No. 986, Docket 82-2004.United States Court of Appeals, Second Circuit.Argued April 6, 1982.
Decided October 25, 1982.
Judith Preble, New York City (William E. Hellerstein, The Legal Aid Society, New York City, of counsel), for petitioner-appellee.
Richard M. Howard, Deputy Asst. Atty. Gen., New York City (Robert Abrams, Atty. Gen. of the State of N.Y., Gerald J. Ryan, Asst. Atty. Gen., New York City, of counsel), for respondent-appellant.
Appeal from the United States District Court for the Southern District of New York.
Before FRIENDLY and NEWMAN, Circuit Judges, and CURTIN,[*]
CURTIN, District Judge:
 This is an appeal from the judgment of the United States District Court for the Southern District of New York, Morris E. Lasker, Judge, granting appellee David Dickerson’s application for federal habeas corpus relief. Dickerson v. Fogg, 526 F.Supp. 1299 (S.D.N.Y. 1981). The court found that the victim’s out-of-court and in-court identifications of Dickerson were unconstitutionally tainted under the fifth, sixth, and fourteenth amendments to the United States Constitution. We agree with the thorough and well-reasoned opinion of the court below, and affirm.
 David Dickerson was charged with the armed robbery of Robert Colon, Jr. in May 1977. Ironically, he was arrested while sitting in a courtroom watching his cousin’s arraignment on stolen car charges linked to the armed robbery. The police arrested him after Colon identified him as one of the four men who had stolen his automobile and personal property almost two days earlier.
 Following a combined Wade-Huntley hearing, the state trial judge ruled that Colon’s pre-trial identification of Dickerson was not suggestive, and denied Dickerson’s motion to suppress. Colon later testified at trial that he had recognized Dickerson at the arraignment, and offered a contemporaneous in-court identification of Dickerson as the culprit. The jury found Dickerson guilty of robbery in the first degree, for which he was sentenced to a prison term of five to fifteen years.
 The New York Supreme Court, Appellate Division, First Department, affirmed Dickerson’s conviction in a split decision People v. Dickerson, 67 A.D.2d 122, 414 N.Y.S.2d 712 (1st Dep’t 1979). Although the majority acknowledged that Colon’s pre-trial confrontation of Dickerson “partook of a certain . . . degree of suggestiveness,” id. at 714, it found that the identification procedures were not unduly suggestive. The “somewhat suggestive” circumstances could not be avoided, the majority explained, because of the “concomitant . . . inability of the police to control the courtroom scene,” and because the confrontation took place only on the hunch of police. Id. The three majority justices also found a reliable independent basis for Colon’s in-court identification of Dickerson. On the other side of the bench, the two dissenting justices concluded that Colon’s pre-trial identification was impermissibly suggestive and should have been suppressed, and that his in-court identification could not be isolated from the corrupting influence of the suggestive pre-trial identification. The New York Court of Appeals affirmed the majority’s decision in a memorandum opinion People v. Dickerson, 50 N.Y.2d 937, 431 N.Y.S.2d 453, 409 N.E.2d 927 (1980).
 Dickerson applied for federal habeas redress in July, 1981. The district court held that the admission of both Colon’s pre-trial and trial identifications of Dickerson deprived him of his due process rights under Manson v. Brathwaite, 432 U.S. 98, 97 S.Ct. 2243, 53 L.Ed.2d 140 (1977), and Neil v. Biggers, 409 U.S. 188, 93 S.Ct. 375, 34 L.Ed.2d 401 (1972). He ordered Dickerson released unless a new trial were held without the use of Colon’s identification testimony. This appeal by the state followed.
 FACTUAL BACKGROUND
 At about 4:00 a.m. on May 19, 1977, Robert Colon, Jr., a security supervisor at Sloan’s Supermarkets, answered a call over his pager telling him to go to one of the chain’s stores. Colon walked out of his house on Jerome Street in New York City to the company car which was parked across the street and a few houses down the street from his own. As he entered the car, a young man with a pistol wrapped in a jacket banged on the window and yanked on the door on the driver’s side. The man ordered Colon to move over and open the front passenger’s door. A second man, later identified as appellee’s cousin and co-defendant, Ronald Brown, entered the car through the passenger’s door as the first man came in on the driver’s side. Simultaneously, Colon sensed other people entering the car through the rear doors.
 The car proceeded to drive off. During the whole time he was in the car, Colon never looked to see who was in the rear seat. Even when the driver handed his gun to the man sitting directly behind him, Colon did not turn around to see who accepted the weapon (H. 166-67). The man who now wielded the gun pressed it against Colon’s neck until a police car came into view, then lowered it and warned Colon to remain silent. As the car moved along, Brown reached over and removed Colon’s jewelry and cash (H. 141-42, 169-70).
