On February 4, 1991, following a four-month jury trial in the Southern District of New York before Judge Lowe, Bernhard Fred Manko and Jon Edelman were convicted of making and subscribing false tax returns in violation of 26 U.S.C. § 7206(1), aiding and abetting in the preparation of false returns in violation of 26 U.S.C. § 7206(2), and conspiring to violate the tax laws and to defraud the United States in violation of 18 U.S.C. § 371. On November 25, 1991, the district court sentenced each defendant to five years imprisonment and fined each $450,000. On appeal, Manko and Edelman raise a host of objections to the district court's conduct of the trial. In addition, Manko and Edelman assert that the government failed to produce sufficient evidence to sustain their convictions and that prosecutorial misconduct denied them a fair trial.

Manko and Edelman were professionals in the once thriving field of tax shelters. In the late 1970s and early 1980s, the two sold interests in partnerships that engaged in complex commodities straddles. These straddles had the effect of creating losses in one tax year and a corresponding gain in the subsequent year, thereby deferring income. See Gardner v. Commissioner, 954 F.2d 836 (2d Cir.) (per curiam), cert. denied sub nom. Falk v. Commissioner, --- U.S. ----, 112 S. Ct. 1940, 118 L. Ed. 2d 546 (1992).

The straddle business evaporated when Congress passed the Economic Recovery Tax Act of 1981 ("ERTA"), Pub.L. 97-34 (1981). ERTA eliminated the tax benefits of straddles by requiring the taxpayer to recognize both the gain and the loss generated by the straddle in the same tax year. Manko and Edelman began to look for new tax deferral techniques.

Eventually, Manko and Edelman devised a way around ERTA. They determined that by engaging in repurchase transactions ("repos") in United States Treasury Bills ("T-Bills"), they could create interest deductions in the first tax year and corresponding gains in the second year, with little or no risk. A T-Bill is an obligation of the United States Government to pay a fixed amount (the face value) at a certain date in the future (maturity). T-Bills are purchased at a discount from face value. The interest is determined by the difference between the face value and the purchase price. A repurchase agreement is an agreement to sell a T-Bill and then repurchase the bill at a later date for a predetermined amount. In effect, a repurchase agreement is a loan in the amount of the proceeds of the original sale, collateralized by the T-Bill, with interest equal to the difference between the sale and repurchase prices. A repurchase agreement may expose the parties to market risk, since the agreed-upon repurchase price may be greater or less than the market price of the T-Bill at the agreed-upon date. A repo to maturity exists when the repurchase date specified in the agreement is the maturity date of the T-Bill. In a repo to maturity, there are no economic variables, since the market value of the T-Bill at maturity is always equal to the face value of the T-Bill.

Repos created tax advantages because, until the tax laws changed in 1984, a securities dealer could deduct the interest payments on the loan as they accrued, but only had to report the corresponding gain from the appreciation of the T-Bill at maturity. If the T-Bill matured in the tax year after the repurchase agreement was made, the taxpayer would be able to defer income equal to the amount of interest accrued in the first year.

One problem with using repos to create tax deductions is that the deduction created in the first year is relatively small compared with the face value of the T-Bill. A repo on a ten million dollar T-Bill would only create a few hundred thousand dollars of interest deductions. Since Manko and Edelman had raised over $25 million from investors with a promise of generating four times that amount in tax deductions, they realized they would have to do huge quantities of repos to maturity.

At trial, the government asserted that Manko and Edelman hatched a fraudulent scheme in order to achieve such volumes. The government alleged that Manko and Edelman appropriated a dormant shell company, TSM Holding Company ("TSMH"), and unilaterally wrote documents to reflect billions of dollars of repos to maturity between TSMH and two clearing brokers controlled by Manko and Edelman--Government Clearing Company and Government Securities Trading Corporation (collectively "GCC"). In turn, Manko and Edelman created documents to reflect mirror transactions between GCC and several investment partnerships, thereby passing on the ersatz tax breaks to numerous unwitting investors. Manko and Edelman represented to their lawyers and accountants that TSMH was an independent entity and that all their transactions with TSMH actually occurred and carried market risk.

