The primary issue on this appeal is whether the prosecution may cross-examine a defendant's character witness by asking questions based on an assumption that the defendant is guilty of the offense charged. This issue arises on an appeal by Michael P. Oshatz and Leonard A. Messinger from their judgments of conviction entered, respectively, on July 14 and May 5, 1989, in the District Court for the Southern District of New York (Robert W. Sweet, Judge) after an eleven-week jury trial on charges arising out of their involvement in a series of fraudulent tax schemes. We reaffirm this Circuit's view that such cross-examination is impermissible, but nonetheless affirm the convictions because our prior pronouncement on this issue was understandably misinterpreted by the District Court and the resulting error was harmless in the circumstances of this case.Background

Between 1979 and 1983, Oshatz, a tax attorney, assisted in the formation of a number of affiliated partnerships known as the "Monetary Group." The offering memoranda for the Monetary Group reported that the partnerships would invest in various financial instruments to secure economic gain, that these investments would involve substantial market risk, and that any losses generated by these transactions would be available as tax deductions. Oshatz and Messinger, his law partner, also formed a number of other partnerships, in which they held an interest, for the purpose of purchasing tax shelter investments from the Monetary Group.

The Monetary Group partnerships engaged primarily in two types of securities transactions on behalf of their limited partner investors. Initially, the partnerships entered into "straddle" transactions in which "short" and "long" positions are simultaneously established in a commodity or a security. At the end of the year, the side of the transaction with a loss is closed out, generating a tax deduction. At the beginning of the next year, the other "leg" of the transaction is closed out, generating a taxable gain. After Congress passed legislation curtailing the use of straddle transactions, see Economic Recovery Tax Act of 1981, Pub.L. No. 97-34, 95 Stat. 172, 323-26, the partnerships entered into repurchase ("repo") agreements as investment vehicles. This type of arrangement involves the purchase of a security with borrowed money, with the security serving as collateral for the loan. In "open" repurchase agreements, the interest charge on the loan fluctuates with the prevailing market rate, entitling the investor to an interest expense deduction since a profit or loss may be realized on the transaction. Open repurchase agreements function much like straddle transactions since the interest on the loan may be deducted immediately, while the gain from the underlying security, generally a Treasury bill, is not realized until the next taxable year.

The Government offered convincing proof that the tax losses reported by the partnerships from these transactions were not the product of legitimate trading. Edward Markowitz, the head trader for the Monetary Group, testified that he falsified trade documents to reflect straddle transactions that never occurred. The partnerships also removed the risks associated with repurchase agreements by fixing the interest rate of the loan to coincide with the interest rate of the securities that collateralized the loan. Though this type of arrangement, known as a "repo to maturity" repurchase agreement, is legal, it provides no basis for claiming an interest expense deduction since no profit or loss can be realized in connection with the interest charges. To generate the desired tax losses, the partnerships financed repurchase agreements by using fixed "repo to maturity" rates but fraudulently documented the transactions as "open" repurchase agreements. Cf. United States v. Atkins, 869 F.2d 135, 138 (2d Cir.), cert. denied, --- U.S. ----, 110 S. Ct. 72, 107 L. Ed. 2d 39 (1989).

The Government brought a sixteen-count indictment against Oshatz and Messinger for their roles in the trading activities of these partnerships. Count One charged both defendants with a conspiracy to defraud the United States, in violation of 18 U.S.C. Sec. 371 (1988), by engaging in fraudulent securities transactions for the purpose of generating tax losses. The remaining counts alleged that one or both of the defendants aided in the filing of false tax returns for various partnerships, in violation of 26 U.S.C. Sec. 7206(2) (1988), and that both defendants knowingly and willfully filed false personal income tax returns, in violation of 26 U.S.C. Sec. 7206(1) (1988). The jury returned guilty verdicts against Oshatz and Messinger on each of the counts in which they were charged. The District Court sentenced Oshatz to forty months' imprisonment and three years' probation. Messinger was sentenced to twenty-eight months' imprisonment and three years' probation.

Appellants challenge the cross-examination of character witnesses, the admission of similar acts evidence, the exclusion of evidence claimed to impeach the credibility of a prosecution witness, and the admission of a summary chart.

Defense counsel questioned Gail Logan as to the reputation of Oshatz for truthfulness and honesty and as to her opinion of his truthfulness and honesty, eliciting testimony favorable to the defense. Logan, who had been an associate in defendants' law firm, had originally been called as a Government witness and became a character witness for Oshatz during cross-examination. For convenience, we will consider the defense questioning that elicited her character testimony to be direct examination and the prosecution's questioning that challenged her character testimony to be cross-examination. On cross-examination, the prosecutor asked, "If I were to show you that Mr. Oshatz knew that the Markowitz transactions were backdated, would that affect your opinion?" Objection was made and overruled. Logan answered, "Yes." Similar questions were asked and similarly answered with respect to other aspects of the wrongdoing alleged to constitute the offenses for which Oshatz was on trial. After the witness was excused, counsel for both defendants challenged the Government's right to cross-examine a character witness with hypothetical questions based on an assumption of guilt as to the pending charges. The next day defense counsel called Judge Sweet's attention to several opinions that appeared to proscribe the challenged cross-examination, including this Circuit's decision in United States v. Morgan, 554 F.2d 31 (2d Cir.), cert. denied, 434 U.S. 965, 98 S. Ct. 504, 54 L. Ed. 2d 450 (1977). In that decision we had said that "[i]nsofar as non-expert character witnesses are concerned," a hypothetical question that assumes the guilt of the defendant "should not be asked." Id. at 34.

