing the cases supporting both the Baldwin principle and the harmless error exception, the court stated that it felt itself bound by its two most recent opinions in Smith and McMurrary, "both of which strictly imposed the general rule." Id. at 1298-1300. Thus, despite the fact that it appeared "very unlikely that the submission of the instruction on [an invalid defense] significantly influenced the jury or prejudiced Farrell's substantial rights," the court "reluctantly" applied the Baldwin principle. Id. at 1300-01.

(99.) Braun was the exception, having been decided in 1984 by the Fifth Circuit. In the early 1980s, the First Circuit, despite proclaiming allegiance to the Baldwin principle, articulated a subset theory of the harmless error exception. In two cases, Brochu v. Ortho Pharmaceutical Corp., 642 F.2d 652 (1st Cir. 1981) and Shepp v. Uehlinger, 775 F.2d 452, 456-57 (1st Cir. 1985), the First Circuit applied an extraordinarily narrow harmless error exception where the elements of the valid theory of liability were a subset of the invalid theory of liability minus the invalid element. I do not, however, consider these applications of the harmless error exception to be of the same genus as those mentioned in the text, largely because they seem to be the result of a fluke. Indeed, not even Judge Kozinski views the First Circuit's subset theory as in the same genus as the discretionary harmless error rules. See Kern, 899 F.2d at 790 (Kozinski, J., dissenting) ("Brochu involved a very rare situation and, in the few cases where it applies, the Brochu rule is fully consistent with Baldwin."). Apart from Brochu and Shepp, the First Circuit has steadfastly adhered to the Baldwin principle. See Levinsky's, Inc. v. Wal-Mart Stores, Inc., 127 F.3d 122, 134 (1st Cir. 1997); Lattimore v. Polaroid Corp., 99 F.3d 456, 468 (1st Cir. 1996); Kassel v. Gannet Co., 875 F.2d 935, 950 (1st Cir. 1989).

(100.) See Ely v. Blevins, 706 F.2d 479, 480 (4th Cir. 1983); Gill v. Rollins Protection Serv. Co., 722 F.2d 55, 59 (4th Cir. 1983); Flowers v. Tandy Corp., 773 F.2d 585, 591 (4th Cir. 1985).

(101.) 59 F.3d 166 (4th Cir. 1995) (per curiam).

(102.) Id. at **5.

(103.) Id.

(104.) The Third Circuit had steadfastly maintained that its Albergo decision in the early years, supported by the Supreme Court's Halecki decision, counseled in favor of the Baldwin principle. See Simko v. C & C Maint. Co., 594 F.2d 960, 967-68 (3d Cir. 1979); Avins v. White, 627 F.2d 637, 646 (3d Cir. 1980); Carden v. Westinghouse Elec. Corp., 850 F.2d 996, 1000-1003 (3d Cir. 1988); Limbach Co. v. Sheet Metal Workers Intl. Assn., 949 F.2d 1211, 1217-18 (3d Cir. 1991); Rush v. Scott Specialty Gases, Inc., 113 F.3d 476, 485 (3d Cir. 1997); Wilburn v. Maritrans GP, Inc., 139 F.3d 350, 361 (3d Cir. 1998); Brokerage Concepts, Inc. v. U.S. Healthcare, Inc., 140 F.3d 494, 533-35 (3d Cir. 1998).

(105.) 174 F.3d 95 (3d Cir. 1999).

(106.) Id. at 120-22.

(107.) Id. at 135-37 (citations omitted).

(108.) Id. at 137-38 (citations omitted). In addition, Judge Cowen noted that: "[t]he invocation of a harmless error rule in this case is particularly misguided because it will produce no benefit to either the judicial system or the parties. As the Supreme Court has recognized [in other contexts], the chief justification of the harmless error doctrine is the conservation of judicial resources, But in this case, as a consequence of our decision to vacate the punitive damages award against the ACPD and the liability judgments against Madamba and in favor of Rifice, the entire matter will have to be re-presented to a new jury. There is thus no efficiency advantage to support the majority's approach." Id. at 138 (citations omitted).

(109.) In a 1996 article, David Axelrad and Loren Kraus had argued that, apart from the First Circuit's subset theory decision, "none of the other courts that have referred to a possible harmless error exception has found it applicable." See David M. Axelrad & Loren Homer Kraus, The Federal General Verdict Rule: Conflict in the Courts of Appeal, 43 Fed. Law. 43, 43 (June 1996). As illustrated, this is no longer the case.

