dissenting:

I.

Some fifteen years ago, petitioner’s trial counsel, an attorney with approximately thirty years of active, continual, criminal experience, and who by that point had been involved in approximately 175 other murder cases, became counsel in this one. Petitioner, Charles Ray Tomlin, had been charged with the shooting death of Daniel Stewart during a hold-up in connection with a drug deal. Two other individuals were acquainted ■with some of the events that took place that evening. One, Laura Leticia Mendez, was sitting in the pickup truck between the assailant and Stewart when the assailant shot Stewart in the head. The other, Charles Tillman, had acted as an intermediary during the drug transaction, and had exited the truck moments before the assailant’s arrival. Tillman had been arrested, and was later convicted of murder and robbery for his participation in the crime.

Tomlin’s trial counsel knew that the prosecution intended to use Mendez as a witness against Tomlin at trial. He had listened to Mendez testify in detail at the preliminary hearing about Tillman’s involvement in the incident, and had cross-examined her at length regarding her identifications of his client. Counsel believed that Mendez was testifying truthfully, but that she was mistaken as to the identification.

Counsel also knew that the prosecution could, if necessary, offer a deal to Tillman in exchange for his testimony against Tomlin. Mendez had identified Tillman as the intermediary, and Tillman had identified Tomlin as the assailant to the police. Counsel was further concerned that because Tomlin was to be tried before Tillman, the prosecution would be in a position to offer Tillman a deal at any point.

In addition, Tomlin had an alibi. His wife and mother-in-law would testify that he was at home that night, with a neighbor recalling that Tomlin’s car remained parked at home at the time of the murder. Accordingly, counsel developed a defense based on mistaken identification and alibi, and pursued that strategy at trial.

When the issue of the illegal line-up arose, counsel would have been faced with a decision. Because the live line-up was conducted *1244 outside of his presence, Mendez’s in-eourt identification of Tomlin would not be admissible unless the prosecution could show, by clear and convincing evidence, that the identification was based on a source independent of the line-up. Counsel could move to suppress Mendez’s identification, which motion, if successful, increased the likelihood that the prosecution would strike a deal for Tillman’s more damaging testimony. What’s more, counsel believed that the Mendez’s in-court identification was based on an independent source, and that the motion would not be successful. Under the circumstances, trial counsel’s judgment could very reasonably have been not to try to suppress Mendez’s in-court identification, at the risk of Tillman testifying for the prosecution. After a full evidentiary hearing, the state court found that it was in Tomlin’s best interests to fore-go a challenge to Mendez’s in-court identification. 1

The majority has taken a somewhat different view of the facts and, based on that, has determined that counsel’s performance was not objectively reasonable. I cannot concur.

II.

The sixth amendment guarantees the right to effective assistance of counsel. Strickland v. Washington, 466 U.S. 668, 686 , 104 S.Ct. 2052, 2063-64 , 80 L.Ed.2d 674 (1984). In order to prevail on the ineffective assistance of counsel claim, Tomlin must show that: (1) the assistance provided by his counsel fell below an objective standard of reasonableness, and (2) but for counsel’s errors a reasonable probability exists that the result of the proceeding would have been different. Id. at 687 , 104 S.Ct. at 2064 . A district court’s dismissal of a claim of ineffective assistance of counsel is reviewed de novo, but the state court’s factual findings are given a presumption of correctness. Sumner v. Mata, 455 U.S. 591, 591-92, 102 S.Ct. 1303, 1303-04 , 71 L.Ed.2d 480 (1982) (per curiam); United States v. Horodner, 993 F.2d 191 ; 194 (9th Cir.1993) 28 U.S.C. 2254(d).