Before the end of the trial, Judge Sweet filed an opinion rejecting the defendants' objection to the cross-examination.1 United States v. Oshatz, 704 F. Supp. 511 (S.D.N.Y.1989). He observed that under Federal Rule of Evidence 405(a) character witnesses may be cross-examined as to their knowledge of specific instances of a defendant's misconduct in order to help the jury determine how much weight to accord the character testimony, id. at 514; see United States v. Birney, 686 F.2d 102, 108 (2d Cir.1982), and that cross-examining about the misconduct at issue in the trial presented "nothing new to the jury," United States v. Oshatz, 704 F. Supp. at 514. Noting Morgan, he said that its distinction between expert and non-expert character witnesses does not appear in Fed.R.Evid. 405(a). He also rejected Morgan 's view that the challenged cross-examination involved hypothetical questions, contending that "the fact that a defendant has yet to be convicted for an unlawful act does not make a question regarding that act 'hypothetical.' "2 Id.

Armed with Judge Sweet's ruling and undeterred either by Morgan or by the risk that disregarding it would imperil any conviction that might be obtained, the prosecutor cross-examined Messinger's character witnesses who had testified as to both his reputation for honesty and truthfulness and their own opinion of these character traits. The form of the cross-examination was even more pointed than with Logan. To one witness the prosecutor asked, "[I]f you found that Mr. Messinger knowingly participated in setting up a phony tax shelter that generated over half a million dollars worth of business for his law firm, would that affect your opinion?" When objection to the form "If you found" was sustained, the question was rephrased, "So let's leave the facts aside and assume that it is found that Mr. Messinger knowingly participated in setting up a phony tax shelter that generated over half a million dollars worth of business for his law firm, would that affect your opinion?" The witness said it would not. Judge Sweet instructed the jury that the question involved a Government contention, that it was allowed as bearing only on the witness's credibility, and that its allowance did not alter the jury's responsibility. Another witness was similarly cross-examined with the question beginning "If the government were to establish that...." With another witness, the prosecutor reverted to the form, "Assume that it were found"; when the witness inquired, "Found by the jury?" and was answered by the prosecutor, "Or found as a fact, correct," the witness replied, "They could be wrong."

We consider whether Morgan proscribes the challenged cross-examination, whether Morgan should have been followed in this case, and whether the failure to follow Morgan warrants reversal.

1. What does Morgan mean? Morgan began its discussion of the scope of cross-examination of character witnesses by noting that the cases cited by appellant for the proposition that questions based on an assumption of guilt on the pending charges are improper involved testimony about the defendant's reputation in the community. Judge Van Graafeiland then pointed out that Rule 405(a) had broadened the prior law to permit a witness to give an opinion as to a trait of character that is in issue. He then noted that opinion testimony of "expert witnesses has traditionally been given in response to hypothetical questions based upon the evidence in the case ... and this form of questioning may properly be used on cross-examination as well as direct." 554 F.2d at 33 (emphasis added) (citations omitted). "Time and again," he continued, "experts are asked hypothetical questions which assume the very facts upon which the defendant's guilt is predicated." Id. (emphasis added). He concluded that the hypothetical question asked in Morgan "was not prejudiciously improper so as to mandate reversal." Id.

Judge Van Graafeiland then added the following:

It does not follow from this holding that we approve of the question which was asked. Because it is too early in the history of Rule 405 to predict how much use ingenious counsel will make of opinion testimony from witnesses who may qualify as experts on traits of character, we are reluctant to prescribe an evidentiary rule which will inhibit full cross-examination of any such expert. Insofar as non-expert character witnesses are concerned, however, we believe that the probative value of a hypothetical question such as the one at issue herein is negligible and that it should not be asked. The jury is in as good a position as the non-expert witness to draw proper inferences concerning the defendant's character from its own resolution of the issue. Cf. Wigmore [on Evidence] Sec. 679 [1940].

Id. at 34.

Judge Mansfield joined the panel opinion and added a concurring opinion to underscore his view of the vice of putting to a character witness questions that "asked the jury to assume the defendant to be guilty of the very charge on trial." Id. (Mansfield, J., concurring).

There is arguably an ambiguity as to the meaning of Morgan, arising from its discussion of expert and non-expert witnesses. Without categorizing the witness in Morgan as either expert or non-expert, the Court acknowledged the latitude to be accorded district judges in permitting cross-examination of expert witnesses with hypothetical questions based on facts on which the defendant's guilt is predicated, yet condemned that very type of question, which was asked of the witness in Morgan. Possibly that ambiguity led Judge Sweet to conclude that Morgan does not bar hypothetical questions, based on an assumption of guilt, asked of a witness who offers an opinion about the defendant's character. We note that the District of Columbia Circuit, though ruling that such hypothetical questions may not be asked of a character witness testifying only as to the defendant's reputation, has cited Morgan for the proposition that such hypothetical questions may be asked of a character witness offering an opinion as to the defendant's character. See United States v. White, 887 F.2d 267, 274-75 (D.C.Cir.1989).

We think Morgan is quite clear in its condemnation of guilt-assuming hypothetical questions asked of lay character witnesses like friends or neighbors, whether testifying about a defendant's reputation for a character trait or expressing an opinion about such a trait. As to such non-expert character witnesses, Morgan is emphatic that a hypothetical question based on an assumption of guilt "should not be asked." 554 F.2d at 34. We note that three circuits have correctly read Morgan as prohibiting use of guilt-assuming hypothetical questions in cross-examination of non-expert character witnesses. See United States v. McGuire, 744 F.2d 1197, 1204-05 (6th Cir.1984), cert. denied, 471 U.S. 1004, 105 S. Ct. 1866, 85 L. Ed. 2d 159 (1985); United States v. Williams, 738 F.2d 172, 177 (7th Cir.1984); United States v. Polsinelli, 649 F.2d 793, 796-97 (10th Cir.1981).

