*1301OPINION OF THE COURT
ALDISERT, Circuit Judge.The major question for decision in this appeal from the district court’s grant of a writ of habeas corpus is whether the state of New Jersey violated the requirements of Brady v. Maryland, 373 U.S. 83, 83 S.Ct. 1194, 10 L.Ed.2d 215 (1963), by failing to disclose to the defendant certain reports of a lie detector test administered to an important prosecution witness. The district court determined that this evidence was material to the defendant’s guilt or innocence and therefore that its suppression denied him due process. We conclude that the district court did not err and therefore will affirm the judgment.
I.
The underlying facts may be distilled from the many recorded opinions in this highly publicized case. At approximately 2:30 a.m. on June 17, 1966, two armed men entered the Lafayette Bar & Grill in Paterson, New Jersey and opened fire. The bartender, James Oliver, and a patron, Fred Nauyaks, were killed immediately. A second patron, Hazel Tanis, died one month later from her wounds and a third patron, William Marins, was partially blinded after being shot in the head.
Patricia Valentine lived above the tavern and was awakened by the sound of gunshots at about 2:30 a.m. She ran to her window and saw two men leave the scene in a white car. At the same time, Alfred Bello and Arthur Bradley were breaking into a nearby factory. Bello, who was standing lookout, was either in or outside the bar, and his location at that time is a main point of contention for the Brady issue. Within minutes, police arrived at the scene and took statements from Marins, Valentine and Bello. The officers transmitted a description of the car on police radio. A few minutes later, a Paterson police officer stopped a white car leased by Rubin Carter, a well-known prize fighter, approximately 14 blocks away from the Lafayette Bar. John Artis was driving, John Royster was sitting in the front seat, and Carter was alone in the back seat. The car had not been speeding and there were no weapons in sight. Carter told the officer that the men were driving to his home about six blocks away to obtain money, and the officer allowed them to proceed. Fifteen minutes later, police observed Carter’s car outside the La Petite Bar, approximately 10 blocks west of the Lafayette. Five minutes later, police sighted the car a third time, with only Carter and Artis inside. This time the police escorted the car and its occupants to the crime scene.
From the evidence, it appears that at the original questioning neither Bello nor Valentine identified Carter or Artis at the scene, and there is considerable dispute as to the identification of the car. The police took Carter and Artis to the station and then to the hospital where the two survivors failed to identify them. They were questioned. Both Carter and Artis were given polygraph examinations. They were released about 7 p.m. on June 17. Meanwhile, police searched the car and allegedly found a live .12 gauge shotgun shell in the trunk and a live .32 caliber shell on the floor of the front seat, the same calibers as the weapons used in the Lafayette shootings.
Subsequently, Carter and Artis voluntarily testified before a Passaic County Grand Jury which did not return any indictments. In July and October, however (and apparently for the first time), Bello told a Paterson police officer that he had seen Carter and Artis at the crime scene. Bello and Bradley then identified Carter as one of the two men they saw coming from the Lafayette Bar with weapons in their hands. Bello, but not Bradley, identified Artis as the second man.
At least in part as a result of these identifications, in 1967 Carter and Artis were indicted, tried and convicted of three counts of first degree murder. On June 29, 1967, the trial judge imposed life sentences after a jury recommendation of mercy. The New Jersey Supreme Court affirmed the convictions on direct appeal. State v. Carter, 54 N.J. 436, 255 A.2d 746 (1969), cert. denied, 397 U.S. 948, 90 S.Ct. 969, 25 L.Ed.2d 130 (1970).
*1302On October 1, 1974, Carter and Artis filed a new trial motion based on the statements of the prosecution’s two key witnesses — Bello and Bradley — recanting their 1967 identification testimony. The original trial judge denied the motion. State v. Carter, 136 N.J.Super. 271, 345 A.2d 808 (1974). A second new trial motion based on other grounds was made on January 30, 1975, and was also denied. State v. Carter, 136 N.J.Super. 596, 347 A.2d 383 (1975). The New Jersey Supreme Court reversed, overturned the convictions, and ordered a new trial. State v. Carter, 69 N.J. 420, 354 A.2d 627 (1976). The court determined that the prosecution had withheld certain evidence from the defense. Such evidence indicated that prosecutors had offered the key identification witnesses both protection of their persons and assistance with criminal charges that were then pending or contemplated against them. The case was remanded to the trial court where, after numerous motions and hearings, the retrial began on October 12, 1976.
