In this appeal, we must review the judgment of the district court granting petitioner a writ of habeas corpus on the ground that he was denied the right to the effective assistance of counsel. Because the district court correctly identified the governing principles of law and properly applied those principles to the facts of this case, we affirm its judgment.

* Facts

The district court described the state trial proceedings in this case in detail. We see no reason to recast what has already been done; accordingly, we set forth the district court's rendition:1

Petitioner, Melvin Sullivan ("Sullivan") was found guilty at a bench trial of murdering Michael Grayson ("Grayson") outside the Pepperbox Lounge ("Pepperbox") on March 18, 1979 at about 2:30 A.M. Sullivan was sentenced to twenty years in prison.

At his trial, the prosecution called three witnesses, the first of whom was Sammy Titus ("Titus"), an employee working at the Pepperbox on the night of the murder. (Transcript of Trial Proceedings at 4) [hereinafter "Tr."]. Titus testified it was his job to patrol the floor of the Pepperbox and to keep confusion to a minimum. The witness recalled having seen Sullivan (who had red hair and a beard) at the bar that night but he could not remember the exact time. (Tr. 8-9). Titus also acknowledged that he did not see Sullivan arrive or depart from the Pepperbox. When questioned as to what Sullivan was wearing, Titus could not recall. (Tr. 12-13). Ms. Elnora Barnes ("Barnes") was also seen by Titus in the bar that evening, but he did not observe Barnes talking with Sullivan. Titus did not witness the murder. (Tr. 14-15).

The prosecution's second witness was William Davis ("Davis"), a patron at the Pepperbox on this St. Patrick's Day. (Tr. 21-22). Davis remembered having spoken to Sullivan at the Pepperbox on the night of the murder at about 12:00 A.M. He could not recall, however, what Sullivan was wearing. (Tr. 28). Davis said he consumed a half pint of liquor that evening. (Tr. 32). At approximately 2:30 A.M., Davis testified he stepped outside the bar, lit a cigarette, and heard a gunshot. (Tr. 22-23). After hearing the shot, he saw two women and a man wearing a long gray coat each running down the street. (Tr. 24-25). Davis recognized one of the women, but could not identify her by name. (Tr. 26-27).

The prosecution's principal witness was Elnora Barnes who had been seeing Sullivan at the Westlane Hotel in the evenings for two weeks prior to the murder. (Tr. 45-46). Barnes testified that she went to the Pepperbox at about 1:00 A.M. (Tr. 47). Barnes was a prostitute and had heard on the street that Sullivan was looking for her. (Tr. 49). Sullivan and Grayson (the victim) were talking when Barnes stated she spotted them in the Pepperbox. (Id.) Barnes then testified that she, Grayson, Sullivan and two others got up and left the bar. (Tr. 50). Barnes testified Sullivan told her, "[t]hey was going to stick up the dude because he had a piece of money," and Sullivan then pointed out Grayson. (Tr. 51). Barnes testified that no one else was around them when Sullivan spoke. Barnes recalled that Sullivan was wearing a long gray coat that evening and had a gun tucked in the waistband of his trousers. (Tr. 53-54). Upon leaving the Pepperbox with Sullivan, Grayson, and the two others (Tr. 52), Barnes testified she heard a gunshot and saw Grayson fall. She then ran back to her hotel room at the Westlane. (Tr. 55).

Barnes stated that Sullivan arrived at the hotel about twenty minutes after the shooting and said, "I just killed the nigger up on the corner." (Tr. 57). When Barnes asked him what he said, he repeated his statement. (Tr. 57). Despite this event, Barnes stayed at the hotel until mid-May when she moved to Mississippi. Several weeks later Barnes returned to Chicago. After allowing almost three months to elapse after the murder, Barnes felt the compulsion to notify the police as to her version of the events that took place on the night of the murder.

After the conclusion of Barnes' testimony the state entered a stipulation as to the cause of death of the victim and then rested. The judge then denied a motion for a directed finding. (Tr. 97). The defense began its case.

