In United States v. Kahaner, 317 F.2d 459 (2 Cir.), cert. denied, 375 U.S. 835, 836, 84 S. Ct. 62, 11 L. Ed. 2d 65 (1963), we affirmed the conviction of James Vincent Keogh, a Justice of the Supreme Court of New York for Kings County, and two co-defendants, Elliot Kahaner, former Chief Assistant United States Attorney for the Eastern District of New York, and Antonio Corallo, for a conspiracy to influence, obstruct or impede justice in violation of 18 U.S.C. 1503. Justice Keogh served eight months of his 2-year sentence and was placed on probation for the remainder. The New York Court of Appeals ruled in a proceeding for his disbarment that he should 'be privileged to relitigate the issue of his guilt by introducing, if he can, evidence which was unavailable to him upon the trial in the Federal court, including proof of recantations or perjury by witnesses who had testified against him,' Keogh v. Richardson, 17 N.Y.2d 479, 266 N.Y.S.2d 984, 214 N.E.2d 163 (1965); the hearing on this is pending. In May 1967 he brought before the District Court for the Southern District of New York a petition for a writ of error coram nobis, supported by numerous affidavits and exhibits, alleging that the prosecution had knowingly suppressed exculpatory evidence and used perjured testimony; the Government answered and also submitted affidavits. Judge Weinfeld, who had presided over the trial, ruled that consideration of the moving papers, weighed against the trial record and files, sufficed to show the petition's lack of merit without need for relying on the answering affidavits or holding an evidentiary hearing. The appeal is from the court's order of dismissal, 271 F. Supp. 1002 (1967).

Although the Government has naturally preferred to answer the charges of prosecutorial misconduct, it has properly called our attention to a procedural point which, if sound, would preclude us from considering the merits. The notice of appeal was filed twenty-seven days after entry of the order of dismissal-- well within the 60 days allowed by F.R.Civ.P. 73(a), which would govern if the district court had denied a petition for habeas corpus or under 28 U.S.C. 2255, but more than the 10 days limited by F.R.Cr.P. 37(a)(2). The Government suggests the Criminal Rule is applicable in light of a footnote to the opinion in United States v. Morgan, 346 U.S. 502, 505, 506 n. 4, 74 S. Ct. 247, 98 L. Ed. 248 (1954), sustaining the continued vitality of the writ of error coram nobis, where Mr. Justice Reed said that a petition for such a writ 'is a step in the criminal case and not, like habeas corpus where relief is sought in a separate case and record, the beginning of a separate civil proceeding.'

We disagree. The problem to which the footnote was addressed was that F.R.Civ.P. 60(b) had abolished writs of error coram nobis in favor of the motion procedure there provided. The answer was that the abolition effected by Rule 60(b) cannot fairly be extended beyond the 'suits of a civil nature' to which alone, under the general provision of Rule 1, the Rules of Civil Procedure apply. We do not read the quoted statement as intending more than that. Indeed the footnote ends by characterizing a motion for coram nobis as 'of the same general character as one under 28 U.S.C. 2255,' which, as indicated above, is under the civil rules as regards time for appeal. See Heflin v. United States, 358 U.S. 415, 418 n. 7, 79 S. Ct. 451, 3 L. Ed. 2d 407 (1959). Compare Wells v. United States, 318 U.S. 257, 260, 63 S. Ct. 582, 87 L. Ed. 746 (1943). The policy considerations supporting prescription of a very short time for appeal in a criminal case are notably absent in coram nobis.

