dissenting.
I dissent from giving retroactive effect to Rose v. Lundy, 455 U.S. 509, 102 S.Ct. 1198, 71 L.Ed.2d 379 (1982).1
The majority’s holding on exhaustion overrules Richards v. Solem, 693 F.2d 760 (8th Cir.1982), cert. denied, — U.S. —, 103 S.Ct. 1898, 77 L.Ed.2d 286 (1983), and Dunn v. Wyrick, 679 F.2d 731 (8th Cir.1982). Although the Supreme Court in Rose v. Lundy left open the retroactivity question, the majority apparently feels its holding is compelled by the three post-Rose v. Lundy summary orders issued by the Supreme Court. The Supreme Court has stated, “the lower courts are bound by summary decisions by this court ‘until such time as the court informs [them] that [they] are not.’ ” Hicks v. Miranda, 422 U.S. 332, 344-45, 95 S.Ct. 2281, 2289, 45 L.Ed.2d 223 (1975), citing Doe v. Hodgson, 478 F.2d 537, 539 (2d Cir.) (brackets in quoted material), cert. denied, 414 U.S. 1096, 94 S.Ct. 732, 38 L.Ed.2d 555 (1973).
The majority’s discussion of Bergman v. Burton, 456 U.S. 953, 102 S.Ct. 2026, 72 L.Ed.2d 478 (1982) (mem.), Duckworth v. Cowell, 455 U.S. 996, 102 S.Ct. 1626, 71 L.Ed.2d 858 (1982) (mem.), and Rodriquez v. Harris, 455 U.S. 997, 102 S.Ct. 1627, 71 L.Ed.2d 858 (1982) (mem.), fails to persuade me that the analysis undertaken in Richards v. Solem is incorrect. See Richards, 693 F.2d at 764. In the first two summary orders, Duckworth and Rodriquez, the Supreme Court vacated the lower court judgment and remanded with directions to dismiss the mixed habeas petition. Yet in the latest case, Bergman, the Court merely remanded the case to the Sixth Circuit for further consideration in light of Rose v. Lundy. This inconsistency demonstrates that the Supreme Court’s three post-Rose v. Lundy orders do not mandate retroactive application of Rose v. Lundy.
Accordingly, I dissent from the majority’s opinion overruling Richards v. Solem and Dunn v. Wyrick.
II. THE MERITS
FAGG, Circuit Judge, with whom ROSS, ARNOLD, JOHN R. GIBSON and BOWMAN, Circuit Judges, join.
Because we hold that the district court failed properly to apply the “fairly supported by the record” standard of 28 U.S.C. § 2254(d)(8) in determining that the victim’s identification testimony was inadmissible, we reverse the judgment of the district court granting the writ of habeas corpus.
The evidence presented at trial shows that the victim and her roommate met with a co-worker on August 22, 1977, to celebrate the co-worker’s return to college. After smoking some marijuana and drinking some beer the three women traveled to a nearby Nebraska restaurant for dinner. Following their meal, the women returned to Yankton and went to a bar. While the victim was talking to an acquaintance, her friends left for a bar next door. Around midnight the victim left the bar to rejoin her friends next door. En route, she stopped to look inside a customized van which was parked on the street. While she was peering inside the open side door of the van, a man shoved her inside the van and forcibly restrained her. Two other men immediately got into the van and drove it *1540away with the victim still inside. The victim later identified the driver as defendant Reiman and the passenger as petitioner Graham. The man who held her in the van was never identified. The van was driven into a building which the victim later identified as the Outasite Paint Shop owned by defendant Reiman. The men assaulted her, forcibly removed her clothing, and then repeatedly raped her for at least four hours. There were at least five men who attacked her, four of whom she later identified as defendants Reiman, Onstott, Elliott and Graham. Her attackers later released her near her home.
Immediately after being released by her captors, the victim went home and unsuccessfully tried to wake her roommate. She then went to the home of two friends and told them about the rape. Afterwards she returned home and spoke to her roommate about the rape. During her discussions about the rape, her roommate and friends mentioned that Reiman and his cohorts might be the rapists. Graham’s name was not mentioned. The victim’s roommate also mentioned that the rape could have occurred at Reiman’s paint shop. That same day the victim went to the Yankton County Sheriff’s office and gave the police descriptions of four of the assailants. She also drew a diagram of the building where the rape occurred.
