This appeal by the owner and operator of a supertanker from a verdict in favor of plaintiff Francis McQueeney, a seaman aboard the vessel, presents three interesting questions in the law of evidence. The first, arising under Fed.R.Evid. 401 and 403, is whether evidence from which it might be inferred that McQueeney has suborned perjury of a proffered witness is admissible as substantive evidence that his claim is unfounded even though the witness never testified. The second, arising under Fed.R.Evid. 901 et seq., is whether McQueeney's Sea Service Records which bore significantly on his damage claim were sufficiently authenticated by circumstantial evidence without the testimony of a witness. The district court excluded the evidence of subornation of perjury and the Sea Service Records, but we conclude that it erred.

Having reached these conclusions, we are confronted with the third evidentiary question of this case, which arises under Fed.R.Evid. 103(a): what standard of review should a court use in analyzing claims of nonconstitutional harmless error in civil suits? We hold that a court can find that such errors are harmless only if it is highly probable that the errors did not affect the outcome of the case. Applying that standard to the facts here, we find that the errors of the trial court in this case were not harmless. Hence we reverse its judgment and remand for a new trial.1

A. Plaintiff's Accident, His Lawsuit, and the Deposition of Mauro De la Cerda

Appellee McQueeney was a second officer on the T T WILLIAMSBURG, a supertanker owned by appellant Wilmington Trust Company and operated by Anndep Steamship Corporation. McQueeney claims that on March 20, 1981, while the WILLIAMSBURG was docked at Hounds Point Scotland, he was knocked to the deck while manning a water hose. McQueeney asserts that his fall was caused by both overpressure of the hose and by oil that had been spilled on the deck, making firm footing impossible, and that as a result of his accident, he suffered a herniated cervical disc. He brought this suit in June, 1982 in the district court for the Eastern District of Pennsylvania. Because the case arose under the Jones Act, 46 U.S.C. Sec. 688 et seq. (1982), jurisdiction was predicated upon 28 U.S.C. Sec. 1333 (1982). The district court conducted a jury trial at the end of which the jury awarded plaintiff a verdict of $305,788.00 against the two defendants. Judgment was entered in the same amount, and the defendants' motions for a new trial and for relief from the judgment were denied. The present appeal followed.

At trial, McQueeney was his only witness on the issue of liability. On the day the trial was scheduled to begin, however, McQueeney's counsel informed the court that he had just located an eyewitness to the accident, a fellow seaman of McQueeney's named Mauro De la Cerda, who was on board a ship in Freeport, Texas, and was therefore not able to appear as a witness. Counsel requested permission to depose De la Cerda.

The district court granted plaintiff's counsel permission to depose De la Cerda on the conditions that (1) defense counsel be given an opportunity to speak with De la Cerda before deciding whether to travel to Texas, and (2) plaintiff pay costs of defense counsel's trip to Texas if defense counsel chose to make the trip. Defense counsel spoke with De la Cerda by telephone that afternoon and chose to go to Houston. The appropriate arrangements were made, trial was recessed, and the next day De la Cerda was deposed in Houston. His testimony corroborated McQueeney's in all significant respects. Defense counsel, claiming to have been surprised by the deposition testimony because De la Cerda had allegedly told him a different version with respect to several significant facts in their telephone conversation, cross-examined De la Cerda about his statements. However, on both direct and then redirect examination at the deposition, De la Cerda either denied making any statements that contradicted his deposition testimony or testified that his statements of the night before were incorrect and that his current statements were accurate.

When the parties returned to trial, defense counsel moved for leave to withdraw his appearance so that he could testify and impeach De la Cerda's deposition testimony, which he presumed plaintiff would offer at trial. Defense counsel also listed plaintiff's counsel and his associate as witnesses. After a colloquy in the chambers of the district court, plaintiff's counsel and his associate signed affidavits stating that they had not discussed De la Cerda's testimony with him prior to his deposition. The court thereupon denied the counsel's motion for leave to withdraw.

B. Evidence of the Falsity of De la Cerda's Deposition, Plaintiff's Decision Not to Offer it, and the District Court's Ruling.

