ILANA DIAMOND ROVNER, Circuit Judge.

Following a 1992 bench trial, an Illinois judge convicted Clifton McFowler of first-degree murder and ordered him to serve a prison term of forty years. The Illinois Appellate Court affirmed McFowler's conviction, judging the testimony of Charlene Meredith, who placed McFowler at the scene of crime with a shotgun in his hand, sufficient to convict him beyond a reasonable doubt notwithstanding conflicts in the evidence as to whom Meredith had picked out of a lineup shortly after the murder. The district court (Hon. Joan B. Gottschall) granted McFowler's petition for a writ of habeas corpus, concluding that the Illinois Appellate Court had unreasonably applied Jackson v. Virginia, 443 U.S. 307, 99 S.Ct. 2781, 61 L.Ed.2d 560 (1979), when it sustained McFowler's conviction based on Meredith's identification testimony. United States ex rel. McFowler v. Pierson, 2003 WL 76861 (N.D.Ill. Jan.8, 2003). In view of the irreconcilable inconsistencies in Meredith's own testimony, we share the district court's doubts about the reliability of her identification. "The vagaries of eyewitness identification are well known; the annals of criminal law are rife with instances of mistaken identification." United States v. Wade, 388 U.S. 218, 228, 87 S.Ct. 1926, 1933, 18 L.Ed.2d 1149 (1967). However, the Antiterrorism and Effective Death Penalty Act of 1996 commands great deference to the decisions of state courts. See 28 U.S.C. § 2254(d); e.g., Bell v. Cone, 535 U.S. 685, 693, 122 S.Ct. 1843, 1849, 152 L.Ed.2d 914 (2002). Because we cannot say that the Illinois Appellate Court was objectively unreasonable in holding that Meredith's testimony provided sufficient support for McFowler's conviction, we are compelled to reverse the grant of McFowler's petition for a writ of habeas corpus.

I.

Sammy Logan was shot on October 9, 1989, when he answered a knock at the front door of his home. He lapsed into a coma and died three days later. An autopsy would reveal that he had been shot in the forehead and that the bullet had penetrated his brain.

Although the Illinois Appellate Court was not explicit on this point, it is undisputed that McFowler's murder conviction is premised not on the notion that McFowler shot Logan himself, but rather on the theory that McFowler is legally accountable for the acts of the person who did. The trial judge neglected to make this clear when he convicted McFowler, but his remarks at sentencing eliminate any doubt on this score. Indeed, the State defends McFowler's conviction solely on an accountability theory. Illinois law provides that "[a] person is legally accountable for the conduct of another when . . . [e]ither before or during the commission of an offense, and with the intent to promote or facilitate such commission, he solicits, aids, abets, agrees or attempts to aid, such other person in the planning or commission of the offense." 720 Ill. Comp. Stat. Ann. 5/5-2(c) (West 2003). As we shall see, there is ample circumstantial evidence from which one reasonably might conclude both that McFowler was present at the scene of the murder and that he fled from the scene in the immediate aftermath. However, mere presence at the scene of a crime — even with knowledge that a crime is being committed — and flight from the scene do not suffice to establish accountability for the acts of a principal under Illinois law. E.g., People v. Williams, 193 Ill.2d 306, 250 Ill.Dec. 692, 739 N.E.2d 455, 472 (2000), cert. denied, 533 U.S. 953, 121 S.Ct. 2599, 150 L.Ed.2d 756 (2001). The State must also prove that the defendant intended to facilitate or promote the principal's criminal activity. E.g., People v. Stanciel, 153 Ill.2d 218, 180 Ill.Dec. 124, 606 N.E.2d 1201, 1210 (1992). The critical evidence pointing to McFowler's complicity in the murder came from Meredith.

In October of 1989, Meredith resided in a second-floor apartment at 3144 West Lexington Street in Chicago along with Logan (her boyfriend), Glenda Roberts (Meredith's sister), Percell Swinney (Roberts' boyfriend), and seven children. At approximately 11:45 a.m. on October 9, there was a knock at the front door of the apartment. At that time, Logan, Meredith, and Meredith's children were congregated in a bedroom opposite the kitchen. Against Meredith's advice, Logan left the bedroom to answer the knock, closing the bedroom door behind him. Through the closed door, Meredith subsequently heard a shot. Meredith's two-year-old son Davion opened the bedroom door, and at a distance of approximately sixteen feet, Meredith would later testify, she saw a somewhat short African-American man with a shotgun in his hand standing in the kitchen. He was wearing blue jeans and a white t-shirt. Next to him, lying on the floor of the kitchen, was Logan. At first, the intruder's face was turned away from Meredith toward Swinney, who was standing in a doorway between the kitchen and the living room. But Meredith got a clear look at the intruder's face when he subsequently tripped over Logan's body. Meredith also heard a voice from somewhere else in the apartment say "if you are going to shoot him[,] shoot him." Tr. G22. Meredith grabbed Davion by the shirt and pulled him into the bedroom closet along with her other children. They remained in the closet for approximately fifteen to twenty minutes, until Meredith's sister came to find her. Meredith later saw that Swinney had been shot in the arm.