The documents introduced at trial described the trades as follows. First, TSMH would sell a T-Bill through GCC to the partnerships. Since TSMH was only a shell company without assets, TSMH would sell the bill short. In a short sale, a party sells a security that it does not own and must therefore find a way to deliver the security to the buyer, such as borrowing it from a third party. Since TSMH sold as much as a billion dollars worth of T-Bills at a time, the partnerships lacked the funds to pay for the T-Bills. Accordingly, the partnerships would borrow the funds from GCC by entering into a repo to maturity with GCC secured by the T-Bill received from TSMH in the short sale. GCC, of course, did not have a spare billion dollars lying around to lend to the partnerships. Instead, GCC would borrow the funds from TSMH with a repo to maturity secured by the same T-Bill that TSMH had sold short to the partnerships and that the partnerships had loaned to GCC.

The return of the T-Bill to TSMH as collateral cancelled TSMH's obligation to deliver the T-Bill promised in the short sale. Therefore, at the end of the day, GCC would owe TSMH the proceeds of the loan plus interest and TSMH would owe GCC a T-Bill. Both obligations would come due at maturity. GCC would have mirror obligations with the partnerships. The terms of the repurchase agreements between GCC and TSMH exactly mirrored the terms of the agreements between GCC and the partnerships. At the end of the first tax year, GCC would "roll up" the outstanding interest payments on the loan by unilaterally adding them to the principal amount of the loan. The mirror transactions between GCC and the partnerships would proceed with a similar interest roll up, thus allowing the partnerships to take the tax deductions.

GCC set the interest rates on both loans so that the loan proceeds plus interest would exceed the face value of the T-Bill by a set amount--$50 per million at the outset. This money, in effect TSMH's fee for providing GCC with a tax deduction to pass on to the partnerships, was the only thing actually to change hands in the transactions.

The government presented a substantial amount of evidence that the transactions described above never took place--that GCC engaged in typing, not trading. First, from July of 1982 to September of 1982, there was no one at TSMH with whom Manko and Edelman could negotiate the deals. Vincent Tese and James Sinclair (former partners of Manko) set up TSMH in June of 1982. Neither, however, ever injected any capital into TSMH or used the company in any way. Both denied any knowledge of the transactions between TSMH and GCC. During this period in which nobody controlled TSMH, Manko and Edelman had their trading room staff write up trade tickets that made it appear that TSMH had sold $500 million in T-Bills to GCC and loaned GCC the money to purchase them. GCC prepared but did not send confirmations reflecting the terms of these trades. Edelman admitted that GCC determined the size of the transactions and set the interest rate on the repo without negotiating with TSMH.

In October of 1982, perhaps in an effort to provide more substantial documentation of the "trades," defendant Manko hired Richard Bandel, a former employee of GCC, to work at TSMH. At Manko's instruction, Bandel picked up the confirmations that GCC had printed and logged into accounting ledgers all the trades that had been written up to that point. With Bandel in place at TSMH, GCC began sending trading confirmations to TSMH, and Bandel logged these transactions into TSMH's books. The defendants admitted that even with Bandel in place, GCC did not negotiate the terms of the transactions with TSMH.

At about the same time that Manko placed Bandel at TSMH, Tese and Sinclair transferred TSMH, free of charge, to Robert Maurus and Charles Genovese. Maurus testified that he did not even learn that Bandel worked at TSMH until March of 1983. Genovese testified that Manko informed him that TSMH needed Bandel in order to make TSMH "look like a real company."

After the transfer, Maurus and Genovese remained largely unaware of the details of the alleged transactions. On redirect examination, Maurus conceded that he understood nothing about the transactions except that GCC wrote up the paper and TSMH received a fee. It was not until the Spring of 1983 that TSMH began to send reciprocal confirmation slips to GCC. By the time these confirmation slips were sent, most of the underlying T-Bills had already matured and the parties' obligations had already been settled. Genovese testified that these confirmation slips were sent for "cosmetic" reasons.