The trial resumed, and McQueeney took the stand. His testimony lasted several days. During cross-examination, and after court had adjourned for the day, defense counsel received crew lists from his client. The lists reflected that De la Cerda had not joined the crew of the WILLIAMSBURG until three months after the alleged accident. The lists proved, therefore, that De la Cerda's "eyewitness" testimony that he had given at his deposition had been fabricated. The next morning, defense counsel brought this information to the attention of the court in a discussion in chambers. After reviewing the crew lists, plaintiff's counsel immediately stated his intention not to use the deposition.2 Defense counsel rejoined that he intended to use the deposition to show fraud on the court. Plaintiff's counsel responded that, so long as he was not using the deposition himself, and so long as there was no evidence that McQueeney had perjured himself on the stand, there had been no fraud and the deposition was irrelevant. The district court agreed with plaintiff's counsel and stated that it would not receive the deposition and the crew lists into evidence.

The district court did not articulate the basis for its ruling at trial. However, as appears from the colloquy at the time, the district court felt that so long as the deposition was not introduced by plaintiff, any perjury associated with the deposition was irrelevant to the suit at bar.3 That this was the court's thinking is evident from its opinion denying defendants' post-trial motion for relief from the judgment or, in the alternative, a new trial. In support of the motion, defendants argued that it was reversible error to bar the deposition and crew lists, but the court ruled that the deposition and crew lists were either irrelevant or only minimally relevant:

It is questionable if Mr. De la Cerda's testimony is even relevant, since defendant seeks admission for the mere purpose to impeach and not for the substance of the testimony. Certainly, the confusion of a person's recollection brings little to show whether or not a set of facts occurred in a specific manner. Such testimony would confuse and mislead the jury, because what is at issue here is whether or not certain events did occur, not whether Mr. De la Cerda can remember or has such knowledge. To admit such testimony would not be for any probative value but merely to prejudice the jury against the plaintiff. Surely the probative value of such testimony would be minimal at best and is outweighed by the severe prejudice it would cause at the time of trial.

Thus the court may fairly be said to have excluded the evidence as either irrelevant under Fed.R.Evid. 401, or as relevant but misleading or unfairly prejudicial, in accordance with Fed.R.Evid. 403.4

During discovery, the defendants requested copies of plaintiff's Sea Service Records. Sea Service Records are official records that the Coast Guard requires seamen on merchant ships to fill out at the end of each voyage. 46 C.F.R. Sec. 14.10-5. Each entry records the time and place that the seaman began and finished his work, as well as the name of the ship and the nature of the voyage. Each entry, signed by both the seaman and the master of the vessel he is leaving, is reproduced in triplicate, the seaman keeping the original, the master and the Coast Guard getting the copies. Id.

Plaintiff supplied the requested copies, and at the close of its case the defense offered them into evidence. The purpose of introducing the Records was to suggest that McQueeney's future earning capacity, which had allegedly been impaired by his injury, should be based on a severely limited work schedule. Taken together, the records showed that McQueeney had, for the preceding several years, worked only four to six months a year, and defendants contended that the jury should be aware of McQueeney's pared-down work schedule in assessing damages.

The district court, however, refused to admit the copies of the Sea Service Records into evidence on the grounds that their authenticity had not been established, stating that they had to be authenticated by a testifying witness.5 The court told defense counsel that he could put plaintiff back on the stand for the purpose of introducing the records. Although plaintiff took the stand again, defense counsel did not seek to authenticate the Sea Service Records through him. When the defense proffered the Records for a second time just before the district court charged the jury, the court again refused to admit them.

D. Defendant's Post-Trial Motions and Appeal

After the jury's verdict in favor of the plaintiffs, the defendants moved for a judgment n.o.v. or in the alternative for a new trial pursuant to Fed.R.Civ.P. 50(b). Defendants argued that the jury verdict was unsupported by the evidence and that the court had erred in refusing to admit the crew lists, defense counsel's testimony, and De la Cerda's deposition. Several months thereafter, before the district court had reached a decision on their motions, the defendants moved for relief from judgment pursuant to Rule 60(b)(2) on account of newly discovered evidence. In this motion, defendants stated that De la Cerda had recently been deposed in his own personal injury case against the tanker operator, Mauro DeLaCerda v. Anndep Steamship Corp. et al., No. E-118-606 (172d Judicial District, Jefferson County, Texas), in a state court in Texas. According to the papers, De la Cerda testified at the deposition that McQueeney had asked him to testify in the present case, that De la Cerda told McQueeney that he had not been on board the ship when the accident took place, but that McQueeney asked him to testify nonetheless. The court consolidated all of the defendants' motions, and denied them.