Police were summoned from a nearby police station by a neighbor who ran into the police station yelling that two men had just been shot. They arrived at the scene within minutes of the shooting. Officer J. Delpilar found Swinney sitting on the front steps of the building with a gunshot wound to his right arm. When he entered the second-floor apartment, Delpilar saw Logan lying on the floor in a pool of blood, unconscious but still breathing. Delpilar spoke with Swinney and the other witnesses and subsequently broadcast a description of three suspects: (1) an African-American male in his late teens or early twenties, approximately six feet tall, 180 pounds, with brown eyes and black hair, wearing a red and white jogging suit and gym shoes; (2) a second African-American male in his twenties, wearing an orange shirt and red jogging pants; and (3) a female African American aged eighteen to twenty, wearing white pants and tall black riding boots.

On further inspection of the scene, Delpilar found a set of automobile keys near the top of the stairs leading to the second-floor apartment where Meredith and Logan lived. Those keys, he discovered, fit the driver's side door and ignition of a gray Ford Escort parked in front of (and across the street from) the property. Officer Edward Kulbida learned from bystanders that the suspects had attempted to use that car as their getaway vehicle. (The keys found in the stairwell explain why the suspects had left the Escort behind and fled on foot.) Because Delpilar failed to handle the keys in such a way as to preserve any fingerprints on them, they were not tested for prints. However, prints lifted from the exterior of both the driver and passenger sides of the Escort matched McFowler and Byndum, respectively. It could not be said how long those prints had been on the car.

Touring around the area to the north of the crime scene, Kulbida and other officers soon spotted a woman — Shanie Flowers — matching the description of the female suspect. Flowers was making a call from a pay phone at 3258 West Flournoy, just two blocks away from the scene of the shooting. Flowers volunteered to the officers that the men they were looking for could be found in the second-floor apartment at that same address. Kulbida, along with Detective Wayne Gulliford and Officer Robert Gomez, entered the Flournoy apartment approximately thirty minutes after the shooting was reported. Inside, they found McFowler and Trenton Byndum, among other individuals. McFowler was lying in bed, wearing only blue jeans. Byndum was in the living room, wearing striped blue jeans, a sleeveless blue top, white socks, and no shoes. Gulliford spotted blood on Byndum's right sock. On the floor next to Byndum were a beige trench coat, a pair of Nike Air Flight athletic shoes, and a Payless shopping bag containing other items of clothing. Among the clothes found in the bag were a pair of red and white jogging pants, a shirt, and a jogging jacket, all stained with what appeared to be blood. Byndum and McFowler were placed under arrest.

During a subsequent search of the apartment, Detective Gregory Baiocchi discovered a black and white Converse athletic shoe for a right foot in the apartment's oven. The shoe had what appeared to be blood spots on it. Baiocchi recovered what appeared to be that shoe's mate in the Escort parked in front of the crime scene. Both shoes were submitted to a laboratory for testing, which confirmed that human blood was present on both shoes. The blood found on the left shoe was Type B; the right shoe contained an insufficient quantity of blood for typing. At trial, McFowler was asked to try the shoes on, and the shoes fit him. They also fit his lawyer. Logan's blood type apparently was Type B. (Type B blood was recovered from the crime scene.)

Also discovered at the Flournoy apartment (secreted behind bricks in the rear stairwell of the basement) were a Ruger Security Six revolver with two live .38-caliber rounds and four spent casings of the same caliber, as well as a Westernfield .20-gauge shotgun with two live rounds. The record before us does not include any ballistics evidence tying either of these weapons to the shooting. It is undisputed, however, that Logan was shot with a revolver rather than a shotgun.

The hands of both McFowler and Byndum were tested for gunshot residue. The results for Byndum were positive and thus consistent with Byndum having recently handled or been in close proximity with a discharged firearm. The results for McFowler reflected high levels of each of the heavy metals associated with firearm use — lead, barium, and antimony — but because the control swab used in the test reflected an elevated level of barium (suggesting contamination), the trace evidence analyst could not confirm that McFowler, like Byndum, had been in close proximity to a recently discharged firearm.

A bloody footprint was found on the first-floor landing at the scene of the shooting. The parties stipulated that an expert in shoe print comparison, if called to testify, would opine that a photograph of the footprint was of the same type, size, and style as the pair of Nike Air Flight athletic shoes observed next to Byndum in the Flournoy apartment, but the expert could not say with a reasonable degree of scientific certainty that those shoes had, in fact, made the footprint. Although the shoes bore what appeared to be blood stains, they apparently were never tested to confirm the presence of blood or the blood type.