Although Carter and Artis contend that their 1976 trial was constitutionally deficient in several respects, we find the Brady issue to be dispositive. At the 1967 trial, Alfred Bello testified that he had seen Carter and Artis carrying weapons on the street outside the bar immediately after the killings. As noted above, however, Bello recanted portions of this testimony in 1974 and continued to revise his story during 1975 and 1976. At a recantation hearing in October 1974, he said he could not identify either Carter or Artis as having been near the Lafayette Bar & Grill after the shootings. During August 1975, Bello made a series of tape recordings with Melvin Ziem and Joseph Miller in which he recounted several additional versions of the events following the Lafayette Bar killings. On October 30, 1975, Bello stated in an affidavit that he had been “in the bar” — not on the street — at the time of the shootings, and that immediately afterward he ran outside where he saw Carter and Artis on the street. Bello said he did not remember seeing them in the bar and he did not believe they were the trigger men, but thought they were somehow involved. In another affidavit executed the next day, Bello said that he had not seen Carter and Artis with weapons or in the bar. In another version of the story — presented to an Essex County grand jury in December 1975 and to the prosecution in June 1976 — Bello said that he was in the bar where he saw two other men, while Carter and Artis waited outside.
Before the second trial of Carter and Artis, the state arranged for a polygraph examination by Professor Leonard H. Harrelson which took place on August 7, 1976. The purpose of the examination was to evaluate Bello’s credibility. As a result of the polygraph, Harrelson concluded that Bello was telling the truth when he said he was in the bar shortly before and at the time of the shooting. He also concluded that Bello saw Carter and Artis outside the bar after the shooting. The same day, Harrelson gave an oral report of his findings to Passaic County Assistant Prosecutor Kayne and Chief of Detectives DeSimone. DeSimone told Harrelson that the polygraph examiner’s conclusion was impossible: that Bello could not have been inside the bar at the time of the shooting.
On August 11, 1976, Harrelson repeated his conclusions by telephone. Two weeks later, Harrelson filed a written report with the prosecutor’s office. That report summarized his conclusions by stating that: “After careful analysis of [Bello’s] poly-grams, it is the opinion of the examiner that [Bello’s] 196[7] testimony at the trial was true, and the statement recanting his original statement is not true.” App. Vol. 4E at 623. As noted ábove, however, Bello testified at the 1967 trial that he had been on the street at the time of the shooting; in his oral reports to the prosecutor and police, Harrelson concluded that Bello was telling the truth when he said that he had been in the bar at the time of the shooting. To this extent, Harrelson’s written report contradicted his earlier oral reports.1 The *1303prosecution disclosed to the defense Harrelson’s written report, but did not disclose the substance of the earlier oral reports. At trial, Bello testified — consistent with his 1967 testimony but contrary to the finding expressed in Harrelson’s oral reports — that he had been on the street during the shootings. No evidence of Harrelson’s polygraph examination was introduced at trial.
On December 22, 1976, the jury returned first degree murder verdicts against Carter and Artis. Carter and Artis appealed to the appellate division of New Jersey Superior Court and the New Jersey Supreme Court. The New Jersey Supreme Court ultimately remanded a new trial motion to the trial court for a hearing on issues surrounding the polygraph test. State v. Carter, 85 N.J. 300, 426 A.2d 501 (1981). The remand hearing lasted 15 days. In an 80-page unreported opinion the judge found against the defense on all issues. On appeal from this decision, the supreme court decided two things: first, and unanimously, that a Brady violation had occurred; second, by a 4-3 majority, that Carter and Artis failed to establish that the suppressed evidence would “have been material to the outcome.” State v. Carter, 91 N.J. 86, 119, 449 A.2d 1280, 1297 (1982). The majority ruled that the suppressed evidence was merely cumulative. Id. at 118, 449 A.2d at 1297. The dissent, however, felt that the evidence was material. Id. at 135-40, 449 A.2d at 1307-09 (Clifford, J., dissenting).