(110.) In the Fifth Circuit, the standard is reasonable certainty. See Braun, 731 F.2d at 1206 ("reasonably certain"); Olney, 885 F.2d at 271 ("reasonably certain"); Ward, 854 F.2d at 790 ("reasonably certain"). The Fourth Circuit concurs in this standard. Henderson, 1995 WL 378602 at **5 ("reasonably certain") (citing Braun).

(111.) The Second Circuit employs a sufficient confidence standard. See Bruneau, 163 F.3d at 759-61 ("sufficient confidence").

(112.) See Asbill, 726 F.2d at 1504

(113.) See Farrell, 866 F.2d at 1301 ("However, because we cannot say with absolute certainty, as required by McMurray and Smith, that the jury was not influenced by the submission of the abnormal use instruction, we must reverse and remand for a new trial."); Anixter, 77 F.3d at 1229 ("[W]e have adhered strictly to the general rule and have remanded cases where we could not say `with absolute certainty' that the jury was not influenced by the submission of the improper or erroneous instruction.").

(114.) See the discussion in Hurley, 174 F.3d at 137 n. 4 (Cowen, J., dissenting).

(115.) 726 F.2d 1499 (10th Cir. 1984).

(116.) Hurley, 174 F.3d at 121-22.

(117.) Asbill, 726 F.2d at 1504.

(118.) Id. (citing Fed. R. Civ. P. 61; 28 U.S.C. [section] 2111 (1976)).

(119.) 328 U.S. 750, 764 (1946).

(120.) Bruneau, 163 F.3d at 760.

(121.) See Wilmington, 205 U.S. at 78 ("And, of course, in a case like the one we are considering we cannot maintain the verdict, as might be done in a criminal case upon a general verdict of guilty upon all the counts of an indictment.") (emphasis added).

(122.) 283 U.S. 359 (1931).

(123.) 462 U.S. 862 (1983).

(124.) Neubauer, 766 F.2d at 1575-76.

(125.) 642 F.2d 652 (1st Cir. 1981).

(126.) Id. at 662.

(127.) 775 F.2d 452, 456-57 (1st Cir. 1985) ("To have found liability under the indemnification theory, the jury, as instructed, would have had to find liability under the `breach of contract' theory; thus, even if we assume some error of presentation as a matter of New York indemnification law (and if we assume a proper objection), any error of the sort here claimed was harmless.").

(128.) It requires that the elements of the valid theory of liability be a subset of the invalid theory of liability, rather than requiring just a similarity between the evidence needed to establish the elements of two different theories of liability.

(129.) Indeed, the Ninth Circuit itself seems to have folded the First Circuit's subset theory within the framework of the Traver rule's factual similarities factor. Consider Bender v. Wholesale & Retail Food Dist., Teamsters Local 63, No. CV-86-1072-JSL, 1991 WL 276452, at **2 (9th Cir. 1991) (affirming general verdict where "not only was the evidence against Local 63 sufficient to support an award for Brown on his state law claims, but these causes of action are also subsets of liability under the [faulty Labor Management Reporting & Disclosure Act] claim"); Kern, 899 F.2d at 778 (exercising discretion under Traver rule where proper claim of breach of contract was possible subset of age discrimination claim, which required additional evidence as to motive); Traver, 627 F.2d at 939 (exercising discretion where [section] 1983 claims are "all but derivative" of the alleged state torts and their corresponding facts and a "jury finding on a section 1983 claim would necessarily encompass a finding of liability on one or more of the state law claims").

(130.) See Part II, supra.

(131.) See Lawndale Natl. Bank v. Am. Cas. Co., 489 F.2d 1384, 1388-89 (7th Cir. 1973) (applying the Baldwin rule to insurance dispute where "it is impossible to determine from the general verdict upon which defense--arson or misrepresentation--the jury relied in making its general verdict") (citing Baldwin).

(132.) 651 F.2d 458 (7th Cir. 1981).

(133.) Id. at 464 ("The case was submitted to the jury on three theories of liability--breach of contract, common law fraud, and violation of the securities laws--but the verdict in favor of plaintiff did not specify under which theory or theories liability for compensatory damages was imposed. Accordingly, to triumph on appeal, defendants must show that under none of the three rationales was plaintiff entitled to the award of compensatory and punitive damages.").

(134.) Id. at 472 ("Even if we assume arguendo that the instruction relating to the securities violation was erroneous it cannot be shown, other than on the basis of speculation or conjecture, to have affected the jury's decision, there being a totally adequate independent theory upon which the verdict may have rested.").