On the day of his arrest, McFowler made a number of statements to Detective Baiocchi, which the detective later recounted at trial. Baiocchi testified that McFowler told him he had gone to the Flournoy apartment the previous evening with a friend and that he had remained there until the police arrived and arrested him the following day. He also indicated that Byndum was his cousin. He had telephoned Byndum earlier that day to let him know where he was, and Byndum had subsequently joined him at the Flournoy apartment. Baiocchi testified that he did not see a telephone in the Flournoy apartment.

At approximately 5:30 p.m. on the day of the shooting, Meredith viewed a six-man lineup at the police station. That lineup included both McFowler and Byndum. At trial, on direct examination, Meredith testified that she told police at the lineup "[t]hat the second person from the end looked like the one who was in my house." Tr. G25. The prosecutor then showed Meredith a photograph of the lineup and asked her to circle the individual that she had identified as the man she saw in her apartment with the shotgun. Meredith circled McFowler, although as it turned out, he was not the second person from either end of the lineup as displayed in the photograph. She also proceeded to identify McFowler in court as the man she had seen. Tr. G28-29. Additionally, when shown the shotgun that had been found at the Flournoy address, Meredith indicated "[t]hat's the shotgun that I saw" in McFowler's hand. Tr. G29-30. On cross-examination, she confirmed that it "looks just like the gun that I saw." Tr. G33.

Detective P. Foley conducted the lineup that Meredith viewed. The parties stipulated that if called to the witness stand, Detective Foley would testify that Meredith had identified Byndum, not McFowler, as the man she had seen with the shotgun. Foley also would have testified that when he spoke with Meredith after the shooting, she indicated to him that she previously had seen the man with the shotgun in her neighborhood.

On cross-examination, defense counsel asked Meredith whether, at the lineup, she had actually identified a man wearing striped pants and a sleeveless blue t-shirt (Byndum). In response, Meredith indicated that she had "picked out both." Tr. G35. Meredith acknowledged, however, that in the photograph of the lineup, neither McFowler nor Byndum was the second person from either end of the lineup. Meredith recalled having spoken with Detective Foley after the shooting, but she did not recall telling him that she had seen the man with the shotgun before around the neighborhood. Indeed, she testified that she had never seen McFowler before the date of the incident.

On redirect, the State brought Byndum into the courtroom. Byndum had pleaded guilty to the murder of Logan prior to trial and was ultimately sentenced to a thirty-year prison term. The State had called Byndum as its first witness against McFowler; however, Byndum had invoked his Fifth Amendment right not to incriminate himself and had refused to answer any of the questions put to him about the shooting. (The trial court had found him guilty of criminal contempt of court and ordered him to serve a six-month prison term consecutive to the thirty-year term for murder.) With both Byndum and McFowler side-by-side in the courtroom, the State asked Meredith which of the two men she had seen with a shotgun on the day of the shooting. Meredith indicated that she had seen McFowler. Tr. G38. She also acknowledged, contrary to her earlier testimony, that the person she had seen in her apartment was not the second person from either end of the lineup as depicted in the photograph she was shown. Tr. G39-40.

On re-cross examination, McFowler's attorney asked Meredith how long a glimpse she had gotten of the person in her apartment. Meredith answered, "Maybe just a few minutes." Tr. G40.

After hearing the evidence, the trial court (Hon. Thomas A. Hett) convicted McFowler of first degree murder. In the course of his ruling, Judge Hett made the following remarks about Meredith's identification testimony:

Now, if I am not mistaken Miss Meredith testified that, here in court she testified Mr. McFowler was the one that she saw in the apartment with the gun . . . the shotgun. She also testified that in the — that she had a photo array, and she said that this defendant's picture was the, like the picture of the person who was in the apartment and that she also identified the defendant in a lineup.

That portion of [her testimony], the latter part is impeached by Foley who[se stipulated testimony] said that she only identifies Trenton Byndum.

Tr. K52-53. But the judge went on to credit her in-court identification of McFowler (Tr. K56), and based on that identification along with the other circumstantial evidence placing McFowler at the scene of the shooting, concluded that he was guilty:

The defendant is identified by Miss Meredith. He has his prints on the car outside of the house. There are shoes that fit him partly in the car and partly hidden. He is found with a person whose clothing and shoes apparently are covered with blood.

Both he and the person who he is with are identified as having been in that apartment when the shooting took place. Sure looks to me like proof beyond a reasonable doubt, if not [of] the actual murder then being the accomplice.

Miss Meredith testified that [there were] two sets of shootings in the apartment and we do note that the victim was killed and [Percell Swinney] was shot so I believe that the State has proven its case beyond a reasonable doubt the crime of murder and the defendant is found guilty thereof.