After speaking to the defendant and his aunt, the defense decided upon an alibi defense. Sullivan's Aunt Kathleen testified first. Kathleen Sullivan ("Kathleen") stated that at around 1:00 A.M. in the morning of March 18, 1979 she heard Sullivan outside and threw him a key from her window to let him enter the home where he resided at times with her. (Tr. 98-99). Linda Lee Sullivan ("Linda"), the defendant's cousin who also lived at this residence, testified that she was sleeping and did not hear Sullivan come home. (Tr. 108-09).

The final witness at the trial was the defendant Sullivan. Sullivan testified that he was at the Pepperbox on the evening of March 17, 1979, where he spoke with Titus and Davis. (Tr. 110-12). However, Sullivan left the Pepperbox at around 12:00 A.M. that evening without seeing Barnes. Id. After leaving the Pepperbox, Sullivan testified he returned to his aunt's home where she tossed him a key to enter the house. Sullivan stated he turned on the television for a few minutes that evening and then went to sleep. (Tr. 112-13).

Sullivan further testified that he and Barnes had been seeing each other for a period of several weeks at about the time the murder occurred. Sullivan stated he and Barnes ended their relationship on bad terms when he told Barnes he was returning to his wife now that he had graduated from college. (Tr. 113-15). Barnes apparently moved south for a period of weeks after hearing it was over between her and Sullivan. (Tr. 115).

At the close of the trial, defense counsel moved for a continuance to allow him to locate and interview several additional witnesses. (Tr. 120-21). The court offered defense counsel the opportunity to state the names of the persons he would call and an offer of proof as to what their testimony would be. (Tr. 12. Defense counsel identified Jedda Sullivan and Vernell Davis as two persons who were named in the police report as witnesses who would offer testimony inconsistent to that of Barnes. (Tr. 12. The court denied the motion, and after final argument found Sullivan guilty and sentenced him to twenty years in prison.

Subsequently, Sullivan moved for a new trial on the basis of newly discovered occurrence witnesses, even though these five witnesses had been named in the State's answers to discovery several months earlier. The motion was denied by the trial judge on the grounds that this evidence would have been discovered but for the lack of diligence exhibited by the defense attorney. (Tr. 201; 184-86).

United States ex rel. Sullivan v. Fairman, No. 82 C 613, order at 4-8 (N.D.Ill. Sept. 12, 1986); R.96 at 4-8 [hereinafter cited as Order].

On appeal to the Appellate Court of Illinois, the judgment of conviction was affirmed. People v. Sullivan, 95 Ill.App.3d 571, 51 Ill. Dec. 60, 420 N.E.2d 474 (1981). The court rejected Mr. Sullivan's claim that he was not proved guilty beyond a reasonable doubt. It also rejected his contention that the trial court had erred in failing to grant him a continuance to procure witnesses on the ground that counsel had failed to show that he exercised due diligence. Id. 51 Ill.Dec. at 63, 420 N.E.2d at 477. It also affirmed the denial of a new trial on the same ground. Id. Finally, it rejected petitioner's claim that he had been denied the effective assistance of counsel. Id. 51 Ill.Dec. at 63-64, 420 N.E.2d at 477-78. On June 15, 1981, the petitioner filed leave to appeal to the Illinois Supreme Court, which was denied by the court on October 19, 1981.

After exhausting the available state remedies, Mr. Sullivan filed a petition for a writ of habeas corpus in the district court. He alleged that the state trial court's failure to grant him a continuance or a new trial deprived him of his right to due process of law. He further alleged that he had been denied the effective assistance of counsel. The district court dismissed the petition. United States ex rel. Sullivan v. Fairman, 564 F. Supp. 575 (N.D.Ill.1983). It ruled that Mr. Sullivan failed to meet the cause and prejudice standard of Wainwright v. Sykes, 433 U.S. 72, 97 S. Ct. 2497, 53 L. Ed. 2d 594 (1977) with respect to the due process claim. Id. at 579. With respect to the ineffective assistance of counsel claim, this court vacated the district court's judgment and remanded the case for further proceedings in light of United States ex rel. Cosey v. Wolff, 727 F.2d 656 (7th Cir.1984), overruled, United States v. Payne, 741 F.2d 887 (7th Cir.1984) (per curiam).2 The court also noted that the district court had disregarded the exculpatory affidavits without the benefit of an evidentiary hearing. United States ex rel. Sullivan v. Fairman, 731 F.2d 450, 455-56 (7th Cir.1984).