Turning to the merits we find, for the most part, no occasion to add to Judge Weinfeld's characteristically thorough opinion. In some instances witnesses have simply changed details of testimony they gave before the grand jury or at trial.1 In another instance, Erdman's chauffeur, Maggio, who testified before the grand jury that he had made deliveries to Keogh in 1959, including medicines, a new automobile, and $1000 in cash, but was not called at the trial, now claims his testimony as to the cash delivery was made at Erdman's suggestion and he had told the prosecutors it was untrue. But he did not aver he had told the prosecutors he had lied at Erdman's suggestion, the only point that would have been useful to defense counsel, if, indeed, one can possibly assume the defense would have wished to open up the history of prior financial dealings between Erdman and Keogh, see 317 F.2d at 470-471. An affidavit of a resigned member of the New York bar, himself convicted of subornation of perjury, asserting that Erdman had told him the case was 'full of perjury' and that the prosecutors had told him to 'forget' facts helpful to the defense, was properly dismissed as hearsay. The claim of prejudice by the prosecutors' suppression of grand jury testimony of Erdman as to participation with Keogh and Kahaner in an effort with respect to the sentence of Joseph Abrams for income tax evasion somewhat similar to that allegedly performed on Moore's behalf, when the prosecutors knew that Abrams would deny this, is, in Judge Weinfeld's apt phrase, utterly extravagant,' 271 F. Supp. at 1008. It is simply incredible that any competent defense counsel would have brought out testimony from Erdman to show such a pattern of criminal conduct in order to have an opportunity to rebut him with Abrams, four times convicted of federal felonies, see 271 F. Supp. at 1007 n. 15, assuming that would have been permissible, when, as the district court stated, 'the undisputed fact that petitioner did receive from Erdman a gift of a car in 1959, and other circumstances which could be deemed corroborative, would permit a fact finder to conclude that Erdman's account of an Abrams fix was not the product of imagination,' 271 F. Supp. at 1008 and n. 24.

We reach a similar conclusion with regard to the Government's non-disclosure of the grand jury testimony of Moore's close associate, Harry Kay. Kay, who could not be located at the time of the trial, had testified before the grand jury that he had been with Moore on March 29, when the final payments, totalling $20,000, were made to Keogh and Kahaner. He said that the money had been stored in the bank safe deposit box of Moore's brother-in-law, Kerner, and that on the morning of March 29 he drove Moore to the bank to get it. They then proceeded to Erdman's office in midtown Manhattan and, shortly afterwards, he drove both Erdman and Moore to Brooklyn. Keogh contends it was impossible to cover that much territory and still arrive in Brooklyn about 10:00 A.M., as both Erdman and Moore had testified at trial. Judge Weinfeld reasoned, 271 F. Supp. at 1017, that since both Erdman and Moore denied that Kay had been with them at all on March 29, and were cross-examined thoroughly as to this, Kay's grand jury testimony that he had been would have added nothing. It is argued against this that though Kay's presence was denied, his version of the events-- especially that concerning the detour to the bank-- was not developed.2 But again it is inconceivable that defense counsel would have chosen to add the grand jury testimony of a third witness supporting Moore's and Erdman's on the basic fact of payment simply in order to bring out an inconsistency, or even incredibility, in details of the third witness' version.

There is no need to mention several other matters that have even less force than those we have recounted. Indeed it was doubtless the very number of wholly unmeritorious claims that distracted the judge's attention from the one item where, as we see it, Keogh's allegations cross the rather low threshold entitling him to an evidentiary hearing on his petition, to wit, the prosecutors' failure to disclose an FBI report as to certain of Erdman's financial transactions that was in their hands.

A few circumstances concerning the indictment and the trial must be recounted to place this matter in proper setting. If Moore paid $35,000 in February and March, 1961, the money must have landed somewhere. Accordingly, as the affidavit of one of the prosecutors attests, he requested the FBI 'to conduct financial investigations of Keogh, Erdman and Kahaner to see if any of the bribe money could be traced.' Concededly none was ever traced to Keogh, who allegedly received $5000 around March 7, 1961, and $17,500 on March 29, or to Kahaner, who allegedly received $5000 in late February, $5000 around March 7, and $2500 on March 29. The defense naturally made much of this at the trial. However, Moore was positive he had paid the money and there was other evidence to support this. Accordingly the theory of the defense3 was, as stated in our previous opinion, 317 F.2d at 476, 'that Erdman intended from the outset to pocket the money, did exactly that, and then, when his iniquity was discovered, decided to implicate Kahaner, Keogh and Corallo in an effort to gain favor with the Government, with Moore joining in the endeavor for the same reason.'4 Determination of this was crucial-- the judge instructed the jury that 'if you should find that no cash payment was made to Keogh or Kahaner * * * there would be no basis for a finding of corrupt endeavor.' 317 F.2d at 475. The state of Erdman's finances thus had an importance almost equal to Keogh's and Kahaner's, and the prosecutors properly required this to be investigated.