Carter and Artis submitted petitions for habeas corpus to the United States District Court for the District of New Jersey on February 13 and 28, 1985, respectively; the actions were later consolidated. On November 7, 1985, the district court granted petitioners’ motion for summary judgment, basing its decision on two of the twelve grounds advanced by petitioners: first, that the prosecution violated the requirements of the fourteenth amendment due process clause by asserting a “racial revenge” motive for the murders; and, second, that the state violated the requirements of Brady v. Maryland by failing to disclose the results of the lie detector test administered to Bello shortly before the second trial. Carter v. Rafferty, 621 F.Supp. 533 (D.C.N.J.1985). Our decision on the Brady issue makes it unnecessary for us to address the first basis of the district court’s judgment.
II.
As a preliminary matter, we hold that appellants have effectively appealed from only that portion of the district court’s judgment relating to Rubin Carter. The notice of appeal, filed on November 8, 1985, specifically limited itself to the order releasing Carter:
Notice is hereby given that [appellants] hereby appeal to the United States Court of Appeals for the Third Circuit from the order granting petitioner Carter’s habeas corpus petition and releasing petitioner Carter from the custody of the State of New Jersey entered in this action on the 8th day of November, 1985.
*1304App.Vol. ID at 79-80. Moreover, although appellants maintain that they “intended the appeal to be taken from the issuance of the respective writs of habeas corpus as to Petitioners Carter and Artis,” they concede that “[b]ecause of the limited time available before the [enlargement] hearing, partly because of the uncertainty as to the status of Petitioner Artis noted in the Opinion of the District Court, and partly through inadvertence, the name of Petitioner Artis was not included in the body of the notice of appeal.” Reply of Respondents to Motion of Petitioner-Appellee John Artis to Dismiss Appeal at 5. (Affidavit of John P. Goceljak.)
Rule 3(c), F.R.App.P., requires that “[t]he notice of appeal ... shall designate the judgment, order or part thereof appealed from____” (emphasis supplied). Because the notice here is silent as to Artis, it effectively challenges only that portion of the district court’s judgment granting relief to Carter. See Elfman Motors, Inc. v. Chrysler Corp., 567 F.2d 1252, 1254 (3d Cir.1977) (per curiam). Appellants have failed to file a timely notice of appeal with respect to the grant of relief to Artis. Their failure deprives this court of jurisdiction: “Where the litigant fails to file a timely notice of appeal within the prescribed period, the litigant loses the right to an appeal on the merits of the predicate controversy.” West v. Keve, 721 F.2d 91, 95 (3d Cir.1983) (footnote omitted). Accordingly, we will dismiss appellants' appeal as it relates to John Artis.
III.
Ordinarily on appeal of a district court’s disposition of an application for writ of habeas corpus, the district court’s findings of fact are reviewed under the clearly erroneous standard and its application and interpretation of legal precepts are subject to plenary review. Sullivan v. Cuyler, 723 F.2d 1077, 1082 (3d Cir.1983). Because this action was decided on a motion for summary judgment, however, the appellate court is required to apply the same test that the district court should have used initially under Rule 56(c), F.R.Civ.P.: summary judgment is properly granted only if “there is no genuine issue as to any material fact and ... the moving party is entitled to judgment as a matter of law.” See King v. Brewer, 577 F.2d 435, 439 (8th Cir.1978), cert. denied, 440 U.S. 918, 99 S.Ct. 1238, 59 L.Ed.2d 468 (1979). In making this determination, all reasonable inferences must be drawn in favor of the non-movant. Goodman v. Mead Johnson & Co., 534 F.2d 566, 573 (3d Cir.1976), cert. denied, 429 U.S. 1038, 97 S.Ct. 732, 50 L.Ed.2d 748 (1977). Moreover, subject to certain exceptions, factual determinations made by a state court after a hearing on the merits “shall be presumed to be correct.” 28 U.S.C. § 2254(d); see Cuyler v. Sullivan, 446 U.S. 335, 341-42, 100 S.Ct. 1708, 1714-15, 64 L.Ed.2d 333 (1980).
IV.