On remand, the district court held a hearing. In the course of this hearing, Mr. Sullivan, his Aunt Kathleen, the defense attorneys for Mr. Sullivan at his original trial, as well as four of the five occurrence witnesses testified. The fifth witness to the murder is deceased. In its opinion, the district court carefully summarized the evidence presented by both sides at this hearing:3

Two witnesses who saw the murder of Grayson testified that they had been with the victim for the entire evening up until the time of the shooting. Both Jedda Sullivan ("Jedda") and Vernell Davis ("Vernell") testified to relatively the same factual occurrences on the evening of March 17, 1979. The two witnesses [and the victim] were walking along the street on which the Pepperbox is located and were all three singing the song "Reunited" as they walked. (Remand Proceedings at [hereinafter "R."]; (Second Remand Hearing at 15-16) [hereinafter "SR."]. An unknown man from across the street yelled at them to shut up. (SR. 16). The three continued singing when the man across the street fired a shot. (R. 9; SR. [17-18]. This same man from across the street was then seen approaching the three individuals a few moments later. (R. 11; SR. 19). [Grayson said], "I want to talk to you," and [the man] and Grayson then argued about the singing. The man then fired a gun at Grayson from a relatively short distance. (R. 11-12).

Both Jedda and Vernell testified that before the gunman ran away they were able to get a look at him. (Jedda being about 15 feet away from him and Vernell being about six feet from the gunman). (R. 11-14; SR. 20-24). Both witnesses identified the murderer as being over six feet tall, with nappy, wild hair, and a mustache. (Id.) Both said he was wearing a long coat. (Id.) Jedda and Vernell were in agreement that if they saw this man again they would be able to identify him. They also agreed that the man who murdered their friend Grayson was not Sullivan. Sullivan was present in the courtroom at these hearings. (R. 13-14; SR. 22-23).

Neither Jedda nor Vernell knew Sullivan before the hearing, but they had seen him around the neighborhood. (R. 13; SR. 22-23). The police had interviewed both witnesses, and their earlier testimony was similar to the testimony they gave at the hearing. (R. ; SR. [24-25]. Neither witness had left the City of Chicago since the murder, nor had they ever been contacted by Sullivan's defense counsel. [R. 19-20; SR. 28-29].

The remaining two witnesses called to testify were Leroy Johnson ("Johnson") and Thomas Brown ("Brown"). Brown testified that he saw two men and one woman walking on the other side of the street, presumably Grayson, Vernell and Jedda. [SR. 44]. Brown heard a shot and then saw a man run past him. Similarly, Johnson heard shots and the gunman subsequently passed by him as well. [SR. 68]. Both Brown and Johnson testified that Sullivan was not the gunman, but that the gunman was taller than Sullivan (who is about 5'8"'), had darker hair and skin, and a mustache, and no beard. ( [S]R. 47-48; 70-74). Johnson further testified that he had seen the murderer about six times on previous occasions and that he believed the gunman was a friend of Art Malone, the deceased fifth occurrence witness. [SR. 71]. Both of these witnesses denied knowing Sullivan or being contacted by Sullivan's defense attorney. [SR. 52; 62-63; 71; 76-77]. The police reports indicate that these two men gave similar statements to the police regarding the night of the murder.

Mr. Weiner, the court-appointed attorney for Sullivan at his original trial, testified at the hearings before this Court on January 17, 1985. (R. 7. Weiner stated that on the basis of interviews with Sullivan that he filed an alibi answer to the Complaint against the defendant. Weiner and Sullivan met on at least six occasions, three of the meetings lasting over half an hour. (R. [72-75]. At the meetings Weiner testified that he discussed with the defendant calling witnesses at trial who would testify he was not at the murder scene. (R. 73-76). Furthermore, Weiner recalls going over the police report and discussing the witnesses interviewed by the police and assuring the defendant that he would attempt to locate these witnesses. Weiner does not remember giving Sullivan a copy of the police report. (R. 75).

The attorney testified that he sent letters to the witnesses in the police report and tried to phone them with no result. Weiner asked Sullivan's aunt (Kathleen) to help locate these witnesses since she lived in the area. (R. [76-77]. Kathleen reportedly returned to Weiner without having located the witnesses. (R. [79-80]. The alibi defense was not altered after the witnesses could not be located; rather Weiner testified that he believed he had a strong defense for Sullivan by calling Sullivan, his aunt and cousin to testify. (R. [79, 98-99].