On August 31, 1977, nine days after the assault, the police drove the victim around Yankton to see if she could identify any of her assailants. While touring the town, the victim pointed out defendants Reiman, On-stott and Elliott immediately upon seeing them for the first time after the rape. The next day the police drove the victim past a construction site where Graham was working and the victim noted Graham’s similarity to her attacker. Later that day, as the police were driving the victim around town they went past an auto garage shop where Graham was located. The victim picked Graham out of a group of men standing in front of the garage and at that time she was “ninety percent sure” that Graham was one of her assailants. The police then brought Graham to the Yankton Public Safety Center for a one-man lineup or showup. At this point, the victim said she was absolutely certain that Graham was one of the rapists. Graham was arrested and charged with kidnapping and rape. Also that day, the victim accompanied the police in their execution of a search warrant at the Outasite Paint Shop. While at the paint shop, the victim recognized a bathroom mirror as one she had seen on the night of the rape. All the defendants admit to being at or near the Outasite Paint Shop during the hours the rape occurred.
The federal district court held that the victim’s in-court identification of Graham was so tainted by the suggestive pretrial identification procedures that it denied Graham due process of law. For the reasons discussed below, we disagree.
A. Overview
In reviewing a petitioner’s state court conviction, a federal court must accord the state court findings of fact a “high measure of deference.” Sumner v. Mata, 455 U.S. 591, 598, 102 S.Ct. 1303, 1307, 71 L.Ed.2d 480 (1982) (per curiam). Under 28 U.S.C. § 2254(d), the state court findings are presumed to be correct. Before such findings may be set aside, a federal court must do more than simply disagree with the state court: “it must conclude that the state court’s findings lacked even ‘fair[] support’ in the record.” Marshall v. Lonberger, 459 U.S. 422, 103 S.Ct. 843, 850, 74 L.Ed.2d 646 (1983); 28 U.S.C. § 2254(d)(8). Conversely, when the state findings have the necessary support, the statute requires “the federal courts to face up to any disagreement as to the facts and to defer to the state court.” Sumner v. Mata, supra, 455 U.S. at 597, 102 S.Ct. at 1307.
In the process of finding the underlying facts, credibility determinations are left for the state courts to decide; we are not permitted to substitute our judgment as to the credibility of witnesses for that of the state court. Maggio v. Fulford, — U.S. —, 103 S.Ct. 2261, 2262, 76 L.Ed.2d 794 (1983); Marshall v. Lonberger, supra, 103 S.Ct. at 851. A federal court must accept credibility, determinations *1541made by a state court just as any appellate court must accept the credibility determinations of a trial court. King v. Strickland, 714 F.2d 1481, 1494 (11th Cir.1983). Federal courts are not entitled to draw inferences which are adverse to, or conflicting with, the state courts’ factual determinations. Marshall v. Lonberger, supra, 103 S.Ct. at 851. It is immaterial that the state court made its factual findings on a cold record instead of live testimony. This does not alter the fundamental principles of federal-state relations on which the limitations of section 2254(d) are founded. See Sumner v. Mata, 449 U.S. 539, 550, 101 S.Ct. 764, 770, 66 L.Ed.2d 722 (1981). A state appellate court may be a fact finder for purposes of 28 U.S.C. § 2254(d). Id. at 545-48, 101 S.Ct. at 768-69; Green v. Zant, 715 F.2d 551, 558 (11th Cir.1983).
We disagree with the approach of both the district court and the dissent in at least two respects. The first disagreement concerns the deference that a federal court must accord to a state court’s factual findings. In this ease, there are conflicting arrays of evidence, one which favors the victim’s identification and another which suggests that the victim was mistaken in identifying Graham. The South Dakota Supreme Court based its findings upon the evidence that supports its determination that the victim’s identification was reliable from a constitutional standpoint. We believe that under the proper balance of federal and state roles in habeas corpus cases, the state courts are afforded the right to make such a choice from competing bodies of evidence. The federal role is simply to ascertain whether the state court’s choice has fair support in the record.