In Brady v. Maryland, the Supreme Court held that a defendant’s due process right to a fair trial is violated when the prosecution withholds evidence that is both favorable to the accused and “material either to guilt or to punishment.” 373 U.S. 83, 87, 83 S.Ct. 1194, 1197, 10 L.Ed.2d 215 (1963). See also Moore v. Illinois, 408 U.S. 786, 794-95, 92 S.Ct. 2562, 2568, 33 L.Ed.2d 706 (1972). The purpose of the Brady rule is not to displace the adversary system as the primary means by which truth is uncovered, but to ensure that a miscarriage of justice does not occur. Thus, the prosecutor is required to disclose only that evidence favorable to the accused which, if suppressed, would deprive the defendant of a fair trial. “[U]nless the omission deprived the defendant of a fair trial, there was no constitutional violation requiring that the verdict be set aside; and absent a constitutional violation, there was no breach of the prosecutor’s constitutional duty to disclose.” United States v. Agurs, 427 U.S. 97, 108, 96 S.Ct. 2392, 2399, 49 L.Ed.2d 342 (1976). Accordingly, “the prosecutor will not have violated his constitutional duty of disclosure unless his omission is of sufficient significance to result in the denial of the defendant’s right to a fair trial.” Id.
*1305The Court has explained that in addition to exculpatory evidence, the Brady rule covers evidence that might be used for impeachment purposes. Giglio v. United States, 405 U.S. 150, 154, 92 S.Ct. 763, 766, 31 L.Ed.2d 104 (1972). Impeachment evidence is thus “evidence favorable to an accused,” Brady, 373 U.S. at 87, 83 S.Ct. at 1196, where, if disclosed and used effectively, it may make the difference between conviction and acquittal. United States v. Bagley, 473 U.S. 667, 676, 105 S.Ct. 3375, 3380, 87 L.Ed.2d 481 (1985). Accordingly,
When the “reliability of a given witness may well be determinative of guilt or innocence,” nondisclosure of evidence affecting credibility falls within this general rule [of Brady ]. We do not, however, automatically require a new trial whenever “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[.”] ... A finding of materiality of the evidence is required under Brady____ A new trial is required if “the false testimony could ... in any reasonable likelihood have affected the judgment of the jury____”
Giglio, 405 U.S. at 154, 92 S.Ct. at 766 (citations omitted).
Notwithstanding the convoluted fact situation and the many facets of the teachings of Brady, this appeal turns on one issue and one issue alone: were Harrelson’s oral reports of Bello’s polygraph examination material? In the view we take this is the centerpiece of this appeal. It bears mention that this was also the controlling issue that faced the New Jersey Supreme Court; as stated before, on a narrow division of 4-3, that court found that the oral reports were not material in the context of Brady.
Appellants apparently concede that the reports should have been turned over to the defense before trial. See br. for respondents-appellants at 43-44. They maintain, however, that the reports were not “material” under the controlling legal standard:
[EJvidence is material only if there is a reasonable probability that, had the evidence been disclosed to the defense, the result of the proceeding would have been different. A “reasonable probability” is a probability sufficient to undermine confidence in the outcome.
Bagley, 473 U.S. at 682, 105 S.Ct. at 3384. See also id. at 685, 105 S.Ct. at 3385 (White, J., concurring in part and concurring in the judgment).
Carter’s theory is that the prosecution used Harrelson’s written report to pressure Bello to return to his 1967 “on the street” testimony. He contends that if he had known of Harrelson’s contradictory oral reports, he would have used this information to impeach Bello’s credibility at trial and thus undermine an essential basis for the guilty verdict. The district court agreed, and, casting its lot with the New Jersey Supreme Court dissenters, held that the oral reports were “material” evidence under Bagley. Appellants offer three grounds in their challenge to the determination of materiality.
A.