What Morgan left open is the possibility that under Rule 405(a), a trial judge might permit an expert witness, for example a psychiatrist, to offer an opinion on a character trait such as truthfulness. If Rule 405(a) might one day be read that broadly,3 see United States v. Pacelli, 521 F.2d 135, 140-41 (2d Cir.1975) (no abuse of discretion to exclude psychiatrist's testimony offered to impeach credibility of witness), cert. denied, 424 U.S. 911, 96 S. Ct. 1106, 47 L. Ed. 2d 314 (1976), the panel did not wish to foreclose the possibility that guilt-assuming hypothetical questions may be asked of the expert character witness, just as other expert witnesses, such as Internal Revenue Service agents in a tax fraud case, may be asked hypothetical questions based on facts offered to establish the defendant's guilt. United States v. Morgan, 554 F.2d at 33.

We do not doubt that to some extent a guilt-assuming hypothetical question might be probative of the credibility of testimony given by a non-expert character witness. Steadfast adherence to a favorable opinion by a witness asked to assume the defendant's guilt might provide some basis for concluding that the witness is simply supporting the defendant, rather than providing credible testimony about his character. But we share the view of all members of the Morgan panel that such cross-examination is nevertheless to be prohibited because it creates too great a risk of impairing the presumption of innocence. Moreover, after a jury has repeatedly heard a prosecutor assure a trial judge that he has a good-faith basis for asking permitted hypothetical questions, the jury might infer from the judge's permission to ask a guilt-based hypothetical question that the prosecutor has evidence of guilt beyond the evidence in the record.

In reaffirming Morgan, we note that every circuit to have considered the issue, except the District of Columbia Circuit,4 see United States v. White, supra, has agreed that guilt-assuming hypothetical questions should not be asked of character witnesses: Fourth Circuit, United States v. Siers, 873 F.2d 747, 749-50 (4th Cir.1989) (opinion and reputation testimony); Fifth Circuit, United States v. Palmere, 578 F.2d 105, 107 (5th Cir.) (per curiam) (same), cert. denied, 439 U.S. 1118, 99 S. Ct. 1026, 59 L. Ed. 2d 77 (1978); Sixth Circuit, United States v. McGuire, 744 F.2d at 1204-05 (same); Seventh Circuit, United States v. Williams, 738 F.2d at 176-77 (same); Eighth Circuit, United States v. Barta, 888 F.2d 1220, 1224-25 (8th Cir.1989) (reputation testimony); Tenth Circuit, United States v. Page, 808 F.2d 723, 731-32 (10th Cir.) (same), cert. denied, 482 U.S. 918, 107 S. Ct. 3195, 96 L. Ed. 2d 683 (1987). In some cases, use of such questions on cross-examination has resulted in reversal of a conviction. See United States v. Polsinelli, supra (reputation testimony); United States v. Candelaria-Gonzalez, 547 F.2d 291 (5th Cir.1977) (same).

2. Should Morgan have been followed in this case? Having concluded that Morgan disapproves of cross-examination of non-expert character witnesses with questions that assume the guilt of the defendant, we next consider whether Morgan should have been followed in this case. Initially, we note that there is some question whether the views of the Morgan panel and the supplemental concurring views of Judge Mansfield are dicta. The structure of the opinion suggests that they are. Judge Van Graafeiland's opinion for the Court first announces that asking the hypothetical question "was not prejudiciously improper so as to mandate reversal," 554 F.2d at 33, then observes that "[i]t does not follow from this holding that we approve of the question," id. at 34, and concludes that "it should not be asked," id. Comments following a holding are often dicta. On the other hand, the opinion can be read to "hold" that the question was improper and to make the further holding that the error in permitting the question did not warrant reversal because the question was "not prejudiciously improper."

Even if the panel's disapproval is regarded as dictum, we think it important to make clear that this is dictum that should have been followed in this case and in subsequent cases. We acknowledge that not every observation contained in an opinion of this Court deserves to be regarded as the law of this Circuit. Opinion authors frequently express thoughts peripheral to the holding of a case, and these thoughts do not bind the Circuit, nor even the concurring judges on the panel. If every phrase in an opinion were accorded binding effect, there would be a tendency either to refine language with such meticulous care as to imperil the prompt disposition of the Court's work or to reduce opinions to bare pronouncements of holdings. The latter course might be welcomed by some members of the bench and bar, but it would be inconsistent with the time-honored tradition of crafting opinions that seek not only to pronounce results but also to explain reasoning, to stimulate informed commentary, and, on occasion, to provoke future consideration of emerging issues.

Nevertheless, in some contexts expressions of views by an appellate court must be regarded as the law of the circuit, even though not an announcement of a holding or even of a necessary step in the reasoning leading to a holding. See United States v. Bell, 524 F.2d 202, 205-06 (2d Cir.1975). One such context is the clear statement of an approved or disapproved aspect of trial court procedure. The orderly administration of justice requires certainty as to many details of trial procedure, yet it would be subversive of such administration to order a reversal every time a disapproved procedure was used. In some cases we tolerate a disregard of clear prohibitions, recognizing that not every opinion of ours is etched in the memory of the trial bench or the trial bar, that unforeseen matters often arise in the courtroom so quickly and require such prompt disposition as to preclude even hurried research, and that a procedural misstep often does not impair "substantial rights," see Fed.R.Crim.P. 52(a); Fed.R.Civ.P. 61. But harmless error rules are not a license to disregard procedural constraints announced by an appellate court.

Morgan illustrates the type of appellate guidance that, even if dictum, is not to be disregarded. The guidance concerns an aspect of trial procedure, one likely to recur with frequency. Moreover, the guidance is not a tentative expression of views, but a forceful declaration: "a hypothetical question such as the one at issue herein ... should not be asked." 554 F.2d at 34. Though it has been said that the only word in a judicial opinion that prosecutors understand is "reversed," we urge them, unless they wish to see this word more often, to add to their lexicon the words "should not."

