United States v. Dennis R. Josleyn, United States of America v. John W. Billmyer, United States of America v. Dennis R. Josleyn and John W. Billmyer

LYNCH, Circuit Judge. Widespread bribery, kickbacks, and other fraud infected the operations of the American Honda Motor Company from the 1970s to the early 1990s. Two American Honda executives and beneficiaries of the illicit largesse, Dennis Josleyn and John Billmyer, were convicted in 1995 in federal court of conspiring to defraud their employer. Josleyn was also convicted of racketeering, conspiracy, and fraud for receiving kickbacks in connection with sales training seminars and dealer advertising programs. Josleyn had most recently served the company as zone sales manager for the West Coast, and Billmyer, after 18 years with the company, had attained the position of national sales manager. Bill-myer retired in 1988 and Josleyn resigned in 1992. This court affirmed their convictions in 1996. See United States v. Josleyn, 99 F.3d 1182, 1185 (1st Cir.1996), cert. denied, 519 U.S. 1116, 117 S.Ct. 959, 136 L.Ed.2d 845 (1997). The matter now before us involves appeals from the denial of motions for new trial based on newly discovered evidence. I. A. The Trial and Post-Trial Proceedings The prosecution’s theory at trial was that the defendants had victimized American Honda by taking kickbacks from dealers in exchange for dealership rights and favorable vehicle allocations, among other things. At the time, Honda cars were difficult to come by: the need for energy efficient cars had enhanced their popularity and Japanese export restrictions had diminished their availability. Demand exceeded supply and profit margins for dealers were high. These economic conditions created a climate ripe for the fraud scheme. The prosecution argued that Bill-myer and a cohort, Stanley James Cardig-es, masterminded a continuous, self-perpetuating kickback scheme by regularly accepting cash and lavish gifts from dealers around the country in exchange for *148 favorable allocation treatment. The prosecution claimed that Cardiges also participated in Josleyn’s kickback schemes involving the dealer advertising programs and national sales training seminars. Car-diges and several other employees were indicted and entered into plea agreements. Cardiges cooperated and became one of the government’s star witnesses. At trial, one of the defenses offered by the defendants was that of condonation: that American Honda (and perhaps even its Japanese parent company) had sanctioned and condoned its executives’ receiving money and lavish gifts from dealers. That another has also done wrong does not usually excuse wrongdoing by a defendant, so at trial there was a question as to the relevance of this condonation theory. In the end, the condonation theory resulted in an instruction to the jury that the evidence of condonation could be considered “to the extent such evidence bears on the issue of whether or not [the defendants] formed the required intent to commit the crimes with which [they are] charged.” After seven days of deliberation, the jury rejected the Iack-of-intent defense and convicted. These appeals are from the 1997 denial of the defendants’ post-trial motions for a new trial, from the 1999 denial of a renewed motion for a new trial, and from the denial of a motion to reconsider the denial of the renewed motion for a new trial. Those motions were based on evidence that at first trickled and then flooded out of civil litigation that had been instituted by various dealers against American Honda. In that multi-district litigation (the “MDL litigation”), a federal district court in Maryland ordered the disclosure of certain information that American Honda had attempted to shield from disclosure through the attorney-client privilege. See In re Am. Honda Motor Co. Dealer Relations Litig., MDL Case No. 1069, slip op. at 1 (D. Md. June 3, 1998). The materials disclosed as a result of this ruling and others were used by Josleyn and Billmyer to support their renewed motion for a new trial in this case and tended to show much more knowledge of corruption by top-level American Honda officials than had previously been shown. The 1997 motions for new trial were denied by the district court on July 17, 1997. After some skirmishing, 1 the defendants filed what was styled as a renewal of their original motions for new trial. The motion included the newly disclosed materials from the MDL litigation, which strengthened their argument. The district court denied the renewed motion for a new trial on the papers, thus denying as well the defendants’ request for an evidentiary hearing on the motion. The court carefully reviewed the newly discovered evidence, both individually and in combination, and concluded that the evidence fell short of showing “that a reasonable probability of a different result exists or that the new evidence undermines confidence in the verdict.” 2 *149 B. The MDL Litigation The évidence on which much of the new trial motions rested came to light as a result of the MDL proceedings. In November 1997, the lead plaintiffs in the MDL litigation moved to set aside American Honda’s attorney-client and work product privileges as to all communications regarding gratuities among American Honda, Honda North America, and any of their attorneys that occurred between 1979 and June 1993. On March 10, 1998, the district court ordered production of these communications between specified persons, finding that the crime-fraud exception vitiated the attorney-client and work product privileges. On March 24, 1998, a judge of the United States Court of Appeals for the Fourth Circuit denied a request to stay that order, finding that the “apparent breadth of American Honda’s design to enlist the assistance, knowing or otherwise, of its attorneys in concealing evidence of gratuities supports the district court’s finding that the privileges asserted in the instant case are negated by the crime-fraud exception.” In June 1998, the district court also granted the plaintiffs’ separate motion to compel the production of two memoranda written by counsel representing American Honda. These memo-randa discussed two “proffer” meetings held in New Hampshire between American Honda and the prosecutors in the criminal case against Josleyn, Billmyer, and other American Honda employees. The MDL court found that American Honda had, at these meetings, attempted to persuade the prosecutor that it was “innocent of any wrongdoing” and that it was “ignorant of the crimes committed by its former executives who were under investigation.” In so doing, the court concluded, the extent of American Honda’s knowledge was concealed from the prosecutor. According to the district court, there was “'prima facie evidence ... that members of American Honda’s upper management did have