They first contend that the district court employed an inappropriate standard for reviewing the state court decisions on the Brady issue. The New Jersey Supreme Court determined that although the evidence of Harrelson’s oral reports should have been disclosed to the defense, it was “neither material nor of the sort that would lead to a change in the jury’s verdict.” State v. Carter, 91 N.J. 86, 121, 449 A.2d 1280, 1299 (1982). The state trial court had previously determined, after an evidentiary hearing, that it was not material because the defense had ample opportunity to impeach Bello at trial and therefore “the undisclosed information was merely cumulative and/or repetitious as to the purpose for which it could have been used____” State v. Carter, No. 167-66, slip op. at 74 (N.J.Super.Ct. Law Div.: Passaic Cty. Aug. 28, 1981), reprinted in app. Vol. 1E at 134; see also id. at 56-57, reprinted in app. Vol. 1E at 116-17. Appellants contend that the district court erred because it failed to accept the state trial court’s determination. The district court, argue the appellants, *1306“did not accord the state trial court the level of deference required by 28 U.S.C. § 2254(d)” on the issue of materiality. Br. for respondents-appellants at 48; see also reply br. for respondents-appellants at 23-24. It now becomes our task to analyze the degree of constraint or freedom possessed by a federal habeas corpus court after a state trial court has made a ruling on materiality in the context of Brady. Our starting point is the federal habeas corpus statute.
Section 2254(d) provides that
In any proceeding instituted in a Federal court by an application for a writ of habeas corpus by a person in custody pursuant to the judgment of a State court, a determination after a hearing on the merits of a factual issue, made by a State court of competent jurisdiction in a proceeding to which the applicant for the writ and the State or an officer or agent thereof were parties, evidenced by a written finding, written opinion, or other reliable and adequate written indicia, shall be presumed to be correct, unless the applicant shall establish or it shall otherwise appear____
28 U.S.C. § 2254(d). Neither this court nor the Supreme Court has precisely decided whether a determination concerning materiality is presumptively correct under section 2254(d). It is clear, however, that the presumption of factual correctness may not apply to mixed questions of law and fact. See Miller v. Fenton, 474 U.S. 104, 113-14, 106 S.Ct. 445, 451, 88 L.Ed.2d 405 (1985).2 Although the Court has acknowledged that “[i]t will not always be easy to separate questions of ‘fact’ from ‘mixed questions of law and fact’ for § 2254(d) purposes,” Wainwright v. Witt, 469 U.S. 412, 429, 105 S.Ct. 844, 855, 83 L.Ed.2d 841 (1984), we are safe in suggesting that the Court adheres to the view that the presumption applies only to “ ‘basic, primary or historical facts: facts “in the sense of a recital of external events and the credibility of their narrators____”’” Cuyler v. Sullivan, 446 U.S. 335, 342, 100 S.Ct. 1708, 1714, 64 L.Ed.2d 333 (1980) (citations omitted); see Wainwright, 469 U.S. at 428 n. 8,105 S.Ct. at 854 n. 8.
Cases from the Fifth, Seventh and Tenth Circuits hold that determinations of materiality of evidence under Brady are mixed questions of law and fact and that state court determinations are therefore not entitled to section 2254(d)’s presumption. See, e.g., Bowen v. Maynard, 799 F.2d 593, 610 (10th Cir.), cert. denied, - U.S. -, 107 S.Ct. 458, 93 L.Ed.2d 404 (1986); Ruiz v. Cady, 635 F.2d 584, 589 (7th Cir.1980); Davis v. Heyd, 479 F.2d 446, 451 (5th Cir.1973). This court recently suggested that this is the better approach. See Daniel v. Warden, State Correction Institute, 794 F.2d 880, 882-83 (3d Cir.1986) (double jeopardy mixed question; therefore, no presumption); see also Universal Minerals, Inc. v. C.A. Hughes & Co., 669 F.2d 98, 102-03 (3d Cir.1981); cf. TSC Industries, Inc. v. Northway, Inc., 426 U.S. 438, 450, 96 S.Ct. 2126, 2133, 48 L.Ed.2d 757 (1976) (materiality of misrepresentation or omission under securities laws mixed question of law and fact). Appellants have not cited any case in which a state court’s determination of Brady materiality was presumed correct under section 2254(d).
Accordingly, we now decide that the materiality of evidence under Brady is a mixed question of law and fact and that state court determinations of the law portions of this mixture are not entitled to a presumption of correctness under section 2254(d). We hold that the district court did not err in failing to apply the presumption to the New Jersey courts’ legal conclusions as to the polygraph reports as cumulative and immaterial.
B.