3. Does Disregard of Morgan Require Reversal? What we have just said might be expected to lead to a reversal in this case, but we conclude that a reversal is not warranted. Looking first at traditional harmless error considerations, we are satisfied that the improper cross-examination created no substantial risk of prejudice. Judge Sweet gave the jury prompt and sufficient cautionary instructions, pointing out the limited purpose for which the questioning was permitted and reminding the jurors of their ultimate responsibility to determine guilt or innocence. Moreover, the evidence of guilt was so substantial as to preclude any reasonable likelihood that the improper cross-examination contributed to the verdicts.

Nevertheless, we are troubled by the disregard of Morgan, especially by the Government's urging the District Court to permit what Morgan condemns, even after that decision was called to the Court's attention by defense counsel. Our readiness to reverse in such circumstances is diminished in this case, however, by our acknowledgement that, though we believe that Morgan is clear, the opinion is susceptible to misinterpretation. Indeed, the District of Columbia Circuit in White has interpreted Morgan to support the very cross-examination that we believe it condemns. Under the circumstances, the Government was entitled to urge an interpretation of Morgan favorable to its position, and the District Judge cannot be faulted for misreading Morgan as he did. A reversal would not be a justified response to what occurred in the trial court. We trust, however, that the matter has now been sufficiently clarified to deny the Government any reason to think that a subsequent disregard of Morgan 's admonition will be overlooked.

Oshatz and Messinger challenge the District Court's decision to admit evidence of their involvement in several other fraudulent tax schemes not charged in the indictment. The defendants contend that the Government failed to show that they participated in these uncharged acts with the knowledge and intent to defraud, and that under cases such as United States v. Afjehei, 869 F.2d 670 (2d Cir.1989), and United States v. Peterson, 808 F.2d 969 (2d Cir.1987), such acts are not to be considered "similar" to the charged crimes absent a demonstration of criminal intent. Alternatively, the appellants contend that even if the evidence were admissible under Rule 404(b) of the Federal Rules of Evidence, it was precluded by Rule 403 because its prejudicial effect outweighed its probative value.

We are satisfied that the evidence sufficed to permit the inference that the defendants knowingly participated in each of the alleged similar acts. The first similar act concerned Joseph Bachman, a stock options trader who claimed to have engaged in a separate tax fraud with Oshatz and Messinger. Bachman testified that in 1980 and 1981 the defendants helped form several limited partnerships, known collectively as the Yardley entities, for the sole purpose of providing tax deductions to investors. Bachman alleged that he had informed the defendants that the Yardley entities would engage in "fully hedged" transactions. The defendants maintain that this disclosure does not establish their knowledge of the fraud, observing that not all hedged transactions are illegal and that nothing else Bachman told them should have alerted them to the fraudulent nature of the trades. This argument, however, ignores Bachman's testimony that at a meeting with the defendants he had clarified that the transactions would involve "no risk of capital." In view of the efforts of Messinger in preparing offering memoranda for the Yardley entities and of Oshatz in soliciting investors, the evidence was sufficient for a jury to conclude that the defendants knowingly participated in a fraudulent trading scheme.

The second similar act stemmed from the defendants' relationship with David Lamb, a London commodities broker. In late 1982 Oshatz and Messinger hired Lamb to replace Markowitz as the trader for Trend Capital Associates, a partnership in which both defendants held an interest. Peter Stefanou, the accountant for Trend Capital, testified that though Lamb was not hired until sometime after September 1982, trading statements indicated that he had made substantial trades on behalf of the partnership as early as July of that year. The Government contends that these trades were fabricated and that the defendants knew of the fraudulent nature of the tax losses generated by Trend Capital. The Government established that the bulk of a half million dollars purportedly forwarded to Lamb as a margin payment had been placed in a joint bank account controlled by Oshatz and Messinger. This evidence sufficed to permit the inference that the defendants knew that these trades, for which little or no margin was posted, were fraudulently backdated.

The third similar act concerned Laurel Capital Associates, a limited partnership formed by Oshatz in December 1983. The Government introduced evidence that Oshatz did not make his allotted $51,000 capital contribution to the partnership in 1983, but that Messinger, along with several others, belatedly made this capital contribution in return for Oshatz's share of the partnership's losses. The defendants do not dispute the illegality of purchasing tax losses after they have occurred. Rather, they argue that this proof does not demonstrate that they knew the Laurel Capital trades were spurious, rendering evidence of their involvement in the Laurel Capital partnership insufficiently similar to the crimes charged in the indictment. We disagree. The appellants' knowledge may be inferred from the improbability of their repeated claims of innocent involvement in the purchase of fraudulent tax losses. See United States v. Kahan, 572 F.2d 923, 933 (2d Cir.), cert. denied, 439 U.S. 833, 99 S. Ct. 112, 58 L. Ed. 2d 128 (1978); 22 C. Wright & K. Graham, Federal Practice and Procedure Sec. 5245, at 507 (1978).

The cases cited by the appellants do not require a contrary result. In United States v. Afjehei, supra, the defendant was arrested entering the United States with a suitcase full of heroin. At trial the Government offered evidence of several prior trips that the defendant had taken. This Court reversed the defendant's conviction on the ground that this other act evidence should not have been admitted since the prior trips were "not shown to be other than innocent." 869 F.2d at 675. In United States v. Peterson, supra, the defendant was charged with possessing a stolen check made out to a third party. After the defendant denied knowledge that the check was stolen, the Government introduced evidence that the defendant had endorsed a second check that was made out to a third party. We reversed on the ground that since the Government failed to demonstrate that the defendant received the proceeds from the second check or knew that it was stolen, there was no evidence that the defendant's possession of the second check was wrongful. Unlike Afjehei and Peterson, where nothing indicated that the other acts were criminal, in this case there was ample proof that the Laurel Capital transactions lacked economic substance. No margin money was ever posted for the Laurel Capital trades, even though the partnership engaged in transactions generating close to $1 million in tax losses. This evidence, coupled with the proof of the defendants' knowing participation in the other similar acts as well as in the crimes charged in the indictment, permitted the inference that they also knew the Laurel Capital losses were fraudulently generated. See Huddleston v. United States, 485 U.S. 681, 691, 108 S. Ct. 1496, 1502, 99 L. Ed. 2d 771 (1988). Having ruled that this evidence was admissible under Rule 404(b), we reject the defendants' contention that the District Court exceeded its discretion in determining that the probative value of this evidence outweighed its prejudicial effect.