knowledge of (or at least wilfully blinded themselves to) allegations of bribe-taking by Honda executives.” These findings led the court to note the possible impact this newly revealed evidence might have on the criminal prosecutions of Josleyn and Bill-myer: The indictment in United States v. Bill-myer broadly charged that the defendants had “devised and intended to devise a scheme to defraud American Honda and its dealers” against the background of allegations relating to American Honda’s conflict of interest policy and its policy regarding the award of new dealerships. If (as plaintiffs contend) the written policies were merely empty verbiage because of widespread knowledge about the illegal conduct among members of the upper management of American Honda ..., the factual underpinnings of the prosecutions would have been brought into question. In re Am. Honda Motor Co. Dealer Relations Litig., slip op. at 10 n. 7. II. There is a preliminary question as to whether this court has appellate jurisdiction over the 1999 denial of the renewed motion for a new trial. The question arises because the defendants filed, but later withdrew, a notice of appeal from the denial of the renewed motion. 3 Defen *150 dants later explained that they withdrew the notice of appeal because they believed the denial of the renewed motion was encompassed within the original notice of appeal they filed when the district court denied their 1997 motions for new trial. 4 There is a split in the circuits over whether, when a direct criminal appeal is pending, a defendant needs to file a separate notice of appeal from a subsequently entered order denying a new trial motion or whether the original notice of appeal from the judgment of conviction suffices to bring before the appellate court the subsequently denied new trial motion. The Fifth, Ninth and Eleventh Circuits have held that a court of appeals has jurisdiction to review the denial of a post-trial motion under Fed.R.Crim.P. 33 based on newly-discovered evidence even though no second notice of appeal from the denial of that motion has been filed, provided an initial timely notice of appeal from the judgment of conviction has been filed and the government has not been prejudiced. United States v. Wilson, 894 F.2d 1245, 1251 (11th Cir.) (“[A] second notice of appeal is not required in order for [the appellants] to challenge the district court’s denial of their motions for a new trial[,]” where no prejudice to the government results[.]), cer t. denied sub nom., Levine v. United States, 497 U.S. 1029, 110 S.Ct. 3284, 111 L.Ed.2d 792 (1990); United States v. Burns, 668 F.2d 855, 858 (5th Cir.1982) (“While it may be desirable for a second notice of appeal to be filed, we must conclude that there is no requirement in the law that this be done to perfect an appeal.”). United States v. Davis, 960 F.2d 820, 824 (9th Cir.1992); see also United States v. Wilson, 894 F.2d 1245, 1251-52 (11th Cir.1990); United States v. Burns, 668 F.2d 855, 858 (5th Cir.1982). The Sixth Circuit, however, has taken the view that by being a distinct appealable order from which a separate appeal must be taken, an appeal from a Rule 33 motion denial is subject to the requirement that the appeal be taken within ten days from the docketing of the district court’s order. See Fed.RApp.P. 4(b). Absent an appeal within this time, or an extension from the district court for filing the notice of appeal, id., this court, being without authority to extend the time for filing a notice of appeal, Fed.R.App.P. 26(b), will lack the jurisdiction to hear the appeal. United States v. Hatfield, 815 F.2d 1068, 1073 (6th Cir.1987) (citations omitted). The issue that has created the circuit split can be distinguished from this case. The present situation is a timely appeal from one new trial motion followed by a failure to appeal from a renewed new trial motion, rather than a timely direct appeal followed *151 by a failure to appeal a new trial denial. We defer to another day a decision on the broader issue that has split the circuits; this case can be decided more narrowly. The facts of this case are idiosyncratic. All counsel and the district court treated the 1999 motion as truly a renewal of the 1997 motions, that is, as simply the 1997 motions supplemented with information that was not available when the 1997 motions were filed. Indeed, the government’s brief does not challenge appellate jurisdiction over the denial of the renewed motion for a new trial. Under these circumstances, and in light of the 1997 order from this court staying the appeals from the denial of the first motions in order to permit the district court to consider the renewed motion, we think that the notices of appeal from the denial of the original motions for new trial suffice to preserve and present to us the issues raised by the denial of the renewed motion. III. A. Standards of Appellate Review for New Trial Motions Review of the denial of new trial motions is for manifest abuse of discretion. See United States v. Montilla Rivera (Mantilla Rivera II), 171 F.3d 37, 40 (1st Cir.1999). Where it is contended that the district court applied an incorrect legal standard, that contention is reviewed de novo. See United States v. Huddleston, 194 F.3d 214, 218 (1st Cir.1999). Even application of an incorrect legal standard, though, does not itself mean that a new trial will be ordered. See United States v. Natanel, 938 F.2d 302, 314 (1st Cir.1991); cf. United States v. Montilla-Rivera (Mantilla-Rivera I), 115 F.3d 1060, 1066-67 (1st Cir.1997). These appellate standards of review operate as an overlay on the standard that the trial judge was required to use in evaluating the new trial motions. The defendants brought their motions for new trial under both Brady v. Maryland, 373 U.S. 83, 83 S.Ct. 1194, 10 L.Ed.2d 215 (1963), which requires the government to produce to defendants exculpatory and impeachment evidence that is in its custody, possession, and control, see Brady, 373 U.S. at 87, 83 S.Ct. 1194; see also United States v. Bagley, 473 U.S. 667, 676, 105 S.Ct. 3375, 87 L.Ed.2d 481 (1985); Huddleston, 194 F.3d at 222, and Federal Rule of Criminal Procedure 33. New trial motions based on newly discovered evidence require the defendant to show four elements: (1) the evidence was unknown or unavailable to defendant at the time of trial; (2) the failure to discover the evidence was not due to a lack of diligence on the part of the defendant; (3) the new evidence must be material; and (4) the evidence would probably produce an acquittal upon retrial of defendant. United States v. Ortiz, 23 F.3d 21, 27 (1st Cir.1994); see also Fed.R.Crim.P. 33. The difference between a new trial motion based on an alleged Brady violation and an ordinary Rule 33 motion is found in the tests for the third and fourth elements. For