 The automobile traveled perhaps seven or eight blocks before it came to a halt. When it did, Brown let Colon out on the passenger’s side. Colon walked away from the car in the direction from where they had just come. When he was roughly a car’s length from the vehicle, he turned and saw the left back seat passenger staring at him, trying to pull a beach hat on his head and over his face. The hat came down to the man’s eyebrows (H. 159-60; T. 86, 110-11).
 This was the only occasion Colon had to view the passenger’s face. Colon testified at the Wade-Huntley hearing that he looked for “[t]wo minutes, I couldn’t really say, about two minutes.” (H. 160). However, he pared his estimate at trial when he guessed he stared at the man for “only a minute or so” while the car was stationary (T. 86, 110).
 Some nineteen hours later, at approximately 10:30 p.m., Police Officers James D. Dugan and Craig Moruzzi were on radio motor patrol when they spotted what they determined to be a stolen car driving in the opposite direction from them on Seventh Avenue in Manhattan. The policemen reversed their direction, pulled alongside the driver’s side of the vehicle, and, at gunpoint, ordered the driver to stop. Instead of yielding, the driver ducked his head below the window and sped away. Officers Dugan and Moruzzi immediately chased the fleeing car, but not for long. The stolen car crashed into a halted car and three men jumped out, fell to the ground, and ran off. Officer Dugan pursued and nabbed Ronald Brown. Officer Moruzzi chased the other two men who had exited the car on the driver’s side, but was unable to apprehend either of them.
 Officer Moruzzi identified Dickerson as the driver of the stolen automobile at both the Wade-Huntley hearing and trial. He testified he saw the driver’s face just before the man slid from view when the squad car pulled up for anywhere from 2 to 20 seconds (T. 150-56), and again when the driver jumped out of the car, falling with his face towards the squad car right before he took off. Moruzzi noted that the driver was seventeen to nineteen years old, short, 110 to 120 pounds, and had close cropped hair and large ears. (H. 86-87). Despite this otherwise detailed description, Moruzzi could not recall what the young man was wearing, nor did he write down a description of the incident.
 The next evening, May 20, 1977, the police phoned Robert Colon, informed him they had his automobile, and asked him to come by to swear out a complaint (H. 151, 173, 191). Police Officer Dugan, wearing the Sloan’s security shield Colon normally kept in his car, greeted Colon at Manhattan Criminal Court where Ronald Brown, Dickerson’s co-defendant, was being arraigned. Earlier, Dugan had learned that two men and two women who were sitting together on the right side of the courtroom were present on Brown’s behalf. Upon meeting Colon, Dugan asked Colon to describe the culprits. Colon told him they were four young male blacks (H. 57-57a, 61). Dugan stated he then pointed to the right side of the courtroom where the people in question were sitting in a row and asked Colon if he could identify anyone (H. 56-57).
 Colon testified at both the Wade-Huntley hearing and trial that when Dugan met him in court he directed Colon’s attention to the right side of the room to see if he could identify anyone in a group of two males and one female sitting there (H. 153, 193-94; T. 91-92). Noticing that the group was sitting towards the front of the courtroom, Colon walked down the aisle to obtain a closer view before returning to the courtroom foyer. There he told Dugan he was “pretty sure” that one of the men was, or “looks like,” the man who had sat directly behind him in the car (H. 156, 194). Colon did not recognize the other man (H. 194-95).
 Officer Dugan urged Colon to “take a better look,” and sent him back into the courtroom with a court officer. When he returned this time, Colon informed Dugan that one of the men “looked just like” the back seat passenger (H. 157). As Dickerson and his male companion left the courtroom and passed within ten feet of Colon, Dugan pressed Colon by asking, “Is it him or not?” Colon responded yes with respect to Dickerson but felt he “couldn’t tell you anything about the other one.” (H. 157). Following this identification, Dugan arrested Dickerson, handcuffed him, and removed him and his companion to nearby court robing room.
 Colon also testified that at one point he had described three of his assailants, including the back seat passenger with the hat, to the police over the phone. Significantly he could not remember whether he furnished his descriptions to the police before or after he identified Dickerson at the Manhattan Criminal Court, or to whom he had given his description, although he suggested he might have spoken with either officer Viggiano or Officer Dugan (H. 177-78, 195). Colon described the man with the hat as a black male with a short afro haircut and a tiny moustache. The driver, according to Colon, was about five feet six inches tall, chubby, and round-faced. The man sitting to Colon’s right in the car, Ronald Brown, was about five feet ten or eleven inches, thin, light-skinned, and dark-haired.