Counsel must be allowed wide latitude with regard to making tactical decisions. Strickland, 466 U.S. at 689-90 , 104 S.Ct. at 2065-66 ; Campbell v. Kincheloe, 829 F.2d 1453, 1464 (9th Cir.1987), cert. denied, 488 U.S. 948 , 109 S.Ct. 380 , 102 L.Ed.2d 369 (1988); United States v. Murray, 751 F.2d 1528, 1535 (9th Cir.), cert. denied, 474 U.S. 979 , 106 S.Ct. 381 , 88 L.Ed.2d 335 (1985) (tactical decisions essentially unreviewable). While we may disagree with counsel’s tactics, such tactics do not fall outside the wide ambit of reasonably representation merely because their wisdom is subject, to second-guessing. See Morris v. California, 966 F.2d 448, 456 (9th Cir.) (defendant failed to overcome presumption that counsel’s failure to object to certain character evidence constituted sound trial strategy), cert. denied, — U.S. -, 113 S.Ct. 96 , 121 L.Ed.2d 57 (1992); Hughes v. Borg, 898 F.2d 695, 703 (9th Cir.1990) (counsel’s conduct was not objectively unreasonable where counsel took into eonsider- *1245 ation all circumstances); United States v. Mayo, 646 F.2d 369, 376 (9th Cir.) (a difference of opinion as to appropriate trial tactics does not rise to the level of denial of effective assistance of counsel), cert. denied sub nom., Dondich v. United States, 454 U.S. 1127 , 102 S.Ct. 979 , 71 L.Ed.2d 115 (1981).

Our precedent teaches that a claim of incompetence of counsel cannot be based upon counsel’s choice not to pursue a constitutional challenge if there is a possibility that, by doing so, evidence harmful to his client may be admitted. In Nelson v. California, 346 F.2d 73 (9th Cir.), cert. denied, 382 U.S. 964 , 86 S.Ct. 452 , 15 L.Ed.2d 367 (1965), the Ninth Circuit held that appellant had received satisfactory legal representation despite trial counsel’s failure to challenge the constitutionality of an incriminating search of appellant’s apartment. The court determined that counsel’s decision to object, on the ground of immateriality, to an inquiry which was relevant to the question of probable cause, did not constitute ineffective assistance of counsel because, if probable cause had been made an issue, additional testimony, possibly harmful to appellant, might have been admitted. Id. at 80.

In Morris, 966 F.2d at 456 , the Ninth Circuit held that counsel’s decision not to call a witness on petitioner’s behalf was within the wide range of professional representation because had the witness been called, the prosecution “could possibly” have linked the witness and the petitioner to prior involvement with drugs. The court also held that petitioner had failed to overcome the presumption of sound trial strategy when counsel decided not to object to the prosecutor’s questions regarding petitioner’s prior drug use. Assuming that petitioner’s testimony was inadmissible, the court nonetheless concluded that an effective advocate could reasonably have decided not to object because petitioner’s admission to prior drug use could have bolstered her credibility with the jury as to her other testimony. However, “[e]ven if that turned out not to be the case, a tactical decision not to object would have been reasonable at the time it was made.” See Williams v. Chrans, 894 F.2d 928, 935 (7th Cir.1990) (failure to challenge reliability of a witness’s identification proper since such a challenge could have called attention to a more detrimental previous identification).

In this ease, the majority pays no deference whatsoever to counsel’s tactical decisions when analyzing whether they were objectively seasonable. First, the majority raises questions about counsel’s assessment of the possible impact of Tillman’s testimony against Tomlin. 2 According to the majority, the fact that Tillman was Tomlin’s accomplice would adversely affect his credibility before a jury. That argument overlooks Tomlin’s defense of mistaken identity and alibi, however. As the state court recognized, no more convincing identification could be made of Tomlin than by his accomplice who personally knew him, who had confessed to functioning as an intermediary during the events leading up to Stewart’s death, who had led Stewart and Mendez to the assailant, and who had identified Tomlin to police. Furthermore, no greater damage could be done to Tomlin’s alibi than to have Tillman detail Tomlin’s participation in the planning and execution of the crime.

*1246 I find it significant that no suggestion has been made that counsel failed to meaningfully test the substance of Mendez’s testimony before deciding to forego a challenge to her in-court identification of Tomlin. Even “strategic choices made after less than complete investigation are reasonable precisely to the extent that reasonable professional judgments support the limitations on investigation.” Strickland, 466 U.S. at 690-91 , 104 S.Ct. at 2066 . Here, trial counsel was involved in the preliminary hearing. He heard Mendez’s detailed account and cross-examined her extensively. His assessment that Mendez was truthful but mistaken in her identification of Tomlin, and his choice of her as a witness, as opposed to Tillman, was therefore an informed one.