Rule 33 motions, “the evidence must create an actual probability that an acquittal would have resulted if the evidence had been available.” United States v. Sepulveda, 15 F.3d 1216, 1220 (1st Cir.1993) (emphasis added). If, on the other hand, the government possesses Brady evidence but does not disclose it, it is usually said — subject to some caveats — that the non-disclosure warrants a new trial if the evidence is “material.” It is “material” only if there is a “reasonable probability” that the evidence would have changed the result. Bagley, 473 U.S. at 682, 105 S.Ct. 3375. The Court has defined “reasonable probability” as a probability “sufficient to undermine confidence in the outcome.” Id. In Sepulveda, this court noted that this test was somewhat “delphic.” See Se- *152 pulveda, 15 F.3d at 1220. Since Sepulveda, the Supreme Court has further expounded on the “reasonable probability” concept. Recently, the Court said, “The question is not whether the defendant would more likely than not have received a different verdict with the evidence, but whether in its absence he received a fair trial, understood as a trial resulting in a verdict worthy of confidence.” Strickler v. Greene, 527 U.S. 263, 119 S.Ct. 1936, 1952, 144 L.Ed.2d 286 (1999) (quoting Kyles v. Whitley, 514 U.S. 419, 434, 115 S.Ct. 1555, 131 L.Ed.2d 490 (1995)). The standard applied to new trial motions based on Brady violations is thus more favorable to defendants, and so the defendants 5 try to shoehorn as much of the new evidence into the Brady category as possible. The defendants make two arguments as to which evidence is- Brady material. First, they say that all of the new evidence is Brady material because, although the government literally did not know of the existence of the evidence at the time of the New Hampshire criminal case, American Honda did. American Honda’s knowledge should be attributed to the government, the defense argument goes, because American Honda and the government portrayed themselves as partners in the criminal prosecution and American Honda cooperated with the government in producing the evidence used to prosecute the two defendants. Second, and in the alternative, they argue that some of the evidence in the first motions for new trial was actually known to the prosecution and was, therefore, Brady material that should have been disclosed. B. The Attribution Theory The factual basis for the defendants’ attribution argument is the close working relationship between the government and American Honda’s lawyers in the prosecution of the criminal case. The defendants rely heavily on the fact that the lead prosecutor referred to American Honda’s lawyers as part of his “team” of “prosecutors working on this case.” That statement is puffery; there is no evidence that American Honda’s lawyers were, in any sense, prosecutors here. Instead, they were trying to dodge prosecution of their client. There were two “proffer meetings” between the prosecution and American Honda held in October of 1993. At the beginning of the first meeting, counsel for American Honda obtained the signature of the lead prosecutor on a “non-waiver of attorney-client privilege agreement.” American Honda created the impression that it was giving the prosecutors the results of a privileged investigation. 6 Counsel for American Honda then made a presentation and turned over a “factual proffer” document; there ensued a discussion of the relevant facts and the credibility of certain involved individuals. The first meeting closed with the prosecutor stating that he wanted to address the issue of venue in New Hampshire “as quickly as possible” and with the scheduling of a second meeting. At the second meeting, conversation continued regarding the evidence offered by American Honda, the terms of cooperation, and the venue issue. We will take as true defendants’ assertion that American Honda’s lawyers helped the prosecution in at least five areas. 7 *153 American Honda had a strong interest in helping the government: it did not want to be named as a target in the criminal investigation nor did it want to subject itself to civil liability. In addition, American Honda apparently did not want its top executives named as targets or conspirators. By the same token, however, it is clear that some of what American Honda knew was withheld from the government. The government decided that the role of the company in this matter was that of the victim. The attribution theory must be put in the context of what constitutes a Brady violation. In Strickler, the Supreme Court, noting that the term “Brady violation” is often loosely used, said that “[t]here are three components of a true Brady violation: The evidence at issue must be favorable to the accused, either because it is exculpatory, or because it is impeaching; that evidence must have been suppressed by the State, either willfully or inadvertently; and prejudice must have ensued.” Strickler, 119 S.Ct. at 1948. For the purpose of evaluating the defendants’ attribution argument, we will assume that the first and third components of a Brady violation are present and will focus on the second component — the suppression of the evidence by the state. The defendants do not accuse the government of literally suppressing the evidence and do not seriously claim that the government should have known of this evidence. Rather, they say that the government should be treated as having American Honda’s knowledge of the evidence. On the facts here, 8 we reject the defendants’ attribution theory. The defendants draw analogies to civil rights law where private action is sometimes held to be so intertwined with government action as to render it state action. See Lugar v. Edmondson Oil Co., 457 U.S. 922, 941, 102 S.Ct. 2744, 73 L.Ed.2d 482 (1982). 9 , But the context here is different. The government did not have access to the evidence at issue and did not suppress it either willfully or inadvertently. American Honda had an interest in not divulging this material to the government and, for a time, was successful in shielding the information both from the government and from the defense. There was no intertwining of private action with government action along the lines of the Lugar case. Nor would the purposes of Brady be furthered by accepting the attribution theory here. Where the government has exculpatory evidence and fails to disclose it, both society’s notion of a fair trial and the trial’s truth-finding function are offended. Such suppression is also contrary to a third interest: “the special role played by the American prosecutor in the search for truth in criminal trials.” Strickler, 119 S.Ct. at 1948. The government’s interest “is not that it shall win a case, but that justice shall be done.” Id. (quoting Berger v. United States, 295 U.S. 78, 88, 55 S.Ct. 629, 