 At trial, Colon conceded he might have initially reported to the police that he had seen and could identify only two of the four men who had jumped into his car. He further testified he could only recall telling the police initially that his assailants were “four black males.” (T. 129-30). Dugan confirmed Colon’s memory on this point, testifying that the only description of the men he received from the precinct was “four male blacks” (T. 204-06).
 The fingerprints found on the automobile were the only other relevant piece of evidence offered at trial. Ronald Brown’s fingerprints were discovered on the inside front window. However, Dickerson’s fingerprints were nowhere to be found (T. 198-204, 230-36, 244-51).
 The respondent ascribes error to the district court opinion in two respects. He contends first that the court failed to apply properly the presumption of correctness to the state court’s findings of fact as required by 28 U.S.C. § 2254(d). He also argues that notwithstanding the district court’s analysis, Colon’s in-court identification was free from the taint of the concededly suggestive criminal court confrontation. Both points merit discussion.
 PRESUMPTION OF CORRECTNESS
 In reviewing an applicant’s state court conviction, a federal court is statutorily commanded to presume that findings of fact made by a state court are correct. 28 U.S.C. § 2254(d). This mandate does not mean a federal court must invariably accept a state court’s factual findings as true beyond question. However, when a federal court chooses to reject a state court’s factual finding, the federal court must explain why the finding was erroneous in spite of the presumption of correctness it was afforded. Sumner v. Mata, 449 U.S. 539, 101 S.Ct. 764, 66 L.Ed.2d 722 (1981).
 Respondent argues that the district court failed to apply this rule when it determined, counter to the state appellate court’s determination, that Colon’s in-court identification of Dickerson was constitutionally unreliable in the wake of unnecessarily suggestive pre-trial identification procedures. In particular, respondent insists the district court did not apply the presumption of correctness to the state appellate court’s conclusion that the indicia of the reliability of Colon’s identification of Dickerson as set forth in Neil v. Biggers, 409 U.S. 188, 93 S.Ct. 375, 34 L.Ed.2d 401 (1972), were adequately satisfied.
 This argument misconstrues the crux of the presumption of correctness doctrine, as well as the nature of the state court’s findings. Though perhaps hazy at times, there is a material distinction between “basic, primary, or historical facts . . . in the sense of a recital of external events and the credibility of their narrators,” and a mixed determination of fact and law which “requires the application of legal principles to the historical facts” of the case. Cuyler v. Sullivan, 446 U.S. 335, 342, 100 S.Ct. 1708, 1715, 64 L.Ed.2d 333 (1980). The presumption of correctness applies only to a state court’s findings of fact, it does not apply to mixed findings of law and fact. Id. at 341-42, 100 S.Ct. at 1714-1715.
 Respondent’s argument mistakenly presupposes that the state court’s conclusions vis-a-vis the application of Neil v. Biggers criteria are the sort of factual findings to which the presumption of correctness attaches. To the contrary, they embody mixed findings of law and fact which are not entitled to the presumption.
 For instance, whether Colon had an adequate opportunity to view the person directly behind him in the car, the first of the fiv Neil v. Biggers factors used to assess an identification’s independent reliability, calls for a conclusion based on the application of legal standards to empirical facts. The facts underlying this determination include, inter alia, how long Colon and the man were in the car together, whether Colon turned around to look at the man, how far away Colon was from the car when he turned around and saw the back seat passenger staring at him, whether a light was on in the car at the time, the condition of the street lighting, what the position of the hat on the man’s head was when Colon first looked at him as well as later on, how long he watched the man’s face, and the transparency of the car’s rear view window for viewing the back seat passenger. Whether Colon had an adequate opportunity to view the man depends not only upon the answers to questions such as these, but upon an evaluation of the answers in relation to each other and in light of applicable legal guidelines; it is not itself a historical fact.
 Accordingly, the district court committed no error when it reexamined the reliability of Colon’s identification of Dickerson without expressly relying on the presumption of correctness.
 On the other hand, the court below did have occasion to review the state court’s factual findings which fall within the purview of 28 U.S.C. § 2254(d). For the most part, Judge Lasker saw no reason to depart from the state appellate court’s findings. However, he did find one critical factual finding to be incorrect, and consonant with Sumner v. Mata, supra, he carefully explained why the finding was unsupported by the record pursuant to 28 U.S.C. § 2254(d)(8).