Weiner testified that he made a conscious choice not to use an investigator to locate the missing witnesses in this case, despite his awareness that one could be hired at no cost to him or the defendant. (R. 99). After reading the police reports, Weiner stated that he thought some of these witnesses were not to be believed. However, Weiner believed Sullivan's and his aunt's testimony would be believable. (R. .

Order at 9-12.

After setting forth the testimony taken at the hearing, the district court embarked upon an analysis of that testimony under the framework of Strickland v. Washington, 466 U.S. 668, 104 S. Ct. 2052, 80 L. Ed. 2d 674 (1984). Under the Strickland analysis, the petitioner must establish both that counsel's performance was deficient and that the deficiency prejudiced the defense so as to deny the petitioner a fair trial. Id. at 687, 104 S.Ct. at 2064. In brief, the district court determined that trial defense counsel's performance was constitutionally deficient because he did not do more to obtain the testimony of the five occurrence witnesses. The court held that, in assessing the reasonableness of defense counsel's investigation, it must consider the availability of the witnesses, the importance of their testimony, and the degree of difficulty in locating them. Order at 14. The court found that their names, addresses and telephone numbers were given in the police report of the crime. They were also identified in the prosecution's response to criminal discovery. The district court stressed that they were the only persons, outside the petitioner's family, who could have offered exculpatory testimony. Their testimony directly contradicted the state's chief witnesses' testimony. Id. at 16. In determining that the performance of trial defense counsel was substandard, the district court emphasized that it relied upon "evidence known to defense counsel as they prepared for trial," id. at 16, and not upon testimony first discovered at the remand hearing or upon hindsight about the witnesses' presence in Chicago after the shooting. Id. at 16.

The district court then turned to an assessment of the prejudice to the defense resulting from counsel's failure to present the uncalled witnesses. The district court analyzed the testimony of the occurrence witnesses and defense counsel together with the trial transcript and the record of post-trial motions from the state proceeding. The court determined that, if believed, the testimony of the occurrence witnesses would have been completely exculpatory. The court found, moreover, that there were significant reasons why they should have been believed: 1) they were at the scene of the shooting; 2) they were disinterested (indeed, two were close friends of the victim); 3) the testimony was consistent in essential respects yet "not so close that an inference of collusion is likely." Id. at 21. The court also noted that the testimony given by the occurrence witnesses at the habeas hearing was consistent with the accounts given by the same witnesses in their police reports. Finally, the court opined that Elnora Barnes' testimony of Mr. Sullivan's admission to her that he had committed the murder was significantly called into question because the five witnesses had contradicted her testimony about the murder itself. Id. at 23-24.

Like the district court, each of the parties addresses the issue of effective assistance of counsel within the framework of Strickland v. Washington, 466 U.S. 668, 104 S. Ct. 2052, 80 L. Ed. 2d 674 (1984). Each addresses whether trial defense counsel's performance was substandard and whether that performance prejudiced the petitioner. We shall set forth their principal contentions.

The respondent maintains that counsel's failure to locate and interview potential witnesses did not fall below the standard of reasonably effective assistance when viewed from the perspective of counsel at the time of representation. The state argues that, unlike the situation in Code v. Montgomery, 799 F.2d 1481 (11th Cir.1986), the manner in which counsel dealt with the investigation of the case did not constitute objectively unreasonable conduct because the failure of counsel did not result in a decision to forego the only available defense. See Keys v. Duckworth, 761 F.2d 390, 392 (7th Cir.1985) ("[T]he decision to forego what may be the only available defense can fall below sixth amendment standards."). Here, argues the state, the defense did present a case based on a viable alibi theory. Moreover, defense counsel actively cross-examined the state witnesses to impeach their credibility. Respondent's Br. at 19-21.