Tr. K56-57. McFowler subsequently filed a motion for a new trial, which the court denied. At sentencing, the State asked the judge to impose the death penalty, but the judge demurred, concluding that McFowler was ineligible in the absence of evidence that he had fired the gun used to kill Logan. Tr. L9. Judge Hett took the opportunity to reiterate, however, that the evidence was sufficient to hold McFowler criminally culpable for the murder under an accountability theory:

There is a wealth of circumstantial evidence. There's an identification process. So, I'm convinced beyond a reasonable doubt that Mr. McFowler was present with Mr. Byn[d]um in that house, and that as a result of that presence, that man was killed. And there was an obvious intention to kill. It wasn't something that was a happenstance.

I have determined also from that evidence that it is apparent, unless they switched guns, and it wasn't apparent that that's the case, that Mr. McFowler did not pull the trigger. Nevertheless, he is accountable for what Mr. Byn[d]um did.

Tr. L11-12. The judge went on to sentence McFowler to a prison term of forty years.

The Illinois Appellate Court affirmed McFowler's conviction, concluding in relevant part that a rational trier of fact could find him guilty beyond a reasonable doubt. People v. McFowler, No. 1-92-2176, Order at 31-38, 276 Ill.App.3d 1123, 231 Ill.Dec.735, 697 N.E.2d 16 (Ill.App.Ct. Sept. 26, 1995) (unpublished) ("App.Ct.Order"). The court found Meredith's identification of McFowler was sufficiently reliable to place McFowler at the scene of the murder with a shotgun in his hand. Meredith testified that she had identified McFowler in a lineup just six hours after the shooting, and her testimony established that at the time of the shooting, she had seen him at a distance of sixteen feet, although the court "greet[ed] with some skepticism her further testimony that she was able to maintain her view of him for perhaps as long as a few minutes." App.Ct. Order at 33.

The court acknowledged that Meredith's testimony reflected some confusion as to where McFowler had been positioned in the lineup and that it conflicted with the stipulation that Detective Foley would have testified that she had identified Byndum and not McFowler. Id. "However, any such confusion or apparently conflicting evidence would not necessarily mandate that the circuit court totally disregard her lineup identification." Id. Meredith, the court noted, ultimately testified that she had identified both Byndum and McFowler in the lineup. Moreover, the circuit court itself had acknowledged that Detective Foley's recollection about the lineup tended to impeach Meredith's testimony on that point.

It was not clear to the appellate court whether the trial court, in view of the conflict, had given Meredith's testimony about the lineup diminished weight or had disregarded it altogether; but the latter possibility would not leave McFowler's conviction without adequate evidentiary support:

Even if Meredith's lineup identification were given no weight whatsoever, there still would have been credible eyewitness testimony in this case. Meredith made an unwavering in-court identification of the defendant. The strength of this identification was only reinforced when Meredith, during redirect examination, viewed both the defendant and Trenton Bynd[u]m together in court and reaffirmed that the defendant was in fact the person she saw trip over the victim on October 9, 1989.

Id. at 34. The court also noted that there was no evidence Meredith had ever indicated any uncertainty in identifying McFowler and that her in-court description of the intruder as a short, African-American male was not challenged and was consistent with McFowler's appearance. Id. at 36. (The presentence report for McFowler indicated that he was African American and five feet, six inches in height. Id. n. 2.)

The court also emphasized that in addition to Meredith's identification testimony, the record reflected "substantial corroborative testimonial and physical evidence" that linked McFowler to the crime. Id. at 36. Meredith had identified the shotgun that had been recovered (along with the revolver) from the Flournoy building as the one she had seen in McFowler's hand on the day of the shooting. The keys to a Ford Escort parked directly across the street from the scene of the crime had been found in the Lexington building; witnesses had told the police that the assailants had attempted to flee the scene in that vehicle; and McFowler's prints (along with Byndum's) were found on the exterior of that car. The Converse shoe found in that car appeared to be the mate of the one found in the oven of the Flournoy apartment. Both of those shoes were stained with blood, and testing revealed that the blood on the left shoe was type B, which apparently was the same blood type as the victim's. The shoes also fit McFowler. The pair of Nike shoes discovered in the Flournoy apartment also bore what appeared to be blood stains, and the pattern of those shoes was consistent with a bloody footprint found at the crime scene.

Thus the single Converse shoe found inside the oven, which apparently matched the Converse shoe found in the Ford Escort, and the Nike shoes found inside the apartment, corroborated the inference that Bynd[u]m and the defendant, who were both barefoot at the time of their arrest just thirty minutes after the shooting, were the individuals who wore those shoes at the crime scene.

Id. at 38. The "cumulative impact" of this evidence was, in the court's view, sufficient to support McFowler's conviction notwithstanding the weaknesses in Meredith's identification testimony. Id. The Illinois Supreme Court denied McFowler's petition for leave to appeal. People v. McFowler, No. 80488, 166 Ill.2d 549, 216 Ill.Dec. 8, 664 N.E.2d 645 (Ill. April 3, 1996) (unpublished).