II. DEFENDANTS' PROFFER THAT MCQUEENEY HAD SUBORNED PERJURY

The stated purpose of defense counsel's proffer of De la Cerda's deposition, the crew lists for the date of the alleged accident, and defense counsel's own testimony about what De la Cerda had told him the night before the deposition, was to show that the plaintiff had suborned perjured testimony. Defense counsel intended to argue that plaintiff's subornation was evidence of his knowledge of the weakness of his case, and that such knowledge could be taken into account by the jury.6

The district court's decision to exclude the evidence as irrelevant is governed by Fed.R.Evid. 401. We review that decision according to the abuse of discretion standard. See United States v. Steele, 685 F.2d 793, 808 (3d Cir.) (once the threshold of logical relevance is satisfied, "the matter is largely within the discretion of the trial court"), cert. denied, 459 U.S. 908, 103 S. Ct. 213, 74 L. Ed. 2d 170 (1982); United States v. Robinson, 560 F.2d 507, 515 (2d Cir.1977) (en banc ), cert. denied, 435 U.S. 905, 98 S. Ct. 1451, 55 L. Ed. 2d 496 (1978); 1 J. Weinstein & M. Berger, Weinstein's Evidence p 401 at 401-9--401-12 (1982). We believe that the district court abused its discretion in excluding the proffered evidence. We base our conclusion on common sense, eminent commentators, case law, and the explicit language of Rule 401.

The intuitive appeal of defendants' proffer is immediate. One who believes his own case to be weak is more likely to suborn perjury than one who thinks he has a strong case, and a party knows better than anyone else the truth about his own case. Thus, subornation of perjury by a party is strong evidence that the party's case is weak. Admittedly the conclusion is not inescapable: parties may be mistaken about the merits or force of their own cases. But evidence need not lead inescapably towards a single conclusion to be relevant, it need only make certain facts more probable than not. The evidence of subornation here does cast into doubt the merits of McQueeney's claim, even if it does not extinguish them.7

There is ample support among both scholars and courts for this line of argument. Wigmore calls the inference "one of the simplest in human experience":

It has always been understood--the inference indeed is one of the simplest in human experience--that a party's falsehood or other fraud in the preparation and presentation of his cause, his fabrication or suppression of evidence by bribery or spoliation, is receivable against him as an indication of his consciousness that his case is a weak or unfounded one; and from that consciousness may be inferred the fact itself of the cause's lack of truth and merit.

2 Wigmore Sec. 278(2) (Chadbourne Rev.1979). McCormick makes a similar point:

[W]rongdoing by the party in connection with his case, amounting to an obstruction of justice[,] is also commonly regarded as an admission by conduct. By resorting to wrongful devices he is said to give ground for believing that he thinks his case is weak and not to be won by fair means. Accordingly, a party's false statement about the matter in litigation, whether before suit or on the stand, his fabrication of false documents, his undue pressure, by bribery or intimidation or other means, to influence a witness to testify for him ... all these are instances of this type of admission by conduct.

McCormick's Handbook on the Law of Evidence Sec. 273 at 660 (2d ed. 1972).

This court has upheld the inference and admitted evidence accordingly. See Newark Stereotypers' Union v. Newark Morning Ledger, 397 F.2d 594, 599 (3d Cir.1968) ("an attempt by a litigant to persuade a witness not to testify is properly admissible against him as an indication of his own belief that his claim is weak or unfounded or false"). Several other courts of appeals and state appellate courts have done the same. See Great American Insurance Co. v. Horab, 309 F.2d 262, 264 (8th Cir.1962) (applying rule and citing cases); see also 2 Wigmore Sec. 278(2) (Chadbourne Rev.1979) (collecting cases).

McQueeney points out that Newark Stereotypers' Union and Horab involved a party's subornation of perjured testimony from a witness who testified at trial, whereas in the case at bar De la Cerda's testimony was never introduced into evidence. This is correct, but it is a distinction of no consequence. Evidence of subornation of perjury is substantive evidence, not mere impeachment material; the inference that one may draw from the subornation does not depend upon anyone else's testimony. The fact that a party suborned perjury is what matters, not the ultimate success or failure of that subornation.

Finally, the explicit language of Fed.R.Evid. 401, which defines as relevant evidence that has "any tendency" to make a difference in the case, provides added reason to permit the materials into evidence. The plain meaning of the Rule demonstrates that the scope of relevant evidence is intended to be broad, and the authorities support such a broad reading. See United States v. Clifford, 704 F.2d 86, 90 (3d Cir.1983); United States v. Steele, supra, 685 F.2d at 808; Carter v. Hewitt, 617 F.2d 961, 966 (3d Cir.1980) ("[t]he standard of relevance established by the Federal Rules of Evidence is not high"). Thus, logic, authority, and the clear language of Rule 401 lead us to the conclusion that the deposition and related testimony were relevant and that the district court abused its discretion in excluding it on relevancy grounds.