The pace of the transactions accelerated in November of 1982. Faced with the need to create almost $100 million in deductions by the end of the year, Edelman ordered John Mann, the chief trader for GCC, to write up $500 million and $1 billion transactions with TSMH. Edelman instructed Mann to set the price without contacting TSMH. Mann refused, with the explanation that he did not participate in transactions in which he did not speak to the other party. Eventually, Mann agreed to limited participation in the transactions, informing other trading room employees of the market price for T-Bills so that the correct amounts could be written on the trade tickets.

Right after the first billion dollars was decided on, it was just quiet right after, and then Joe Bryant [a trading room employee] said "this is the one that's going to get them." And John Mann and [another employee] and myself all agreed to that, and a few minutes later John Mann said, "it's gone too far now. The boys have gone too far."

At other times, trading room employees referred to the billion-dollar trades as "the straw that's going to break the camel's back." Employees also discussed the financial reward for turning Manko and Edelman in to the IRS.

After the first three days of the billion dollar trades, Mann informed Edelman that he "had legal reservations about these trades and didn't want to do them anymore." Mann testified that Edelman replied that the trades were "none of [Mann's] concern" and "that it was not [Mann's] job to be policeman for the world."

In fifteen days in November of 1982, GCC wrote $25 billion worth of T-Bill repurchase transactions with TSMH. After that, the pace of the scheme slackened, but GCC continued to write repurchase transactions until the tax laws changed in 1984. At the time of the billion dollar trades in November of 1982, TSMH had $33 in cash, $50 in office expenses, and no telephone bills. TSMH did not even have a bank account until the end of that month.

Eventually, Barbara Kanter, Edelman's personal assistant, opened an account for TSMH. Kanter then transferred millions of dollars into and out of TSMH's account. The defense alleged at trial that this money reflected interest payments from GCC to TSMH. However, Edelman acknowledged on cross-examination that he could not correlate any particular payment to interest owed on any loan. When Maurus and Genovese learned in the spring of 1983 that Kanter had been transferring money into and out of their account, they made no efforts to stop her. Genovese explained at trial that he felt the money did not belong to TSMH.

Faced with this powerful evidence that the purported transactions were shams that never took place, the defense contended at trial that TSMH had granted GCC discretion to make trades on TSMH's behalf, and thus that there was no need for GCC to seek TSMH's approval of the terms of the repo transactions. The defense never presented any direct evidence of such an agreement apart from Edelman's testimony that Manko told him that TSMH had granted GCC discretion to make trades.

The defense further argued that there was nothing strange about a corporation with no assets making billions of dollars of loans because it was common in the government securities markets for undercapitalized entities to engage in highly leveraged, low-risk transactions.

The trial also proceeded on a second front. The government argued that even if the jury were to determine that the transactions actually occurred, the transactions were devoid of risk and thus lacked economic substance. Accordingly, the government contended that the transactions could not form the proper basis for tax deductions. The defense countered that so long as the repurchase loan represented a real debt, interest paid on that debt could be deducted even if the transactions as a whole lacked economic substance. The court sided with the government and charged the jury that it could convict if it found either that the transactions did not occur or that they lacked economic substance.

The trial in this extremely complex matter took four months. At the close of trial, the jury convicted Manko and Edelman on all counts. After the jury returned the verdict, the district court gave the jury a supplemental interrogatory that asked the jury to indicate whether it concluded that the transactions never took place or that the transactions lacked economic substance. After deliberating briefly, the jury announced that it had determined that the transactions never took place. This appeal followed.

On appeal, Manko and Edelman principally challenge the district court's handling of the trial. They contend that the court questioned witnesses in a way that suggested disbelief of the defendants' position, improperly held an ex parte conference with the prosecutor near the end of the trial, permitted the prosecutor to make improper comments in summation, and issued confusing instructions to the jury. Defendants suggest that the combined force of these asserted infirmities denied them a fair trial.