McFowler subsequently filed a petition for a writ of habeas corpus pursuant to 28 U.S.C. § 2254. Upon review of the parties' briefs and the record, the district court granted the petition. McFowler, 2003 WL 76861. The court proceeded from the premise that McFowler could be liable for Logan's murder under an accountability theory if, in fact, he was the man with a shotgun that Meredith had seen next to (and tripping over) Logan's body. Id. at *6. (The court separately considered whether McFowler could have been convicted as a principal, but concluded the facts would at most support McFowler's conviction under an accountability theory. Id. at *11.) But in the court's view, the evidence was insufficient to establish that McFowler was the intruder that Meredith had seen.

First, "[n]o rational trier of fact could ascribe any weight to Meredith's testimony concerning her identification of McFowler from the lineup on the day of the crime." Id. at *7. The Illinois Appellate Court seemed to think that the trier of fact could have given the lineup at least some weight, because when confronted with Foley's stipulated testimony that she had identified Byndum at the lineup, Meredith said that she had identified both Byndum and McFowler. But Meredith testified that she had only seen one assailant, and Byndum and McFowler did not look alike. So it was "nonsensical" to suppose that Meredith could have identified both of them at the lineup. Id.

On the other hand, the lineup could not simply be disregarded in assessing the balance of Meredith's identification testimony, as the Illinois Appellate Court had supposed. Id. Rather, the evidence underlying McFowler's conviction must be viewed as a whole. Although Meredith had identified McFowler in the courtroom, the fact that she had evidently identified someone else (Byndum) either in addition to or instead of him in the context of a non-suggestive lineup conducted hours after the crime was committed cast doubt on the value of Meredith's in-court identification. Id. The court noted that the circumstances relevant to the reliability of the identifications in this case suggested strongly that the lineup was the more reliable of the two identifications that Meredith had made: the courtroom identification was made under suggestive circumstances (McFowler had been seated at table with his counsel); even when Byndum was summoned into the courtroom during Meredith's redirect examination, she was called upon to make an identification from just two people as opposed to the six she had viewed in the lineup; and a good deal of time had passed before Meredith made her in-court identification in 1992. Id. at *8-*9. "Given Meredith's same-day identification of Byndum, the suggestiveness of the in-court identification procedure, and the very long time delay between the two, it was objectively unreasonable for the Illinois Appellate Court to find credible Meredith's albeit `unwavering,' in-court identification of McFowler." Id. at *9.

One could infer from the other record evidence that McFowler was present at the scene of the murder, the court acknowledged ( id. at *9-*10), but that evidence was insufficient to support his conviction. In order to prove McFowler's guilt under an accountability theory, the State had to establish that McFowler, either before or during the commission of the crime, and with the intent to promote or facilitate that crime, in some way solicited, aided, abetted, agreed or attempted to aid someone else in the planning or commission of the crime. Id. at *11, quoting 720 Ill. Comp. Stat. Ann. 5/5-2(c). In this case, there was no evidence that the perpetrators had planned the crime, let alone that McFowler was aware of that plan. Id. Beyond McFowler's presence at the scene of the crime and his flight therefrom, the most that the evidence established was his presence in the same apartment as Byndum a half hour after the murder and his failure to report the crime. Id. at *12. In the district court's view, this was not enough to show that McFowler took some act to promote or facilitate the commission of the crime or that he intended to do so. Id.

The district court therefore granted McFowler's request for a writ of habeas corpus and ordered his release within twenty days. Id. at *13. On the State's request, the district court stayed its order pending resolution of the State's appeal. R. 40.

II.

"The Antiterrorism and Effective Death Penalty Act of 1996 modified a federal habeas court's role in reviewing state prisoner applications in order to prevent federal habeas `retrials' and to ensure that state-court convictions are given effect to the extent possible under law." Bell v. Cone, supra, 535 U.S. at 693, 122 S.Ct. at 1849, citing Williams v. Taylor, 529 U.S. 362, 403-04, 120 S.Ct. 1495, 1518, 146 L.Ed.2d 389 (2000). Under the AEDPA, in order to obtain habeas relief, a petitioner must establish that the proceedings in state court resulted in a decision (1) "that was contrary to, or involved an unreasonable application of, clearly established Federal law, as determined by the Supreme Court of the United States," or (2) "that was based on an unreasonable determination of the facts in light of the evidence presented in the State court proceeding." 28 U.S.C. § 2254(d). The relevant decision, for purposes of this assessment, is the decision of the last state court to rule on the merits of the petitioner's claim-here, the order of the Illinois Appellate Court. E.g., Schultz v. Page, 313 F.3d 1010, 1015 (7th Cir. 2002), cert. denied, ___ U.S. ___, 123 S.Ct. 2220, 155 L.Ed.2d 1107 (2003).