Oshatz contends that the District Court improperly excluded certain documents offered during his cross-examination of Markowitz. On direct examination Markowitz testified that he had forfeited to the Government all of his possessions, including whatever money he had acquired through his affiliation with the Monetary Group. Oshatz cross-examined Markowitz about several transfers of substantial sums of money he allegedly had made to members of his family or to family-controlled businesses. When Markowitz could not recall the transfers, Oshatz sought to introduce a number of checks Markowitz had drawn on the account of the Monetary Services Corporation, an entity owned by Markowitz and his sister. In concluding that Rule 608(b) barred the admission of this extrinsic impeaching evidence, Judge Sweet rejected Oshatz's claim that the evidence was offered not to attack Markowitz's credibility but to establish that he had a motive to testify falsely on behalf of the Government--namely, to hide assets the Government would otherwise require him to forfeit.

On appeal, Oshatz argues that Judge Sweet excluded this evidence in the mistaken belief that he had no discretion to admit it. He claims that this ruling should be reviewed not under an abuse of discretion standard but under the plenary review appropriate for the application of an erroneous legal standard. A fair review of the record, however, reveals that Judge Sweet understood, as he said, that this issue was "committed to the discretion of the trial court." In light of the wide latitude permitted Oshatz in cross-examining Markowitz and the speculative nature of the "motive" theory, Judge Sweet acted within his discretion in excluding the proffered evidence.

Oshatz also contests the admission of a summary chart prepared by the Government. This chart listed both the gross losses generated by the defendants' activities and the annual losses suffered by each entity involved in the scheme. Oshatz challenges both the Government's failure to produce, prior to trial, the computer program on which the chart was based and the District Court's conditional decision to admit the chart before ascertaining its accuracy. We have previously acknowledged the "great desirability" of making computer programs used in the preparation of a summary chart available to the defense a reasonable time before trial, see United States v. Dioguardi, 428 F.2d 1033, 1038 (2d Cir.), cert. denied, 400 U.S. 825, 91 S. Ct. 50, 27 L. Ed. 2d 54 (1970), and have stressed that the Court should independently determine whether the summary chart "fairly represent[s] and summarize[s] the evidence upon which [it] is based," see United States v. Citron, 783 F.2d 307, 316 (2d Cir.1986), modified, 853 F.2d 1055 (2d Cir.1988). In this case, though the Government did not produce the computer program before trial, Oshatz had ample time during trial to check the validity of the program and to cross-examine the Government's witness concerning errors in his calculations. And though the Court conditionally admitted the summary chart prior to determining the accuracy of its underlying figures, the Court permitted the Government to amend any incorrect figures in the chart before ultimately admitting the exhibit. Consequently, we do not believe Oshatz suffered any prejudice from the admission of the chart.

We have carefully considered the other claims of error and have concluded that they are without merit. The judgments of conviction are affirmed.

I concur in parts II, III and IV of the majority opinion, and in the judgment. However, I believe respectfully that in part I the majority imposes a rule that excludes relevant evidence, in order to avoid only in part a danger that is in any event illusory, and does so on highly debatable authority. For those and other reasons set forth below, I write separately.

A short discussion of basic principles is necessary to an appreciation of what the majority has withdrawn from a jury's consideration. After the effective date of the Federal Rules of Evidence in July 1975,1 it became permissible in federal courts under Fed.R.Evid. 405(a) to offer evidence of a person's character by testimony not only as to reputation among others than the witness, but also as to the witness' own opinion.2 Particularly as applied to evidence of good character offered by a criminal defendant, the ultimate point of both kinds of testimony is the same: to suggest to the jury that because the defendant is a person of good character--as shown either by a generally held opinion that he is, or by the witness' own opinion that he is--he would not have committed a crime, and therefore did not commit the crime charged in the indictment. See, 1A Wigmore, Evidence, Sec. 52 (Chadbourn Rev. 1974) (hereinafter "Wigmore"); 5 Wigmore Sec. 1608. The character judgment here thus has to do at its essence with whether a defendant adheres to a moral and ethical standard, and how likely it is that his behavior has conformed to that standard.

A jury evaluating the testimony of a reputation witness for the defendant in a criminal case must determine how accurately the witness has reported what others think, presumably as evidenced by what they say. Accordingly, such a witness may be cross-examined about well-founded rumors circulating in the community concerning events having nothing to do with the crime on trial in order to determine whether the witness has reported accurately a community consensus. 3A Wigmore, Sec. 988. Because rumors of the crime on trial would prove nothing about how likely it is that the defendant committed that crime, which is the ultimate point of character testimony, a reputation witness may not be cross-examined about such rumors. 5 Wigmore, Sec. 1618.

A jury evaluating the testimony of an opinion witness, however, has a different focus. That jury must determine two things: how well the witness knows the defendant, and by what standard the witness judges the defendant. Both are essential in order for the jury to weigh the testimony. If the witness does not know the defendant well, it is unlikely the witness will have seen enough of the defendant's behavior to judge his character. If the witness' judgment is distorted either by such partisanship that the witness would think highly of the defendant despite misbehavior, or by a warped ethical standard, the witness' opinion may be correspondingly discounted. A strong enough partisan would swear truthfully that the defendant is a person of good character even if he has committed the crime on trial; a witness who thinks the crime on trial is not inconsistent with good character would do the same. The question at issue in this case probes both the witness' bias and the witness' own standards by asking whether the witness would retain a favorable opinion of the defendant even if the evidence at trial proved guilt. In either eventuality--bias or distorted ethics--the excluded testimony would be bound to alter the value the jury sets on the opinion. Thus the testimony the majority withdraws from the jury is valuable and relevant in determining by what standard the witness has judged the defendant.