The district judge took an active role in the trial of this case, asking numerous and probing questions of witnesses. Such conduct is unquestionably proper--indeed, often required--in extremely complex cases. United States v. Gurary, 860 F.2d 521, 527 (2d Cir.1988), cert. denied, 490 U.S. 1035, 109 S. Ct. 1931, 104 L. Ed. 2d 403 (1989); United States v. DiTommaso, 817 F.2d 201, 221 (2d Cir.1987); United States v. Pisani, 773 F.2d 397, 403 (2d Cir.1985). However, such questioning carries with it the risk that the judge will appear partial to one side or the other. In order to avoid the appearance of partiality, a trial judge should limit questioning to inquiries necessary to clarify ambiguities, correct misstatements, or obtain information necessary to make rulings. United States v. Mazzilli, 848 F.2d 384, 388 (2d Cir.1988).

The fact that a judge asks a question the purpose of which is not so limited will not always require reversal. Judges are human, and it is inevitable that in the course of long, complex, and hotly contested trials, a judge may on occasion react intemperately. A defendant is not entitled to a perfect trial, only a fair one. United States v. Robinson, 635 F.2d 981, 984 (2d Cir.1980), cert. denied, 451 U.S. 992, 101 S. Ct. 2333, 68 L. Ed. 2d 852 (1981). Our ability to discern from a cold record whether particular statements of a trial judge rendered the trial unfair is quite limited. "[I]t is difficult to glean tone of voice, whether sarcastic or neutral, facial expression, whether reflecting disbelief or sober thoughtfulness, or physical demeanor, whether self-contained or expressing opinion and emotion. Therefore, it is only after an examination of the entire record that we can come to a conclusion about the conduct of the district court." Mazzilli, 848 F.2d at 389 (2d Cir.1988).

The defendants point to several incidents in the trial that we quite frankly find disturbing. Nonetheless, after a careful review of the entire record, we conclude that the district court's conduct did not deny defendants a fair trial.

1. The District Court's Questioning of Jon Edelman

A district court must show particular restraint in questioning a criminal defendant during the defendant's direct testimony. At this critical phase of the trial, the court must scrupulously insure that its questions do not indicate that the court doubts that the witness is telling the truth. Mazzilli, 848 F.2d at 388-89; United States v. Victoria, 837 F.2d 50, 55 (2d Cir.1988). Impeaching the defendant is the job of the prosecution, not the court.

Here, the district court vigorously interceded in Edelman's testimony. Most of the court's questioning took the form of inquiry into precise details of the transaction. The court repeatedly sought to clarify whether GCC had made certain payments to TSMH in cash or by book entry. While the relevance of this distinction to whether the transactions took place or had economic substance is not entirely clear, our view upon reading the record is that the court's questions were aimed at pinning down this factual issue, not at casting doubt on defendant Edelman's credibility. Accordingly, we think this portion of the court's questioning of Edelman was proper.

We are more troubled by the court's questioning of Edelman with regard to financing arrangements between GCC and the investor partnerships. At one point, Edelman compared GCC's willingness to enter into a repurchase agreement with an investor partnership to allowing a customer to purchase securities on margin. The court asked a series of questions that undermined the analogy:

[Edelman]: Yes, there is financing here and the best analogy I could make simply is to a financing on a margin account where you might have 100 shares of IBM you paid for, no money in your account, but fully paid for and you want to buy another 100 shares. You can do that.

The Court: Yes, you can, but if it's called you are expected to pay.

Edelman: If there is a margin call.

The Court: Yes.

[Edelman]: And the same thing is here, if my equity declines--

The Court: At any time in these whole 2 years we are talking about, did GCC or TSM ever call upon any of these partnerships to put money on any of these financing transactions?

[Edelman]: Can I answer that generically first?

The Court: Any way you want to.

[Edelman]: When an account at GCC has deficit equity, that is, that its value is less than the amount that it has borrowed, it has deficit equity, then GCC would be in a position to ask for more money. It may ask for more money, it may not.

The Court: That isn't my question. This is 1990. We are seven years down the road of the last transaction. In any of those two years, '82 or '83, did GCC or TSM ever ask any of these partnerships to put up money such as you were talking about a margin call in a stockbroker's account?

[Edelman]: With respect to the financing repo to maturity transactions, no other transactions?

The Court: Yes.

[Edelman]: I don't think so. GCC would be the person who would ask for money and I don't recall any situation where that was required.

[The Court]: That is the question I was asking.

