We have before us appeals from a number of post-trial orders of Judge Blumen-feld in the District Court for Connecticut arising out of the conviction of James Miller for conspiring to import heroin in violation of 21 U.S.C. §§ 173-74. We affirmed the conviction, 2 Cir., 381 F.2d 529 (1967), in an opinion noting that the appeal, was “rather unusual these days in that Miller claims he was innocent of the crime charged,” 381 F.2d at 531; in the same opinion we affirmed the denial of Miller’s first and somewhat pro forma motion for a new trial. After holding a petition for certiorari for many months, the Supreme Court denied this on the last day of its 1967 Term, 392 U.S. 929, 88 S.Ct. 2273, 20 L.Ed.2d 1387 (1968). Meanwhile the district court had entered an order, 277 F.Supp. 200 (1967), denying a second motion for a new trial based on confessions made under highly peculiar circumstances by one Mario Natalizio, who was not brought before the court and had repudiated the confessions. A further order, 296 F.Supp. 422 (1968), denied two other motions, one seeking a new trial on grounds relating to the merits, and another based on a supposed threat to one of the jurors, see 381 F.2d at 538-540; 284 F.Supp. 220, aff’d, 403 F. 2d 77 (2 Cir. 1968).1 We are constrained to reverse so much of the last order as refused a new trial on the ground of the Government’s failure to disclose to the defense the pre-trial hypnosis of its prin-cipaj witness, in part by the attorney who examined him at trial. This ruling and our concomitant direction of a new trial render the other issues moot.
Our first opinion made plain how heavily the case against Miller rested on the identification of him by Joseph Michel Caron and the latter’s wife, Ida, see 381 F.2d at 531-533, 537-538. Apart from these identifications there were five pieces of evidence linking Miller with the crime undoubtedly perpetrated at Bridgeport by someone. These were (1) that Rivard, Caron’s boss in Canada, had referred to a Bridgeport contact as “Frank,” an alias Miller was shown to have used fifteen years earlier, see 381 F.2d at 536-537; (2) that Rivard had said “Frank” was a hairdresser, which Miller concededly was; (3) that Rivard had also characterized “Frank” as hanging out in New York City between 44th and 48th Streets and Miller was shown to have been a patron of the Luxor Baths on 46th Street; (4) that “Frank” had excused his failure to keep an appointment with Caron on the ground of having been at a party the night before and Miller was proved to have been up late; and (5) that, on the day Caron was arrested while attempting to bring a shipment of heroin across the border into Texas, Miller left his beauty parlor, although solidly booked with appointments, and checked in at the Luxor Baths between 2 A.M. and 6 A.M. the next day. While these afforded impressive substantiation of the identifications, only items (4) and (5) were independent of what Rivard allegedly told Caron and only item (5) had significance completely independent of Caron’s testimony. The correctness of the identifications and Caron’s veracity were therefore crucial.
In May 1968, at the request of defense counsel, the Government arranged to have Caron attend at the New York City office of Dr. Herbert Walker for a psychiatric interview including hypnosis in the presence of counsel for both sides. Just before being placed under hypnosis, Caron disclosed he had been hypnotized while in custody in Texas. The circumstances are set forth in a contemporaneous memorandum dated February 23, 1966, by William B. Butler, Assistant United States Attorney for the Southern District of Texas, a copy of which had *828been forwarded along with other papers in the case to the United States Attorney for Connecticut,2 and in Butler’s testimony at a hearing held before Judge Blumenfeld on the fourth new trial motion. We summarize the facts as developed in these sources:
Caron had been brought to the Southern District of Texas shortly before the trial of Miller’s codefendants in September 1965 and had been extensively interviewed by the prosecutors, including Butler. Caron had then gone into some detail about the car used by the recipients of the second Bridgeport delivery that Butler had not previously heard; he identified the car as an old Buick and indicated that he had taken particular note of the license number.3 Since the identity of the recipients at Bridgeport was of no great importance in the Texas trial, the prosecutors had not pressed him on that point.