Second, we accept as a natural attribute of the adversary system that internal contradictions within the testimony of individual witnesses, as well as contradictions between witnesses, are inevitable, and are matters properly to be resolved by the fact finder based upon its sense of credibility. The victim of a rape has gone through a traumatic experience that is ripe for confusion, contradiction, and self-doubt. It is inherent in the fact finding process for the state court to determine the extent to which these considerations diminish the weight of the victim’s testimony. The district court and the dissent are skeptical of the presence of any evidentiary contradictions, and they have relied upon evidentiary contradictions in the record to discredit the victim’s testimony, which the state court found credible, and to find a set of facts considerably different from those found by the state court. We believe this methodology is at odds with the limited federal role under section 2254(d).
B. Reliability of the Identification
Graham’s due process challenge to the victim’s in-court identification is examined under a two-step test:
The first step is to determine whether the challenged confrontation between the witness and the suspect was “impermissi-bly suggestive.” Simmons [v. U.S.], 390 U.S. [377] at 384, 88 S.Ct. [967] at 971 [19 L.Ed.2d 1247]. If so, the second inquiry is whether, under the totality of the circumstances of the case, the suggestive confrontation created “a very substantial likelihood of irreparable misidentification.” Manson [v. Brathwaite], 432 U.S. [98], at 116, 97 S.Ct. [2243] at 2254 [53 L.Ed.2d 140].
United States v. Henderson, 719 F.2d 934, 936 (8th Cir.1983). See also United States v. Amrine, 724 F.2d 84, 87 (8th Cir.1983). Pared to its essence, the second inquiry is whether the identification is reliable. Manson v. Brathwaite, 432 U.S. 98, 114, 97 S.Ct. 2243, 2253, 53 L.Ed.2d 140 (1977); United States v. Amrine, supra, 724 F.2d at 87. In Neil v. Biggers, 409 U.S. 188, 93 S.Ct. 375, 34 L.Ed.2d 401 (1972), the Supreme Court provided criteria for an orderly marshaling of the facts incidental to an application of the legal standard of reliability. The following factors are to be considered:
the opportunity of the witness to view the criminal at the time of the crime, the witness’ degree of attention, the accuracy of the witness’ prior description of the criminal, the level of certainty demonstrated by the witness at the confronta*1542tion, and the length of time between the crime and the confrontation.
Id. at 199-200, 93 S.Ct. at 382. Against these factors, the court must weigh the corrupting effect of the suggestive identification itself. Manson v. Brathwaite, supra, 432 U.S. at 114, 97 S.Ct. at 2253; United States v. Henderson, supra, 719 F.2d at 937.
The ultimate question of the constitutionality of admitting the in-court identification is not governed by section 2254. The questions of fact that underlie this conclusion, however, are governed by the statutory presumption. Maggio v. Fulford, supra, 103 S.Ct. at 2264-65; Sumner v. Mata, supra, 455 U.S. at 597, 102 S.Ct. at 1307 (1982). “Thus, whether the witnesses in this case had an opportunity to observe the crime or were too distracted; whether the witnesses gave a detailed, accurate description; and whether the witnesses were under pressure * * * are all questions of fact as to which the statutory presumption applies.” Sumner v. Mata, supra, 455 U.S. at 597, 102 S.Ct. at 1307.
Considering the first part of the two-step test, there is little doubt in this case that at least the final out-of-court confrontation, the showup, was impermissibly suggestive. We have previously noted that a showup is “the most suggestive, and therefore the most objectionable method of pre-trial identification.” United States v. Henderson, supra, 719 F.2d at 937, quoting United States v. Cook, 464 F.2d 251 (8th Cir.), cert. denied, 409 U.S. 1011, 93 S.Ct. 457, 34 L.Ed.2d 305 (1972). Once the victim had selected Graham from a distance as a person resembling her attacker, there was no necessity for this suggestive procedure of showing Graham to the victim singly. See Stovall v. Denno, 388 U.S. 293, 302, 87 S.Ct. 1967, 1972, 18 L.Ed.2d 1199 (1967).