The FBI report of its investigation was rendered November 28, 1961, some ten days before the indictment. It recited that Erdman's bank deposits of $243,590.58 from January through August 1961 had been examined and fully accounted for-- with one exception:

'The following unidentified items as to source could not be identified by DR. ROBERT ERDMAN or his accountant.

                                    Date
        Identity of Bank          Deposited    Amount
Chemical Bank New York Trust Co.    2- 8-61  $ 1,500.00
               "                    2-10-61    5,539.94
               "                    2-17-61    7,500.00
Chase Manhattan Bank                2- 6-61    1,000.00
                                             -----------
                                             $15,539.94"

The Chemical account was a new one, opened with the $1,500 deposit of 2-8-61; the three Chemical deposits were indicated as check deposits. Another section of the report disclosed that 'A confidential source at the First Federal Savings and Loan Association, 595 W.235th Street, Bronx, New York' advised that Caroline W. Erdman, the doctor's wife, had opened three accounts, of $9500 each, on June 26, 1961, which had a seemingly curious history of withdrawals, and also recited the facts as to another account, as summarized in the margin.5

The defendants made no pre-trial request for Erdman's financial records. In the course of cross-examination by Keogh's trial counsel on May 21, 1962, Erdman stated he had no objection to the latter's reviewing his bank records from February to June 1961, which had been lodged with the clerk of the court on May 18. Apparently they were turned over shortly thereafter. On June 8, 1962, Keogh's counsel offered a number of Erdman's deposit slips, stating he was interested only in currency deposits. The prosecutor objected on the ground that these 'are obviously offered to try and substantiate the inference that the pay-off money was put in Dr. Erdman's account' whereas 'in the present state of the record' they were not susceptible of any such inference. After the court had indicated it would allow the tickets 'for whatever probative value they may have,' the prosecutor sought and received permission to recall Erdman and to keep his case open to that end. Defense counsel countered that he had been handicapped by lack of bank records in cross-examining Erdman and that if Erdman resumed the stand, he would 'conduct an examination of his financial transactions during this period, not only with regard to currency but with regard to other financial transactions and deposits of checks in substantial sums.' After an interval discussion was resumed, to much the same effect as before.6 The judge indicated concern that recall of Erdman and extended cross-examination on his financial affairs would greatly prolong the trial which, having lasted some four weeks, was then in its concluding stages, and requested defense counsel to submit a memorandum on the admissibility of the deposits. Apparently this was not done and the deposit tickets were never offered.

We start from what seems the incontestable proposition that, with the evidence pointing to delivery of $15,000 by Moore in February and/or March, 1961, a defense lawyer would have been glad to have information that in the fall of that year Erdman and his accountant could not identify deposits of approximately this amount in February-- especially when three were in a new account and the last was in the not insignificant sum of $7,500, exactly half of the money allegedly paid. And we put aside the suggestion that if the non-disclosure of the report would otherwise entitle Keogh to relief, the delivery of Erdman's records to his counsel during the trial would defeat this. While the unexplained deposits were reflected in the records, defense counsel's attention had naturally been directed rather to the cash deposits, whereas the FBI report showed that any pay-dirt was in the four deposits of early February, all but one of them checks. Indeed, if we were to suppose that the prosecutor remembered the report of November 28, 1961, on June 8, 1962-- which we do not suggest-- his threat to recall Erdman to explain the inexplicable would have a troubling aspect, unless, of course-- which we have not been told-- an explanation had turned up in the interval. So we examine the two considerations sustained by the district court, to wit, that the deposits, or at least those in the Chemical, were check deposits whereas the money was paid in cash, and that the unexplained deposits antedated the payments. We think these factors minimize but do not wholly obliterate the value the evidence might have had to the defense.