After the Texas trial Butler heard that Caron was threatening not to cooperate in the separate Miller trial, the continuance and prospective transfer of which to Connecticut had apparently become known to him. Butler went to see Caron in jail on February 16, 1966, “to assure him I would do the best I could to help him with his problems,” and also to revive the question of his memory of the license number of the Buick. The best Caron could do was to recall that the number included 26. Having “a slight knowledge of hypnotism,” Butler asked Caron “whether he would be willing to submit to hypnotism to see whether he could recall the license plate.” Caron agreed.
Butler was back on February 21 with Edward B. Cushing whom he regarded as an expert in hypnosis. Cushing got the subject into hypnosis quickly. Caron remembered that the car was a green Buick and that the plates were black on yellow, see fn. 3. He muttered the letters AM and then some numbers Butler thought were 526 but Cushing identified as 1826, as Caron confirmed. Since Caron was manifesting some distress, Cushing brought him out of hypnosis. After an interval, a second attempt by Cushing failed to produce hypnosis. Butler decided to take over, thinking that despite his “very limited knowledge of hypnotic techniques,” he might have more success because of Caron’s confidence in him. He asked Caron to reconstruct the early part of his second trip to Bridgeport in his mind and to tell what he recalled. Caron began this recital substantially as in his interviews with the prosecutors but when he reached the point of being in the restaurant for lunch with his family, 381 F.2d at 532, Butler’s questions became more searching and Caron gave details Butler had not heard before. He described that the Buick would not start, that Frankie and the other man had to raise the hood, and. that one of them complained over the choice of the car. There were six digits on the license plates but, while Caron put forward a considerable number of letters and numbers, he was uncertain about them. After suggesting that Caron dream about the license plates that night, Butler brought Caron out of hypnosis.
Butler and Cushing did not let go that easily. After lunch, Cushing again put Caron into hypnosis, and Butler took over the interrogation, this time proceeding to the license plates rather swiftly; Caron was now quite sure about “AM 1826.” Cushing sought a description of the second man and got *829one consistent with what Caron had previously given.
Once the initial surprise over use of hypnosis has subsided, the incident seems rather colorless. Butler appears to have acted in entire good faith in subjecting Caron to hypnosis. He was endeavoring to revive Caron’s memory about the license number, and a correct recollection of this could have tended to lead away from Miller as much as toward him.4 Neither Butler nor Cushing suggested anything to Caron he had not already said.5 The Government made no use of Caron’s dim recollection of the license number at the trial, and the defense does not assert it would have been assisted by this. To the contrary, the one point on which Caron was clear, namely, the colors of the plates, points away from the defense’s current theory, see fn. 4.
Naturally the defense did not leave the matter in this stance. It adduced, through affidavit or testimony, the evidence of three psychiatrists expert in hypnosis.6 They claimed in the first instance that repetition under hypnosis of the story Caron had already told tended to imprint this on his mind; in other words, if Caron had once entertained doubts whether Miller was “Frank,” these would have been dispelled by “seeing” Miller as Frank in his reconstruction of the episode under hypnosis.7 After the hypnosis, they contended, Caron would always behold Miller as Frank and would be immune to defense suggestions that he had picked the wrong man. Still more damage was wrought in their view by Butler’s having acted as hypnotist. This, they said, increased the influence Butler already possessed as the man who could help to relieve Caron of the many problems besetting him, and further reduced the reliability of Caron’s testimony at the Connecticut trial, where Butler interrogated Caron and acted for the Government during the protracted cross-examination. They asserted there was a real possibility that Caron may have testified under a mild trance unwittingly induced by his previous hypnotist and that, at minimum, the defense should have had an expert on hand to observe. They claimed also that the ease with which Caron was hypnotized showed a high degree of suggestibility, and proceeded from that to the firm conclusion that his previous recollection of Rivard’s having characterized Frank as a hairdresser 8 was false, being a “gift” he knew the interrogators would like.9
*830While the expert called by the Government regarded many of these claims as exaggerated, he conceded on cross-examination that if the hypnosis was not1 conducted according to expert techniques, what happened “could have had an effect” on Caron’s subsequent ability to correct a mistake.10