The focus in this case is on the second step of the test: whether the in-court identification was reliable, even though an identification procedure was suggestive. The South Dakota Supreme Court made factual findings with regard to each of the Biggers factors and these findings must each be examined properly within the context of section 2254(d) to determine whether they have fair support in the record. Once each of these factors has been considered, giving proper deference to the state court findings, then the federal court must determine the legal question of whether the pretrial identification procedure has caused a very substantial likelihood of irreparable misidenti-fication.
1. The Opportunity of the Victim to View the Criminal at the Time of the Crime
The South Dakota Supreme Court found that “[t]he victim in this case had an unusual opportunity to view the defendant over a period of several hours.” State v. Reiman, 284 N.W.2d 860, 871 (S.D.1979). Further, the South Dakota Supreme Court noted the victim’s motivation carefully to observe the appearance of her attackers. Id. at 872. Thus, putting opportunity and motivation together, the court found that a mental image of the attackers was indelibly fixed in the victim’s mind during the extended assault. Id.
The record amply supports the supreme court’s finding. At trial the victim testified as follows:
Q. Were you able ever while they were having intercourse with you, either orally or otherwise, to see their faces?
A. Yes.
Q. Did you have an opportunity to see them more than once?
A. Yes.
Q. And for more than a brief time?
A. Yes.
Q. For quite a long period of time?
A. Yes.
Q. Is there any doubt in your mind that these are four of the men?
A. There is no doubt. These are the four.
Tr. 829-30. At another point during the trial she stated, “I’m positive it’s these four. I can never forget their faces.” Tr. at 860 (emphasis added).
*1543Confronted with the state court findings and the clear evidence that supports them, the federal courts are in no position to make different findings or to draw adverse inferences to the effect that the victim’s ability to perceive or recall was impaired. The victim was the only person to testify about her opportunity to see her attackers, and she stated that she had the ability to observe the men’s faces and that she did so. The victim stated specifically that she saw Graham both in the van and when he forced her to have intercourse. Tr. at 811, 827.
The district court and the dissent choose to focus on the evidence that the victim’s perception was blurred due to the consumption of alcohol and marijuana. The South Dakota Supreme Court specifically noted the victim’s smoking and drinking. State v. Reiman, supra, 284 N.W.2d at 863. Nevertheless, the state court found that the victim had an excellent opportunity to view Graham, that she did so, and that the observations were stamped into her memory. This finding subsumes the question of the victim’s blurred perception. The state court necessarily took into account the victim’s smoking and drinking, and the state court did not deem these factors to be impairments to the victim’s perception. There is evidence to support the state court’s findings concerning the victim’s perception. The victim testified at her deposition that she “was not drunk” and that she was acting normally. Dep. at 449. One of the victim’s companions did not think she was intoxicated. Tr. at 704. The other companion did not think she was acting unusual or high. Tr. at 770. Moreover, there was no testimony that the victim’s senses were dulled during the course of the rape, which lasted at least four hours — long after the time frame when any witness testified as to having seen her. The victim's testimony shows that the sordid details of the prolonged attack were committed to her memory.
The dissent also emphasizes that the victim did not have her glasses and that this impaired her perception. The victim is nearsighted, not farsighted, and only needs her glasses for driving. “Q. That means you can see things real close but not very far away. A. Yes.” Tr. at 897. Hence, it is unlikely that she would have needed her glasses to view her attackers, whom she viewed at necessarily close range. In any event, the specific finding of the state court that she was able to view her attackers subsumes the question of her impaired vision, and the federal court is not permitted to draw such an adverse inference.
The district court and the dissent emphasized the fact that the rape occurred in darkness. Once again, the state court necessarily took the lighting conditions into account in finding that the victim was able adequately to view her attackers. The only testimony that the room was dark was by the victim, who nevertheless stated that she could view her attackers and that a light was shining in the bathroom. Tr. at 821, 829. Finally, it should be noted that the Supreme Court has found victims of violent crime able to view their attackers sufficiently in lighting conditions that have been far from ideal. See Neil v. Biggers, supra, 409 U.S. at 193-94, 93 S.Ct. at 379-80 (victim viewed assailant in dark kitchen lighted only by adjoining bedroom and also outdoors in woods, by moonlight; entire incident lasted 15-30 minutes); Coleman v. Alabama, 399 U.S. 1, 4, 90 S.Ct. 1999, 2000, 26 L.Ed.2d 387 (1970) (brief view on dark highway lighted only by passing ear’s headlights).