The decision to rely on an alibi defense and not to pursue further the uncalled witnesses was, continues the state, "made in the exercise of reasonable professional judgment and well within the wide range of reasonable assistance." Id. at 21. While defense counsel admittedly relied upon letters and telephone calls to the possible witnesses--and did not hire an investigator--the issue is not, argues the state, whether it is advisable or "good practice" for defense counsel to hire an investigator as a matter of course. "Rather, the issue is whether that failure to do so amounts to an omission which is deficient." Id. at 22 (citing United States ex rel. Rivera v. Franzen, 794 F.2d 314 (7th Cir.), cert. denied, --- U.S. ----, 107 S. Ct. 588, 93 L. Ed. 2d 590 (1986)).

The government also stresses that defense counsel testified that he spoke with the petitioner on five or six occasions from the time of his appointment until trial. On three of those occasions, he met with the petitioner for extended periods of time. During those meetings, counsel discussed with petitioner the defense of the case and reviewed the police reports. In addition, petitioner put counsel in touch with his aunt, Kathleen Sullivan, who petitioner indicated he lived with at the time of the offense.

The government argues that the attempt to locate witnesses by mail and telephone and the failure to hire an investigator did not fall outside the wide range of professionally competent assistance when these actions are evaluated in light of the possibility of an alternative defense or evaluated in light of the strength of the defense based upon the available witnesses. Respondent's Br. at 23-25. Finally, the state argues that, if this court were to affirm the district court's issuance of the writ of habeas corpus, it would be requiring counsel to choose one particular form of investigation over another in order to render effective assistance of counsel. Rather, argues the state, the form of investigation chosen by counsel meets the sixth amendment standard if it is not "patently unreasonable." Id. at 28 (citing Goins v. Lane, 787 F.2d 248, 254 (7th Cir.), cert. denied, --- U.S. ----, 107 S. Ct. 165, 93 L. Ed. 2d 103 (1986)).

With respect to the "performance" component, the petitioner stresses that trial defense counsel's representation may not be explained on the ground of trial tactics. Trial tactics only become a factor, he argues, when there has been a reasonably thorough investigation of the evidence available. Only then, he argues, may alternate approaches to the defense be adequately evaluated. Petitioner acknowledges that the performance of trial defense counsel may not be evaluated with the benefit of hindsight. Petitioner's Br. at 6-8. However, he argues that, when the record is reviewed from the perspective of Mr. Sullivan's trial counsel at the time of trial, the performance falls below a minimally acceptable standard of professional competence. He points to the following:

1. The existence of five potentially available defense witnesses whose identity was known well before trial through the state's response to discovery and through police reports.

2. These known witnesses were neither friends nor relatives of Sullivan and were therefore disinterested.

3. The testimony of these disinterested witnesses was potentially the strongest and most impressive proof at trial. Their testimony was not repetitive and each witness's testimony tended to corroborate that of the others. Further, the testimony of the five disinterested witnesses was exculpatory.

4. Trial defense counsel had at his disposal a means of contacting the witnesses because their addresses and phone numbers were listed in the police reports.

5. The defendant did not have copies of the police reports until the date of trial.

6. The defendant was incarcerated and therefore could not be expected to help his defense.

7. A paid investigator was available to locate, interview and subpoena witnesses. Such services had been used by trial counsel in other cases.

8. An alibi defense would not be nearly as strong as the one based on the police reports and state discovery responses. Unlike the witnesses who were not called, the witnesses in support of the alibi defense were relatives who had limited knowledge of Mr. Sullivan's activities on that particular night and could provide no direct testimony as to his whereabouts at the time of the shooting. On the other hand, the witnesses who would have testified at trial were unrelated in blood or friendship to the defendant and indeed somewhat close friends of the victim and, therefore, had an interest in seeing his murderer punished. Id. at 8-9.

The petitioner also emphasizes that these witnesses were located by and cooperated with substitute counsel immediately after Mr. Sullivan's trial and that, five years later, Mr. Sullivan's counsel for his habeas corpus petition was still able to find and locate these witnesses and secure their cooperation (with the exception of the one eyewitness who died in the interim). He also points out that the testimony of the eyewitnesses at the habeas corpus hearing was in substantial conformity with the information contained in the police reports that trial defense counsel had in his possession prior to Mr. Sullivan's trial.4 Id. at 9-10.