Whatever the rights and wrongs of the expert testimony may be, it is undeniable that disclosure of the hypnosis would have added another arrow to the rather large quiver trial counsel for the defense shot at Caron — limited opportunity for observation; initial alleged misdescription and picking the photograph of another man;11 a suggestion through the inclusion of three photographs of Miller in the dozen initially exhibited; knowledge that the prosecution was pleased with his identification; a lengthy criminal record; a letter written from jail telling his wife he had lied in identifying Frank; the Government’s failure to place any charge against his wife, or any charge against him for the Bridgeport crimes; the Government’s maintenance of his family during his incarceration; his fear of deportation; and his hope for pardon or commutation of sentence on the charge of smuggling heroin from Mexico to which he had pleaded guilty. It is not difficult to hear the ringing tones in which defense counsel would have told the jury that, not satisfied with “buying” Caron with money for his family and promises of leniency, the Government had blocked the door to repentance under cross-examination and the sanctity of the oath by hypnotism carried out by none other than the examining prosecutor himself. It is altogether plain that if the defense had known of the hypnosis incident and the court had refused to allow its use at the trial, the amount of other impeaching evidence against Caron would not have avoided a reversal. See United States v. Zborowski, 271 F.2d 661, 667 (2 Cir. 1959).
■ The issue here differs since the request is for a new trial on the ground of newly discovered evidence.12 We have noted, Kyle v. United States, 297 F.2d 507, 512 (2 Cir. 1961), the two classical formulations of the criterion — whether the new evidence “is so material that it would probably produce a different verdict, if the new trial were granted,” Berry v. State, 10 Ga. 511, 527 (1851), and whether it might have produced a different verdict, Larrison v. United States, 24 F.2d 82, 87 (7 Cir. 1928)— a test that has been stated to be limited to cases of “recantation or where it has been proved that false testimony was given at the trial.” See United States v. Hiss, 107 F.Supp. 128, 136 (S.D.N.Y. 1952), aff’d, 201 F.2d 372 (2 Cir.), cert. denied, 345 U.S. 942, 73 S.Ct. 830, 97 L.Ed. 1368 (1953). In United States v. Johnson, 327 U.S. 106, 111 n. 5, 66 S.Ct. 464, 90 L.Ed. 562 (1946), the Supreme Court also noted the two tests but did not determine their respective areas of application. While we do not think the hypnosis evidence would meet the Berry test, we find it unnecessary to decide this or to consider whether the weaker test of Larrison is ever applicable in the ordinary case of evidence newly discovered by the defense. For we think developments during the trial placed a duty on the Government to disclose the hypnosis; that, where such a duty has not been discharged, *831a motion for a new trial must be granted if there is a signficant possibility that the undisclosed evidence might have led to an acquittal or a hung jury;13 and that such a possibility exists here.
Miller argues that the Government had a duty to disclose the hypnosis, quite apart from any request, on the basis that this was “evidence whose high value to the defense could not have escaped the prosecutor’s attention.” See United States v. Keogh, 391 F.2d 138, 146-147 (2 Cir. 1968), and authorities there cited. The Government responds that the value is discernible only by hindsight in the bright light allegedly shed by the defense’s psychiatric experts. It points, as a demonstration of its good faith, to its acquiescence in the hypnosis of Caron by a defense expert — something it would hardly have done if it had supposed this would “contaminate” him, to use the defense’s favorite phrase, as a witness in the new trial Miller was seeking, see fn. 7. We are not required to decide the question thus posed since we find that the combined effect of two episodes in the trial was to create a duty of disclosure even if none existed before.
Repeatedly during several days of cross-examination, Caron was questioned about the license number of the green Buiek. He was asked particularly about a letter he had written Butler in March 1966, giving various possible license numbers and other details with respect to the Buiek car. In answer to a rather general question Caron said he came to give the numbers because “they ask me if I can recall numbers. I give them an idea what was on that license plate.” When asked “Were these numbers suggested to you, or did you write — indent the numbers yourself that are on this exhibit?”, Caron answered, “This is my own thought.” What Caron said may have been the truth but it was hardly the whole truth as known to Butler, and should have brought to his mind the hypnosis performed only three months before and the interest the defense could have in this.