There is no contention that the Appellate Court's decision is "contrary to" clearly established law. The court acknowledged that due process requires the State to establish beyond a reasonable doubt that a defendant committed each element of the charged offense. App.Ct. Order at 31, citing In re Winship, 397 U.S. 358, 361-63, 90 S.Ct. 1068, 1071-72, 25 L.Ed.2d 368 (1970). The court also recognized that, under Jackson v. Virginia, supra, 443 U.S. at 319, 99 S.Ct. at 2789, a court examining the sufficiency of the evidence underlying a defendant's conviction must construe that evidence in the light most favorable to the State and determine whether any rational trier of fact could have found the elements of the crime beyond a reasonable doubt. App.Ct. Order at 31-32. Moreover, we note that in evaluating the reliability and sufficiency of Meredith's identification testimony as the key evidence supporting McFowler's conviction, the court looked to the same factors that the Supreme Court identified as pertinent in Neil v. Biggers, 409 U.S. 188, 199, 93 S.Ct. 375, 382, 34 L.Ed.2d 401 (1972). See App.Ct. Order at 32.

The question posed by McFowler's petition is whether the Illinois Appellate Court's decision amounts to an unreasonable application of the Jackson sufficiency standard. "A state-court decision that correctly identifies the governing legal rule but applies it unreasonably to the facts of a particular prisoner's case certainly would qualify as a decision `involv[ing] an unreasonable application of . . . clearly established Federal law.'" Williams, 529 U.S. at 407-08, 120 S.Ct. at 1520. "Unreasonable" means something more than "mistaken," however. Woodford v. Visciotti, 537 U.S. 19, 25, 27, 123 S.Ct. 357, 360, 361, 154 L.Ed.2d 279 (2002) (per curiam); Williams, 529 U.S. at 410, 120 S.Ct. at 1522; Hardaway v. Young, 302 F.3d 757, 762 (7th Cir. 2002), cert. denied, ___ U.S. ___, 123 S.Ct. 1802, 155 L.Ed.2d 668 (2003). A state court decision is unreasonable for purposes of section 2254(d)(1) if its application of Supreme Court precedent "l[ies] well outside the boundaries of permissible differences of opinion." Id. at 762, citing Williams, 529 U.S. at 411, 120 S.Ct. at 1522; see also Visciotti, 123 S.Ct. at 361 ("[t]he federal habeas scheme . . . authorizes federal-court intervention only when a state-court decision is objectively unreasonable"). Within this framework, the precise legal question that we must answer is the one that the district court itself posed: "Was it objectively unreasonable for the Illinois Appellate Court to conclude that any rational trier of fact, after viewing the evidence in the light most favorable to the state, could have found the essential elements of first degree murder beyond a reasonable doubt." 2003 WL 76861, at *5.

It is undisputed that McFowler was convicted of Logan's murder not as a principal but on an accountability theory. As we have noted, neither the trial judge (in finding McFowler guilty) nor the Illinois Appellate Court (in affirming his conviction) stated this expressly. Neither court, for that matter, identified the elements that would have to be established in order to find McFowler guilty under either theory. However, when he sentenced McFowler, Judge Hett remarked that "Mr. McFowler did not pull the trigger. Nevertheless, he is accountable for what Mr. Byn[d]um did." Tr. L12. Judge Gottschall considered whether the evidence would have supported McFowler's conviction as a principal and concluded that it would not: Logan was killed with a revolver; Meredith testified that she saw McFowler holding a shotgun, not a revolver; there was no evidence that the shotgun had been fired; the inconclusive results of the gunshot residue test on McFowler were consistent with the possibility that he had handled the revolver, but shed no light on whether he had fired the gun or merely handled it after the shooting; and there were at least two and possibly three intruders present in the apartment at the time of the murder, rendering it impossible to conclude beyond a reasonable doubt that McFowler was the one who fired the shot that killed Logan. 2003 WL 76861, at *11. For its part, the State does not suggest that McFowler could have been convicted as a principal. On the contrary, it asserts that McFowler was convicted on an accountability theory ( e.g., State's Opening Brief at 10) and it defends his conviction on that basis alone.