79 L.Ed. 1314 (1935) (internal quotation marks omitted)). Thus, the law makes it easier for defendants to obtain a new trial where the government has engineered an unfair trial by withholding material exculpatory evidence. It is quite another matter when both the government and the defendants are the victims of an interested private third-party (here, American Honda’s top management and attorneys) withholding information even while it feigns full cooperation with the government. American Honda, despite the pros *154 ecutor’s hyperbole, was not part of the prosecution team. While prosecutors may be held accountable for information known to police investigators, see Kyles, 514 U.S. at 437-38, 115 S.Ct. 1555, we are loath to extend the analogy from police investigators to cooperating private parties who have their own set of interests. Those private interests, as in this case, are often far from identical to — or even congruent with — the government’s interests. American Honda, at the time of its disclosures to the government, was advised that it was a subject of the investigation and could become a target. It had an interest in seeing that the government knew some information about the kickback schemes, but not too much. Thus, we agree with the district court that the newly discovered material may not be attributed to the government for new trial motion purposes. As a result, only the evidence that the defendants have shown was actually known to the government is subject to the Brady standard. C. The Two Condonation Theories The effect of the new evidence must be evaluated in context. The condonation defense went primarily to the theory that the defendants lacked the intent to defraud American Honda because they knew that American Honda was aware of and implicitly condoned their activities. It was in American Honda’s interest to condone the fraud, the defendants say, because the kickbacks and bribes worked to subsidize the compensation of American Honda’s sales managers, enabling American Honda to pay them lower salaries. A second theory regarding the relevance of American Honda’s alleged condonation — that there could be no scheme to defraud their employer where American Honda knew of the fraud and, in fact, benefitted from it — emerged most clearly at oral argument. It had been alluded to in the moving papers. This second theory, unlike the first, does not depend on whether the defendants knew that American Honda had knowledge of the fraudulent activities. This second condonation theory had been discussed in various trial and pretrial motions. On direct appeal from the conviction, Josleyn argued that the trial court had erred in not instructing the jury that American Honda’s condonation was a defense, apart from any impact that the condonation may have had on the defendants’ intent. This court responded to that argument by stating that “[s]ince Jos-leyn neither cites authority, nor demonstrates, that any condonation by Honda was relevant to an element of the charged offenses other than intent, we conclude that the jury instruction given by the district court was adequate.” Josleyn, 99 F.3d at 1194 (citations omitted). This second theory is less present in the various filings supporting the motions for new trial. The memoranda of law filed in support of the 1997 motions for new trial referred to the second theory, but did not discuss it fully or separately from the first theory. The memorandum in support of the 1999 renewed motion for a new trial adverted to a variant of the second theory in a footnote, but the district court’s opinion denying the renewed motion did not discuss the second theory. Nor do the briefs in these appeals clearly delineate that there were two separate theories of relevance (although the distinction came up at oral argument on questioning by the court) or provide any supporting authority for the second theory. On this record, the second theory has not been properly presented and preserved. 10 See *155 United States v. Bongiorno, 106 F.3d 1027, 1034 (1st Cir.1997). IY. A. The Brady Material The defendants suggest that there were four pieces of evidence that were known to the government but not disclosed to the defense and that should, therefore, be evaluated under the Brady standard. The government concedes an arguable Brady violation as to one piece of evidence — the Novelly material — but disputes any Brady violation as to the other three items. 11 The Novelly evidence is simply insufficient to lead to a finding that the district court abused its discretion in denying the new trial motions. The defendants say that the government failed to turn over to them information collected through interviews with Roger Novelly and other involved individuals regarding a car obtained from a Steubenville, Ohio dealership. The information obtained by the FBI showed that Koichi Amemiya, the president of American Honda, contacted Novelly, informed him that a friend needed to trade in her car for a used car at the Steubenville dealership, and told him that he did not want his friend to have to make any payment beyond the value of the trade-in. The defendants characterize this information as showing that Amemiya accepted a gratuity from a dealer and also that he had knowledge of that dealer’s violation of the conflict of interest policy. Their theory is that Amemiya himself received a gratuity in the form of a favorable deal for an acquaintance and that Amemi-ya only made the request of Novelly, who had no other connection to the Steuben-ville dealership, because he knew of Novellas hidden and improper ownership interest in the dealership. The district court held as to this evidence: [T]he evidence is in large part duplica-tive of the testimony suggesting Japa *156 nese condonation of the defendants’ conduct .... [and] has little bearing on the defendants’ good-faith belief that American Honda condoned their behavior.... [T]he court finds that the defendants have failed to sustain their burden of demonstrating that there is a reasonable probability that the introduction of the Novelly testimony would have yielded not guilty verdicts. Indeed, the court finds that it is highly improbable that the verdicts would have been different. If we move beyond the verdict-outcome test erroneously utilized in this Brady context by the district court and apply the standard the Supreme Court has prescribed for Brady cases, see Strickler, 119 S.Ct. at 1952, the defendants still have made no showing “that the favorable evidence could reasonably be taken to put the whole case in such a different light as to undermine confidence in the verdict.” Kyles, 514 U.S. at 435, 115 S.Ct. 1555. Such confidence may be undermined even if there is “less than an even chance that the [new] evidence would produce an acquittal,” Sepulveda, 15 F.3d at 1220, but that is not the situation here. This evidence, even drawing the inferences as defendants would have them, was about events after Billmyer had left the company, is very weak evidence of possible knowledge of Novelly’s improper dealings by the American Honda president, and added even less to the more direct evidence of condonation that defendants put on at trial. Further, the fact that there is no suggestion that either Josleyn or Bill-myer knew of this alleged act of condonation gives this evidence limited relevance to the issue of intent. It fails the Strickler test. B. The Rule 33 Evidence Apart from the alleged Brady violations, the defendants outline seven areas of new evidence, all of which were raised in the renewed motion for new trial and derived from the MDL litigation. We describe the evidence here in terms most favorable to the defendants. 