That the evidence was relevant does not necessarily mean that it should have been admitted. As noted above, the district court, after voicing its doubts that the evidence had any relevance, ruled that even assuming it was relevant, the evidence should not have been admitted because its prejudicial impact would likely outweigh its probative value. Supra pp. 919-920. This is a standard Fed.R.Evid. 403 balance which we review with substantial deference. See United States v. Lebovitz, 669 F.2d 894, 901 (3d Cir.) (trial judge has "very substantial discretion"), cert. denied, 456 U.S. 929, 102 S. Ct. 1979, 72 L. Ed. 2d 446 (1982). Despite this deferential standard, we find that the district court erred, for it underestimated the probative value of the evidence, and misevaluated its prejudicial impact.

The district court assigned virtually no probative value to the evidence of subornation. See supra pp. 919-920. It should be clear from what we said above, see supra pp. 921-923, however, that evidence of subornation of perjury may be quite valuable to the defendant in this case. Intuition and the unanimity of the commentators and numerous courts that have considered it suggest not only that subornation of perjury is relevant but that it is powerful evidence indeed. Evidence that McQueeney suborned perjury might well have made the jurors re-evaluate McQueeney's case. Of course, it is not certain that McQueeney did suborn perjury; De la Cerda may have had other reasons for making up his story. But the circumstances of the case and the correlation between McQueeney's story and De la Cerda's deposition suggest that subornation of perjury by McQueeney is a possibility that the jury should have been allowed to consider.

By contrast, although the district court referred to potential confusion and "severe prejudice," supra at p. 920, and although there was danger of prejudice--the mere suggestion that McQueeney suborned perjury might have led the jury to reflect on his character in an improper manner--it is unlikely that the evidence would result in the "unfair prejudice" proscribed by Rule 403. Cf. Dollar v. Long Mfg., N.C. Inc., 561 F.2d 613, 618 (5th Cir.1977) (" 'unfair prejudice' as used in Rule 403 is not to be equated with testimony simply adverse to the opposing party. Virtually all evidence is prejudicial or it isn't material. The prejudice must be 'unfair.' "). Moreover, it appears to us that the danger of improper influence was not sufficient to "substantially outweigh[ ]" the probative value of the evidence, as required by Fed.R.Evid. 403. The court did not articulate any reasons for its finding of prejudice, and this does not appear to us to be the kind of evidence with obvious or overwhelming potential for unfair prejudice. In the absence of a showing of particularized danger of unfair prejudice, the evidence must be admitted. Were we to rule otherwise, evidence could be excluded on an unfounded fear of prejudice and we would effectively preclude all evidence of subornation of perjury.8

In sum, the district court misconstrued both elements of the Rule 403 balance hence its balancing exercise was skewed. The court's Rule 403 ruling was thus an abuse of discretion.9

Fed.R.Evid. 103(a) states that an evidentiary ruling may not be reversible error "unless a substantial right of a party is affected." See also Fed.R.Civ.P. 61 ("No error ... is ground for granting a new trial or for setting aside a verdict ... unless refusal to take such action appears to the court inconsistent with substantial justice"); 28 U.S.C. Sec. 2111 (1982) (appellate court should give judgment "without regard to errors or defects which do not affect the substantial rights of parties"). It remains for us to consider, therefore, whether the court's decision substantially affected the rights of appellants.

At the outset, we must establish the appropriate standard of review.10 The analysis is complicated because although there are several cases in this circuit dealing with harmless error in the criminal context,11 there are none dealing with the standard of review of harmless error in civil cases, and some courts and commentators have argued quite forcefully that the two standards are not the same. We proceed, therefore, by first discussing this circuit's standard of review in harmless error questions in criminal cases. We then consider whether the same standard should apply in civil and criminal cases, and conclude that it should. Finally, we apply the uniform standard to the facts of this case.

1. Criminal cases--In Government of the Virgin Islands v. Toto, 529 F.2d 278 (3d Cir.1976), this court stated that an appellate court's harmless error determination requires "perforce [a] resort to probabilities." Id. at 283.12 Following the lead of former Chief Justice Roger Traynor, the Toto court discussed three possible standards of review for harmless error:

The reviewing court might affirm if it believes: (a) that it is more probable than not that the error did not affect the judgment, (b) that it is highly probable that the error did not contribute to the judgment, or (c) that it is almost certain that the error did not taint the judgment.