The majority concedes that its rule excludes relevant testimony, supra, at 539, but invokes the specter of greater evils--prejudice to the defendant from two ideas that might infect the jury: first, that the presumption of innocence is attenuated or inapplicable; and second, that the prosecutor has undisclosed evidence of the defendant's guilt. Id.

The first is utterly illogical. There is no chain of reasoning by which a rational jury could conclude that a question calling for a witness to indulge an assumption for the purpose of testing that witness' opinion invites the jury to indulge the same assumption when weighing the evidence of which that opinion is a part. Moreover, consider the setting in which the jury would succumb to this hypothesized illogic. As part of their orientation in this Court House, an orientation that is not unique either among or to federal courts, jurors are shown a film designed to impress upon them the importance of their duties. That film includes reference to the presumption of innocence in criminal cases. Also, it is a routine part of the voir dire in criminal cases to explain this presumption to jurors, to tell them they are obligated to apply it, and to exclude for cause those who cannot promise upon oath that they will apply it. Further, and most important in my view, a trial judge would instruct the jury specifically, as did the trial judge here, that the question is limited to testing the witness' views and does not bear on guilt or innocence, which is for the jury to decide. Finally, the presumption of innocence is a central and mandatory feature of the jury charge, Taylor v. Kentucky, 436 U.S. 478, 486 n. 13, 98 S. Ct. 1930, 1935 n. 13, 56 L. Ed. 2d 468 (1978), and one often alluded to by defense counsel in their summations even before the charge is delivered. To fear that a jury so oriented, so sworn and repeatedly so instructed would be impelled by the hypothetical question at issue here to the illogical conclusion suggested by the majority is to fear a phantasm.

Next, the majority urges that "after a jury has repeatedly heard a prosecutor assure a trial judge that he has a good-faith basis for asking permitted hypothetical questions, the jury might infer from the judge's permission to ask a guilt-based hypothetical question that the prosecutor has evidence of guilt beyond the evidence in the record." Supra at 539. Although no explicit reference is provided, the majority seems to be talking about questions of the "Have-you-heard" variety posed to a reputation character witness, which the court should be assured have some factual basis before the court allows them. 3A Wigmore, Sec. 988. See, Michelson v. United States, 335 U.S. 469, 481 and n. 18, 69 S. Ct. 213, 221 and n. 18, 93 L. Ed. 168 (1948); United States v. Birney, 686 F.2d 102, 108 (2d Cir.1982). However, the jury in this case never heard any such assurance from the prosecutor to the trial judge; nor would a jury in any other case hear it, because no capable trial judge would permit such assurances to be given in the jury's presence. Michelson, 335 U.S. at 481 and n. 18, 69 S. Ct. at 221 and n. 18. Again, we are dealing with a phantasm.

But even giving the majority the benefit of all the arguendos, its rule does not apply to a category of witness it believes was recognized in United States v. Morgan, 554 F.2d 31 (2d Cir.1977)--expert character witnesses. The Morgan panel, as the majority reads the opinion, "did not wish to foreclose the possibility that guilt-assuming hypothetical questions may be asked of the expert character witness, just as other expert witnesses, such as Internal Revenue Service agents in a tax fraud case, may be asked hypothetical questions based on facts offered to establish the defendant's guilt." At 539. Yet if the danger to be avoided is a virus of illogic in the jury box, what is the difference to the jury whether the question is asked of an expert or a lay witness? There is no difference, and so the proposed solution does not cure thoroughly even the imaginary problem the majority fears.

The other prop supporting the majority position--authority--is not much stronger than the policy arguments discussed above. The main authority, of course, is Morgan, and the majority expends considerable energy decrying the alleged disregard of that case--a decision the majority describes in the same sentence as both "clear" and "susceptible to misinterpretation." At 541. As set forth below, Morgan is not without ambiguity and there is another possible answer to the majority's question, "What does Morgan mean?" Supra, at 538. It bears mention, however, that even if Morgan means exactly what the majority says it means, Morgan itself contemplated at least the possibility of change in light of experience under the Federal Rules of Evidence, which were brand new when Morgan was decided. Morgan, 554 F.2d at 34. Although the reexamination suggested in Morgan was in the direction of greater restriction, there is no reason to put a one-way ratchet on experience, and every reason to reexamine the logical basis for Morgan--if Morgan is to be read as the majority reads it.

But again, that reading is not the only one possible. The character witness in Morgan testified, as Fed.R.Evid. 405(a) permits, as to both the defendant's reputation and the witness' own opinion of the defendant's honesty, integrity and truthfulness. Opinion testimony was an innovation introduced by the Rule; prior practice had permitted only reputation testimony. On cross-examination, the witness was asked if his opinion (as distinguished from the defendant's reputation) would change if he knew that facts specified in the prosecutor's questions, comprising all or part of the crime charged, were true. 554 F.2d at 32. On appeal, the defendant argued that the questions were improper inter alia because they asked the jury to assume guilt. The panel majority in Morgan first distinguished cases involving reputation character testimony from the case then at hand, which dealt with opinion character testimony:

"Here, the matter being pursued was the opinion of the witness concerning the defendant's character. When a witness is permitted to state his own opinion on a matter in issue, as he is now under Rule 405 of the Federal Rules of Evidence, some latitude in cross-examination must be allowed."