1. The Smoot Investigation Attorney Roland “Bud” Smoot was a senior partner at the law firm of Lyon & Lyon. That firm served as American Honda’s outside general counsel, and Smoot did much of the work. Smoot was also on the American Honda Board of Directors in the 1980s and 1990s — pertinent time periods — and served on American Honda’s ethics committee, and internal audit committee. In 1984, while wearing one of his American Honda-affiliated hats, Smoot hired Emmett Doherty, a private investigator, to look into rumors that Billmyer and his predecessor had received bribes. From Smoot’s comments to Doherty, it was evident that Smoot had heard sufficient stories of corruption to warrant investigation. Further, after Doherty made his initial report, Smoot terminated the investigation. 2. Shredding of the 1983 Allocation Audit At trial, evidence was presented that J.D. Power, an independent automobile industry analyst and author, informed the then executive vice president of American Honda of rumors of wrongdoing at the company. American Honda’s then president Tetsuo Chino ordered an internal audit of the automobile allocation system. This was known to at least one of the defendants at the original trial. The audit apparently showed- flaws in the allocation system that left the system ripe for abuse and recommended changes to that system. What was newly discovered was that the audit was never completed and the initial results were shredded. It was far. from usual practice to shred the results of a partially-completed audit. The evidence showed that the shredding was done after Smoot told the auditors that if the audit “did not exist,” his life would be much simpler. Like King Henry II’s comments about Thomas á Becket, Smoot’s instructions were followed by subordinates. *157 3. 1986 Conflict of Interest Survey In 1986, American Honda did an employee survey as part of an audit of the conflict of interest policy. The responses of four of the employees were to the effect that the existing conflicts policy was disregarded. At least one employee suggested that a new policy was needed. 4. Nissan Information In 1986, Nissan’s director of administration told American Honda’s audit division chief, Douglas DiVall, that Honda dealers were complaining about having to make payments to sales managers for allocation of vehicles. DiVall testified that when he brought this information to Chino, American Honda’s then president, he was told not to worry about it, that it was “just dealers wanting more product.” 5. Memorandum to American Honda’s President A memorandum entitled “For Mr. Ame-miya” discussed the problem that “there may have been wide spread [sic] bribe taking by Honda auto reps — both on the Honda and Acura side.” The parties agree that the memorandum was prepared in 1991 or 1992. The unknown author of the memo, most likely a Lyon & Lyon attorney, spoke of certain corrupt activities as being “common practice” at the company and said “[i]t has long been believed that Honda dealerships are obtained by paying off the right people.” 6. Limits on the Sherry Cameron Investigation At trial, American Honda’s human resources manager, Sherry Cameron, testified about an investigation she conducted in 1992 into the receipt of gratuities by Cardiges and others. She flatly denied that American Honda had any prior knowledge of receipt of gratuities by its sales force. She further denied that her investigation had been limited in any way and, in fact, testified that it was conducted according to a protocol prepared by outside accountants Coopers & Lybrand. New documents suggest, however, that Coopers & Lybrand’s protocol had been circumscribed by Lyon & Lyon, who revised the protocol to, among other things, remove its name from the title of the investigation, delete the word “kickback” from the protocol, and remove a proposed phase of the investigation, which would have looked into the activities of Billmyer and others. Cameron apparently did not know that Lyon & Lyon had circumscribed the protocol. 7.American Honda’s “Cover-Up” Finally, the defendants say that the fact that American Honda went to some lengths to cover up its knowledge — while simultaneously seeming to cooperate with the government' — is, in and of itself, evidence that American Honda knew of the fraud but chose to condone it. C. Effect of the New Evidence We focus on the original theory that the defendants lacked the requisite intent and turn to an evaluation of the new evidence in juxtaposition to the evidence actually admitted at trial. We review the trial judge’s decision to determine whether he abused his discretion in finding that the new evidence did not create an “actual probability” of acquittal and whether he used erroneous legal standards. At trial, the government put on evidence that the defendants’ practices violated American Honda policy and that these violations of company policy occurred without the knowledge of American Honda executives. Defendants countered with evidence that management knew of, and blinked at, the gratuities scheme. The government also put on evidence that the defendants went to great pains to keep their activities secret, suggesting that they wished to keep American Honda management in the dark and that - they knew that what they were doing was wrong. The defendants agree that they kept their ac *158 tivities discreet but argue that they did not do so out of fear that their scheme would be uncovered by management. Instead, they say that the scheme was well-known among management and that it was kept quiet to avoid a mass of litigation and a besmirched reputation for the company. The defendants point to the MDL litigation as proof of the reasonableness of these fears. The court’s assessment and denial of the renewed motion for new trial was primarily based on the items of proof listed in numbers three, four, five, and seven above, that is, those items of new evidence that did not involve attorney Smoot or Lyon & Lyon. 1. The Smoot Evidence