Appellants next contend that Harrelson’s oral reports were not material because “the only use the defense could have *1307made of [them] would be to impeach Bello — a merely cumulative and repetitious endeavor since numerous prior, contradictory statements of Bello saying he was in the bar had already been brought out.” Br. for respondents-appellants at 47. The district court rejected appellants’ characterization of Harrelson’s oral reports. Instead, it accepted Carter’s theory that the police or prosecution confronted Bello with the results of the polygraph test to persuade Bello to return or adhere to the “on the street” testimony of the 1967 trial. Carter v. Rafferty, 621 F.Supp. at 553-54; see also br. for petitioners-appellees at 38-40. Appellants challenge this determination and argue that the New Jersey trial court was correct when it concluded that any further cross-examination of Bello would have been cumulative and repetitious and, therefore, not material under Brady.
We note at the outset the New Jersey trial court’s determination that “[t]he prosecution did not cause Bello to recant his recantation through coercion, deceit or threats of prosecution for perjury; nor did the State misrepresent to Bello the results of the polygraph test to get Bello to go back to the ‘on the street’ version.” State v. Carter, No. 167-66, slip op. at 55, reprinted in app. Vol. 1E at 115. This finding is clearly a “fact”, presumptively correct under section 2254(d). See Cuyler v. Sullivan, 446 U.S. at 342, 100 S.Ct. at 1714-15. But it is only a factual component of the mixed question of fact and law. Accordingly, the trial court’s determination does not compel the conclusion that the defense could not have effectively used knowledge of Harrelson’s oral reports to further impeach Bello’s testimony.
Read in context, the trial court’s determination that the prosecution did not “misrepresent” the polygraph results suggests only that the prosecution acted in good faith. See, e.g., State v. Carter, No. 167-66, slip op. at 26, reprinted in app. Vol. 1E at 86 (“I conclude then that it was not the State’s misuse of the polygraph test that caused Bello to revert back to the 1967 trial testimony.”); id., slip op. at 31-32, reprinted in app. at 91-92 (Although “[t]he [polygraph pre-] test interview was what triggered Bello to return to his original 1967 trial version, the prosecution did not misrepresent the test results to turn Bello around nor did the State threaten Bello with prosecution for perjury in order to coerce him into recanting the recantation.”).
But the good faith of the prosecution is not in issue. As the district court noted,
a finding of intentional misconduct on the part of the prosecution ... makes no real difference under the factual circumstances of this case. Whether the conduct was deliberate or negligent, the consequences to the petitioners were the same: they were deprived of a vital opportunity to totally discredit the key and only eyewitness to the crime. Indeed, if the trial court knew and was satisfied that Bello finally selected one of his many versions merely because he was told that it was independently confirmed by the polygrapher (albeit mistakenly), it might well have stricken his entire testimony.
Carter v. Raffetty, 621 F.Supp. at 553.
The New Jersey trial court made ample findings of fact to support this theory. The court noted that at the hearing on remand, “[w]hen referring to the testimony he had given at the 1976 trial, Bello indicated that the polygraph was what brought him back to the 1967 trial version.” State v. Carter, No. 167-66, slip op. at 26, reprinted in app. Vol. 1E at 86. The trial court determined that “[t]he test interview was what triggered Bello to return to his original 1967 trial version.” Id. at 31, reprinted in app. Vol. IE at 91. Moreover, there was other evidence presented at the remand hearing not mentioned in the trial court’s opinion that conclusively demonstrated the importance of the polygraph results in shaping Bello’s testimony at the 1976 trial. See State v. Carter, 91 N.J. 86, 135-37, 449 A.2d 1280, 1307-08 (1982) (Clifford, J., dissenting) (after reviewing the record evidence, “[t]he conclusion is iiies*1308capable that ... the ‘wrong’ test result was fed to Bello to ‘break’ him”).3
As the district court noted, under these circumstances, the introduction of Harrelson’s oral reports would have been “far more than another attack upon Bello’s credibility.” Carter v. Rafferty, 621 F.Supp. at 553. If Carter’s counsel had “had the relevant information, they would have had the means to convince the jury that Bello selected one of several versions, possibly all untrue, merely because he mistakenly believed it had been confirmed by a polygraph test.” Id. Reduced to its essence, the issue comes to this: had the prosecution properly disclosed Harrelson’s oral reports Carter would have been in a position to argue that the prosecution had persuaded Bello to return to the “on-the-street” version not because it was true, but because that is what the “lie detector” results demanded. This is more than mine-run cumulative evidence. Carter could have argued to the jury that Bello was such a malleable witness that he would have testified to either the “on-the-street” or the “in-the-bar” version depending upon what he was told the polygraph test showed. Moreover, we will not accept appellants’ contention that the added inconsistency concerning Bello’s vantage point at the time of the shootings — in the bar or on the street — was merely cumulative. That inconsistency goes to the very heart of Bello’s testimony. It goes to his opportunity and ability to identify Carter and Artis, and to describe their actions and movements.