554 F.2d at 33. In order to measure how much latitude is appropriate, the Morgan majority then examined the pre-Federal Rules of Evidence practice with respect to experts, who are asked all the time to give opinions and answer hypothetical questions, on both direct and cross-examination, and concluded as follows:

"We conclude, therefore, that the asking of the hypothetical question at issue herein, based as it was upon testimony already offered, was not prejudiciously improper so as to mandate reversal. It introduced nothing into the case which was not already before the jury."

Id. However, the opinion went on to say that because the Rule was new insofar as it permitted opinion testimony, the Court was reluctant to impose an evidentiary rule "which will inhibit full cross-examination of any such expert." 554 F.2d at 34. It was during this discussion that the panel majority first introduced the concept of "witnesses who may qualify as experts on traits of character." As I read the case, that does not necessarily refer literally to experts, whose testimony is governed in any event by the rules in Fed.R.Evid. Article VII, but rather to those who testify in the manner of experts--i.e., by expressing an opinion after they have been qualified to do so by evidence of their acquaintance with the defendant, something only experts had been permitted to do before Fed.R.Evid. 405(a). Indeed, nothing in the facts of Morgan makes any other reference to experts necessary or relevant; Morgan did not involve expert testimony as that term is commonly understood and as the majority appears to use it here. Thus the statement in Morgan that a guilt-assuming question "should not be asked" of "non-expert character witnesses," 554 F.2d at 34, can be read to mean that the question should not be asked of reputation witnesses, as distinct from opinion witnesses.

Lest this reading of Morgan be dismissed as entirely idiosyncratic, I hasten to add that one of the circuit court decisions cited by the majority here in support of its position, United States v. Polsinelli, 649 F.2d 793 (10th Cir.1981), reads Morgan precisely that way, to draw a distinction between witnesses who testify in the manner of experts to personal opinion, and those who testify as non-experts to reputation, as follows:

"Stated differently, we are not here concerned with the situation faced by the Second Circuit in United States v. Morgan, discussed supra, namely cross-examination of a so-called 'expert' character witness who has testified, on direct examination, as to his personal opinion of the defendant's character. In our case the so-called 'non-expert' character witnesses only testified as to Polsinelli's community reputation. We have now held that the Government's cross-examination of these 'non-expert' character witnesses was improper and, under the circumstances, prejudicial. In thus holding, we do not intend to imply that such cross-examination would have been proper had the character witnesses expressed their personal opinion of Polsinelli's character. Resolution of that particular matter must await a different fact situation."

Polsinelli, 649 F.2d at 799. To be sure, the panel in Polsinelli, while conceding that the issue was not before it, then went on--in dictum--to express opposition to such a question posed to an opinion witness.

It bears mention also that in Lopez v. Smith, 515 F. Supp. 753, 756 (S.D.N.Y.1981), Judge Weinfeld wrote that when a character witness testified to opinion, "it was not error to allow the attempt to impeach his opinion" with a guilt-assuming hypothetical question; his authority for that proposition was a "cf." citation to Morgan. Id. n. 13.

Of the three cases the majority cites as having "correctly" read Morgan to prohibit the question at issue here when addressed to a "non-expert," supra at 539, two fail to live up to that billing: Polsinelli, as set forth above, did not read Morgan to bar the question at issue when asked of an opinion character witness; United States v. McGuire, 744 F.2d 1197, 1204-05 (6th Cir.1984), cert. denied, 471 U.S. 1004, 105 S. Ct. 1866, 85 L. Ed. 2d 159 (1985), cited Morgan and Polsinelli for the general proposition that questions to character witnesses which assume guilt of the charge on trial were improper, while pointing out that the question at issue in that case was of a different sort. Only United States v. Williams, 738 F.2d 172, 177 (7th Cir.1984) focused on the distinction between reputation and opinion witnesses and found "no reason to treat reputation and opinion witnesses differently in this regard." 738 F.2d at 177. Notably, Williams did not rely on Morgan in reaching that conclusion, but described Morgan simply as follows: "(allowing conclusory cross-examination of expert character witness, but, in dictum, disapproving similar questioning of a non-expert character witness)." The majority tells us that "every circuit to have considered the issue, except the District of Columbia Circuit ... has agreed that guilt-assuming hypothetical questions should not be asked of character witnesses." Supra, at 539. That is literally true, but the impression it creates--of widespread support for the majority's position as to opinion witnesses--seems extravagant in view of what those cases actually considered. Of the six cases cited, each from a different circuit, only Williams supports the rule the majority would impose.

The majority dispatches in a footnote United States v. White, 887 F.2d 267, 274-75 (D.C.Cir.1989), which holds that a question of the sort at issue here to an opinion character witness is entirely permissible. That decision, we are told, relied on cases that do not support the proposition for which they are cited, supra at 539, n. 4, but that conclusion is itself questionable for two reasons. First, White cited three cases: Polsinelli, supra, which read Morgan as I do to suggest a distinction between opinion ("expert") and reputation ("non-expert") character witnesses; United States v. Palmere, 578 F.2d 105 (5th Cir.1978) (per curiam ), cert. denied, 439 U.S. 1118, 99 S. Ct. 1026, 59 L. Ed. 2d 77 (1979), which reiterated the holding in United States v. Candelaria-Gonzalez, 547 F.2d 291 (5th Cir.1977) that guilt-assuming questions to reputation witnesses are impermissible; and Morgan itself. For the reasons set forth above, it is not unreasonable to read those cases as supporting a distinction between opinion and reputation character witnesses, such that the question challenged here would be permissible for the one and not for the other. Second, the majority's dismissal of White seems too casual also because that case does not simply cite authority; it is authority.