The district court largely disregarded the new evidence that involved Smoot and Lyon & Lyon because it concluded that attorney Smoot’s conduct and knowledge could not be attributed to the company. The court gave two explanations for this conclusion. The first suggested a legal principle that Smoot’s knowledge could not be attributed to American Honda management if the defendants’ assertion that the “Japanese management of American Honda [were] the controlling figures of the company” was correct. The second was that there were no facts in the record to support the defendants’ factual assertions about Smoot’s role. Our view of these matters differs from the district court’s. The record — particularly those portions reflecting the conclusions drawn by the district court in the MDL litigation — provides support for the defendants’ factual assertions about Smoot’s role. In the MDL litigation, the court found that both Lyon & Lyon and American Honda “acknowledged the close relationship the law firm ... had with the company,” even while they tried to emphasize that Lyon & Lyon was outside counsel. The picture of the firm’s minimal involvement initially put forth by American Honda and Lyon & Lyon is at odds with the description of their relationship that emerged from the MDL litigation. In a memorandum disclosed as a result of that litigation, American Honda’s attorneys related the following discussion with the prosecutors: That led to a discussion about Lyon & Lyon’s outside counsel’s role to the company, and we talked a little about Bud Smoot and how he started with the company in the 1970’s when it was a motorcycle company and eventually ended up on the Board of Directors and actually came in-house for a while, and how, as a result of the Lyon & Lyon relationship, that the institutional memory of the company’s problems and successes really resided with the law firm because, traditionally, the top Japanese management of the company was not there for 20-30 years at a time, and they rotated out maybe in the five to eight year range, so that nobody was left within the company who really had a historical perspective on problems that may have arisen over the years within the company but for Lyon & Lyon and Bud Smoot. Connolly [the lead prosecutor] was ... [informed that] there was not a strong general counsel’s office in the company and that Lyon & Lyon literally had authority to deal with line personnel, people in the field could call Lyon & Lyon with the most insignificant of matters, and that their relationship was such that all dealer litigation and corporate litigation went directly to Lyon & Lyon and really was not handled or managed in-house at all. The important role that Lyon & Lyon played in the company suggests the magnitude of Smoot’s role as the attorney whom the parties acknowledge handled most of American Honda’s legal work. In fact, the evidence suggests that Smoot — at the very least in his role as counsel for American Honda — had sufficient authority to hire an outside private investigator to investigate allegations of bribe-taking by company managers. In addition to his role as counsel, Smoot served the company directly in several roles. It is undisputed that Smoot *159 was a member of American Honda’s Board of Directors during relevant periods. He also served on American Honda’s ethics committee and internal audit committee. In these roles, Smoot apparently had sufficient authority in the company to bring about the destruction of the results of an internal audit that had been ordered by the company president. As to the legal principle, we clarify that there is no requirement that a person be a “central figure” at a company in order for that person’s knowledge to be imputed to the company. 12 The person whose knowledge is to be imputed must have some relationship to the company— whether director, officer, agent, or employee — which allows the person to obtain the knowledge in the course of the engagement with the company and within the scope of his or her authority. See 3 William Meade Fletcher, Fletcher Cyclopedia of the Law of Private Corporations § 790, at 15 (Beth A. Buday & Gail A. O’Gradney, eds.1994). Determining whether a given relationship is sufficient to support the imputation of knowledge may, in some cases, depend on an assessment of the person’s “level of responsibility.” Continental Oil Co. v. Bonanza Corp., 706 F.2d 1365, 1376 (5th Cir.1983); see also 3 Fletcher, supra, § 807, at 61 (“[W]hen notice to an agent is relied upon to bind a corporation, the nature of the agency must be such that the law will presume that the agent carried the notice to the principal ..., for notice to every employee of a corporation would not be held notice to the employer.”). Second, that other individuals may be controlling figures within the company does not mean that a person is not considered an agent for purposes of imputing his or her knowledge to a corporation. Whether knowledge can be imputed depends on a fact-specific assessment of the particular individual’s authority within the company. See id. § 807, at 62 (“The best test of imputation of knowledge is not whether the agent ... is president, treasurer, and so forth, but whether the condition and facts known were within the sphere of the authority of the particular agent.”). Smoot clearly occupied a role in the company that allows his knowledge of these pieces of newly discovered evidence to be imputed to American Honda. Whether in his capacity as director, see Dinco v. Dylex Ltd., 111 F.3d 964, 972 (1st Cir.1997) (stating that a director’s knowledge can be imputed to the company); Baker v. Latham Sparrowbush Assocs., 72 F.3d 246, 255 (2d Cir.1995) (same), or as attorney, see 3 Fletcher, supra, § 807.10, at 71 (“Whether notice to or knowledge of an attorney is imputed to his or her client is governed by the general rules .... ”), or as general agent, the facts clearly show that he had authority to act for the company and, in the instances relevant for these appeals, that he was acting in that capacity with the interests of the corporation in mind, see United States v. Route 2, Box 472, 60 F.3d 1523, 1527 (11th Cir.1995); United States v. 7326 Highway 45 North, 965 F.2d 311, 316 (7th Cir.1992). See also United States v. Bank of New England, N.A., 821 F.2d 844, 856 (1st Cir.1987) (indicating that the “collective knowledge” of a corporation’s employees can be attributed to the corporation). 13 Thus, the district *160 court should not have disregarded these pieces of newly discovered evidence in its assessment of the new trial motion. 