2. The Victim’s Degree of Attention.
The South Dakota Supreme Court concluded that “[i]n view of the circumstances, we can assume the witness’ degree of attention was very high.” State v. Reiman, supra, 284 N.W.2d at 872. The district court and the dissent agree with this finding.
3. Accuracy of the Victim’s Prior Descriptions of the Criminal
The South Dakota Supreme Court found that there was no inaccurate prior description “because only four descriptions were given when there were at least five assailants. Graham could thus have been completely omitted from the group of suspects described.” Id. There is ample record sup*1544port for this finding. From the outset, she claimed rape by “six guys.” Tr. at 736. On August 23 she gave the sheriff four descriptions, not including Graham. Dep. at 454-55. The victim testified at trial that she did not identify Graham’s characteristics and features to the sheriff on August 23 because “[a]t that time” she “couldn’t remember.” Tr. at 892 (emphasis added). The victim provided a plausible explanation for her omission: she stated that when she saw the sheriff on that day, she was still “in shock.” Dep. at 454; Tr. at 892. The victim explained her reaction: “Maybe you can’t understand, but I was blocking everything and that was all I could remember at that time.” Dep. at 454. She stated that “within the next few days” she began to remember Graham’s identity: “I just started remembering. And I could remember the other ones clearer and that is when he came to mind.” Dep. at 456.
Of course, there are inconsistencies in the victim’s testimony but the fact finding role of selectivity and credibility is for the state court. Based upon the record, the South Dakota Supreme Court was in a position fairly to find that the victim did not misdes-cribe Graham before she saw him at the construction site. Hence, the accuracy of her description cannot fairly be eroded on the basis of the descriptions given to the sheriff on August 23.
4. The Level of Certainty Demonstrated by the Victim at the Pretrial Confrontations.
In principle, one-on-one confrontations are suggestive. Nevertheless, the record demonstrates that the pretrial confrontations had as a common theme the victim’s recognition of Graham tempered by restraint to avoid a hasty, mistaken identification. There is ample record support for the South Dakota Supreme Court’s determination that the confrontations did not destroy the reliability of the victim’s identification.
The difference in nature between two of the three pretrial confrontations and the victim’s encounter with her attackers at the time of the gang rape is important. Of necessity, the victim’s focus was upon the heads, faces, shoulders and arms of her attackers. Hence, any reluctance by the victim in identifying a male viewed from a distance in a head to toe profile does little realistically to dispel the reliability of an identification that was based upon a face to face encounter.
a. The first confrontation
The victim first viewed Graham at a construction site from a distance of forty yards. She noted Graham’s resemblance to her attacker but she stated that “I couldn’t tell for sure because the man had on a coverall suit and he had on a hard hat. I said if he could take that hat off, I could tell better because the man I remembered was balding towards the front of the head * * *.” Tr. at 854. The South Dakota Supreme Court found from the circumstances that the victim’s reluctance to make a positive identification was prompted by caution, not doubt, and indicated a high degree of reliability. State v. Reiman, supra, 284 N.W.2d at 872. The officers did not point Graham out or tell the victim who he was. She was driving “all over” with the officers and only told to “keep your eyes open.” Tr. at 853. It was the victim that saw the two men at work and noticed the similarity between Graham and her attacker. Tr. at 854.
b. The second confrontation
The victim next viewed Graham as he was standing in front of a garage with a group of other men. It was dark, making an identification based upon a close quarters encounter with the assailants somewhat difficult. The victim was riding in a car driven by an officer without any indication from him where they were going or who they might see. Tr. at 854; Dep. at 415. The officer stated, “we are going to just drive around again, and just keep your eyes open.” Tr. at 854. The victim testified that they drove around the streets while she viewed people. Tr. at 854. They drove by an auto garage shop and there were a group of men standing in the door. The victim picked Graham out of the group. *1545They drove by the garage a number of times, according to the victim, because it was not until the third time that “he [Graham] was standing in the light and I could see him.” Tr. at 855. The victim said she was “ninety percent sure” at that time that Graham was the man. Tr. at 855. The state trial court felt the victim had made a positive identification at this time. The victim did indicate a desire, however, to get a closer look at the man. Tr. at 855.