 This was the finding, conspicuously absent from the dissenting justices’ version of the facts which was otherwise adopted by the majority, that Colon provided his more detailed description of the back seat passenger to the police after the hold-up but prior to the courtroom confrontation. See People v. Dickerson, supra
414 N.Y.S.2d at 713. The district court found instead that the evidence produced at trial demonstrated that Colon’s description of the rear seat passenger as a man with a short afro haircut and a little moustache could have been furnished to the police only after he saw Dickerson at Brown’s arraignment. Respondent does not seriously protest the district court’s rejection of the state court’s finding as to the timing of Colon’s description, nor do we find any reason to reverse it. Judge Lasker’s chronology of events is not only plausible; it is the only one consistent with the testimony.
 Even when pressed, Colon could not recall whether he offered his detailed description before or after the criminal court incident (H. 177-79, 195; T. 134). Officer Dugan, who set up the courtroom viewing, testified that when he spoke to Colon earlier by phone, Colon had described the perpetrators to him merely as “four young male blacks.” (H. 57-57a, 58, 61; T. 204-06). Finally, Colon maintained that the only time he
actually saw the back seat passenger during the commission of the crime was after he had left the car, walked away a few feet, turned around, and saw the man trying to pull a hat down over his eyes (H. 188; T. 86, 138-39).
 Because Colon saw the man only with a hat covering the top of his head to his eyebrows, the circumstances leave no room for the possibility that he could have seen the man’s short afro hairstyle as he turned to see the man attempting to pull his hat over his face. No evidence was produced to counter this unavoidable deduction, such as the testimony of another police officer who might have taken Colon’s description before Brown’s arraignment, or records of any telephone conversations between Colon and any police officers. Thus, simple logic compels the conclusion that Colon offered his more elaborate description of the back seat assailant only after he saw Dickerson at the arraignment. The state appellate court’s finding to the contrary is therefore not supported by the record, 28 U.S.C. § 2254(d)(8), and is not entitled to the presumption of correctness.
 RELIABILITY OF COLON’S IDENTIFICATION OF DICKERSON
 Long cautious about the accuracy of eyewitness identifications, the Supreme Court has steadfastly stressed that the reliability of the identification is the “linchpin in determining the admissibility of identification testimony.” Manson v. Brathwaite, supra 432 U.S. at 114, 97 S.Ct. at 2253. Under the due process of law afforded to state criminal defendants by way of the fourteenth amendment to the United States Constitution, eyewitness identification evidence is not reliable and must be suppressed “if suggestive identification procedures have led to `a very substantial likelihood of irreparable misidentification.'” Simmons v. United States, 390 U.S. 377, 384, 88 S.Ct. 967, 971, 19 L.Ed.2d 1247 (1968), quoted in Manson v. Brathwaite, supra 432 U.S. at 116, 97 S.Ct. at 2254; Solomon v. Smith, 645 F.2d 1179, 1185 (2d Cir. 1981).
 Thus, whether eyewitness identification evidence which ensues from a pre-trial identification procedure is constitutionally infirm requires a two-step inquiry. First, it must be determined whether the pre-trial identification procedures were unduly suggestive of the suspect’s guilt. If so, then the suggestiveness of the identification procedures must be balanced against factors indicating that the in-court identification was independently reliable. “Even grossly suggestive procedures will not require suppression of a witness’ identification testimony if it is clearly reliable, independent of improper procedures.” Styers v. Smith, 659 F.2d 293, 297 (2d Cir. 1981). See Neil v. Biggers, supra 409 U.S. at 199-200, 93 S.Ct. at 382-383; Manson v. Brathwaite, supra 432 U.S. at 114, 197 S.Ct. at 2253.
 Biggers and Brathwaite cite five factors to consider in evaluating the likelihood of a mistaken identification. These include:
the opportunity of the witness to view the criminal at the time of the crime, the witness’ degree of attention, the accuracy of his prior description of the criminal, the level of certainty demonstrated at the confrontation, and the time between the crime and the confrontation.
 Manson v. Brathwaite, supra at 114, 97 S.Ct. at 2253; Neil v. Biggers, supra 409 U.S. at 199-200, 93 S.Ct. at 382-383.
 There is no need to belabor the first prong of the reliability litmus test. Respondent does not challenge the district court’s conclusion that the pre-trial identification procedures utilized at Brown’s arraignment were highly suggestive. Indeed, the district court’s finding was not substantially different from that of the appellate division, which acknowledged but excused as “unavoidable” the suggestive identification procedures used by the police. People v. Dickerson, 414 N.Y.S.2d at 714.