14 The question is whether this error made any difference. We think not, because even if the new Smoot evidence were considered on remand, see Montilla-Rivera I, 115 F.3d at 1066-67, the outcome would undoubtedly be the same. The district judge, who presided over this three-month trial, made it clear in both of his denials of the new trial motions that he felt the evidence of guilt was “overwhelming.” Indeed, this court also called the evidence against Josleyn and Billmyer “overwhelming.” See Josleyn, 99 F.3d at 1189. In fact, the district judge made clear his conclusion that, whatever the degree to which American Honda knew of and ignored the corruption, the defendants had the requisite intent: they knew that what they did was wrong, they went to great lengths to conceal their activities from upper-level management, and they had no belief that their activities were condoned by American Honda. 2. The Other New Evidence As to the other new evidence, there were no errors of law and there are simply too many gaps in the logic of the defense’s lack-of-intent theory to find manifest abuse of discretion in the district court’s decision to deny the new trial motions and the motion for reconsideration. 15 See Montilla-Rivera II, 171 F.3d at 40. We take as a given that the defendants accepted bribes and kickbacks, but find little solid evidence that American Honda’s upper management directly received gratuities or expressly approved of the corruption. The defendants’ argument requires a jury firstly to infer tacit approval of the corruption from the company’s failure to act on rumors and warnings of corruption; secondly, to infer that the defendants knew of this tacit approval; thirdly, to infer that the defendants therefore believed the company had given its blessing to their activities; and fourthly, to infer that the defendants thus lacked the requisite intent. But the very point of this being new evidence is that the defendants did not know of these alleged acts of condonation and thus could not, based on them, have formed a belief that their activities were tacitly approved. These are perilously large leaps in logic. The coup de grace is that the evidence is clear that the defendants went to some effort to conceal their activities from American Honda management and others. V. The jury heard evidence as to the management’s alleged condonation and testimony that American Honda consistently paid lower salaries and then conveniently looked the other way and allowed bribe-taking to supplement those salaries. Yet the jury rejected the lack-of-intent theory as to each defendant. Although the new evidence is more powerful than that offered at trial, it is the same type of evidence. It does little to strengthen the tenuous link between any evidence of con-donation and the defendants’ intent. After individually considering each piece of evidence offered in the new trial motions, as well as assessing its combined effect, we conclude that the evidence does not provide a basis for concluding that the trial judge erred in not upsetting the jury’s guilty verdict. 16 *161 We affirm, the denials of the 1997 and 1999 new trial motions and of the 1999 motion for reconsideration. So ordered. Footnotes [1] The defendants filed notices of appeal from the district court’s denial of their motions for new trial in July 1997. They then notified this court that they had obtained newly discovered evidence and intended to file a renewed motion for new trial. They sought a stay of their appeals until the district court had ruled on the renewed motion. On September IS, 1997, this court granted the stay until such time as the district court ruled on the renewed motion. In their December 2, 1997 status report to this court, the defendants noted that they had been informed that a motion to set aside the attorney-client privilege had been filed in the MDL litigation. The information that eventually emerged from the MDL litigation as a result of this motion largely formed the basis of the renewed motion for new trial that was filed with the district court. [2] As will be discussed, the court reviewed the combined effect of the new evidence under the more defendant-friendly Brady standard. The court said that it did so “in an abundance of caution’’ because it was considering both Brady and non-Brady evidence. Thus, if anything, the court's standard favored the defendants in applying the Brady standard to evaluate the combined effect of Brady and non-Brady evidence. [3] There is another problem: the withdrawn notice of appeal, filed January 22, 1999, may not have been timely to appeal from the January 7, 1999 order denying the renewed motion for a new trial. See Fed. R.App. P. 4(b)(1); id. 26(a). That notice's timeliness may depend on whether we would allow the motion for reconsideration, which was filed on the same date, to suspend the time for filing a notice of appeal, compare United States v. Morillo, 8 F.3d 864, 867-69 (1st Cir.1993), with United States v. Rapoport, 159 F.3d 1, 3 n. 3 (1st Cir.1998), and United States v. Marsh, 700 F.2d 1322, 1325 (10th Cir.1983), and whether we would consider the motion for reconsideration to itself have been timely filed, see, e.g., United States v. Brewer, 60 F.3d 1142, 1144 (5th Cir.1995) (discussing whether Federal Rule of Appellate Procedure 26(a) or Federal Rule of Criminal Procedure *150 45 governs the counting of days for determining the timeliness of motions to reconsider). Given our holding, we need not decide whether the withdrawn January 22, 1999 notice of appeal was timely. [4] Concerned by this court’s inquiry regarding the jurisdictional issue and “out of an abundance of caution,” the defendants filed a new notice of appeal (from both the denial of the renewed motion for new trial and the denial of the motion for reconsideration) and a motion for extension of time to file a notice of appeal on May 3, 1999. The motion to extend time was granted by the district court "in view of the fact that the Clerk's office failed to send out a copy of the Court's order dated Feb. 23, 1999 [denying the motion to reconsider] to counsel.” However, Federal Rule of Appellate Procedure 4(b)(4) only allows district courts to extend the time to file a notice of appeal for 30 days beyond the expiration of the original time to file. Even making the assumptions for timeliness discussed in footnote 3, supra, May 3, 1999 was well beyond the allowable 30-day period. That the Clerk’s office failed to notify the defendants that their motion to reconsider had been denied “does not affect the time to appeal or relieve or authorize the court to relieve a party for failure to appeal within the time allowed, except as provided by Rule 4(b).” Fed.R.Crim.P. 49(c). Therefore, we cannot, as the defendants argue, consider the 40-day period to have begun running on the day the Clerk's office first notified the defendants that the motion had been denied. [5] In fact, it is defendant Josleyn who has largely created the record on these issues. Given the outcome, we do not analyze Bill-myer’s claims separately. [6] The