By forcing Edelman to concede that GCC never made a margin call, the court discredited Edelman's analogy to ordinary brokerage accounts. By undermining Edelman's testimony in this fashion, the court may have cast some doubt on Edelman's credibility. We note, however, that the court's questioning related to a collateral issue and that Edelman's lawyer, upon resuming questioning, elicited testimony explaining that GCC would never need to make a margin call because the securities held as collateral for the repos, the T-Bills, were by their very nature always increasing in value as maturity approached. Thus, any doubts raised by the court's questions were quickly dispelled. While the court should not have challenged Edelman's analogy, since the challenge reflected as much on credibility as it did on factual clarification, we do not think this incident undermined the fairness of the trial.

We are also disturbed by the district court's conduct during the cross-examination of Richard Bandel, who testified for the government. The government's theory was that Bandel had been placed in TSMH by Manko to record the paper transactions in TSMH's books and that Bandel had no substantive role in determining the dimensions of the transactions.

The defense sought to discredit the government's view of Bandel's role in several ways. First, the defense asked Bandel to explain his testimony that Manko had hired him at TSMH in light of Bandel's prior statement in an affidavit that someone other than Manko had hired him to work at TSMH. Bandel, with some prodding from the district court, indicated that he stood by his testimony in court and that his present recollection was more accurate than that found in his prior affidavit. The defense also asked Bandel about his view of the transactions, perhaps in an effort to show that Bandel had some role in negotiating the deals:

Q. [The Prosecutor] asked you about the large trades that occurred in November of 1982.

A. Yes, he did.

Q. And it's your testimony that you viewed those trades no differently than any others?

A. Just that I was told--

The Court: By whom?

The Witness: By Joe Bryant, usually the day before I would get the ticket, that there were large tax trades going through.

Q. And it's fair to say the word "tax trade" was used openly?

A. Yes.

Q. And that based on your experience on Wall Street that many trades had a tax component?

Mr. Robinson: Objection.

The Court: This man was a scrivener, he wasn't a policy maker.

While prosecutor Robinson's objection was well taken, in that Bandel had not been qualified as an expert, the court's observation that Bandel "was a scrivener" was unnecessary and prejudicial. The government's theory was that Bandel was a scrivener, and evidence in the record supports this view. The district court, however, should not have endorsed the government's position.

The court repeated the error moments later:

Q. Did you ever tell Mr. Maurus or Mr. Genovese that you had any problems with these trades?

The Court: He has told us that when he found there might be errors, he spoke to Mr.--who did you speak to about the errors in calculating the fees?

The Witness: The financing, Joe Bryant.

The Court: And Mr. Bryant and you got together and corrected whatever errors?

The Witness: Yes.

Q. My question was, did you ever tell Mr. Maurus or Mr. Genovese, when they took control of TSM Holdings, that you had any problems or concerns about these trades?

The Court: What kind of problems are you talking about? The man was a scrivener. That's the third time I repeated it. He knew nothing about the underlying transaction, and you're not going to ask questions to give the inference that he did when he has told you repeatedly he did not know.

The court thus explicitly and repeatedly endorsed the government's view of Bandel's role. Such a breach of impartiality is a matter of grave concern.

We cannot, however, conclude that the court's conduct denied Manko and Edelman a fair trial. First, defendants did not argue to the jury that Bandel negotiated the transactions. In closing argument defendants all but adopted the court's characterization, impeaching the accuracy of Bandel's testimony by stressing that after leaving Wall Street, Bandel went to work as a part-time starter at a golf course. Thus, defendants do not really dispute the accuracy of the scrivener label.

Second, the court properly instructed the jury that the jury's view of the evidence controlled, regardless of what the court said:

With respect to factual matters, it is your recollection that controls. Nothing that counsel for either the government or the defendant may have said with respect to matters of evidence, either in their openings or in their summations, is evidence in this case; nor are any remarks that I may make during the course of this charge, or have made during the course of the trial--that is not evidence.