It is true that for an extortionist to convert cash into checks would be unusual behavior. Yet, if it were to be assumed that Erdman was trying to conceal his pocketing of the money, it would not be altogether fantastic for a defense lawyer to argue that a transfer of the cash to some 'friend' of the sort Erdman had, who would then furnish him with a check to be deposited in a new bank account, would occur to him as a desirable course of action.

The point as to the dates is much more impressive, and requires detailed scrutiny of the various statements by Moore and Erdman on that score. In Moore's first statement to the FBI, given on June 29, 1961, he told of a series of discussions 'in the latter portion of February, 1961, but prior to February 24, 1961' which ended up in his bringing to Erdman '$15,000 in cash, $10,000 of which Kerner and I borrowed from a finance company in New York City,' to wit, Esco Trading Company, and '$5,000 I obtained in cash from the Meadowbrook National Bank which was sent to me by Gabe Forman.' In a supplementary statement of July 5, he placed his first meeting with Erdman, at which nothing was paid, 'about February 10, 1961.' He also revised his first statement to reflect that the $15,000 all emanated from Forman, with two transfers of $5,000 and $10,000, and that 'The $15,000 was paid to Erdman at least a week before we were indicted, which indictment was about February 24, 1961.' An August 1 FBI report of an interview with Erdman recited that the first $15,000 was paid 'in the early part of the year,' the money-- apparently equally divided-- being in two envelopes one of which was given to Kahaner and the other to Keogh. In an August 29 interview under oath Erdman placed the episode as 'in March.'

When Moore first testified before the grand jury on July 12, he again told of a single first payment of $15,000, with no date fixed, to be divided equally between Keogh and Kahaner. On reappearing on October 10 he corrected his previous testimony to make clear there were two payments-- 'I gave him five thousand, then a few days later ten thousand.' Again no precise dates were fixed. Two weeks later he testified as follows:

'The first payment was five; then maybe a week later, the ten; and then maybe a month later the twenty in one shot. 'So I gave him the money probably the last week of February, the five and then * * *. 'In fact, come to think of it now, it must have been the first week in March, because he actually didn't get the ten thousand until after, a day or two after, the indictments. He trusted me.'

When Erdman testified before the grand jury on August 2, 1961, he fixed his meeting with Moore, Deutsch and Cohen, see 317 F.2d at 477-78, as 'in the month of February,' saying that he couldn't be more specific about dates because he hadn't checked his appointment book. He also testified that the $35,000 was paid in two installments-- 'At one time I thought it was three but I was probably mistaken'-- and that Moore made his first payment, divided between Kahaner and Keogh, about the time Erdman first brought up moore's problem with Keogh. (At the trial Erdman dated this near the end of February, 317 F.2d at 464.)7

In his direct testimony at trial, Erdman fixed his meeting with Moore and Kahaner as on February 21 and said that two days later Moore came with an envelope for Kahaner, who picked it up. He remembered this was the day before the indictment. On cross-examination he added that he was able to fix the date of the meeting with Moore and Kahaner as February 21 by office records showing a visit by Kahaner and a Dr. Feldman whom Dr. Erdman was to examine. This date was corroborated by Dr. Feldman, a disinterested witness. But Moore told a rather different story in his direct testimony. He said that Forman had sent the first $5,000 from England around February 5 and that Moore brought Dr. Erdman $5,000 in cash 'about I would say the 10th of February-- yes about the 12th or 13th of February.' As to the second $10,000, he adhered to his previous version of March 7. Moore's placing the delivery of the first $5,000 so early in February created a problem for the prosecution unrelated to that here discussed, since Kahaner admittedly was out of the country from February 2 to 13, see 317 F.2d at 477-478. When cross-examined the next day, Moore said he had received $5,000 from Forman around February 3 and had given it to Erdman 'around the 20th, I think it was.' On further cross-examination he said he had delivered $5,000 'about February 10th-- no, I am sorry, I am wrong on that date, February 20th.' February 10, he added, was the date when Erdman had asked for the money. He thought he had always told the FBI agents and the prosecutors the delivery was on the 20th; his direct testimony as to an earlier date of delivery was a mistake.