In short, petitioner argues that this is a case where the failure to investigate amounts to ineffective assistance of counsel because "the record establishes that counsel had reason to know, from an objective standpoint, that a possible defense ... [was] available." Rivera, 794 F.2d at 316. The petitioner notes that police statements generally do not serve as an adequate substitute for a personal interview. "[A]n attorney who fails even to interview a readily available witness whose noncumulative testimony may potentially aid the defense should not be allowed automatically to defend his omission simply by raising the shield of 'trial strategy and tactics.' " Crisp v. Duckworth, 743 F.2d 580, 584 (7th Cir.1984), cert. denied, 469 U.S. 1226, 105 S. Ct. 1221, 84 L. Ed. 2d 361 (1985).

With respect to the prejudice component of Strickland, the respondent quite properly points out that prejudice is not to be presumed even if the errors of counsel were professionally unreasonable. It is still incumbent upon the defendant to establish that there is a reasonable probability that, but for counsel's unprofessional errors, the result of the trial would have been different. "A reasonable probability is a probability sufficient to undermine confidence in the outcome." Respondent's Br. at 29 (quoting Strickland, 466 U.S. at 694, 104 S.Ct. at 2068).

Respondent submits that the petitioner failed to demonstrate this degree of prejudice when the evidence presented at trial is compared to the testimony presented at the evidentiary hearing before the district court. The respondent indicates that the testimony at the evidentiary hearing was conflicting regarding the number of shots fired and the distance the assailant was from the victim at the time of the murder. Respondent's Br. at 30-31. The respondent also argues that the witnesses at the evidentiary hearing before the district court gave conflicting testimony regarding the identity of Mr. Grayson's assailant. Jedda Sullivan testified that the assailant had black nappy hair and another witness testified that he "had a very large Afro, and it was kind of reddish brown, kind of dark brown." Another witness testified that he had black hair. Id. at 31.

Here, the petitioner relies heavily on the district court's analysis. The petitioner stresses that, in making its decision, the district court considered the testimony, on direct and cross-examination, of four independent witnesses who were identified to Mr. Sullivan's original trial defense counsel through police reports and state discovery responses. Additionally, the district court reviewed the affidavit filed in the state court of one other independent eyewitness who was deceased when the district court held the evidentiary hearing. The district court also heard the testimony of Mr. Sullivan's trial counsel, the testimony of Mr. Sullivan himself, and the testimony of his aunt, Kathleen Sullivan. The petitioner stresses that we are bound by the factual findings of the district court unless we find that they are clearly erroneous.

Our decision in this case does not require that we decide any novel issue of law. The governing principles have been set forth by the Supreme Court in Strickland v. Washington, 466 U.S. 668, 104 S. Ct. 2052, 80 L. Ed. 2d 674 (1984), and have been applied regularly by this circuit ever since. See, e.g., United States ex rel. Link v. Lane, 811 F.2d 1166, 1170-72 (7th Cir.1987); United States ex rel. Kleba v. McGinnis, 796 F.2d 947, 953-58 (7th Cir.1986); United States ex rel. Smith v. Lane, 794 F.2d 287, 290-94 (7th Cir.1986); Keys v. Duckworth, 761 F.2d 390, 392-94 (7th Cir.1985); United States v. Noble, 754 F.2d 1324, 1335-36 (7th Cir.), cert. denied, --- U.S. ----, 106 S. Ct. 63, 88 L. Ed. 2d 51 (1985); United States v. Payne, 741 F.2d 887 (7th Cir.1984) (per curiam). The burden on the petitioner is a heavy one. In order to establish that he was denied his sixth amendment right to counsel, he must affirmatively establish both "components" of the Strickland analysis--the "performance" component and the "prejudice" component. Strickland, 466 U.S. at 687, 104 S. Ct. at 2064; see United States ex rel. Cross v. DeRobertis, 811 F.2d 1008, 1011 (7th Cir.1987).

With respect to the "performance" component, "the proper standard for attorney performance is that of reasonably effective assistance." Strickland, 466 U.S. at 687, 104 S. Ct. at 2064. "[T]he defendant must show that counsel's representation fell below an objective standard of reasonableness." Id. at 688, 104 S.Ct. at 2064. In evaluating counsel's performance, "all the circumstances," id., must be considered.

No particular set of detailed rules for counsel's conduct can satisfactorily account for the variety of circumstances faced by defense counsel or the range of legitimate decisions regarding how best to represent a criminal defendant. Any such set of rules would interfere with the constitutionally protected independence of counsel and restrict the wide latitude counsel must have in making tactical decisions.