MDL court found, however, that American Honda made misrepresentations about the “adequacy and good faith of [its] general investigation.” [7] The defendants allege that American Honda's lawyers gave the Assistant United States Attorney their own “prosecution memo,” discussed a proposed “target list,” gave assessments of the credibility of various witnesses, said they would look for overt acts in New Hampshire that would resolve any venue problems, and agreed to give as much help as possible, as quickly as possible. [8] We stress that there is no claim that the government was willfully blind to exculpatory evidence that it should well have known was available. That situation is simply not before us. Indeed, even the defendants acknowledge that American Honda’s selective disclosures resulted in a fraud on the government. [9] Defendants also cite to Lustig v. United States, 338 U.S. 74, 69 S.Ct. 1372, 93 L.Ed. 1819 (1949), a case that has little in common with this situation. In Lustig, the Court held that when federal and state agents jointly participate in a search that is illegal by federal standards, a motion to suppress should be granted in a federal prosecution. See id. at 78-79, 69 S.Ct. 1372. [10] In any event, the second theory is in some conflict with the notion that the shareholder owners of a corporation may generally be the victims of fraudulent schemes — and thus the corporation a victim — even if corporate management is itself corrupt. See United States v. Wallach, 935 F.2d 445, 464 (2d Cir.1991), cited in Josleyn, 99 F.3d at 1194; cf. United States v. Rackley, 986 F.2d 1357, 1361 (10th Cir.1993) (upholding bank fraud conviction where the owner and director knew of the fraudulent activity); United States v. *155 Weiss, 752 F.2d 777, 783-84 (2d Cir.1985) (upholding a mail fraud conviction where the defendant argued that the illegal scheme was "presumptively used for the benefit of the corporation’’); United States v. Yarmoluk, 993 F.Supp. 206, 209 (S.D.N.Y.1998) ("[A]n institution may be defrauded even if its employees allow or participate in the fraudulent practices.”); 3 William Meade Fletcher, Fletcher Cyclopedia of the Law of Private Corporations § 826, at 150-52 (Beth A. Buday & Gail A. O’Gradney, eds.1994). But we do not resolve these issues. [11] We have considered these three pieces of evidence, all offered in the 1997 new trial motions. We agree with the district court that this evidence — the testimony of John Conway, a regional sales manager, that Bill-myer was aware that an industry analyst had informed management about the kickback scheme; Cardiges's statements regarding a letter shown to the president of American Honda; and American Honda's plea to price-fixing charges in the 1960s — does not provide a basis for a new trial, whether considered under the Brady or the Rule 33 standard and whether considered alone or in combination with the other newly discovered evidence. The defendants argue that the Conway and Cardiges evidence shows that Cardiges perjured himself at trial and they, therefore, suggest that the court subject this evidence to a different standard, presumably a standard similar to that adopted in Larrison v. United States, 24 F.2d 82, 87-88 (7th Cir.1928). This court has recently repudiated the Larrison standard, stating that the discovery of perjury should be treated like any other newly discovered evidence. See United States v. Huddleston, 194 F.3d 214, 219-20 (1st Cir.1999). The Huddleston opinion left open, however, the possibility that a different standard might apply in cases where "the government's use of perjured testimony was not unwitting.” Id. at 221. In this case, we agree with the district court's analysis of the issue: "the new evidence proffered by the defendants is insufficient to establish that Cardiges gave deliberately false testimony at trial.” Because the defendants have not shown the statements to be perjury, we need not further consider application of the Larrison standard. Cf. United States v. Natanel, 938 F.2d 302, 313-14 (1st Cir.1991). We also reject the defendants' arguments regarding other alleged perjury at trial. The defendants failed to show, as regards these claims, that the testimony was deliberately false and that the government's use of the testimony was not unwitting. [12] The district court did not explicitly adopt a "central figure" test, although it did quote the "central figure” language used by the defendants. Although the defendants did not explicitly argue that a "central figure” test was appropriate, their use of this language may have misdirected the district court's analysis. [13] The general rules for determining when an individual's knowledge can be imputed to a corporation are subject to a number of exceptions. See, e.g., 3 Fletcher, supra, § 790, at 16 (noting general exceptions); id. § 808, at 72-73 (referring to information "casually acquired” by directors not acting in their official capacities); id. § 807.10, at 71 (referring to information obtained by an attorney-officer “in his or her professional capacity as a member of the bar”). The parties have not pointed us to, nor has our independent examination of the record revealed, any facts indicating that any of these exceptions might apply in this case. [14] Contrary to the suggestion of all parties, we do not consider Smoot’s deposition as it is not a part of the record in this case. [15] Although the usual Rule 33 standard is, as described, one of showing that there is an actual probability that the evidence will lead to an acquittal, there is some flexibility. Rule 33 also allows for a new trial where "the interests of justice so require.” Fed. R.Crim.P. 33. In this case, we conclude that there is neither a probability of acquittal nor a sense that the defendants did not receive a fair trial. [16] The defendants also suggest that the prosecutor’s comment during closing arguments that the condonation defense was "[a]bsolutely made up out of whole cloth for the benefit of you people in the hopes that [they] might *161 get acquitted” was improper. They argue that the newly discovered evidence “reveals the untruth of the prosecutor’s words” and that the government,, engaged in "malfeasance.” These remarks do not rise to the level of the behavior condemned in United States v. Udechukwu, 11 F.3d 1101, 1105-06 (1st Cir.1993). We agree with the district court that the new evidence "is at best speculative as to the issues of Japanese condona-lion and, ultimately, the defendants’ intent to defraud American Honda, and does not directly contradict ... the position that the government took during closing argument.” In fact, immediately after the above-quoted language, the prosecutor said: “Nobody thought this was okay. Nobody was open about it. Nobody thought it was condoned by anybody else at American Honda....”