 Briefly, the procedure was suggestive in several respects. By directly focusing Colon’s attention on two specific male blacks, Dugan, in effect, conducted an unjustifiable show-up. Foster v. California, 394 U.S. 440, 89 S.Ct. 1127, 22 L.Ed.2d 402 (1969) Jackson v. Fogg, 589 F.2d 108, 111
(2d Cir. 1978). Moreover, Dugan illegitimately pressured Colon into taking a second and third look at Dickerson after his first tentative identification. See Solomon v. Smith, supra at 1185. Finally, Dugan acted to reinforce any possible misidentification of Dickerson by arresting him in Colon’s presence, thereby confirming Colon’s selection of Dickerson as the assailant to his face. See United States v. Leonardi, 623 F.2d 746, 754-55 (2d Cir.), cert. denied, 447 U.S. 928, 100 S.Ct. 3027, 65 L.Ed.2d 1123 (1980); United States v. Jarvis, 560 F.2d 494, 499-500 (2d Cir. 1977), cert. denied, 435 U.S. 934, 98 S.Ct. 1511, 55 L.Ed.2d 532 (1978). While none of these aspects of the confrontation alone necessarily would have invalidated the identification, the combination of them all comprised a highly suggestive identification procedure.
 Given these impermissible procedures, an examination of the reliability of Colon’s identification of Dickerson next requires weighing the five Biggers criteria against the extent of damage ensuing from the suggestive procedures. Each factor merits separate consideration.
 1. Opportunity To View The Criminal At The Time Of The Crime
 The first and only time Colon saw the back seat passenger’s face was after he exited the car, walked down the street, and turned around to face the rear of the car. Although nothing in the record reveals what the lighting conditions were like, it is undisputed that the crime occurred in the dead of the night, hours before daybreak.
 Although his estimate was characterized as “unrealistically long,” see People v. Dickerson, 414 N.Y.S.2d at 717 (dissenter’s narration of facts expressly adopted by the majority at 713), Colon guessed he stared at the man for one or two minutes. Even so, the district court perceptively observed:
[W]hatever the duration of Colon’s observation of the back seat passenger, his opportunity to form an accurate mental picture was impaired by the fact that it was nighttime, by the tension of the moment, by the distance between himself and the car, by the rear window of the car, and by the hat itself which Colon testified that the man pulled over his head.
 Dickerson v. Fogg, 526 F.Supp. at 1307. Consequently, even though identifications have been properly admitted where the observation time was less than that here, see, e.g., Mysholowsky v. People, 535 F.2d 194 (2d Cir. 1976), Colon’s opportunity to observe the back seat passenger was minimal at best under the circumstances. This factor fails to lend support to a determination of reliability.
 2. Witness’s Degree Of Attention
 The state court records reveal that Colon neither paid, nor could have paid, a great deal of attention to the back seat passenger’s physical traits. Unlike the victims in Chavis v. Henderson, 638 F.2d 534 (2d Cir. 1980), cert. denied, 454 U.S. 842, 102 S.Ct. 152, 70 L.Ed.2d 125 (1981), or Mysholowsky v. People, supra, Colon had no disinterested pre-crime view of the man. Moreover, there is no indication that Colon’s position as a security supervisor heightened his attentiveness to his assailant’s features.
 To the contrary, it is reasonable to infer, as the district court did, that Colon was frightened and agitated when he was released from the car, having just had his life threatened and a gun at his neck. Colon was so shaken by the experience he could not remember what he told the police when they finally arrived (T. 129-30). Lastly, any possibility that Colon’s training as a security supervisor contributed to his attentiveness is belied by his inability to describe the back seat passenger to the police with any degree of specificity.
 3. Accuracy Of Colon’s Prior Description
 Following the robbery, Colon described his assailants to the police only as “four young male blacks.” (Appendix 59-60, 64; T. 129-30, 204-06). As far as it goes, this general description was not inaccurate. Nevertheless, in the context of this case, Colon’s inability to provide a minimally detailed description when he talked with the
police plainly devalues his description as a factor demonstrating the reliability of his identification.
 In Chavis v. Henderson, supra, the court posited:
While an inaccurate description may be an indication of unreliability, the absence of a detailed description in a situation where prompt police work permits little time for detailed inquiry does not carry the same connotation.
 Id. at 537. In that case, the petitioner was caught and arrested within seconds of the crime. The police immediately returned him to the victim’s home for an identification. With the suspect in tow, the police were understandably under pressure to carry on with the identification rather than to take the time to secure a detailed description first. In addition, because the Spanish-speaking victim could communicate with the police only through a translator, obtaining an immediate precise description prior to the identification would have been exceedingly difficult. Even under these circumstances, the victim was able to describe her assailant as a “male black wearing dark clothing.”Id.