Defense counsel had earlier requested notes of all interviews with Caron in the Galveston jail. After the judge had remarked that he understood the Government had furnished all the Jencks Act material, Mr. Butler volunteered that he had notes of interviews, particularly those made in preparation for the Texas trial, but that he did not consider these within the Jencks Act, 18 U.S.C. § 3500. Recognizing differences of opinion on the subject, defense counsel requested “exactly what Mr. Butler says he has, and has done, and has dictated as a result of the interview of this man, Joseph Michel Caron.” The judge directed that such notes be submitted for in camera inspection. When the court asked whether “everything that was conceivably Jencks Act material in this case had been furnished to defense counsel or to the Court,” Mr. Butler answered in the affirmative, adding “We made a special effort to search the files.” But his special effort had not unearthed his memorandum describing the hypnosis, although admittedly it was there. We find it hard to disagree with the defense’s contention that at least parts of the memorandum came within 18 U.S.C. § 3500(e) (2). In any event there was a sufficient likelihood of this that the memorandum should have been delivered to the court for inspection, and we think it plain that the judge would then have required disclosure of the hypnosis even if he did not consider the memorandum producible under the Jencks Act.
After full allowance for the problems confronting counsel in the stress of a long and heated trial which we noted in United States v. Keogh, supra, 391 F.2d at 147, the prodding furnished by these episodes must be taken to have served the “valuable office” of *832“flagging” the pretrial hypnosis, see id., thus placing upon Mr. Butler a duty of disclosure he did not discharge. Even if he did not recall the hypnosis or remember or locate his memorandum of it, he should have, and negligence of the prosecutor in failing to make evidence available to the defense reduces the standard of materiality needed to require the granting of a new trial below the formulations applicable where no prose-cutorial misconduct exists. See United States v. Consolidated Laundries Corp., 291 F.2d 563, 570-571 (2 Cir. 1961); cf. Kyle v. United States, supra, 297 F.2d at 515.
To reduce but not to eliminate it. Negligent or even intentional failure of the prosecutor to disclose an additional item of impeaching evidence would not invariably require that a verdict be set aside. Here, as we have noted, the defense already possessed an abundance of impeaching material, the hypnosis incident seems without much force apart from the elaboration by the defense’s psychiatric experts, and we have our own views how seriously a jury of Connecticut Yankees would have been likely to regard that. Still, as the record stands, the hypnosis had arguably placed at least some obstacle in the way of one of the most valuable protections accorded Miller by the Sixth Amendment — the possibility that the sanctity of the oath and effective cross-examination might lead Caron to recant his identification or at least to admit doubt. The defense could and very likely would have made this the capstone of its attack on the crucial witness for the prosecution — with what effect we cannot confidently say. Under all the circumstances, as Judge Swan wrote in Consolidated Laundries, supra, 291 F.2d at 571, “we are content to rest our decision on our conviction that the denial of a new trial here is inconsistent with the correct administration of criminal justice in the federal courts, which it is our duty as an appellate court to supervise.”
We have reached this conclusion with some reluctance, particularly in light of the considered belief of the able and conscientious district judge, who has lived with this case for years, that review of the record in light of all the defense new trial motions left him “convinced of the correctness of the jury’s verdict.” We, who also have had no small exposure to the facts, are by no means convinced otherwise. The test, however, is not how the newly discovered evidence concerning the hypnosis would affect the trial judge or ourselves but whether, with the Government’s case against Miller already subject to serious attack, there was a significant chance that this added item, developed by skilled counsel as it would have been, could have induced a reasonable doubt in the minds' of enough jurors to avoid a conviction.14 We cannot conscientiously say there was not. With the panoply of rules that our legal system has devised to insure fairness in criminal trials, the least appropriate instance for a grudging attitude in applying them would be one where a defendant supposedly leading a blameless life has steadfastly asserted that the prosecution got the wrong man and the proof consisted almost wholly of an identification by an accomplice in the conspiracy and the latter’s wife and other testimony by the former. If the price of our decision should be the ultimate escape of a guilty man rather than the vindication of an innocent one, this is the kind of case where that price is worth paying.15
Since we hold that a new trial is required because of the failure to disclose *833Caron’s pre-trial hypnosis, in part by Mr. Butler himself, it is unnecessary to consider the many other grounds urged by the defense.