(135.) See Kern, 899 F.2d at 790-91 (Kozinski, J., dissenting).

(136.) The Sixth Circuit has consistently applied the Baldwin principle in cases arising under federal law. See Arthur Langenderfer, Inc. v. S.E. Johnson Co.,729 F.2d 1050, 1058 (6th Cir. 1982); Doherty v. Am. Motors Corp., 728 F.2d 334, 344 (6th Cir. 1984); Brandenburg v. Cureton, 882 F.2d 211,214 (6th Cir. 1989); Virtual Maint., Inc. v. Prime Computer, Inc., 957 F.2d 1318, 1326 (6th Cir. 1992); Virtual Maint., Inc. v. Prime Computer, Inc., 11 F.3d 660, 667 (6th Cir. 1993).

(137.) Compare Royal Typewriter, 719 F.2d at 1098-99 ("In diversity cases in this circuit, an appellate court applies the federal, rather than the state, standard in reviewing challenges to a general verdict"), with Michigan Abrasive Co., 805 F.2d at 1005 ("Since the jury returned a general verdict, Alabama law requires that this court affirm both of the theories presented to the jury in order to sustain its verdict.").

(138.) See Kern, 899 F.2d at 777 n. 2 (9th Cir. 1990) ("[Judge Kozinski's] dissent attacks the Traver decision as contrary to century-old Supreme Court authority. We disagree, but even if true it is not for this panel to decide. The dissent would have us overturn not only decisions of the Supreme Court of California but of our own court as well. We decline the invitation.") (emphasis added). See also Niles v. U.S., 520 F. Supp. 808, 811-12 (N.D. Cal. 1981), aff'd, 710 F.2d 1391 (9th Cir. 1983) (treating the evaluation and review of general verdicts as a question of state law in diversity cases).

(139.) In the Fourth Circuit, see Barber v. Whirlpool Corp., 34 F.3d 1268, 1278 (4th Cir. 1994) (refusing to apply state "two issue" rule and accompanying waiver provision because "federal law controls verdicts") and Flowers v. Tandy Corp., 773 F.2d 585, 591 (4th Cir. 1985) ("Without regard to what state law may provide in such circumstances, a question we do not decide, federal procedural law controls on this point."). The Fifth Circuit cases include King. See 597 F.2d at 439 C In diversity cases in this circuit, a district court applies the federal, rather than the state, standard for determining whether a party's evidence is sufficient to defeat a motion for a directed verdict or judgment n.o.v."). It is also worth pointing out that the Federal Circuit has a distinction similar to the Erie doctrine whereby it applies the law of the originating circuit to procedural issues, and in Embrex, Inc. v. Service Engineering Corp, 216 F.3d 1343 (Fed. Cir. 2000), it held that "[p]reservation of appeal rights is a procedural issue." Id. at 1350.

(140.) Some courts seem to go even further to suggest that the use of a general verdict, despite Rule 49(a)'s discretionary nature, may be per se error in certain situations. See Nowell, 792 F.2d at 1312 n. 1 (suggesting that "the use of a general verdict to probe disparate allegations of negligence and contributory negligence, if properly objected to, would in all likelihood meet with our decisive disapproval"). But see Mueller, 573 F.2d at 1038 n. 13 (8th Cir. 1978) ("We do not hold that the trial court's refusal to submit the special verdict here was an abuse of discretion, much less that such refusal constituted reversible error. But we do suggest that cases such as this, where multiple theories of liability are asserted, are the ones most suited to the use of special verdicts, because special verdicts will often obviate the necessity of deciding difficult legal questions which are not essential to an appropriate disposition of the controversy.").

(141.) See Brown, supra n. 21, at 338-39.

(142.) 873 F.2d 1271 (9th Cir. 1989).

(143.) Id. at 1274.

(144.) See e.g. Anixter, 77 F. 3d at 1231-32; Henderson, 1995 WL 378602 at **5; Pan E., 855 F.2d at 1124; O'Neill, 839 F.2d at 12. But see Richards v. Michelin Tire Co., 21 F.3d 1048, 1056 n. 15 (11th Cir. 1994) ("Richards contends that because Appellant did not object to the special verdict forms, it waived its right to appeal. Alternatively, he argues that the failure to object authorizes this Court to sustain the verdict based on either claim. We reject these assertions. The failure to object to a proposed verdict format does not waive the right to object to the sufficiency of the evidence.... [I]t was the court's responsibility to ensure that the interrogatories were not deficient; i.e., to guarantee that they differentiated between the design and warning claims.") (citations omitted).