Although the principal government witnesses thus ultimately agreed at trial taht the first payment was not made before February 20-- indeed on Erdman's testimony not until February 23-- knowledge of the FBI report could have significantly affected Keogh's trial strategy. See Note, The Prosecutor's Constitutional Duty to Reveal Evidence to the Defendant, 74 Yale L.J. 136, 145-46 (1964). A cross-examiner, armed with knowledge that Erdman had made an unexplained check deposit of $5,539.94 on February 10 and another of $7,500 on February 17, would almost certainly have inquired about them. Even if the FBI report had not been furnished until Erdman's records themselves were delivered, trial counsel might not have so lightly abandoned the effort to make something out of Erdman's bank deposits, and Mrs. Erdman's deposits and withdrawals at the First Federal Savings & Loan Association, See fn. 5, might also have come up for inquiry. If Erdman had remained unable to recollect the source of the four deposits, subpoenas to the banks might have issued-- with what results we cannot say. At the very least it would have been open to defense counsel to argue the implausibility of Erdman's lack of recollection and also to suggest that the overnight change in Moore's trial testimony as to the date of delivery had been prompted not simply by the prosecution's need to get out of the box created by his first having placed this at a time when Kahaner could not have received the money, a point heavily emphasized in summation by Kahaner's counsel, but also by its fear that the earlier date would strengthen the central defense theory that Erdman had pocketed the money. While we in no way intimate that such a suggestion would have been well-founded, it would hardly have outrun the bounds of legitimate advocacy.

Contending that the conviction must be set aside if use of the FBI report by the defense 'might have led the jury to entertain a reasonable doubt about (defendant's) guilt,' Levin v. Katzenbach, 124 U.S.App.D.C. 158, 363 F.2d 287, 292 (1966), and adopting the expansive conception of the term 'might' applied in Levin v. Clark (D.C.Cir. 1967), petitioner argues that, on the basis of what we have just said, he has surmounted what he conceives to be a rather low barrier to issuance of the writ. On the other hand, the Government's position, with which the district judge seemingly agreed, appears to be that unless the FBI report, considered alone and without regard to its effect on trial strategy, would create a fair doubt as to the propriety of the conviction, which it does not, there was no need for the judge even to conduct an evidentiary hearing. We can hardly be dogmatic on the issues 'whether the prosecution's constitutional duty to disclose extends to all evidence admissible and useful to the defense, and the degree of prejudice which must be shown to make necessary a new trial,' which, in its last encounter with the subject, the Supreme Court explicitly declined to determine. See Giles v. State of Maryland, 386 U.S. 66, 74, 87 S. Ct. 793, 797, 17 L. Ed. 2d 737 (1967) (opinion of Mr. Justice Brennan). But we doubt that either of the extreme positions urged upon us is where the balance should, or will, ultimately be struck. Too many opposing forces are at play. In seeking a workable solution courts must consider not only the maximizing of protection to convicted defendants but the avoidance of impossible burdens on prosecutors and the need to preserve the finality of convictions rendered after trials as nearly faultless as human frailties will permit.

Since the issue has most frequently arisen in habeas corpus and its federal statutory equivalent, 28 U.S.C. 2255, we shall begin our discussion in these terms, reserving the question whether the rule in federal coram nobis is different. The easy cases-- at least they now seem so-- are where the prosecutor's suppression is 'deliberate,' by which we include not merely a considered decision to suppress, taken for the very purpose of obstructing, but also a failure to disclose evidence whose high value to the defense could not have escaped the prosecutor's attention. It suffices to cite as examples People v. Savvides, 1 N.Y.2d 554, 154 N.Y.S.2d 885, 136 N.E.2d 853 (1956); Napue v. People of State of Illinois, 360 U.S. 264, 79 S. Ct. 1173, 3 L. Ed. 2d 1217 (1959); Miller v. Pate, 386 U.S. 1, 87 S. Ct. 785, 17 L. Ed. 2d 690 (1967); and United States ex rel. Meers v. Wilkins, 326 F.2d 135 (2 Cir. 1964), and the decisions referred to in Judge Marshall's opinion there. Such cases rarely present a problem as to 'the degree of prejudice which must be shown'; almost by definition the evidence is highly material.