The majority next asserts that if Tillman’s testimony were used to secure Tomlin’s conviction, the State, under California law, would need corroborating evidence connecting Tomlin to the offense. That position unjustifiably assumes that the authorities would have gained no additional corroborating evidence as a result of Tillman’s cooperation. What’s more, the majority so much as acknowledges that even absent Mendez’s in-court identification, her description of the assailant to police and her identification of him in the photo line-up themselves provide the corroboration necessary for admissibility of Tillman’s testimony. 3

Finally, the majority points out that during the state evidentiary hearing some seven years later, counsel stated that “in retrospect” he should have objected to the live line-up. The majority further seizes, as did the district court, upon counsel’s admission that he must have been “derelict” for not making a motion on those grounds. But it completely passes over the reason why counsel professed such dereliction: by not moving to suppress the line-up he had failed to preserve the issue for appeal. Evidentiary Hearing Transcript at 259. The question of whether counsel was deficient for failing to preserve the line-up issue for appeal, however, is a new ground for relief not before this panel. And the analysis of that question raises the same substantive line-up issues that we have already declined to consider. Counsel’s singular and retrospective concern was that he failed to preserve an appellate issue. He never once suggested that his decision to not object to the identification, at the time it was made, was not in Tomlin’s best interests. Under those circumstances, I differ with the majority’s conclusion that counsel’s testimony was not consistent with an objectively reasonable strategy.

In sum, counsel’s selection of mistaken identity and alibi as defenses was consistent with his decision not to risk the introduction of Tillman’s testimony if Mendez’s identification were suppressed. I cannot say that such a decision is professionally unsound. See Henry v. Mississippi, 379 U.S. 443 , 85 S.Ct. 564 , 13 L.Ed.2d 408 (1965) (defendant could not successfully maintain an ineffective assistance claim on the ground that counsel had eschewed one of several reasonably and mutually exclusive litigation alternatives in favor of another); Martin v. McCotter, 796 F.2d 813 (5th Cir.1986) (where defense strategy was not one of suggestiveness of photographic line-up, but rather alibi, counsel’s admission that photographic line-up was not suggestive, was not ineffective assistance), cert. denied, 479 U.S. 1057 , 107 S.Ct. 935 , 93 L.Ed.2d 985 (1987).

There is another reason that deference should be given to counsel in this case. In the evidentiary hearing, trial counsel was asked how he would have proceeded in terms of the entire identification issue had the in-person line-up been suppressed, but not the photographic lineup. He stated:

Evidentiary Hearing Transcript at 258. Regarding why he did not make a motion to suppress only the photographic line-up, counsel testified:

Id. at 252-53.

Thus, while counsel, after eight years, might not have remembered exactly why he didn’t object, his best explanation was that it would not have been effective in light of the existence of the previous untainted photographic line-up. 4 Indeed, we need not determine the actual explanation for counsel’s failure to object, so long as his actions fall within the range of reasonable representation. Morris, 966 F.2d at 456 .

In United States v. Smith, 551 F.2d 348 (D.C.Cir.1976), the defense attorneys were apprised that the government had obtained pretrial photographic identifications from the witnesses who were expected to identify defendants in court. They declined to object based on their judgment regarding the constitutionality of the photographic identifications even though the weight and credibility of the evidence might be subject to dispute. The court of appeals held that counsel’s failure to object did not constitute ineffective assistance, remarking:

Id. at 353-54 (footnotes omitted).

In this ease, there is nothing in the record that may be used to suggest that the photographic line-up given to Mendez was tainted in any way. I do not see how we can say, in view of the photographic identification, that counsel’s judgment was faulty for anticipating that the trial court would find in it an independent basis for the in-court identification; nor how, assuming that such a question is at best a “close case,” Tomlin has overcome the strong presumption in favor of counsel’s effective performance.

III.