Anyone who reads Morgan literally to distinguish between lay and expert character witnesses must then consider who would qualify as an expert on the character of his fellow beings--a topic ripe with possibilities, none of them attractive. The majority seems to suggest, hesitantly but assumptively, the candidacy of psychiatrists as experts on character. Supra, at 539. The assumption seems characteristic of the faith our age reposes in science in general and psychiatry in particular; the hesitation is entirely warranted. To assume that psychiatrists, because they understand more than most people about certain forces that govern human behavior, are therefore experts in human character, is sublimely logical, but quite wrong--rather like assuming that because Albert Einstein understood more than Ted Williams about the laws of physics that govern the path of a moving sphere, he must also have been a better hitter. Psychiatrists, as psychiatrists should be the first to concede, are experts in pathology, not in morality. They are limited by their discipline to recognized patterns of mental illness. See, e.g., American Psychiatric Association, Diagnostic and Statistical Manual of Mental Disorders (Third Edition--Revised), p. xxix (1987 ed.) (hereinafter "DSM-III-R").3 Yet we have all seen criminals who are glowing portraits of mental health, and saints who bristle with neuroses.

A stock swindler or a tax cheat may act based on a recognized personality disorder, e.g. DSM-III-R 301.70, or he may act based on simple greed and the belief that he can get away with it. Psychiatrists may be able to detect mental conditions that generate criminality, and thus would qualify as experts to testify about such conditions. But that does not make them experts on human character.

That is not to say there will be any shortage of candidates for the title of expert. Among the first probably would be the clergy, who deal regularly with sin and virtue, good and evil, and the like. Of course, once we recognize the clergy as experts in character, we would have to avoid offense to the First Amendment, cf., Welsh v. United States, 398 U.S. 333, 338, 90 S. Ct. 1792, 1795, 26 L. Ed. 2d 308 (1970), by recognizing also a Coxey's army of lay moralists, ethicists, and pretty much anyone else willing to set up a booth on the fair ground--including, as a matter of professional courtesy if nothing else, judges. All in all, this is not an appetizing prospect.

The simple truth is that the notion of expert character testimony, although it has been bandied about in the literature for some time, compare, e.g., Curran, "Expert Psychiatric Evidence of Personality Traits," 103 U.Pa.L.Rev. 999, 1013 (1955) (endorsing such evidence) with, Falknor and Steffen, "Evidence of Character: From the Crucible of the Community to the Couch of the Psychiatrist," 102 U.Pa.L.Rev. 980, 994 (1954) (criticizing such evidence), is one of those ideas whose time has not yet come, and with common sense and a modicum of luck it never will. Rule 702 of the Federal Rules of Evidence provides that, "[i]f scientific, technical, or other specialized knowledge will assist the trier of fact to understand the evidence or to determine a fact in issue, a witness qualified as an expert ... may testify thereto." As this Court wrote in Andrews v. Metro North Commuter R. Co., 882 F.2d 705, 708 (2d Cir.1989), "For an expert's testimony to be admissible under this Rule, however, it must be directed to matters within the witness' scientific, technical, or specialized knowledge and not to lay matters which a jury is capable of understanding and deciding without the expert's help." Character is a subject on which jurors do not need and therefore should not get expert guidance. United States v. Webb, 625 F.2d 709, 710-11 (5th Cir.1980). To suggest otherwise is at best unhelpful.

Finally, while fashioning an absolute rule about what sort of cross-examination should be permitted at a trial, the majority itself appears to have disregarded a bit of guidance on the wisdom of such appellate rule-making that the Supreme Court provided in Michelson, supra, albeit in the form of dictum. That case and its history are instructive. The case originated in this Circuit, and appears to have been heard by the Supreme Court at least in part at the request of the author of the Circuit opinion. That author, Judge Clark, urged that the rule permitting a reputation character witness to be cross-examined about other bad acts of the accused in order to test the witness' grounds for knowledge, should be changed. He doubted that trial judges could fashion effective instructions:

"Because, as Wigmore says, the jury almost surely cannot comprehend the judge's limiting instruction, the writer of this opinion wishes that the United States Supreme Court would tell us to follow what appears to be the Illinois rule, i.e., that such questions are improper unless they relate to offenses similar to those for which the defendant is on trial."

United States v. Michelson, 165 F.2d 732, 735 n. 11 (2d Cir.1948). The Supreme Court took the case, but left the prevailing rule undisturbed. It wrote that despite criticism, the law as it stood "has proved a workable even if clumsy system when moderated by discretionary controls in the hands of a wise and strong trial court." 335 U.S. at 486, 69 S.Ct. at 223. As the majority opinion illustrates, those discretionary controls worked well here; lest anyone miss the point, the trial court made the obvious explicit by instructing the jury specifically on the limited purpose for which the evidence at issue was received. At 538, 541; App. 1422, 1433, 1461. Those discretionary controls should not be abandoned in favor of an inflexible rule. Justice Frankfurter, concurring in Michelson, put the matter even more forcefully:

"Despite the fact that my feelings run in the general direction of the views expressed by MR. JUSTICE RUTLEDGE in his dissent, I join the Court's opinion. I do so because I believe it to be unprofitable, on balance, for appellate courts to formulate rigid rules for the exclusion of evidence in courts of law that outside them would not be regarded as clearly irrelevant in the determination of issues. For well-understood reasons this Court's occasional ventures in formulating such rules hardly encourages confidence in denying to the federal trial courts a power of control over the allowable scope of cross-examination possessed by trial judges in practically all State courts. After all, such uniformity of rule in the conduct of trials is the crystallization of experience even when due allowance is made for the force of limitation. To reject such an impressive body of experience would imply a more dependable wisdom in a matter of this sort than I can claim.

"To leave the District Courts of the United States the discretion given to them by this decision presupposes a high standard of professional competence, good sense, fairness and courage on the part of the federal district judges. If the United States District Courts are not manned by judges of such qualities, appellate review, no matter how stringent, can do very little to make up for the lack of them."

335 U.S. at 488-89, 69 S. Ct. at 224-25.

For the foregoing reasons, as to part I of the majority opinion, I concur only in the judgment.