Id. See also R. Traynor, The Riddle of Harmless Error 35 (1970) (offering the same three options). The Toto court chose the middle option and thus held that errors in a criminal case are not harmless unless it is "highly probable" that they did not affect a party's substantial rights. Id. at 284. See also United States v. Jannotti, 729 F.2d 213, 219-20 & n. 2 (3d Cir.1984) (following Toto ); United States v. Cook, 538 F.2d 1000, 1005 (3d Cir.1976) (same); United States v. Savage, 430 F. Supp. 1024, 1033 (M.D.Pa.1977) (same).

Recent developments have cast doubt on Toto's holding. In Government of the Virgin Islands v. Joseph, 685 F.2d 857, 864 (3d Cir.1982), this court, without citing Toto, held that an error could be harmless only if the appellate court found " 'beyond a reasonable doubt' " that it did not affect a party's substantial rights (quoting Chapman v. California, 386 U.S. 18, 24, 87 S. Ct. 824, 828, 17 L. Ed. 2d 705 (1967)). Because "beyond a reasonable doubt" is a more stringent standard than "high probability," the Joseph test would make it harder for an appellate court to rule that error is harmless than it has been under the prevailing Toto standard.13 We believe, however, that the Toto standard is the law of the circuit, for it is the practice of this court that a panel confronted with conflicting precedents must follow the earlier precedent. See O. Hommel Co. v. Ferro Corp., 659 F.2d 340, 354 (3d Cir.1981). As a matter of the Third Circuit appellate practice, therefore, Toto prevails.14

2. Civil cases--There are no cases in this this circuit analyzing the appropriate standard of review for harmless error in civil cases. Because Toto was a criminal case, there is some question whether its holding applies with full force in the civil context. Although Toto itself did not distinguish between civil and criminal cases, some courts and commentators argue that it should be more difficult to prove harmless error in the criminal context than in the civil. McIlroy v. Dittmer, 732 F.2d 98, 105 (8th Cir.1984); Haddad v. Lockheed California Corp., 720 F.2d 1454, 1459 (9th Cir.1983); United States v. Valle-Valdez, 554 F.2d at 915; Saltzburg, supra n. 14 at 994-98. Others argue that there should be no difference between the standards used in civil and criminal cases. See 1 Wigmore, Evidence Sec. 21 (3d ed. 1940) (no distinction between civil and criminal); R. Traynor, supra at 48 (same).

Although the courts and esteemed commentators who espouse the position cannot be lightly contradicted, we are unpersuaded by the view that the harmless error standard of review in civil cases should be different from the analogous standard in criminal cases. The case for differentiating the two situations was put most comprehensively by the court in Haddad v. Lockheed, supra:

[A] crucial first step in determining how we should gauge the probability that an error was harmless is recognizing the distinction between civil and criminal trials. This distinction has two facets, each of which reflects the differing burdens of proof in civil and criminal cases. First the lower burden of proof in civil cases implies a larger margin of error. The danger of harmless error doctrine is that an appellate court may usurp the jury's function, by merely deleting improper evidence from the record and assessing the sufficiency of the evidence to support the verdict below. This danger has less practical importance where, as in most civil cases, the jury verdict merely rests on a more probable than not standard of proof.

The second facet of the distinction between the errors in civil and criminal trials involves the differing degrees of certainty owed to civil and criminal litigants. Whereas a criminal defendant must be found guilty beyond a reasonable doubt, a civil litigant merely has the right to a jury verdict that more probably than not corresponds to the truth.

The civil litigant's lessened entitlement to veracity continues when the litigant becomes an appellant. We conclude that a proper harmless error standard in civil cases should reflect the standard of proof.

720 F.2d at 1459.

The Haddad court thus makes two arguments for imposing a laxer harmless error standard in civil cases than in criminal. The first starts from the undeniable premise that a finding of harmless error by an appellate court infringes on the function of the jury because it puts the appellate court in the position of weighing the evidence before the trial court. See Field, Assessing the Harmlessness of Federal Constitutional Error--A Process in Need of a Rationale, 125 U.Pa.L.Rev. 15, 33-34 (1976). The court then proposes that the infringement of the jury's role is less significant in civil than in criminal cases, and that therefore appellate courts reviewing civil cases should be more willing to find errors harmless than they would be if they were reviewing criminal cases. The Haddad court's second argument is that the lower burden of proof in civil than in criminal cases implies that society tolerates a greater risk of error in civil than in criminal cases, and that that toleration for error should be reflected in the appellate process as well.15

We find neither argument persuasive. The crux of the first argument is that jury verdicts are less worthy of respect in civil cases than in criminal. This is so, explains the Haddad court, because the civil jury verdict usually rests on a more probable than not standard of proof, whereas the criminal verdict rests on a beyond a reasonable doubt standard. We fail to see the force of the distinction. Verdicts are the result of the same serious deliberation by the jurors no matter what the standard of proof. This is especially so today when, with the proliferation of cases involving civil rights, huge damage awards and forfeitures, the stakes in civil cases are so high, and the infringement of the jury verdict caused by a harmless error finding may be as important in the civil context as in the criminal.