To show prejudice, Tomlin must demonstrate that, but for counsel’s errors, a reasonable probability exists that the result of the proceeding would have been different. Strickland, 466 U.S. at 687 , 104 S.Ct. at 2064 . Here, that determination rests upon whether the in-court identification was tainted by the illegal line-up. If clear and convincing evidence shows that the courtroom identification rests on an independent basis, then the courtroom identification would be admissible *1248 and the results of the proceedings no different. Again, it is Tomlin's burden to show that, had a Wade hearing been conducted, the court would have excluded Mendez’s testimony. There is no license here to make assumptions in favor of a finding of prejudice.

As the majority admits, Mendez identified Tomlin in a legally valid photographic line-up conducted before the illegal in-person lineup. That, coupled with Mendez’s opportunity to view the assailant, should be sufficient for finding an independent basis for Mendez’s in-court identification. See Simmons v. United States, 390 U.S. 377, 384 , 88 S.Ct. 967, 971 , 19 L.Ed.2d 1247 (1968) (convictions in which an eye-witness identification follows a previously held pretrial photographic identification will not be set aside unless the “photographic identification procedure was so impermissibly suggestive as to give rise to a very substantial likelihood of irreparable misidentification.”); United States ex rel. Burke v. Illinois, 465 F.2d 268, 270 (7th Cir.1972) (in independent basis determination, great weight placed on non-suggestive photographic line-up).

The object of a Wade hearing is to determine whether the in-court identification had an independent source. United States v. Wade, 388 U.S. 218, 242 , 87 S.Ct. 1926, 1940 , 18 L.Ed.2d 1149 (1967). The photo line-up singularly provides the “means sufficiently distinguishable to be purged of the primary taint [of the illegal lineup].” Id. at 241 , 87 S.Ct. at 1939 . The photographic line-up occurred before the illegal line-up. It is completely distinguishable from the description that Mendez offered to police, and remains rehable. It seems somewhat inconsistent that the photographic line-up could not be the basis for setting aside Tomlin’s conviction under Simmons , while, at the same time, it could fail to form an independent basis for the in-court identification under Wade .

The district court below found that Tomlin had not been prejudiced. I believe that, in its disagreement with that conclusion, the majority draws some unwarranted conclusions, without which, affirmance on the prejudice aspect would be appropriate.

The majority points to the district court’s mistaken conclusion that Mendez identified Tomlin from the yearbook as evidence that the court may not have taken into account the government’s burden of proof in seeking to show that the in-court identification was constitutional. In fact, since Mendez’s yearbook identification of Tomlin never occurred, and therefore would not have been considered in a Wade hearing, I am at a loss to see how its “shadow” could extend to the in-court identification. In the end, our focus must be on what could have made a difference at a Wade hearing, not what would not have made a difference. 5

Similarly, the majority hypothesizes that a Wade hearing might have elicited something suggestive about the photographic line-up. Such speculation is contrary to Strickland’s requirement that Tomlin show actual prejudice, contrary to the majority’s acceptance of the state court’s credibility determination, and contrary to the record as we have it.

In determining whether an in-court identification has an independent source, a court should consider (1) the opportunity of the witness to view the suspect at the time of the crime, (2) the existence of any discrepancy between a pre-line-up description and the *1249 defendant’s actual appearance, (3) any identification prior to the line-up of another person, (4) the identification by picture of defendant prior to the line-up, (5) the failure to identify the defendant on a prior occasion, and (6) the lapse of time between the crime and the confrontation. Wade, 388 U.S. at 242 , 87 S.Ct. at 1940 ; see also Neil v. Biggers, 409 U.S. 188, 199-200 , 93 S.Ct. 375, 382-83 , 34 L.Ed.2d 401 (1972) (due process analysis including as a factor the level of certainty demonstrated by the witness).

The majority concedes that the reliability of the photographic identification remains unequivocal, that only a short time passed between the crime and the first identification of Tomlin, and that Mendez never identified another individual as the assailant. 6 As for the accuracy of the description, the majority correctly discounts the difference between Mendez’s description and Tomlin’s actual height and build. See United States v. Smith, 473 F.2d 1148, 1149 (D.C.Cir.1972) (description and height difference of 4-5 inches understandable where victim viewed defendant while he was sitting in the car). What is left is an appraisal of Mendez’s opportunity to view the assailant, and the alleged discrepancy between Tomlin’s hair style and Mendez’s description to police.