It is also settled that 'the suppression by the prosecution of evidence favorable to the accused upon request violates due process where the evidence is material either to guilt or to punishment, irrespective of the good faith or bad faith of the prosecution,' Brady v. State of Marylnad, 373 U.S. 83, 87, 83 S. Ct. 1194, 1196, 10 L. Ed. 2d 215 (1963). But that principle does not help Justice Keogh. The defense had made no request for investigative reports of Erdman's finances. We cannot agree with petitioner that where there has not been deliberate suppression in either of the senses we have outlined, the absence of a request is irrelevant.8 It serves the valuable office of flagging the importance of the evidence for the defense and thus imposes on the prosecutor a duty to make a careful check of his files. This consideration is of peculiar weight in a case, such as here presented, of a long trial with many witnesses and much documentary evidence. While the open penknife in Griffin v. United States, 336 U.S. 704, 69 S. Ct. 814, 9 L. Ed. 993 (1949), see 87 U.S.App.D.C. 172, 183 F.2d 990 (1950), or the extreme sexual proclivities of the complaining witness in Giles v. Stae of Marylnad, supra, could scarcely have failed to attract and sustain prosecutorial awareness, it is by no means clear that the FBI report would necessarily have struck Keogh's prosecutors as being of appreciable aid to the defense when it was received in November, 1961, or in their preparation for trial, and still clear that they would have recollected its importance when Moore first testified at trial to a delivery date earlier than his final grand jury testimony. Any lawyer who has prepared and conducted a trial like this one, lasting 23 trial days, knows how easy it is to forget items in his files that an unexpected turn in the evidence has made helpful to his own case, let alone those that might aid his opponents-- something which, under our adversary system, is not exactly in the forefront of his mind. Moreover, while Brady eliminated the issue of good faith when a suitable request has been made, the Court retained the requirement of materiality, as had earlier decisions of ours, see United States v. Consolidated Laundries Corp., 291 F.2d 563, 570-571 (2 Cir. 1961); Kyle v. United States, 297 F.2d 507, 513-515 (2 Cir. 1961), without, however, being obliged by the facts to define with precision just what this meant.

There remains a third category of cases-- where the suppression was not deliberate in either of the senses we have included and no request was made, but where hindsight discloses that the defense could have put the evidence to not insignificant use. While we do not dispute that relief may sometimes be granted even in such cases, the standard of materiality must be considerably higher. This is because, as we said in Kyle v. United States, supra, 'The pans contain weights and counterweights other than the interest in a perfect trial.' Deliberate prosecutorial misconduct is presumably infrequent; to invalidate convictions in the few cases where this is proved, even on a fairly low showing of materiality, will have a relatively small impact on the desired finality of judgments and will deter conduct undermining the integrity of the judicial system. The request cases also stand on a special footing; the prosecution knows of the defense's interest and, if it has failed to honor this even in good faith, it has only itself to blame. Failure to appreciate the use to which the defense could place evidence in the prosecution's hands, or forgetfulness that it exists when a development in the trial has given it a new importance, are quite different. Since this must happen to the most scrupulous prosecutors and the issue of deterrence scarcely arises, the problems of the courts and the wider interests of society unite to require a substantially higher probability that disclosure of the evidence to the defense would have altered the result. To invalidate convictions in such cases because a combing of the prosecutors' files after the trial has disclosed evidence possibly useful to the defense but not likely to have changed the verdict would create unbearable burdens and uncertainties.