As we have noted, the Illinois accountability statute provides that an individual is criminally liable for an offense committed by another person "when . . . [e]ither before or during the commission of an offense, and with the intent to promote or facilitate such commission, he solicits, aids, abets, agrees or attempts to aid, such other person in the planning or commission of the offense." 720 Ill. Comp. Stat. Ann. 5/5-2(c). The pertinent statutory terms here are "aids" and "abets," as there is no evidence that McFowler solicited Logan's murder, that he agreed to aid anyone in the planning or commission of the crime, or that he participated in the planning of the offense. As Judge Gottschall recognized, the defendant's mere presence at the scene of the crime is insufficient to establish that a defendant aided or abetted the commission of the offense, even when coupled with his knowledge of the crime and his flight from the crime scene. People v. Williams, supra, 250 Ill.Dec. 692, 739 N.E.2d at 472; People v. Shaw, 186 Ill.2d 301, 239 Ill.Dec. 311, 713 N.E.2d 1161, 1173 (1999); see also United States v. Johnson, 612 F.2d 305, 308 (7th Cir. 1980) (applying Illinois law). The evidence must additionally establish the defendant's intent to facilitate or otherwise promote the commission of the offense, either by showing that the defendant shared the principal's criminal intent or that there was a common criminal design. People v. Perez, 189 Ill.2d 254, 244 Ill.Dec. 371, 725 N.E.2d 1258, 1264-65 (2000); People v. Stanciel, supra, 180 Ill.Dec. 124, 606 N.E.2d at 1210. Of course, "[i]ntent may be inferred from the character of the defendant's acts as well as the circumstances surrounding the commission of the offense." Perez, 725 N.E.2d at 1265, citing Stanciel, 180 Ill.Dec. 124, 606 N.E.2d at 1210.

Meredith's testimony unquestionably was essential to the finding that McFowler aided or abetted Logan's murder. Although abundant circumstantial evidence placed McFowler at the scene of the crime, it was Meredith's testimony and hers alone that placed him next to Logan's body with a shotgun in his hand. The shotgun was not the murder weapon, of course; but we shall assume (as Judge Gottschall did) that McFowler's presence at the scene of the crime with a firearm, next to Logan's body, just after Logan was shot, would suffice to establish his complicity in the murder.

The dispositive issue, then, is the reliability of Meredith's testimony. After describing the man she had seen trip over Logan's body with a shotgun in his hand, Meredith indicated that it was McFowler whom she had picked out of the lineup on the day of the shooting. Twice during her testimony — first on direct and then again on redirect, when Byndum was brought into the courtroom to stand beside McFowler — Meredith identified McFowler as the person she had seen. However, Detective Foley's stipulated testimony was that Meredith had identified Byndum in the lineup, not McFowler. When confronted with Foley's conflicting testimony on cross-examination, Meredith testified that she had picked "both" Byndum and McFowler. It is Meredith's puzzling testimony that she picked two individuals as the single person that she saw at the scene of the crime which calls into doubt the reliability of her identification testimony.

We begin our analysis by noting that, so far as the record reveals, the lineup that Meredith viewed on the day of the murder was untainted by any suggestive or otherwise inappropriate circumstance or procedure. This is in contrast to each of the Supreme Court's decisions dealing with the reliability of eyewitness identification testimony. Manson v. Brathwaite, 432 U.S. 98, 97 S.Ct. 2243, 53 L.Ed.2d 140 (1977); Neil v. Biggers, supra, 409 U.S. 188, 93 S.Ct. 375, 34 L.Ed.2d 401; Coleman v. Alabama, 399 U.S. 1, 90 S.Ct. 1999, 26 L.Ed.2d 387 (1970); Foster v. California, 394 U.S. 440, 89 S.Ct. 1127, 22 L.Ed.2d 402 (1969); Simmons v. United States, 390 U.S. 377, 88 S.Ct. 967, 19 L.Ed.2d 1247 (1968); Stovall v. Denno, 388 U.S. 293, 87 S.Ct. 1967, 18 L.Ed.2d 1199 (1967); Gilbert v. California, 388 U.S. 263, 87 S.Ct. 1951, 18 L.Ed.2d 1178 (1967); United States v. Wade, supra, 388 U.S. 218, 87 S.Ct. 1926, 18 L.Ed.2d 1149. In those cases, the witness's in-court identification was preceded by a lineup or other identification made under suggestive conditions. However, the factors on which the Court has focused in that context remain relevant even when a witness has made an identification untainted by suggestive identification procedures. See Kennaugh v. Miller, 289 F.3d 36, 46-47 (2d Cir.) (collecting cases), cert. denied, 537 U.S. 909, 123 S.Ct. 251, 154 L.Ed.2d 187 (2002). "[R]eliability is the linchpin in determining the admissibility of identification testimony. . . ." Brathwaite, 432 U.S. at 114, 97 S.Ct. at 2253. And although here we are not concerned with the admissibility of Meredith's identification of McFowler (McFowler's counsel filed a motion to suppress the results of the lineup but later withdrew the motion), the reliability of her identification (both at the lineup and at trial) is of prime importance to the sufficiency of the evidence underlying McFowler's conviction.

We therefore must assess the reliability of Meredith's identification under the "totality of the circumstances." Biggers, 409 U.S. at 199, 93 S.Ct. at 382. The factors that bear on the likelihood that the witness has identified the wrong person are "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 confrontation, and the length of time between the crime and the confrontation." Ibid.