I am very troubled by the majority’s treatment of Mendez’s opportunity to view the assailant. It again draws unwarranted conclusions and overlooks uncontroverted evidence in the record that Mendez looked directly at the assailant’s face at a distance of one foot. The majority points out that Mendez had an opportunity to observe the assailant for four to five minutes and that Mendez testified at trial that she had a good chance to look at the assailant’s face. Despite that, the majority second-guesses the extent to which Mendez may have seen what she was looking at. It remarks that the assailant sat “next” to Mendez in the pickup truck, brandished a gun and fired it once, and that Mendez would have to turn her head to see him. Under such circumstances, the majority states, “it’s doubtful Mendez would then have spent much time turned towards him, calmly noting his facial features and other aspects of his appearance.” For all its supposition, that statement cannot obscure the law and facts that should be guiding our analysis.

Mendez had an extraordinary opportunity to view her assailant, and took advantage of it. The opinion neglects to point out that Mendez first saw the assailant outside the passenger side door next to her even before she knew he had a gun and before a shot was fired. Trial Transcript at 185. Nor does the opinion mention that Mendez and the assailant conversed and interacted during the encounter. The assailant, who was sitting next to Mendez, asked her if she had money; Mendez replied that she thought she had five dollars; she found it in her purse and handed it to him. Id. at 188. And the opinion does not explain that most of the encounter did not take place in a “dark alley,” but while driving out on the streets. 7

As if her trial testimony weren’t enough, however, Mendez’s preliminary hearing testimony should settle the question of whether she got a good look at the assailant’s face during the crime. That testimony reads:

Preliminary Hearing Transcript at 47-48.

Nothing, absolutely nothing, in the record disputes Mendez’s close-up and unobstructed view of the assailant’s face. The fact that Tomlin raises nothing in the record to point out exactly how long Mendez may have looked directly at the assailant severely handicaps his attempt to meet his burden of showing prejudice. I cannot accept the majority’s suggestion that there is something insignificant or unreliable about Mendez’s observation of the assailant because she may not have “spent much time turned towards [the assailant], calmly noting his facial features and other aspects of his appearance.” The time factor in any case is a relative matter. Indeed, any close-up look at a face could leave an indelible impression in the mind of a victim of a violent crime. The impression upon Mendez could arguably have been all the deeper in view of the fact that she witnessed the assailant fire a gun across her, shooting dead the man she loved. Preliminary Hearing Transcript at 12. Laying supposition aside, though: there is simply nothing here that supports the majority’s degradation, and almost literal dismissal, of so vital an observation.

In United States v. Monks, 774 F.2d 945 (9th Cir.1985), the Ninth Circuit found sufficient indicia of reliability in the positive identifications by two tellers of a bank robber, even though the photographic line-ups were unduly suggestive. The robber was only two feet away from one teller, who looked at his face for about three to four minutes. That is less than the amount of time that Mendez had to view the assailant in this case. Because that teller interacted with the robber, as Mendez did with the assailant in this case, the court held that she was necessarily paying full attention to him. Id. at 956 . The other teller observed the robber from a slightly greater distance (“between a customer and a teller”) for about two minutes. The court held that her focus would necessarily be on the robber once she realized a robbery was taking place, just as Mendez’s focus would be on the assailant once the episode involving her began. Id. In this ease, however, the majority finds just the opposite— that because of the stress of the situation, Mendez was not attentive in her observation of the assailant. That presumption is wholly unjustified, and reverses the burden on Tomlin to show actual prejudice.

The hair style issue. This, in my view, is perhaps the best-directed point of the majority’s prejudice analysis. Yet it, too, falls short. When Mendez described the assailant to the police soon after the attack, she said that he had an inch and a half to two inch afro. Tomlin, however, is said to have had a straightened, permanent, shoulder-length hair style at the time of the attack. The majority remarks that this discrepancy cannot be discounted. I do not agree.