 No such exigent circumstances demanding a prompt identification at the expense of a prior detailed description were present in the case at bar. There is no reason why, had Colon been able to provide one, a more precise detailed description could not have been elicited or offered after the crime was committed. The sole purpose of the meeting between Colon and the police at the scene of the crime was to conduct an investigation. At the time, no suspect had been caught. Colon was not injured, in danger, in any hurry, or otherwise precluded from providing as much descriptive detail as possible to the investigating police when his memory of the assailants was the clearest. See Chavis v. Henderson, supra
at 537. Similarly, had he been able to provide a more precise description before he confronted Dickerson at Brown’s arraignment, he was not prevented from doing so when he met Officer Dugan at the courtroom doorway. In sum, not only does the technically accurate but unduly vague description of the back seat passenger furnished to the police before the confrontation diminish the reliability of his later identification, but, considering Colon did not describe Dickerson in greater detail until after he viewed Dickerson in the courtroom, his second specific description was unmistakably the product of the viewing, and questionably reliable as well.
 4. Level Of Certainty Demonstrated By The Witness At The Confrontation
 Certainty entails confidence in one’s identification of a suspected perpetrator. Hesitancy, the inability to be positively sure about a suspect, and the extent to which an affirmative identification is the product of prodding by other are signals which undermine the certainty of the witness’s identification of the suspect at the pre-trial confrontation. All these factors are present in this case.
 Colon was reluctant to commit himself to a positive identification the first time he stared at Dickerson at Brown’s arraignment. He told Dugan only that he was “pretty sure” that one of the men was the back seat gun-wielding passenger (H. 156), or that one of the two men “looks like” the man who had sat in the rear seat. Not content with Colon’s initial equivocal response, Officer Dugan sent Colon back for a second look (H. 156). Colon returned and this time told Officer Dugan that Dickerson “looked just like” the back seat passenger. Not until Dickerson and his companion emerged from the courtroom and walked past Colon and Officer Dugan did Colon firmly identify Dickerson, but only in response to Dugan’s demand for a “yes” or “no” answer.
 In short, Colon’s identification of Dickerson in the courtroom was not offered with the self-assured aura of confidence which might have been expected considering the suggestive surrounding circumstances. We therefore agree with the district court’s assessment that while Colon was not totally unsure about his initial confrontational identification, his ultimate positive confrontation
does not significantly buttress the independent reliability of his identification under the circumstances. See Foster v. California, supra.
 5. Length Of Time Between The Crime And The Confrontation
 As a final factor, Neil v. Biggers calls upon the court to consider the impact of the time between the crime and the confrontation on the reliability of the witness’s identification of the suspect. According to the state court records, the holdup took place at 4:00 a.m. on May 19, 1977. The courtroom confrontation occurred at approximately 8:00 p.m. on May 20, 1977, roughly forty hours after the crime was committed. The interval between the crime and the confrontation is critical for “immediacy makes it much more likely that the witness will have a fresh recollection of the appearance of the suspect and hence that the identification will be accurate.” Chavis v. Henderson, supra at 537, quoting United States ex rel. Springle v. Follette, 435 F.2d 1380, 1383 (2d Cir. 1970), cert. denied, 401 U.S. 980, 91 S.Ct. 1214, 28 L.Ed.2d 331 (1971).
 The time interval here was not so long it would have completely obscured Colon’s memory. On the other hand, it was several hours after the initial time frame, when an eyewitness is able to retain the sharpest image of the suspect. As such, although the forty hour period from the crime to the confrontation may not have decreased Colon’s recall facilities to any great extent, neither does it contribute to a showing that there was an independent basis of reliability for Colon’s identification.
 When balanced against the highly suggestive pre-trial identification procedures, these patently weak indicia of reliability cannot erase the corrupting influence of Colon’s improper confrontation identification of Dickerson. It is evident that under totality of the circumstances, there was a “very substantial likelihood of irreparable misidentification.” Manson v. Brathwaite, supra 432 U.S. at 116, 97 S.Ct. at 2254. We are therefore constrained to agree with the district court that Dickerson’s right to due process under law was abridged by the introduction of Colon’s eyewitness identifications.
 HARMLESS ERROR
 The admission of evidence of Colon’s identification of Dickerson was not harmless error in this case. Chapman v. California, 386 U.S. 18, 87 S.Ct. 824, 17 L.Ed.2d 705 (1967). Colon’s identifications were the only direct nexus linking Dickerson to the robbery. Officer Moruzzi’s testimony establishes that Dickerson was in the nonexclusive possession of Colon’s automobile hours after it was stolen, and little else. The identification evidence was therefore crucial to the prosecution, and damaging, considering the outcome, to Dickerson.
 For these reasons, testimony relating to both Colon’s out-of-court and in-court identifications of Dickerson were unconstitutionally admitted at trial. The judgment below is affirmed.
hearing transcript. References to “T” are to the trial transcript.