* * * * * *

I remind you again that nothing I may have said, even my questions to witnesses during the trial when they were on the stand, those questions should be given no undue weight for their sole purpose was to clarify and aid you in understanding what the content of the testimony was to be. So you are to place no undue weight, one way or another, on the fact that I may have asked questions of witnesses.

In light of this extensive instruction and the fact that the court's statement concerned what proved to be an uncontested issue, we do not think that the court's error denied Manko and Edelman a fair trial.

In sum, our review of the transcript confirms that on a few occasions the court crossed the line in questioning witnesses. In the context of a four month trial, we think the court's conduct did not indicate to the jury that the court believed that the prosecution should win. Accordingly, we reject defendants' contention that the trial court's conduct denied them a fair trial.

Towards the end of the government's summation, one prosecutor initiated a conference with the district court, without the defense team. The conference was transcribed, but the district court sealed the record of the conference and refused to unseal the transcript until after the government's rebuttal summation. In this ex parte conference, the prosecutor advised the district court of his plan to use a chart in the rebuttal summation:

The Court: What is it you have handed me, Mr. Cleary?

Mr. Cleary: Judge, I just handed you a document I typed up headed "Excerpts of testimony." It lists witness's names and either quotations or paraphrases of their testimony. It's what I propose to use. It's a mockup of a chart I propose to use in my rebuttal summation.

I bring this to you ex parte because I didn't want to let the defense know at this time before they sum up exactly what I am proposing to do in my rebuttal and I did want to apprise the court that I am planning on doing this so that your Honor could allot whatever time is necessary between the end of the last defense summation and the beginning of the rebuttal summation should there be objection to this. Because at that time I propose to give the defense a copy of this mockup which contains references to the page of the transcript.

I should add, Judge, if you think it's more appropriate to disclose this information to them now, although I prefer not to I, of course, would go along with that.

The Court: No, I don't think you should disclose it. What I will do is I will check the places that you expect to use quotations or summations of testimony with the record so I will be in a position to rule if there are objections to your use.

Mr. Cleary: Thank you, Judge.

The Court: All right.

At the close of the defense summation, the government informed the defense of its intent to use the chart on rebuttal, and the court allowed the defense forty-five minutes to review the twenty-eight quotations and paraphrases in the chart for accuracy. Defendants objected to the use of the chart and requested surrebuttal, but did not point to any specific inaccuracies in the chart. The court overruled defendants' objection and denied the request for surrebuttal. On appeal, the defendants assert that the ex parte conference denied them due process.

We note at the outset our inability to fathom why the prosecutor would jeopardize a four month trial with a wholly unnecessary ex parte conference. Such conferences "should not be conducted except in extraordinary circumstances." United States v. Walsh, 700 F.2d 846, 858 (2d Cir.), cert. denied, 464 U.S. 825, 104 S. Ct. 96, 78 L. Ed. 2d 102 (1983). No such extraordinary circumstances are present here.

The government lamely contends that the conference was for a proper purpose--to advise the district court of the government's plan to use a chart so that the court could allot proper time for argument between the defense summations and the government's rebuttal summation. But if that were all that the government wished to accomplish in the meeting, then there would have been no need to show the court the chart the government planned to introduce. The government could have indicated, with defense counsel present, that it likely would have a chart in rebuttal which would require prior review. The only purpose for showing the court the precise quotes the government wished to use was to enable the court to review the chart for accuracy before the defense had an opportunity to present any arguments to the court. In short, the prosecution initiated this ex parte proceeding in a blatant attempt to win a tactical advantage, to get a "leg up." Unhappily, the district court went along with the prosecution.

That this ex parte conference was improper and unnecessary does not, of course, establish reversible error. United States v. Persico, 349 F.2d 6, 13 (2d Cir.1965). Rather, "[t]he fact that such a meeting occurs ... does not require reversal if it does not affect the fairness of the trial." Walsh, 700 F.2d at 858; see also United States v. Moten, 564 F.2d 620, 630 (2d Cir.), cert. denied, 434 U.S. 959, 98 S. Ct. 489, 54 L. Ed. 2d 318 (1977).