C.
We now turn to the strength or fragility of the state’s case against Carter as a whole. We must do this because in United States v. Agurs, the Court teaches that “[i]f there is no reasonable doubt about guilt whether or not the additional evidence is considered, there is no justification for a new trial. On the other hand, if the verdict is already of questionable validity, additional evidence of minor importance might be sufficient____” 427 U.S. 97, 112-13, 96 S.Ct. 2392, 2402, 49 L.Ed.2d 342 (1976). Thus, the materiality of the withheld evidence may depend on the closeness of the case. Appellants argue that “[gjiven the overwhelming evidence of guilt presented by the prosecution, the fact that it was not what might be called a close case, ... the district court had no justifiable basis upon which to find a reasonable probability of a different trial result.” Br. for respondents-appellants at 47. They contend that the prosecution’s failure to disclose Harrelson’s oral reports should not “undermine confidence in the outcome.” Bagley, 473 U.S. at 682, 105 S.Ct. at 3384; see br. for respondents-appellants at 4-37. As can be expected, Carter argues the reverse. He contends that the case against him was weak and that further impeachment of Bello’s testimony would have been decisive. See br. for petitioners-appellees at 3-27.
The district court substantially adopted Carter’s characterization of the evidence. After reviewing the record evidence as a whole, the district court concluded that
*1309[e]ven at its strongest links, the government’s chain of evidence has been substantially called into question by petitioners. Therefore, considering the totality of the circumstances, the court concludes that had the evidence withheld by the state involving Harrelson’s oral report been disclosed to the defense, there is a reasonable probability that the result of the trial would have been different.
Carter v. Rafferty, 621 F.Supp. at 558.
Before us, both parties issue a joint invitation requesting us to follow the lead of the district court and the various New Jersey courts in reviewing and re-evaluating the totality of evidence presented at the 1976 trial. Yet such an exercise necessarily requires the reviewing court to engage in speculation. See United States v. Pflaumer, 774 F.2d 1224, 1241 (3d Cir.1985) (Gibbons, J., dissenting), cert. denied, 475 U.S. 1046, 106 S.Ct. 1263, 89 L.Ed.2d 572 (1986). But the general uncertainties inherent in this endeavor are minimized here, because under any reasonable characterization of the 1976 trial, the critical importance of Bello’s testimony to the prosecution’s case clearly looms large and commanding. Bello’s eyewitness identification testimony was the only direct evidence placing Carter and Artis at the Lafayette Bar & Grill. Bello’s credibility, as well as his opportunity and ability to observe the gunmen at the crime scene, has to be considered an issue that is crucial to the guilt-finding process. We are not confronted by a situation in which the suppressed evidence — here, information impeaching Bello’s credibility and challenging his professed vantage point — -was of only “minor importance.” See Agurs, 427 U.S. at 113, 96 S.Ct. at 2402. Justice Clifford noted in dissent, and we agree, that a complete account of Harrelson’s polygraph examinations and the prosecution’s use of his conflicting oral and written conclusions had
the real capacity ... to bring about the utter destruction of by far the most important witness in the State’s arsenal, with the fallout levelling the vaunted polygraphists and casting doubt on the tactics of the prosecution[.] Never before [this information was uncovered] could defendants argue so persuasively that Bello was in all respects a complete, unvarnished liar, utterly incapable of speaking the truth.
State v. Carter, 91 N.J. 86, 139, 449 A.2d 1280, 1309 (1982) (Clifford, J., dissenting).