 NEWMAN, Circuit Judge, concurring:
 Judge Friendly’s dissent contends that the circumstances under which Colon identified Dickerson at the arraignment of Dickerson’s cousin were not sufficiently suggestive to warrant assessment of the reliability of the identification under the criteria set forth in Neil v. Biggers, 409 U.S. 188, 199-200, 93 S.Ct. 375, 382-383, 34 L.Ed.2d 401 (1972). I share the dissent’s concern that the police should not be obliged to act in a “wholly unnatural fashion,” see infra at 249, but I am satisfied that no such requirement has been imposed by the decision of this case.
 In the first place, the brief from the office of the New York Attorney General, challenging the District Court’s grant of habeas corpus relief, does not even ask us to reject Judge Lasker’s conclusion that the circumstances were sufficiently suggestive to warrant analysis under the standards of Neil v. Biggers.
 On the merits of the issue, I agree with Judge Friendly that some of the police actions criticized by the District Court are relatively innocuous. These include informing
Colon, prior to the identification, that his car had been recovered and displaying the security shield obtained from the car. However, Judge Curtin’s opinion for the panel majority places no reliance on these circumstances. The cardinal circumstance indicating undue suggestiveness is the police action in focusing the witness’s attention on just two males, out of all the people seated in the courtroom. Surely a twoman line-up would have been unduly suggestive, and what occurred here was the functional equivalent. Perhaps Officer Dugan’s second and third invitations to the witness to make an identification do not deserve to be called pressure, but such conduct had a suggestive tendency, when coupled with the limitation of viewing to just two males. Of course, the police are entitled to have a witness express his degree of certainty about an identification, but the undisputed facts here go beyond affording such an opportunity. After the witness was able to say at most that the suspect “looks like” and “looked just like” the culprit, the police officer asked, “Is it him or not.” Since the first two responses were more positive than negative, the presentation of absolute alternatives fairly invited an absolutely positive response. We need not decide the significance of such persistent probing when a witness views a properly conducted line-up, but the suggestiveness of the two-man show-up leaves Officer Dugan’s questioning vulnerable to criticism.
 The suggestiveness of the police conduct, however, does not by itself render Colon’s testimony inadmissible. Following the analytical framework of Manson v. Brathwaite, 432 U.S. 98, 97 S.Ct. 2243, 53 L.Ed.2d 140 (1977), the District Court and this Court conclude only that the circumstances of the identification were sufficiently suggestive to warrant careful consideration of reliability under the criteria identified by the Supreme Court. I fully agree with Judge Curtin that such analysis is warranted and that it indicates “a very substantial likelihood of irreparable misidentification,” Simmons v. United States, 390 U.S. 377, 384, 88 S.Ct. 967, 971, 19 L.Ed.2d 1247 (1968). I therefore join Judge Curtin’s opinion.
67 A.D.2d 122, 125, 414 N.Y.S.2d 712, 714 (1979). We are not told how many others were seated in the entire courtroom. The State offers no reason why Colon could not have been asked to survey all persons in the courtroom, or, at the very least, the 15 or 20 people at the right front.
 FRIENDLY, Circuit Judge, dissenting:
 I respectfully dissent. This seems to me an instance where commendable judicial desire
to prevent the risk of misidentification has turned into hypercritical scrutiny of police conduct and the reversal of a jury’s verdict, founded on substantial evidence and observation of the witnesses, by federal judges working only from a cold record. See Foster v. California, 394 U.S. 440, 446-47, 89 S.Ct. 1127, 1130-1131, 22 L.Ed.2d 402 (1969) (dissenting opinion of Justice Black). I agree with Justice Lupiano’s opinion in the Appellate Division that “[w]hile the pretrial identification of defendant by Mr. Colon, the complainant, partook of a certain, albeit unavoidable, degree of suggestiveness, it was not unduly suggestive.” 67 A.D.2d 122, 125, 414 N.Y.S.2d 712, 714 (1st Dept. 1979). There is thus no basis for suppressing Colon’s pretrial identification and no need to subject his identification at Dickerson’s trial to the Biggers criteria. The suggestiveness of what was done here was considerably less than of that i Foster v. California, supra, a case in which the majority’s finding of suggestiveness was unpersuasive to Justices White, Harlan, Stewart and Black. The facts are still further from those in Jackson v. Fogg, 589 F.2d 108, 109-10 (2 Cir. 1978), a decision in which I joined.