(145.) I am aware of only one other extended article on this subject, see Elizabeth Cain Moore, Note, General Verdicts in Multi-Claim Litigation, 21 Mem. St. U. L. Rev. 705 (1991), but it does not take a position on the merits of the arguments advanced on each side.

(146.) Sunkist Growers, 370 U.S. at 30.

(147.) Id. at 29-30 ("Since we hold erroneous one theory of liability upon which the general verdict may have rested ... it is unnecessary for us to explore the legality of the other theories") (citing Baldwin).

(148.) Asbill, 726 F.2d at 1504.

(149.) Am. Airlines, 418 F.2d at 195.

(150.) Wilmington, 205 U.S. at 79 (emphasis added).

(151.) Halecki, 358 U.S. at 619 ("for there is no way to know that the invalid claim of unseaworthiness was not the sole basis for the verdict"); Baldwin, 112 U.S. at 493 ("but its generality prevents us from perceiving upon which plea they found").

(152.) See Axelrad & Kraus, supra n. 109, at 43 (pointing out that the Baldwin principle "derives from an undeniably sound rationale--that a reviewing court can never confidently know either the basis for a general verdict or whether inclusion of an erroneous theory prejudiced the jury's deliberations. Nothing in the record can illuminate this inquiry, or render it more than pure guesswork. The Supreme Court has reaffirmed its general verdict rule three times, without question or qualification.").

(153.) Asbill, 726 F.2d at 1504.

(154.) Fed. R. Civ. P. 61.

(155.) Wilmington, 205 U.S. at 79.

(156.) Fed. R. Civ. P. 1 ("These rules govern the procedure in the United States district courts in all suits of a civil nature whether cognizable as cases at law or in equity or in admiralty, with the exceptions stated in Rule 81.") (emphasis added). Rule 61's harmless error rule thus only applies to post-judgment motions for new trial and things of that sort.

(157.) 464 U.S. 548 (1984).

(158.) Id. at 554.

(159.) Id.

(160.) Asbill, 726 F.2d at 1504.

(161.) 28 U.S.C. [section] 2111.

(162.) Bruneau, 163 F.3d at 760.

(163.) Neubauer, 766 F.2d at 1575-76.

(164.) 328 U.S. 750 (1946).

(165.) Id. at 755-56 (quoting the circuit court as holding that the trial judge "was plainly wrong in supposing that upon the evidence there could be a single conspiracy; and in the view he took of the law, he should have dismissed the indictment").

(166.) Id. at 775.

(167.) Id. at 759. See generally Charles Alan Wright, Federal Practice & Procedure Criminal vol. 3A, [section] 852 (2d ed., West 1982) ("There was a time when appellate courts, in criminal cases especially, were `impregnable citadels of technicality'") (emphasis added).

(168.) See Wilmington, 205 U.S. at 78 ("And, of course, in a [civil] case like the one we are considering we cannot maintain the verdict, as might be done in a criminal case upon a general verdict of guilty upon all the counts of an indictment.") (emphasis added).

(169.) 283 U.S. 359 (1931).

(170.) Id. at 367-368. See also Williams v. N.C., 317 U.S. 287 (1942); Cramer v. U.S., 325 U.S. 1 (1945); Yates v. U.S., 354 U.S. 298, 311-12 (1957).

(171.) 502 U.S. 46 (1991).

(172.) Id. at 53-56.

(173.) There is a broader question here as to whether a constitutional error can ever be considered harmless. See generally Yale Kamisar, Wayne R. LaFave, Jerold H. Israel & Nancy King, Modern Criminal Procedure: Cases-Comments-Questions 1589-1612 (9th ed. West 1999). See also William M. Landes & Richard A. Posner, Harmless Error, 30 J. Leg. Stud. 161 (Jan. 2001); Charles F. Campbell, Jr., An Economic View of Developments in the Harmless Error and Exclusionary Rules, 42 Baylor L. Rev. 499 (Summer 1990). That debate, however, is beyond the scope of this paper.