The Haddad court's second argument, which also relies on the different standards of proof in civil and criminal cases, is similarly uncompelling. The court once again begins from a sound premise: that the lesser burden of proof in the civil context than in the criminal demonstrates that the impact of falsity is less damaging in civil than in criminal suits for only in the latter is personal liberty at stake. The court goes on to assert that "[t]he civil litigant's lessened entitlement to veracity continues when the litigant becomes an appellant." There is no logical reason, however, that the tolerance for error in the civil context must be compounded by a less stringent standard of harmless error review. Society's tolerance for risk of error in civil cases may have been subsumed in the decision to establish a lower burden of proof in civil cases; further enlargment of the margin of error would therefore distort, rather than reflect, society's wishes.16

The argument for a differentiated standard must rest, then, on policy, not logic. Professor Saltzburg recognizes this point, see Saltzburg, supra n. 14, at 995, but his main policy argument is circular. He argues that "a uniform standard based on the criminal norm would result in the reversal of an inordinate number of civil cases." Id. at 997. It is true that applying the criminal standard in civil cases would result in more reversals than would occur if a lesser standard were used, but to say that the number of reversals would be "inordinate" begs the question; the issue is precisely how many reversals on this ground we will tolerate. Saltzburg further argues that "whatever the standard of proof used in a tribunal, the test of errors on appeal should be designed to complement rather than undercut it." Id. at 998. It should be clear by now, however, that a relatively stringent harmless error standard does not "undercut" the standard of proof at the trial stage; the standard of proof and the societal values it represents remain intact.

We are thus unpersuaded by the arguments that the civil and criminal standards of review for harmless error should be different. Moreover, we believe that there are three compelling reasons that the standards should be the same. First, neither Federal Rule of Evidence 103(a) nor 28 U.S.C. Sec. 2111 (1982) (the harmless error statute directly applicable to the appellate courts) distinguishes between criminal and civil cases. This implies that Congress intended uniform treatment, and that differentiating civil and criminal standards of review would be contrary to congressional intent.17

Second, differentiating the standards might lead to unexpected complications when it is recognized that not all civil cases are alike. Some, like fraud or defamation, require clear and convincing evidence. Because the Haddad court and Professor Saltzburg apparently believe that the standard of appellate review should track the burden of proof at trial, it is unclear whether they would require different standards for fraud, defamation, and like cases.18

Third, broad institutional concerns militate against increasing the number of errors deemed harmless. Although it is late in the day to pretend that all trials are perfect, perfection should still be our goal. Judge (now Chief Judge) Robinson put the point well:

The justification for harmless-error rules is singleminded; they avoid wasting the time and effort of judges, counsel and other trial participants. Other considerations enter into the picture, however, when we set out to ascertain what is harmless and what is not. Wisdom of the ages counsels against appellate erosion of the stature and function of the trial jury. Societal beliefs about who should bear the risk of error in particular types of proceedings deserve weight in decisions on harmlessness. Respect for the dignity of the individual, as well as for the law and the courts that administer it, may call for rectification of errors not visibly affecting the accuracy of the judicial process. And the prophylactic effect of a reversal occasionally might outweigh the expenditure of effort on a new trial.

United States v. Burton, 584 F.2d 485, 512-13 (D.C.Cir.1978) (Robinson, J., dissenting). By maintaining a moderately stringent, though not unreasonably high, standard in civil as well as criminal cases, we preserve a strong incentive for the district courts to minimize their errors, and we thereby bolster the integrity of the federal judicial process.