The facts underlying the first and second encounters dispel any notion of a very substantial likelihood of misidentification. The police merely drove the victim around the streets of Yankton without any coaching or suggestive remarks. The victim made the identification responses on her own, without any prompting by the officers. It was the victim herself, not the police, who requested a closer viewing of Graham, an opportunity she received in the third encounter.
c. The third confrontation
The victim’s third viewing of Graham occurred at the public safety center. Graham was in the center for questioning and the victim observed him through a glass door. This was the first time the victim had viewed Graham at close range. At this viewing, the victim stated she was sure of Graham’s identity as one of the rapists:
Q. Could you recognize him at that time?
A. Yes, that’s when — because I was sure before, but there was still some doubt in my mind. And I knew if I could just see him closer than I had been seeing him, I would know. And the thing that really made me sure was when I saw the tattoo on his left arm. I don’t remember what it is, but I remembered it from that night.
Tr. at 855.
The South Dakota Supreme Court recognized the practical nature of the exposure the victim had to Graham when she was attacked, when it made the point that when she saw his bald head and tattooed arm, “this crystallized the identification in her mind.” State v. Reiman, supra, 284 N.W.2d at 871. Notwithstanding the victim’s testimony, and the state court’s finding, that this completed the identification in her mind, the district court stated, “I do not consider the mention of the tattoo to be a significant part of the identification of Mr. Graham.” We cannot agree that the victim’s recognition of distinctive physical characteristics was entitled to so little weight.
The victim’s remarks confirm the state court finding that before she had the confrontations the rapes had already indelibly printed the faces of the attackers in her mind. The victim’s reactions under the circumstances comport with our common-sense knowledge of how memory works. “The lack of a ‘graphic description’ or even the failure to describe the assailant is not decisive since ‘most persons immediately after or a short time after viewing a person, particularly under circumstances that occasion an interest in his appearance, are able mentally to photographically recognize that person.’ ” Jones v. Director, Patuxent Institution, 351 F.Supp. 913, 940 (D.Md.1972), quoting Gallagher v. United States, 406 F.2d 102, 105 (8th Cir.), cert. denied, 395 U.S. 968, 89 S.Ct. 2117, 23 L.Ed.2d 756 (1969). The South Dakota Supreme Court, after reviewing the evidence, found that the victim was being cautious about her identification. That finding has support in the record and must therefore be accepted by this court. Some initial hesitancy on the part of a victim does not necessarily render the identification unreliable. See Smith v. Perini, 723 F.2d 478, 482 (6th Cir.1983); Summitt v. Bordenkircher, 608 F.2d 247, 252-53 (6th Cir.1979), aff’d sub nom., Watkins v. Sowders, 449 U.S. 341, 101 S.Ct. 654, 66 L.Ed.2d 549 (1981). Indeed, we should encourage thoughtful reserve by a victim to the end of sparing suspects a hasty, ill-considered, positive identification.
The advice the victim received from her friends concerning the possible identity of her attackers, while suggestive, was not crucial to Graham’s identification. The victim did not know any of the attackers prior to the rape. Tr. at 961. She had not seen them before and was not aware of their *1546names. Tr. at 962. Because she was in a state of shock the victim did not describe Graham to her friends. Tr. at 892. The victim’s friends mentioned that the attackers might be “Bud Reiman,” “Onstott,” and “Bud Reiman’s crew.” Dep. at 377; Tr. at 899. There is no evidence that Graham’s name was mentioned by the victim’s friends as a suspect; indeed, the victim testified that Graham’s name was never mentioned. Tr. at 899. Moreover, when the victim spotted Graham on the construction site and later viewed him at the garage and the public safety center, she was not informed that Graham was associated with Reiman or Onstott.
5. The Length of Time Between the Crime and the Confrontations
The South Dakota Supreme Court found that the time which elapsed between the rape (August 22) and the victim’s identification (August 31) was not unreasonably long. State v. Reiman, supra, 284 N.W.2d at 872. The district court and the dissent agree that this was “short enough for the victim to have any memories of the rape still fresh in her mind.”