When we consider a coram nobis petition such as this, we find further reason for insisting on a relatively high showing of materiality. In habeas corpus or its federal statutory counterpart, the prosecution has a significant interest in retrying the petitioner, if it practically can, in order to return him to prison for the balance of his term. But Justice Keogh has served his sentence, and, apart from constitutional questions, no judge would impose a new sentence if he were to be convicted again. It would be hard to justify the devotion of the prosecutorial or judicial effort required for another long trial where the only consequence of interest to the Government would be a continuation of Keogh's civil disabilities and vindicating what it had accomplished before. While this does not mean that Justice Keogh's interest in restoration of his civil rights, thereby eradicating the stigma of his conviction, is not deserving of every appropriate consideration, the unlikelihood of a new trial where he could utilize the undisclosed evidence gives special force to the Supreme Court's caution that coram nobis relief should be granted 'only under circumstances compelling such action to achieve justice.' United States v. Morgan, 346 U.S. 502, 511, 74 S. Ct. 247, 252, 98 L. Ed. 248 (1954). It is thus not surprising that successful coram nobis petitions in the federal courts have generally involved such fundamental defects as deprivation of counsel, as in Morgan, or coerced guilty pleas. See Note, Postrelease Remedies for Wrongful Conviction, 74 Harv.L.Rev. 1615, 1619 and n. 35 (1961); Frank, Coram Nobis, ch. 5 (1953). While in Garrison v. United States, 154 F.2d 106 (5 Cir. 1946), a hearing was ordered when the coram nobis petition alleged that two 'material' witnesses perjured themselves as a result of the threats made by officers of the United States, no attempt was made to delineate the applicable standards of materiality. And in Kyle v. United States, supra, both parties argued to us as if 2255 were applicable and we treated the case on that assumption. On a coram nobis petition such as this, it is only when the court concludes that the undisclosed evidence would have permitted the defendant so to present his case that he would probably have raised a reasonable doubt as to his guilt in the mind of a conscientious juror that justice compels the invalidation of the conviction. Indeed it is arguable that, absent deliberate prosecutorial misconduct, coram nobis should be denied unless the undisclosed evidence would raise such a doubt in the court's own mind.9

Our difficulty is that the record is not sufficient to tell us how the case should be decided under either of the standards we have sought to delineate. If the utmost Keogh's defense could have accomplished was to engage in the forensic endeavors outlined early in our discussion and the nondisclosure was an excusable oversight, we would readily sustain the dismissal of the petition by the district judge. But we cannot be certain from the record that these 'ifs' are so. However unlikely at this late date, it is not altogether impossible that petitioner might be able to show that Erdman's unexplained check deposits came from a suspicious source. If that were proved, the judge might conclude the testimony of Moore and Erdman as to a later delivery date was not so impregnable that a conscientious juror might not have more than a capricious doubt of Keogh's guilt. Per contra it is possible that further work on Erdman's and the banks' records might indicate an innocent source of the deposits and thereby put the entire controversy to rest. Records of the Meadowbrook National Bank may either confirm or negate the possibility that a payment to Moore could have been made early enough for Erdman to have translated it into a check and deposited it by February 17. Still other lines of inquiry relevant to the limited issue of the effect that knowledge of the unexplained bank deposits by the defense might have had upon the trial may suggest themselves; simply to take one example, the date when Moore appeared before Referee Castellano, see fn. 7, should be ascertainable. Furthermore the record furnishes no information as to what thought, if any, the prosecutors gave to making the FBI report available to the defense. The judge properly did not consider their affidavits, and these do not go to that issue in any event.

We recognize that possibly, perhaps even probably, the district court will know little more after a limited evidentiary hearing that it did when it rendered its decision. In the last analysis Judge Weinfeld, with his intimate knowledge of the trial including the portion no record can convey, cf. Dyer v. Macdougall, 201 F.2d 265, 268-269 (2 Cir. 1952), must determine whether the Government's failure to turn over the report was sufficiently serious in its motivations or consequences to warrant the extraordinary relief of coram nobis. We hold only that he should have had the benefit of an evidentiary record on this single issue concerning the FBI report of Erdman's finances before making that decision. Cf. United States v. Roberts, 388 F.2d 646 (2 Cir. 1967).

The order denying the petition is affirmed save with respect to paragraph 5C as to which it is vacated with instructions to conduct an evidentiary hearing consistent with this opinion.