Defendants do not point to any inaccuracies in the chart. Nor do defendants identify any way in which the chart went beyond the scope of proper rebuttal. Cf. United States v. Giovanelli, 945 F.2d 479, 493-96 (2d Cir.1991) (Newman, J., concurring). Defendants' contention that a chart prepared before the defense summation is necessarily beyond the scope of rebuttal is unpersuasive. The quotations reflected on the charts were all central to the case. The government could easily predict that defense summation would tread the same ground as the quotations. Thus, we are convinced that the district court acted properly in allowing the government to use the chart.

Of course, the propriety of using the chart cannot be the measure of whether the ex parte proceeding constituted reversible error. Otherwise, the fact of an ex parte proceeding would be irrelevant so long as the court ultimately made the right ruling. Such a rule would not properly reflect our deep distrust for one-sided proceedings. Here, however, two facts convince us that reversal is not required. First, because the chart was clearly accurate and within the scope of rebuttal, there is no reason to believe that the district court would have ruled differently absent the ex parte conference. Moreover, even if the conference did influence the judge's decision to allow the chart, in the context of this four month trial, we doubt that this chart highlighting damaging testimony played even a minimal role in influencing the jury verdict. The prosecution presented substantial evidence that the transactions with TSMH never occurred. The chart at issue did not present any new information or any information that could not have been conveyed easily to the jury without the chart. Accordingly, we conclude that this highly improper ex parte conference does not constitute reversible error.

Defendants raise a series of objections to the court's instructions to the jury. Principally, defendants argue that the court should have given all of the proposed charges submitted to the court by the defense on the afternoon of the charging conference. Most of the proposed charges concerned tangential legal issues that the government never disputed. In this lengthy trial, the court properly avoided burdening the jury with this extraneous information. Accordingly, we find most of defendants' contentions meritless. Defendants do raise three objections to the charge that warrant discussion.

The defense argued at trial that the transactions between TSMH and GCC were real and binding even if TSMH did not grant discretion to GCC to perform trades, because TSMH subsequently ratified GCC's trading. The district court, as part of a general charge on defense contentions, instructed the jury that "the defendants contend that TSM Holding ratified all transactions by subsequent conduct." The defense requested an additional instruction defining ratification. The court refused to give the proposed instruction. Defendants assert this refusal constitutes reversible error. We disagree.

A court may properly refuse to give a charge that fails to accurately state the law. United States v. Paccione, 949 F.2d 1183, 1200 (2d Cir.1991), cert. denied, --- U.S. ----, 112 S. Ct. 3029, 120 L. Ed. 2d 900 (1992). Here, defendants' requested charge stated that TSMH could ratify the acts of GCC "either by sending back confirmations to GCC, by failing to object to the transactions after TSMH received confirmations from GCC, or by any other conduct, such as recording the transactions as if they were done, indicating an intention to be bound by the transactions." The problem with this proposed charge is that under New York law, "[a]n agent's acts that are not valid unless performed pursuant to writing can only be ratified in writing." Weston Associates, Inc. v. Niagara Properties, Inc., 130 A.D.2d 964, 965, 516 N.Y.S.2d 381, 382 (4th Dept.1987). Here, the acts of the purported agent, GCC, contracting for the purchase of securities, would not be valid absent "some writing signed by the party against whom enforcement is sought." N.Y.U.C.C. § 8-319(a). Accordingly, TSMH could ratify GCC's authority only through a writing. The defendants' proposed instruction suggesting that TSMH could ratify through any conduct indicating an intent to be bound by the transactions failed to state New York law accurately. The district court correctly refused to give the proposed instruction.

Defendants claim that they were prejudiced by a reversal by the trial court during the course of the trial on the issue of the proper definition of economic substance. On the eve of trial, the court ruled that if the transactions were not "economically substantive indebtedness," the interest would not be deductible. Defendants claim that the trial court, at this time, rejected a two-part test for economic substance advanced by the government, instead adopting an open definition of economically substantive indebtedness, but then reversed itself in its charge to the jury by adopting the two-part test previously rejected. The two-part test with which the court charged the jury stated, in accordance with our decision in United States v. Atkins, 869 F.2d 135, 139-40 (2d Cir.), cert. denied, 493 U.S. 818, 110 S. Ct. 72, 107 L. Ed. 2d 39 (1989), that a transaction lacks economic substance if it lacks a reasonable possibility of profit and lacks market risk. Defendants contend that the ambiguous term "economically substantive indebtedness" adopted on the eve of trial would encompass purely formal indebtedness, and that they relied on this notion in presenting their case. The defendants claim that this reversal constitutes a violation of the law of the case doctrine and they were prejudiced by it, since they relied on the earlier ruling in formulating their trial strategy.