Id. at 688-89, 104 S.Ct. at 2065. In assessing counsel's performance, our scrutiny must be "highly deferential." Id. at 689, 104 S.Ct. at 2065. We must make "every effort ... to eliminate the distorting effects of hindsight" by indulging in the "strong presumption that counsel's conduct falls within the wide range of reasonable professional assistance...." Id.

When, as here, the inquiry centers on whether trial defense counsel adequately investigated the case, it must be remembered that:

strategic choices made after thorough investigation of law and facts relevant to plausible options are virtually unchallengeable; and strategic choices made after less than complete investigation are reasonable precisely to the extent that reasonable professional judgments support the limitations on investigation. In other words, counsel has a duty to make reasonable investigations or to make a reasonable decision that makes particular investigations unnecessary.

Id. at 690-91, 104 S.Ct. at 2066. In reviewing counsel's decision not to investigate, the court must apply a "heavy measure of deference to counsel's judgments." Id. at 691, 104 S.Ct. at 2066; see also Cross, 811 F.2d at 1014.

With respect to the "prejudice" component, it "is not enough for the defendant to show that the errors had some conceivable effect on the outcome of the proceeding." Strickland, 466 U.S. at 693, 104 S.Ct. at 2067. Rather, "[t]he defendant must show that there is a reasonable probability that, but for counsel's unprofessional errors, the result of the proceeding would have been different. A reasonable probability is a probability sufficient to undermine confidence in the outcome." Id. at 694, 104 S. Ct. at 2068.

Finally, throughout the analysis, we must remember that "the ultimate focus of inquiry must be on the fundamental fairness of the proceeding whose result is being challenged." Id. at 696, 104 S.Ct. at 2069. We must determine "whether, despite the strong presumption of reliability, the result of the particular proceeding is unreliable because of a breakdown in the adversarial process that our system counts on to produce just results." Id.

As our cases in the wake of Strickland amply demonstrate, the foregoing principles erect a high hurdle for the petitioner. Indeed, we expect that few petitioners will be able to pass through the "eye of the needle"5 created by Strickland. On the other hand, the Supreme Court certainly did not intend the Strickland analysis to be a total barrier to relief based on ineffective assistance of counsel claims. Rather, it is designed to focus the inquiry and discipline the analytical methodology by which such claims are evaluated. Our study of this case convinces us that it is one of those very few that should pass through the eye of the needle. In our view, the analysis of the district court demonstrates, with respect to both the "performance" and the "prejudice" components, a scrupulous sensitivity to the concerns outlined by the Supreme Court in Strickland and by this court in its cases following Strickland.

B. Application of Governing Principles to this Case

With respect to the "performance" component, we agree with the district court that counsel's performance did not conform with the standard of professional conduct which, under all of the circumstances, might reasonably be expected of trial defense counsel. It is undisputed that, prior to trial, defense counsel was aware, through the police reports and discovery, that there were five witnesses, with no apparent reason to help the defendant, who made statements to the police that were exculpatory or inconsistent with the prosecution witnesses' statements. The names and addresses of these witnesses were available to defense counsel; yet his attempts to locate and to interview them were perfunctory at best. Given the importance of these witnesses to the defendant's case and the comparative weakness of his other possible defense--an "alibi" supplied by a relative that did not completely preclude the possibility that he was the perpetrator--it was not reasonable for trial defense counsel to rely on his own letter and telephone attempts or the attempts of the defendant's aunt to contact the witnesses.

Again, we stress that we do not hold that trial defense counsel must track down every lead or must personally investigate every evidentiary possibility before choosing a defense and developing it. We simply hold that, under the circumstances presented here, it was not reasonable for defense counsel to permit his client to stand trial for murder without a more thorough investigation of the available evidence. In reaching this conclusion, we are aware of our obligation "to eliminate the distorting effects of hindsight, to reconstruct the circumstances of counsel's challenged conduct, and to evaluate the conduct from counsel's perspective at the time." Strickland, 466 U.S. at 689, 104 S. Ct. at 2065 (emphasis added). However, "strategic choices made after less than complete investigation are reasonable precisely to the extent that reasonable professional judgments support the limitations on investigation." Id. at 690-91, 104 S.Ct. at 2066. Trial defense counsel did not exercise reasonable professional judgment.