We therefore conclude that the standard of review in civil cases is the same as in criminal, and that, as Toto is the law of this circuit with regard to criminal cases, so should it be our guide here.19

3. Application of the Toto standards to the facts of this case--Fortunately, the application of the appropriate standard is far easier than the determination of that standard. The district court's refusal to admit the evidence of subornation impaired the defendants' ability to discredit a central element of the plaintiff's case. Moreover, the evidence was critical for the defendants to make their argument on defense of liability; it was potentially the defendants' best evidence and not cumulative. Cf. United States v. Basic Construction Company, 711 F.2d 570 (4th Cir.) (evidentiary ruling was harmless where evidence in question was cumulative of evidence already properly admitted), cert. denied, 464 U.S. 956, 1008, 104 S. Ct. 371, 527, 78 L. Ed. 2d 330 (1983); Taylor v. Curry, 708 F.2d 886 (2d Cir.), cert. denied, 464 U.S. 1000, 104 S. Ct. 503, 78 L. Ed. 2d 694 (1983). We cannot say that it is highly probable that the trial court's refusal to enter evidence of plaintiff's subornation of perjury did not affect the defendants' substantial rights. Hence, the court's erroneous ruling is grounds for reversal.

III. THE AUTHENTICATION OF MCQUEENEY'S SEA RECORDS

Defendants sought to introduce McQueeney's Sea Records to limit his damages for lost future earnings capacity. The parties' briefs on the authentication of the Sea Records focused, in both the district court and on this appeal, on Local Rule 21(d) of the District Court for the Eastern District of Pennsylvania,20 and on the question whether McQueeney had waived the right to object to the authentication of the Sea Records. It is plain from the record, however, that regardless of these considerations the copies of the Sea Service Records were sufficiently authenticated by circumstantial evidence.21

Rule 903 of the Federal Rules of Evidence states that "[t]he testimony of a subscribing witness is not necessary to authenticate a writing unless required by the laws of the jurisdiction whose laws govern the validity of the writing." See also McCormick's Handbook on the Law of Evidence Sec. 222 at 548 (2d ed. 1972) ("authentication by circumstantial evidence is uniformly recognized as permissible"). There are no cases in this circuit to the contrary, and thus we hold that circumstantial evidence may, in principle, suffice to authenticate a document. The burden of proof for authentication is slight. "All that is required is a foundation from which the fact-finder could legitimately infer that the evidence is what the proponent claims it to be." In re Japanese Electronics, 723 F.2d 238, 285 (3d Cir.1983), cert. granted, --- U.S. ----, 105 S. Ct. 1863, 85 L. Ed. 2d 157 (1985). See also United States v. Goichman, 547 F.2d 778, 784 (3d Cir.1976) ("Once a prima facie case is made, the evidence goes to the jury and it is the jury who will ultimately determine the authenticity of the evidence, not the court.") (per curiam )

Several pieces of circumstantial evidence support the conclusion that the material in question is indeed what the defendant claims--copies of plaintiff's Sea Service Records.22 First, the contents of the documents tend to support their claim to authenticity. See Fed.R.Evid. 901(b)(4) ("[a]ppearances, contents, substance, internal patterns, or other distinctive characteristics, taken in conjunction with circumstances" may establish authenticity). The documents appear to be copies of standard official forms of the Department of Transportation, United States Coast Guard. Each is signed and dated by the plaintiff. Each includes his personal United States Merchant Mariners Document Number. The specificity, regularity, and official appearance of the documents increase the likelihood of their being authentic. See Fed.R.Evid. 901(b)(4); In re Japanese Electronics, supra, 723 F.2d at 286 (court referred to "the appearance, content, and substance" of purported minutes from a board meeting in upholding their authenticity).

Second, the fact that the copies were produced by the plaintiff in answer to an explicit discovery request for his Sea Service Records, while not dispositive on the issue of authentication, is surely probative. See Burgess v. Premier Corp., 727 F.2d 826, 835-36 (9th Cir.1984) (all exhibits found in defendant's warehouse were adequately authenticated simply by their being found there); In re Japanese Electronics, supra, 723 F.2d at 286 (documents "have the appearance, content, and substance typical of minutes. They were produced by the defendants pursuant to a discovery order in this proceeding. They come from a source where such minutes are likely to be kept.... No more evidence was needed to establish a prima facie case of authenticity than the record contains.") (citations omitted); Alexander Dawson v. N.L.R.B., 586 F.2d 1300, 1302 (9th Cir.1978) (circumstances of discovery, along with content of documents, were sufficient to demonstrate authenticity); United States v. Natale, 526 F.2d 1160, 1173 (2d Cir.1975), cert. denied, 425 U.S. 950, 96 S. Ct. 1724, 48 L. Ed. 2d 193 (1976); McCormick, supra, Sec. 224 at 552 ("proof of private custody, together with other circumstances, is frequently strong circumstantial evidence of authenticity"). It is unlikely that plaintiff misunderstood the discovery request, for it was quite specific and plaintiff was presumably well-acquainted with his own Sea Service Records.