During cross-examination at trial, Mendez testified that when she described the assailant’s hair as an short afro, she was indicating how far it came out from the assailant’s head, not whether his hair was straight or curly. 8 Trial Transcript at 210. Mendez’s preliminary hearing testimony was consistent:

Preliminary Hearing Transcript at 48-49.

In addition, at the state evidentiary hearing, the investigating and arresting officer, *1251 Detective Vincent, testified that at the time of Tomlin’s arrest several days after the crime, he had his hair combed in such a way that it could look like a short natural in the dark. The majority admits that it is possible to comb shoulder-length straightened hair so that it looks like a short afro, but contends that the hair styles are vastly different. However, the record contains no evidence that Mendez’s description was anything other than what she meant it to be, or that Tomlin’s hair style was different than that at the time of the crime. Nor were we present to look for ourselves at the assailant’s hair, or judge Mendez’s and Vincent’s credibility when they testified. Considering the record, the alleged hair style discrepancy very arguably is not a discrepancy at all.

Even assuming that the hair style discrepancy is material, it is not substantial enough to change the finding of an independent basis. The majority correctly cites the dangers inherent in eyewitness identification and the suggestiveness of pretrial identification, and points out that Mendez’s testimony was crucial to the prosecution’s case as it was presented. The testimony of a single uncorroborated witness, however, can be sufficient to support a conviction. See United States v. Smith, 563 F.2d 1361, 1363 (9th Cir.1977), cert. denied, 434 U.S. 1021 , 98 S.Ct. 747 , 54 L.Ed.2d 769 (1978). And, as the majority recognizes, a physical description need not be perfect to be rehable. In this case, where all of the other factors favor the reliability of Mendez’s identification, the effect of the possible discrepancy in hair style must be considered inconsequential. See United States v. Dring, 930 F.2d 687, 693 (9th Cir.1991) (government agent’s identification found reliable despite fact that he failed to identify suspect’s beard), cert. denied, — U.S. -, 113 S.Ct. 110 , 121 L.Ed.2d 68 (1992); see also United States v. Cook, 464 F.2d 251, 253 (8th Cir.) (finding independent bases for in-court identification despite witnesses’ descriptions of differences in hair color and length), cert. denied, 409 U.S. 1011 , 93 S.Ct. 457 , 34 L.Ed.2d 305 (1972).

Finally, the certainty of Mendez’s identifications of Tomlin must also be considered. Neil v. Biggers, 409 U.S. at 199-200 , 93 S.Ct. at 382-83 ; United States v. Gregory, 891 F.2d 732, 735 (9th Cir.1989). A high degree of certainty about an identification is one of the most important considerations in the reliability determination. Monks, 774 F.2d at 956-57 (certainty at photo line-up and at trial). It is indisputable from the record before us that Mendez had no doubt regarding her identification of Tomlin. Without reservation, she picked him out of the non-suggestive photographic line-up, and consistently reaffirmed the accuracy of that identification at the preliminary hearing and at trial.

Mendez had an undeniably excellent opportunity to view the assailant. She identified no other person besides Tomlin. She made a reliable identification of Tomlin in a photographic line-up. She never failed to identify Tomlin on any prior occasion. The lapse of time between the crime and the confrontation was short. There are no major discrepancies in Mendez’s description and Tomlin’s actual appearance. Even if the hair style description were inaccurate, it would be inconsequential in view of the other considerations. And Mendez could not have been any surer about her identifications. I have real difficulty holding that, under these circumstances, Mendez’s in-eourt identification of Tomlin lacks an independent basis.

IV.

Counsel was not ineffective for failing to raise the Wade issue. Nor has Tomlin demonstrated prejudice. Denial of the writ should be affirmed.

I fear that the majority decision will be a regrettable obstacle to jurists who must weigh legal burdens in claims of ineffective assistance of counsel, who must respect lower court findings, or who must decide whether an eyewitness identification should be undone by the perhaps inexact description of a single physical characteristic such as the way a suspect combs his hair.

. The state court reasoned:

Opinion and Order, No. 2926 at 8-10 (October 23, 1986).