6. Other Indicia of Reliability
Under the Biggers test for the admissibility of an eye-witness identification, other evidence of guilt does not play a formal role in the analysis. See Manson v. Brathwaite, supra, 432 U.S. at 116, 97 S.Ct. at 2253. Yet, it is difficult to ignore additional facts which indicate that the actual likelihood of misidentification in this case was slim. Indeed, it seems unnatural to set such evidence aside when one considers that the ultimate purpose of the Biggers factors is to avoid eyewitness testimony where there is “a very substantial likelihood of irreparable misidentification.” Neil v. Biggers, supra, 409 U.S. at 198, 93 S.Ct. at 381; Simmons v. United States, 390 U.S. 377, 384, 88 S.Ct. 967, 971, 19 L.Ed.2d 1247 (1968). The following factors do not seem wholly irrelevant to whether, in fact, there was such a substantial likelihood of misidentification in this case: (1) the victim was raped by four or more men; (2) all four men convicted of the rape admit to being together at the Outasite Paint Shop during the time the rape occurred; and (3) the rape victim identified the Outasite Paint Shop as the location of the rape. Although it played no part in its analysis, the Supreme Court noted in Manson that the reliability of the identification was “hardly undermined” by facts which connected the defendant with the scene of the crime. Manson v. Brathwaite, supra, 432 U.S. at 116, 92 S.Ct. at 2254.
The dissent concludes that the victim’s identification of the Outasite Paint Shop as the scene of the crime is “of questionable reliability.” The state court specifically found that this identification was reliable, and this finding is amply supported by the record. On direct examination, the victim was asked what she remembered about the appearance of the building where she was raped:
Right away I knew it was some type of garage. You could smell the oil and the dirt. I got the impression at first that it was a dirty place. It was cluttered. I remember seeing cars all over in there, cars being worked on. I don’t know how many, but they had different parts taken out and being put in, but it was so dark in there. I remember the cement floor with grease all over it.
Tr. at 820. The victim also testified that she remembered the mirror in the bathroom at the Outasite Paint Shop:
I remember it well because when I went into the bathroom, I ran there, I don’t know why, to get away; because I was bleeding from my nose, and I was just running. There was blood all over, and I remember looking above the bathroom sink in the mirror. The mirror was small, it was narrow, and there were red paint designs on it, weird ones, kind of. And I remember looking in and seeing my bloody face in that red painted mirror, and I just don’t think I can ever forget it.
Tr. at 821. Finally, the victim testified in her deposition that she remembered the distinctive smell of the garage, the windows of the garage door, the feel of the steel posts which she had held onto, the location of the toilet, and the general appearances of the *1547sink and the light in the bathroom. Dep. at 424 — 26. It is difficult to conceive what better evidence could be required fairly to support the state court finding than the detailed and realistic testimony of the victim.
C. Weighing Reliability and Suggestiveness
To determine the ultimate legal question in this case — whether the victim’s in-court identification violated Graham’s rights under the due process clause — we must balance the suggestiveness of the identification procedure against the reliability of the eyewitness identification. The reliability of the identification is “the linchpin in determining the admissibility of the identification testimony.” Manson v. Brathwaite, supra, 432 U.S. at 114, 97 S.Ct. at 2253.
In assessing the relative reliability of an identification, it is crucial to remember that the court is not deciding the truth or falsity of the testimony, but only whether there is such a substantial likelihood of irreparable misidentification that the testimony should not be heard by the jury. The testimony need not be without fault: “evidence with some element of untrustworthiness is customary grist for the jury mill. Juries are not so susceptible that they cannot measure intelligently the weight of identification testimony that has some questionable feature.” Id. at 116, 97 S.Ct. at 2254. Indeed, a cornerstone of our judicial system is the confidence we place in the ability of jurors to determine the truthfulness and accuracy of a witness’ testimony. Nowhere is Justice Black’s admonition against robbing the jury of this vital function more appropriate:
[I]t is an incontestable fact in our judicial history that the jury is the sole tribunal to weigh and determine facts. That means that the jury must, if we keep faith with the Constitution, be allowed to hear eyewitnesses and decide for itself whether it can recognize the truth and whether they are telling the truth, It means that the jury must be allowed to decide for itself whether the darkness of the night, the weakness of a witness’ eyesight, or any other factor impaired the witness’ ability to make an accurate identification. To take that power away from the jury is to rob it of the responsibility to perform the precise functions the Founders most wanted it to perform. And certainly a Constitution written to preserve this indispensable, unerodible core of our system for trying criminal cases would not have included, hidden among its provisions, a slumbering sleeper granting the judges license to destroy trial by jury in whole or in part.