We therefore conclude that Bello’s testimony was critical to the prosecution’s case. His credibility was a crucial issue for the jury. From this, it must necessarily follow that there is a “reasonable probability” that the “result of the [trial] would have been different” had the prosecution properly disclosed to Carter Harrelson’s oral reports, see Bagley, 473 U.S. at 682, 105 S.Ct. at 3384, because Bello’s believability was all important. Accordingly, we hold that the district court did not err in concluding that the reports were “material” under Brady.4
VI.
To the extent that this appeal challenges the district court’s judgment with respect to John Artis, it will be dismissed for lack of jurisdiction. The district court’s judgment with respect to Rubin Carter will be affirmed.
. There appears to be some reason for this discrepancy. Harrelson had never read Bello's 1967 testimony. Harrelson had been told that Bello identified Carter and Artis in 1967; the *1303examination supported the veracity of that identification. Harrelson apparently did not know that in 1967 Bello testified that he had been on the street at the time of the shooting and therefore was unaware that the polygraph results contradicted this aspect of Bello’s 1967 testimony. In his dissent to the opinion of the New Jersey Supreme Court, Justice Clifford described the attendant confusion:
For reasons that can charitably be described as unfortunate, in his later written report of his test the polygraphist summarized his findings with the opinion that Bello’s 1967 trial testimony (which contained the "on-the-street” version) was "true": unfortunate, because Harrelson had never read Bello’s 1967 testimony and no representative of the prosecution had enlightened him as to that testimony and hence he was plainly — and grievously — mistaken as to the location from which Bello said, in 1967, that he had witnessed the events before and after the slayings; doubly unfortunate, because although the State continued to promote the notion that Harrelson’s "in-the-bar" conclusion was only tentative, Harrelson specifically and adamantly insisted that he never used those or any similar words or ever made the statement to "anyone at all on the face of the earth that [he] was unsure of Bello’s test results * * * ”; and, most unfortunate of all, because the prosecution never told the defense the critical finding of Harrelson's test — that Bello was in the bar.
State v. Carter, 91 N.J. 86, 134, 449 A.2d 1280, 1306 (1982) (Clifford, J., dissenting) (footnote omitted).
. Miller effectively overruled Patterson v. Cuyler, 729 F.2d 925, 930-32 (3d Cir.1984) (section 2254(d) presumption applies to mixed questions of law and fact). See Miller v. Fenton, 474 U.S. 104, 108 n. 3, 106 S.Ct. 445, 449 n. 3, 88 L.Ed.2d 405 (1985); Holland v. Attorney General of New Jersey, 777 F.2d 150, 153 n. 1 (3d Cir.1985).
. Similarly, the New Jersey appellate courts recognized the importance of the polygraph results in shaping Bello’s 1976 trial testimony.
On August 7, 1976, again at the direction of the prosecutor, Harrelson examined Alfred Bello using standard polygraph procedure. The purpose of the examination was to ascertain whether Bello’s testimony at the 1967 trial was true or whether his statement recanting his original testimony was true. After administering the tests, Harrelson submitted a report to the prosecutor in which he stated his conclusion:
After careful analysis of this Subject’s poly-grams, it is the opinion of the examiner that his 1966 [sic] testimony at the trial was true, and the statement recanting his original statement is not true.
Bello was confronted with this result and admitted it. Prior to the polygraph Bello had maintained the in-the-bar version which he had related to the Essex County Grand Jury. After the interrogation which followed the test, he returned to his 1967 trial testimony____ Bello stayed with this 1967 on-the-street version and testified to it at the 1976 trial.
State v. Carter, Nos. A-5166-76, A-5167-76, slip op. at 19-20 (Super.Ct.N.J.App.Div. Oct. 11, 1979) (emphasis supplied), reprinted in app. Vol. 1E at 19-20. See abo State v. Carter, 85 N.J. 300, 308-09, 426 A.2d 501, 505 (1981).
. Appellants additionally contend that even if Harrelson's oral reports had been disclosed, Carter would not have introduced this evidence at trial. Br. for respondents-appellants at 43-44, 46-48. This argument is supported only by appellants’ speculation and misconstrues Carter’s theory of how he could have used the reports to impeach Bello. Accordingly, we discern no merit in appellants’ contention.