. The majority points out that, during the eviden-tiary hearing, Tillman denied having identified Tomlin. Officer Vincent testified, to the contrary, that Tillman had identified Tomlin by his street name, Treetop. I am somewhat surprised that the majority would raise the dispute between the testimony of Tillman and Officer Vincent in light of the state court's express adoption of Vincent’s version. See Opinion and Order No. 2926 at 3 (October 23, 1986). That finding must be given deference if fairly supported. 28 U.S.C. 2254(d). One need only look at the substance of Tillman’s evidentiary hearing testimony to understand why the state court rejected it. At the beginning of that testimony, Tillman asserted the fifth amendment privilege, to no avail. Eviden-tiaiy Hearing Transcript at 261. What followed was a failure of memory and denial by Tillman regarding virtually all of the events related to the crime for which he was convicted. Tillman did not recollect telling Vincent that he set up the drug deal between Stewart and "Treetop.” He did not recollect writing "Treetop” on the paper. He did not remember telling the police that he had been involved in arranging a drug deal at all. He denied meeting Mendez and Stewart and directing them to the alley where the drug deal was to happen. He denied that he arranged for another person to meet them. He denied having had any contact with Stewart or Mendez. He denied having known Tomlin or having heard of him at all. Id. at 278-79 .

. The majority challenges the corroboration by pointing out (1) the alleged differences in Mendez’s description and Tomlin's appearance, and (2) Mendez’s evidentiary hearing testimony in which she testified that the photo line-up was suggestive. The description/appearance discrepancies, to the extent any exist, go to weight, and not admissibility, of the corroborating evidence. As to Mendez's evidentiary hearing testimony, the majority accepted the state court's credibility determination rejecting it. In my view, we should stay true to that rejection, and forthrightly disregard Mendez's recantations.

. Counsel had substantial basis for determining before the trial that the photographic line-up was untainted. During the preliminary hearing, at which counsel was present, Mendez testified that the photographic line-up was not suggestive. Preliminary Hearing Transcript at 4-5. Counsel cross-examined her on the conduct of the photographic line-up and the makeup of the photographic array. Id. at 48-49. Counsel also examined Detective Vincent, who had presented the photographic line-up to Mendez. Id. at 82-84. Counsel himself moved to have the photographs marked for identification, and used them in questioning Vincent. Id. at 83.

. The majority brings up another point that, as it ends up, doesn't matter much. It raises the issue of the introduction of testimony regarding the illegal line-up, then drops it because it is not one of Tomlin's claims. But not before an obiter dictum proclamation of a prosecutor’s obligation to confine his response when the illegal line-up testimony is elicited in the first instance by the defense. In this case, the majority announces that the prosecutor went too far by asking questions about Mendez's line-up identification of Tomlin and the conduct of the line-up, and by arguing to the jury that the line-up was fair. However, after the line-up issue was raised, even in a minor way, the prosecutor was compelled to capitalize upon it. The jury could have perceived a line-up identification as a crucial piece of evidence. Yet the prosecutor had not raised the issue on direct examination, and could not explain to the jury the reason that he hadn't. Instead, at that point, the prosecutor had to completely dispel any implication that he had not been forthcoming by making the live line-up identification part of his case. Once pretrial evidence is placed before the jury by the defense on cross-examination, the per se exclusionary rule is no longer applicable. United States v. Patton, 721 F.2d 159 , 162 n. 4 (6th Cir.1983).

. The majority asserts that the prosecutor had the duty to conduct additional live line-ups in order to mitigate the effect of the illegal line-up. Perhaps if the illegal line-up had been the sole pretrial identification, that argument would cany weight. I have never seen it suggested until now, however, that a prosecutor is so encumbered when a reliable photographic identification also exists. See United States v. Johnson, 820 F.2d 1065, 1072 (9th Cir.1987) (due process is preserved so long as some pretrial identification procedure is not impermissibly suggestive).

. For that matter, there is no suggestion in the record that it was too dark for Mendez to see the assailant’s features.

. In Monks, 774 F.2d at 956 n. 10 (9th Cir.1985), we accepted a witness’s explanation of what she meant by the term “pock marks" in her description of the defendant.