(174.) As will be explained infra, the substance-procedure dichotomy is not entirely relevant to this debate because the Supremacy Clause renders 28 U.S.C. [section] 2111 controlling. For a useful discussion of the substance-procedure dichotomy, however, see Paul D. Carrington, "Substance" and "Procedure" in the Rules Enabling Act, 1989 Duke L. J. 281; Stephen B. Burbank, Comment, Hold the Corks: A Comment on Paul Carrington's "Substance" and "Procedure" in the Rules Enabling Act, 1989 Duke L.J. 1012.

(175.) See Wilmington, 205 U.S. at 78-79 ("Nor does [section] 57 of the Illinois Practice Act ... support the contention that errors of the character of those we have just been considering must be treated as not prejudicial.... This section has been held not to relate to counts which are vitally defective, but as only providing that where a declaration consists of several counts, and some of the counts contain defects not vital ... a party cannot wait until after the close of the evidence at the trial, and, a fortiori, after verdict, and then for the first time question the sufficiency of the counts.") (citing Illinois cases).

(176.) See supra nn. 136 & 139 and accompanying text.

(177.) 487 U.S. 22 (1988).

(178.) Id. at 32 (citing Hanna v. Plumer, 380 U.S. 460, 473 (1965)).

(179.) See Jeffrey Abramson, We, The Jury: The Jury System and the Ideal of Democracy (179) (Basic Books 1994). See also Jeremy Osher, Note and Comment, Jury Unanimity in California: Should It Stay or Should It Go?, 29 Loy. L.A.L. Rev. 1319 (1996).

(180.) 406 U.S. 404 (1972).

(181.) Id. at 407. Justice White posited four possible theories for the birth of the unanimity requirement. One theory is that the rule was designed to compensate for the lack of procedural safeguards ensuring that the defendant received a fair trial. A second theory is that the rule developed from the practice of afforcement of the jury that was firmly established by the late fourteenth century. This term merely meant that a sufficient number were to be added to the panel until twelve were at last found to agree in the same conclusion. Third, it is possible that unanimity developed "because early juries, unlike juries today, personally had knowledge of the facts of a case; the medieval mind assumed there could be only one correct view of the facts, and, if either all the jurors or only a minority thereof declared the facts erroneously, they might be punished for perjury." Finally, unanimity may have arisen out of the medieval concept of consent which carded with it the idea of unanimity. Id. at n. 2.

(182.) 406 U.S. 356 (1972).

(183.) Id. at 369 (Powell, J., concurring) (citations omitted) (emphasis in original).

(184.) See Richard A. Primus, When Democracy is Not Self-Government: Toward a Defense of the Unanimity Rule for Criminal Juries, 18 Cardozo L. Rev. 1417 (1997). See also James Wm. Moore, Moore's Federal Practice vol. 9, [section] 48.0511]-[4] (3d ed., Matthew Bender 2001) ("Unanimity provides greater certainty in the result and reinforces each juror's agreement with the verdict. In short, unanimity protects the parties and the jury."); Edward P. Schwartz & Warren F. Schwartz, Decisionmaking by Juries Under Unanimity and Supermajority Voting Rules, 80 Geo. L.J. 775 (1992).

(185.) See Apodaca v. Ore., 406 U.S. 404 (1972); Johnson v. La., 406 U.S. 356 (1972). Although Justice Powell's opinion advocating a constitutional unanimity rule was controlling in Johnson, it should be noted that a unanimous verdict rule was rejected by a plurality in Apodaca. See Apodaca, 406 U.S. at 406. Both of these cases, however, were criminal, and it should be further noted that, prior to the Apodaca plurality's decision, there was a long line of Court precedents that advocated unanimous verdicts. See Am. Publg. Co. v. Fischer, 166 U.S. 464 (1897); Thompson v. Utah, 170 U.S. 343 (1898); Patton v. U.S., 281 U.S. 276 (1930); Andres v. U.S., 333 U.S. 740 (1948). The criminal jury verdict rule, however, is premised on the Sixth Amendment, whereas the civil jury verdict rule is premised on the Seventh Amendment, and the distinction may make a difference. See Charles Alan Wright & Arthur R. Miller, Federal Practice & Procedure Civil 2d, vol. 9A [section] 2492 (2d ed. West 1995) (arguing that "[t]he language of the Sixth Amendment differs from that of the Seventh Amendment" such that these decisions are inapplicable in the civil context).