Foster v. California, 394 U.S. 440, 447, 89 S.Ct. 1127, 1131, 22 L.Ed.2d 402 (1969) (Black, J., dissenting).
We are unable to conclude that there was such a substantial risk of irreparable mis-identification in this case that the state trial judge committed constitutional error by admitting the victim’s testimony identifying Graham. Without doubt, the identification procedure employed by the state was suggestive. Nevertheless, there were sufficient indicia of reliability to warrant the admission of the identification testimony. Giving proper deference to the state courts, we are faced with these facts concerning the reliability of the identification: the victim had an adequate opportunity to view her attackers at painfully close range for at least four hours; as a victim of this most degrading crime her degree of attention was very high; the victim’s level of certainty in identifying Graham was high; and finally, she identified Graham reasonably soon after the crime. The only Biggers factor which arguably detracts from the reliability of the identification is the absence of a prior description of Graham and this is satisfactorily explained by the victim’s initial emotional reaction to the brutal encounter with the rapists.
In assessing the importance of the victim’s failure to give a prior description of Graham, consideration must be given for the victim’s real life situation when she is forcibly raped by a gang of strangers. As explained earlier, it has not been shown that the victim misidentified Graham and she provided a plausible explanation for her omission — shock. The approach to the test of reliability must have a measure of flexi*1548bility; otherwise, the victim is in an impossible situation. If common sense and practical experience are drawn upon, the victim’s identification does not seem improbable or unreliable for the purposes of our constitutional analysis. Indeed, it is a common experience that after encountering strangers, an individual may not be able meaningfully to describe them to others. Yet, the faces may be etched in his or her memory, and upon seeing them again there is an immediate recognition. This appears precisely to have been the situation in this case. As the victim stated, she could “never forget their faces.” Tr. at 860.
Considering the totality of the circumstances in this case, the mere failure to give a prior description does not render the identification testimony constitutionally infirm. Cf. United States ex rel. Hudson v. Brierton, 699 F.2d 917, 925 (7th Cir.), cert. denied, — U.S. —, 104 S.Ct. 114, 78 L.Ed.2d 115 (1983); Brayboy v. Scully, 695 F.2d 62, 65-66 (2d Cir.1982), cert. denied, — U.S. —, 103 S.Ct. 1505, 75 L.Ed.2d 935 (1983). Likewise, the intensity and duration of the victim’s encounter with the rapists clearly overrides the suggestiveness of her post-rape confrontations with Graham. Any remaining concerns about the suggestiveness of the identification procedure or the reliability of the identification was properly to be considered by the jury in assessing the weight to be given the victim’s testimony. See United States v. Singleton, 702 F.2d 1159, 1166 (D.C.Cir.1983); Summitt v. Bordenkircher, supra, 608 F.2d at 253. We conclude that the identification by the victim was reliable, there was not “a very substantial likelihood of irreparable misidentification,” and therefore, Graham’s constitutional rights were not violated by the state court’s admission of the victim’s identification testimony.
Accordingly, the judgment of the district court granting the petition for writ of habe-as corpus is reversed.
LAY, Chief Judge, dissenting, with whom HEANEY, Circuit Judge, joins.
I join in Judge McMillian’s analysis of the merits. I also point out that Justice Dunn, an able and experienced jurist, vigorously dissented in the South Dakota Supreme Court in State v. Reiman, 284 N.W.2d 860, 874 (S.D.1979). His analysis is keenly perceptive. He concluded his dissent by saying: “After this police procedure, it seems incredulous that the complaining witness’ in-court identification had an origin independent of and untainted by these pretrial procedures.” Id. I couldn’t agree more.
. Specifically, I dissent from all but the final paragraph of section I. B.