 The majority follows the district court in condemning Officer Dugan and his colleagues for not conforming to a code of conduct which defies human nature. The first fault found by the district judge was that when Colon was asked to come down to the Criminal Court, he was told that the police had recovered his car. Surely Colon was entitled to know this important fact. What could have been more natural and less suggestive than to tell him? If the police had simply requested him to come to the courthouse, would he not have asked why? If the police had said they wanted him to look at someone, that would have been viewed as just as bad or worse. Suggestiveness could have been avoided, on the district court’s view, only by refusing to give any reason. On that hypothesis, would not Colon, on his journey to the courthouse, have been thinking that the police had found someone? In other words any suggestiveness in telling Colon about the recovery of the car was inherent in the situation; we must not forget that what the due process clause condemns is “unnecessarily suggestive identification.” Stovall v. Denno, 388 U.S. 293, 302, 87 S.Ct. 1967, 1972, 18 L.Ed.2d 1199 (1967).
 The district court also stressed that when Colon got to the courthouse, Officer Dugan was wearing the Sloan security shield which Colon had kept in his car. This added nothing in the way of suggestiveness to the good news that the car had been recovered. Dugan is then faulted by the district court for telling Colon that Brown had been arrested and was being arraigned. Again any other course would require the police to act in a wholly unnatural fashion. Colon had been requested to come to the Criminal Court and was about to be asked to go into the courtroom. It would be altogether natural for him to want to know why. Was Dugan required to say simply “Something is going on in there”?
 Although I agree that it would have been better if Colon had been asked to observe the entire right hand side or perhaps even the whole of the courtroom and had not had his attention focused on a group which supposedly had come down for Brown’s arraignment, I do not think this alone made the identification impermissibly suggestive. Both the police and Colon acted with care. On his first viewing Colon was only “pretty sure”. The majority’s statement, echoing the district court’s opinion, that by sending Colon back so he “could take a better look” Officer Dugan “necessarily encouraged Colon to make a positive identification of Dickerson and implicitly communicated his belief that Colon’s leaning was correct” seems to be made up of whole cloth. The majority of the Appellate Division considered Colon’s reluctance to make a firm identification on his first viewing and Dugan’s insistence on his having “a better look” not as pressure but as “evinc[ing] a reflection which only seems to buttress the probability of a reliable identification”, 67 A.D.2d at 126, 414 N.Y.S.2d at 715. “Although the ultimate question of impermissible suggestiveness” is a mixed question of
law and fact that is not governed by [28 U.S.C.] § 2254″, Sumner v. Mata (II), 455 U.S. 591, 597, 102 S.Ct. 1303, 1306, 71 L.Ed.2d 480 (1982), the Court expressly held that the question “whether the witnesses were under pressure from prison officials or others” is one of the “questions of fact to which the statutory presumption applies.” Id. The majority’s statement that Officer Dugan “illegitimately pressured Colon into taking a second and third look”, thus cannot stand since there was as much or, in my view, more evidence to support the inference drawn by Justice Lupiano as to support the majority’s. I find equally little cause for objection in Dugan’s arrest of Dickerson. Just what was Dugan to do? Neither United States v. Leonardi, 623 F.2d 746, 754-55 (2 Cir.), cert. denied, 447 U.S. 928, 100 S.Ct. 3027, 65 L.Ed.2d 1123 (1980), no United States v. Jarvis, 560 F.2d 494, 499-500 (2 Cir. 1977), cert. denied, 435 U.S. 934, 98 S.Ct. 1511, 55 L.Ed.2d 532 (1978), holds that taking necessary action upon an identification which has become as positive as Colon’s renders the identification suspect. Indeed, this court noted i Leonardi that it is almost inevitable that, after a positive identification, “a witness may be asked to give a sworn statement, or else to appear before a grand jury, or to aid the police in other ways, all of which might confirm the `correctness’ of his choice.” 623 F.2d at 755.
 In an opinion subsequently reversed by the Supreme Court as too favorable to defendants who claim to be the victims of mistaken identification, Brathwaite v. Manson, 527 F.2d 363, 371 (2 Cir. 1975), rev’d, 432 U.S. 98, 97 S.Ct. 2243, 53 L.Ed.2d 140 (1977), I spoke of the necessity for rules that will “force police administrators and prosecutors to adopt procedures that will give fair assurance against the awful risks of misidentification.” Strongly adhering to that view, I would not carry it so far as to condemn what seems to me to have been the proper, if not perfect, police conduct in this case. Several of the majority’s criticisms seem to lie beyond our power under 28 U.S.C. § 2254(d), as construed in Sumner v. Mata (II), supra.
Even on the ultimate issue of impermissible suggestiveness I would pay considerable deference to the denial of a suppression motion by a justice of the Supreme Court of New York who saw and heard the witnesses, to the careful (although divided) analysis by the Appellate Division of the First Department, and to the unanimous affirmance of that decision by the Court of Appeals, 50 N.Y.2d 937, 431 N.Y.S.2d 453, 409 N.E.2d 927 (1980), even though that court was confined to questions of law.