While a trial court's reversal of its own ruling during the course of a trial may raise serious issues of prejudice, see United States v. Uccio, 940 F.2d 753, 758 (2d Cir.1991), we doubt whether such concerns are raised here. Our review of the trial record suggests that, at the outset of the trial, the judge's ruling was limited to acknowledging that the question of economic substance was an issue in the case after first saying off the record that it would not be an issue. The judge did not provide the parties with a precise definition of economic substance until her instructions to the jury, when she employed the two-prong test of Atkins. On the eve of trial, the judge simply noted that the defense objected to the government discussing profit motive and market risk, but stated to the parties that in establishing whether or not there was economically substantive indebtedness, the parties may present "whatever factors you believe are relevant in order to reach that legal conclusion...." In light of this, the eventual jury charge defining economic substance as profit potential or market risk does not appear to us to constitute a reversal of a prior decision.

In any event, we need not decide whether there was a prejudicial reversal of a prior ruling of law by the trial court. In the special interrogatory, the jury found that the transactions "were fictitious and did not take place." That finding rendered the jury's understanding of the term "economic substance" irrelevant to the outcome.

Defendants contend that the court's charge implied that prearranged transactions are necessarily fictitious. Defendants argue that since all repurchase transactions are in some sense prearranged--in that the borrower sells the security to the lender and agrees in advance to repurchase the security at a later date for a set price--the court's charge effectively forced the jury to conclude that the transactions were fictitious.

Defendants focus particularly on the court's instructions with respect to the knowledge element of the filing false returns charge. The court advised the jury that the government did not have to prove actual knowledge. Instead, the court stated:

you may find that a defendant knew that the transactions were a sham or without economic substance if you find, 1, that the defendant you are considering was aware of a high probability that the transactions did not take place or were prearranged and, 2, that the defendant acted with deliberate disregard for how or whether those transactions were carried out with a conscious purpose to avoid learning the truth on this subject. (emphasis added)

The defendants argue that this charge equated transactions that "did not take place" with transactions that "were prearranged."

However, we think the more natural reading of the above passage is that it equates transactions that "did not take place" with transactions that "were a sham" and equates transactions that "were prearranged" with transactions "without economic substance." Viewed in this light, we doubt the court's instructions prejudiced defendants. Elsewhere in the instructions the court advised the jury that "transactions may be arranged in an attempt to minimize taxes" and expressly stated that in order to lack economic substance, a transaction must "lack[ ] a reasonable possibility of profit apart from tax deductions, and ... lack[ ] market risk."

More importantly, even if the charge confused the jury on this point, we are unable to see how this confusion prejudiced defendants, for the same reason that we find that defendants could not have been prejudiced by the trial court's alleged reversal on her instructions on economic substance. As noted above, the jury expressly found that the transactions "were fictitious and did not take place." This response to the special interrogatory removed the economic substance issue from the case. Any lingering confusion about whether prearrangement was synonymous with a lack of economic substance was irrelevant to the verdict. Accordingly, we reject defendants' challenges to the jury instructions.

Finally, defendants assert that the government failed to produce sufficient evidence to support the conviction and that prosecutorial misconduct in summation denied them a fair trial. We have considered these and the remainder of defendants' arguments and find them to be without merit.

The trial in this case challenged the resources of the district court. Through the months of testimony and the thousands of documents, the district court grappled with vigorous and experienced defense counsel and diligently struggled to ensure that the jury could follow the evidence and render a fair verdict. Not unexpectedly in a long trial, the court made some mistakes, and defendants did not receive a perfect trial. However, our thorough review of the entire record convinces us that the defendants did receive a fair one.

Affirmed.