The order of November 26, 1968, is reversed insofar as it denied a new trial on the ground of nondisclosure of Car-on’s hypnosis, and a new trial is directed. The other appeals are thereby mooted and are dismissed.
. The order denied a defense request to take testimony of three jurors and of a Connecticut state trooper. The court held that nothing new had been presented; rather the testimony of the theretofore unidentified man who conveyed the “threat” was that there had been no threat at all and reduced the importance of the episode close to the vanishing point.
. Miller was indicted along with six other defendants in Texas and was originally to be tried there. When Miller’s case was transferred for trial to Connecticut, Mr. Butler, who had participated in the Government’s pre-trial investigative work and in the Texas trial of the codefendants, came to Connecticut to act as prosecutor of Miller. Mr. Butler handled the first government witnesses, including Caron, before turning over the remaining prose-cutorial duties to the United States Attorney for Connecticut.
. Caron had previously stated that he remembered the plates as being yellow and black — New York’s colors.
. The current theory of the defense is that the Buiok used in the second Bridgeport delivery was a Florida car registered in the name of Natalizio’s landlord.
. The tempest stirred up by defense psychiatrists and counsel over the 526- v. 1826 incident recited in the text seems to be in something even smaller than a teacup.
. One of these was Dr. Walker, whom the Government had allowed to hypnotize Caron after the trial.
. The Government cogently asks why, if Dr. Walker knew this to be the probable effect of questioning Caron under hypnosis, the defense, when ignorant of the previous episode, wanted Caron to be hypnotized by him in anticipation of a new trial at which Caron would testify again. The response based on Dr. Walker’s greater skill is unconvincing.
. Caron had reason to know Miller was a hairdresser from the circumstances of his identification of Miller at Bridgeport and possibly from other sources as well. The defense claims that Hiyard’s alleged characterization first came out during the intensive interviewing of Caron in September 1965, prior to the Texas trial, and contends that the Government’s failure to disclose the lateness of this narration is an added ground for a new trial. Apparently none of the § 3509 material of previous interviews included such a statement and Butler did not recall it as having been made any earlier in the nearly two years during which Caron had been in custody. However, at the examination in Dr. Walker’s office, Caron claimed he had recounted this from the beginning. Characterizing this as a “patent falsehood,” the defense says it is further proof of the damaging effect of the hypnosis.
. This is asserted with confidence despite the fact that the interrogation before the Texas trial when Caron is claimed to have made the “gift” was not particularly con*830cerned with Miller, whose case had been severed.
. Further testimony makes it arguable that the government witness’ concession of the possibility of harm was limited to cases where a suggestion was “implanted” during hypnosis. However, on recross-examination he made an affirmative answer to an inquiry whether “repeating what had been said when previously the man had some doubts, under the hypnosis, will at least lessen if not remove these.” The examination then drifted off in a way that leaves considerable question just what the witness did think.
. Although with a reservation, see 381 F.2d at 537.
. The Government does not contend that the defense could reasonably have been expected to discover the hypnosis incident.
. AVhile many of the, leading decisions on nondisclosure have arisen on collateral attack, the standard of materiality with respect to a motion for a new trial obviously cannot be higher. See generally 8A Moore, Federal Practice § 33.02 [3] (2d ed. 1968).
. We put it that way because of evidence that “it requires a massive minority of 4 or 5 jurors at the first vote to develop the likelihood of a hung jury.” Kalven & Zeisel, The American Jury 462 (1966). \ However, at the last vote, one juror for acquittal prevented a verdict in 24% of the hung jury cases studied. Id. at 460.
. Nothing in our decision is to be taken as implying any view that Caron’s hypnosis disqualifies him from testifying on a new trial.