Under the mandate of Strickland, even if the petitioner is able to convince the court that the conduct of his defense counsel was not reasonable under all the circumstances, he must shoulder the independent burden of establishing that counsel's conduct resulted in "a reasonable probability that, but for counsel's unprofessional errors, the result of the proceeding would have been different." Strickland, 466 U.S. at 694, 104 S.Ct. at 2068. In other words, the proper inquiry is whether the factfinder would have had "a reasonable doubt concerning guilt." Id. at 695, 104 S.Ct. at 2068. As stated in Strickland:

[A] court hearing an ineffectiveness claim must consider the totality of the evidence before the judge or jury. Some of the factual findings will have been unaffected by the errors, and factual findings that were affected will have been affected in different ways. Some errors will have had a pervasive effect on the inferences to be drawn from the evidence, altering the entire evidentiary picture, and some will have had an isolated, trivial effect. Moreover, a verdict or conclusion only weakly supported by the record is more likely to have been affected by errors than one with overwhelming record support. Taking the unaffected findings as a given, and taking due account of the effect of the errors on the remaining findings, a court making the prejudice inquiry must ask if the defendant has met the burden of showing that the decision reached would reasonably likely have been different absent the errors.

Id. at 695-96, 104 S.Ct. at 2069; see also Cross, 811 F.2d at 1015-16.

In cases, such as this one, where counsel has failed to investigate, we must exercise extreme caution in the evaluation of the prejudice component. It is all too easy--after an unsuccessful trial effort--to hypothesize about "what might have been" if a different trial tactic had been adopted. Therefore, in Cross, we stressed that, in order to meet his burden, the petitioner must make:

a comprehensive showing as to what the investigation would have produced. The focus of the inquiry must be on what information would have been obtained from such an investigation and whether such information, assuming its admissibility in court, would have produced a different result. Under usual circumstances, we would expect that such information would be presented to the habeas court through the testimony of the potential witnesses.

Cross, 811 F.2d at 1016.

Here, the district judge was presented with the testimony of the four surviving witnesses who were not called at trial. He also considered the affidavit of the fifth witness, now deceased. He carefully evaluated their testimony and found that "there are significant reasons to conclude that it would have been believed." Order at 20. He noted that they were all eyewitnesses, that they were "disinterested," that their testimony was consistent in essential respects but sufficiently dissimilar in minor respects to make the possibility of collusion remote. The district judge also evaluated the testimony of these witnesses in light of the government's testimony and found that it conflicted on several essential matters.6 He further noted that, given the fact that Ms. Barnes' testimony about the shooting was now suspect, her account of the defendant's damaging admission is also in doubt. We must accept the findings of fact of the district court when they are not clearly erroneous. Strickland, 466 U.S. at 698, 104 S.Ct. at 2070; United States ex rel. Smith v. Lane, 794 F.2d 287, 289 n. 3 (7th Cir.1986). We also must remember that the district court, and not this court, saw the witnesses and observed their demeanor.

As the respondent points out, the evidence submitted at the habeas hearing does not establish petitioner's innocence. There are some inconsistencies in the testimony of the witnesses and the district court did not have an opportunity to evaluate fully the credibility of the government's witnesses. At retrial, the factfinder could determine that the government has met its burden of establishing guilt beyond a reasonable doubt. However, in this habeas proceeding, the petitioner's burden, while a high one, does not require that he establish his innocence or even demonstrate "that counsel's deficient conduct more likely than not altered the outcome in the case." Strickland, 466 U.S. at 693, 104 S.Ct. at 2068. It is sufficient that the petitioner show that there is a "reasonable probability that, but for counsel's unprofessional errors, the result of the proceeding would have been different." Id. at 694, 104 S. Ct. at 2068.

"[T]he performance and prejudice components of the ineffectiveness inquiry are mixed questions of law and fact." Id. at 698, 104 S. Ct. at 2070. They are therefore subject to our independent review. Our own study of this case convinces us that both the district court's methodology and reasoning were sound, and that its conclusions are correct. Accordingly, the judgment of the district court granting the writ of habeas corpus and requiring retrial within 120 days is affirmed.

AFFIRMED.