Third, we note that the information in each of the purported records, although not secret by any means, was not widely held. In addition to plaintiff's personal Merchant Mariners Number, the records also listed his position on each voyage, and the dates and places that he began and finished each of his jobs.23 Wigmore states that when information is in the sole possession of a person, and a document purportedly written by that person contains that information, there is sufficient evidence of the authenticity of the document. 7 Wigmore Sec. 2148 at 745 (3d ed. 1940) Although Wigmore maintained that so long as more than one person had the information the inference of authenticity was devoid of all force, id. at 746, we believe that the better rule is that the inference does not vanish but rather diminishes as more people know the information contained in the documents. See 5 J. Weinstein & M. Berger, Weinstein's Evidence p 901(b)(4) at 901-49 (1982 ed. & Supp.) ("The force of the inference decreases as the number of people who know the details and may have written the letter increases.") Although we do not know precisely how many people had the information contained in the proffered evidence, we suspect, as noted above, that the number is small. Therefore, the nature of the information in the documents further supports their authenticity.

The sum of this circumstantial evidence easily shoulders the slight burden of proof that authentication requires. It is of course open for the plaintiff at trial to argue that the documents are not genuine, see Goichman, supra, 547 F.2d at 784, or that they are somehow not worthy of great weight in the jury's deliberations. The requisite prima facie showing has, however, been made.24 Having established that the copies of the Sea Records are authentic, we consider the remaining evidentiary hurdles, i.e., relevancy and hearsay. Although the plaintiff did not object on these grounds, in view of our disposition it is necessary that we consider them.

It is immediately apparent that the Sea Service Records were relevant. They showed how much McQueeney had worked in the last few years, and would therefore have been germane to the jury's damage determination. There could be no plausible suggestion that the records could be unduly prejudicial, confusing, or misleading.

Furthermore, there can be no hearsay objection, for the Sea Service Records are admissions of a party opponent admissible under Fed.R.Evid. 801(d)(2). Clearly, if McQueeney had filled out the Records themselves they would come under Rule 801(d)(2)(A) for they would be McQueeney's own statements.25 Upon inspection of the Records it appears, however, that the Records were filled out by several different people, most likely officials of the shipping companies employing McQueeney. This shifts our inquiry to Rule 801(d)(2)(B), adoptive admissions, but the result is the same for we believe that McQueeney's signature on each of the cards is an unequivocal adoption of the contents therein. See Pillsbury Co. v. Cleaver-Brooks, 646 F.2d 1216, 1218 (8th Cir.1981) (signing each page of a report satisfies Rule 801(d)(2)(B) requirement for adoption); United States v. Smith, 609 F.2d 1294, 1301 n. 7 (9th Cir.1979) (records containing defendant's signature "may more appropriately be regarded as non-hearsay admissions under Fed.R.Evid. 801(d)(2)(A) and 801(d)(2)(B).").

Plaintiff also argues that the admissibility of the Sea Records notwithstanding, defense counsel had the opportunity to enter the Sea Records when McQueeney was on the stand, and that defense counsel's failure to do so acts as a waiver of the issue of admissibility. We cannot accept this argument. As shown above, defense counsel was entitled to proffer the Sea Records without any authenticating testimony or other foundation. Counsel cannot be barred from using the evidence because he failed to comply with an erroneous ruling of the trial judge.

Finally, we must consider once again whether the court's error was harmless. The Sea Service Records might have persuaded the jury to award significantly less damages in lost earnings than it in fact did. Although there had been no testimony on plaintiff's work habits, in his closing argument plaintiff's counsel suggested that the jury might base its damage calculations on the assumption that plaintiff would work 8 months per year or even 10 months per year. It is reasonable to believe that the jurors took that figure into account in reaching their damage verdict. The Sea Service Records established, however, that from 1976 through 1980, the plaintiff worked an average of approximately 5 months per year. Applying the Toto standard, see supra pp. 925-928, it is evident that the court's error in excluding the Records was not harmless.

Because the district court erred in refusing to admit (A) relevant, non-prejudicial evidence of the subornation of perjury by a prospective witness, and (B) the Sea Service Records, although they were authentic, relevant, and admissions of a party-opponent, and because the errors were not harmless, the judgment of the district court will be reversed and the case remanded with directions to grant a new trial.26