(186.) In recent years, many commentators have begun to criticize the unanimity requirement. See e.g. Jere W. Morehead, A "Modest" Proposal for Jury Reform: The Elimination of Required Unanimous Verdicts, 46 U. Kan. L. Rev. 933 (1998); Michael H. Glasser, Comment, Letting the Supermajority Rule: Non-unanimous Jury Verdicts in Criminal Trials, 24 Fla. St. U. L. Rev. 659 (1997); James Kachmar, Comment, Silencing the Minority: Permitting Non-unanimous Jury Verdicts in Criminal Trials, 28 Pac. L. J. 273 (1996).

(187.) See Wright & Miller, supra n. 185, at [section] 2492 ("In the absence of a stipulation, however, the requirement of unanimity prevails in federal civil actions, and courts have gone to great lengths to insist that a verdict be in fact the verdict of a unanimous jury."); see also Moore, supra n. 184, at [section] 48.05 [1]-[4] ("The verdict must be unanimous"). Prior to 1991, there was some doubt as to whether there was a federal unanimity rule in civil cases. Compare Wieser v. Chrysler Motors Corp., 69 F.R.D. 97 (E.D.N.Y. 1975) (arguing that there is no Seventh Amendment requirement of unanimity in a civil case) with Masino v. Outboard Marine Corp., 652 F.2d 330 (3d Cir. 1981) (arguing that there is a federal unanimity rule). However, with the 1991 amendments to Rule 48, the issue was resolved. See Moore, supra n. 184, at [section] 48.05 [1]-[4] ("Prior to the 1991 revision of Rule 48, courts disputed whether a federal jury verdict must be unanimous.... Revised Rule 48 resolves that dispute by clarifying that `the verdict shall be unanimous' unless the parties stipulate to the contrary.").

(188.) Fed. R. Civ. P. 48.

(189.) See Murray v. Laborers Union Local No. 324, 55 F.3d 1445, 1450-51 (9th Cir. 1995) ("The Seventh Amendment requires jury verdicts in federal civil cases to be unanimous") (citing Justice Powell's Johnson concurrence, Andres, and Fisher).

(190.) If the unanimity rule is constitutionally based, as the Ninth Circuit asserts, and I believe to be true, then an appellate court can't abridge the Seventh Amendment any more than a district court could--even if it wanted to.

(191.) See Primus, supra n. 184, at 1436.

(192.) See Douglas W. Rae, Decision-Rules and Individual Values in Constitutional Choice, 63 Am. Pol. Sci. Rev. 40, 52-53 (1969). Rae's argument is made specifically in the criminal context, but it applies to civil cases as well.

(193.) See McCord, 873 F.2d at 1274 (recognizing that a waiver rule would "not allow litigants to play procedural brinkmanship with the jury system and take advantage of uncertainties they could well have avoided"). See generally Part III.C, supra (discussing the modern courts' reasoning with respect to a waiver rule).

(194.) The Fifth Circuit, as noted above, has intimated that the use of a general verdict for multiple claims may be error in and of itself. See Nowell, 792 F.2d at 1312 n. 1 (suggesting that "the use of a general verdict to probe disparate allegations of negligence and contributory negligence, if properly objected to, would in all likelihood meet with our decisive disapproval"). But see Mueller, 573 F.2d at 1038 n. 13 (emphasizing discretionary nature of specialized verdicts to reject a Nowell argument). Although a per se reversible error rule would not change the outcome that the Baldwin principle requires in the particular case, it is likely to deter district courts in the future from employing general verdicts to probe disparate theories of liability. It therefore would have the desirable effect of avoiding the Baldwin scenario altogether. See Brown, supra n. 22 (encouraging the use of special verdicts); but for another point of view, see David A. Lombardero, Do Special Verdicts Improve the Structure of Jury Decision-Making? 36 Jurimetrics J. 275 (Spring 1996) (noting that special verdicts, depending on how they are structured, can be biased toward the plaintiff and can lead to judicial inefficiency). Still, because special verdicts are perceived as aiding judicial economy, it is perhaps unsurprising that many appellate courts routinely encourage district courts to use special verdicts, even if they perceive themselves to be powerless to force them to do so because of Rule 49(a). Yet, a reversible error per se rule would not necessarily conflict with Rule 49(a). It could be based on the idea that the use of a general verdict in Baldwin situations is an abuse of the discretion that Rule 49(a) commits to the district court.

(195.) 487 U.S. 22 (1988).

Ryan Patrick Phair, J.D. University of Chicago, M.A. University of Chicago, B.A. Boston College. The author is an appellate litigator at Kirkland & Ellis in Washington, D.C. I would like to thank Professor Emily Buss of the University of Chicago Law School, Christopher Landau, and Allison Landis.