As the Illinois Appellate Court pointed out, one can readily infer from Meredith's testimony that she had a sufficient opportunity to observe the individual with the shotgun. App. Court Order at 33. The man was standing sixteen feet away from her, and although his face was not turned toward her at first, she was able to get a good look at it as he stumbled over Logan's body. Id. Meredith suggested that she may have had the intruder in her sight for as long as a few minutes. That seems quite doubtful given her testimony that she slammed the bedroom door shut and hid in the closet with her children upon seeing Logan's body and the armed man next to it. Tr. G20-23. The Illinois Appellate Court itself viewed this testimony with "some skepticism." App.Ct. Order at 33. Nonetheless, the record supplies us with no reason to doubt that Meredith had an ample opportunity to view the intruder. And although the record is silent as to the degree of attention with which Meredith viewed that individual, she certainly was not a casual bystander, and it is fair to assume that she looked at the man with the shotgun with a high degree of attention. The third Biggers factor does not apply here. There is no evidence that Meredith ever supplied the police with a description of the intruder before viewing the lineup. The fourth factor yields conflicting signals: the record sheds no light on the degree of certainty with which Meredith identified McFowler as the intruder at the lineup, but her later testimony that she picked "both" McFowler and Byndum may suggest uncertainty ( see n. 4, infra); as for her in-court identification, the record does not affirmatively indicate that Meredith identified McFowler with confidence, but on the other hand there are no indicia of hesitation or doubt, and Meredith consistently identified McFowler as the intruder she saw. The final Biggers factor — the passage of time between the crime and the confrontation — weighs differently depending on whether one is addressing the lineup or Meredith's in-court identification of McFowler. The fact that the lineup was conducted within six hours of the shooting weighs rather strongly in favor of reliability. See Chavis v. Henderson, 638 F.2d 534, 537 (2d Cir. 1980) ("`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'"), quoting United States ex rel. Springle v. Follette, 435 F.2d 1380, 1383 (2d Cir. 1970). By contrast, the passage of twenty-nine months before the trial occurred detracts significantly from the weight to be given that identification. See Biggers, 409 U.S. at 201, 93 S.Ct. at 383 (lapse of seven months between crime and confrontation "would be a seriously negative factor in most cases").

What the Illinois Appellate Court did not consider, and what must be considered given the divergent testimony about whom Meredith identified at the lineup, is which of the two identifications is the more reliable; and on that point, the circumstances favor the lineup over the in-court identification, as Judge Gottschall pointed out. The lineup was conducted within hours of the crime, whereas the trial occurred more than two years later. The lineup called for Meredith to choose from among six individuals, whereas the suggestive circumstances of the courtroom would have focused her attention on the (single) defendant. See United States v. Matthews, 20 F.3d 538, 547 (2d Cir. 1994) (noting that "the circumstances at trial may themselves be tantamount to a showup"). True, on redirect examination, Byndum was brought into the courtroom, and Meredith was asked whether he or McFowler was the intruder that she saw on the day of the murder. By then, of course, she had already identified McFowler once in court and naturally would have been inclined to reaffirm her identification. That point aside, choosing from between two individuals (one of them the defendant) in court is necessarily a less reliable identification than choosing from among six individuals in a properly conducted lineup. We hasten to add that this is not a case in which the record discloses some additional fact that would tend to bolster the witness's courtroom identification. See id. (noting types of measures, e.g., seating defendant somewhere other than at counsel table, that can mitigate suggestiveness of in-court identification).

Although the lineup is the more reliable of the two identifications, the evidence is hopelessly inconsistent as to whom Meredith actually picked out of the lineup. Had Meredith stood by her initial testimony that she picked McFowler out of the lineup, then the factfinder would have been presented with a straightforward credibility choice between her testimony and the stipulated testimony of Detective Foley that she picked Byndum. The trial judge would have been free to disbelieve Foley, for the testimony of law enforcement officials is entitled to no greater weight than the testimony of any other witness ( see, e.g., United States v. Martin, 507 F.2d 428, 432-33 (7th Cir. 1974)), and the parties' stipulation as to what a witness would testify does not obligate the trier of fact to credit the witness's would-be testimony ( e.g., Tucker v. Brady, 305 F.2d 550, 552 (9th Cir. 1962)). But instead, on redirect examination, Meredith indicated that she picked "both" Byndum and McFowler at the lineup. As the record stands, that answer is nonsensical.

In only one circumstance would Meredith's answer both make sense and give the trier of fact reason to trust her ability to identify the man with the shotgun, and that is if she saw two intruders rather than one. It is this scenario that the State presses upon us with vigor and it is one, we acknowledge, that is possible in the abstract. The evidence does indicate that there were two if not three people involved in the crime. Moreover, Byndum's guilty plea reveals that he was one of them. Thus, if Meredith saw two people in her apartment, she could have identified McFowler as the man with the shotgun and Byndum as the second intruder. This would have been entirely consistent with her courtroom identification of McFowler and her initial testimony about previously having identified McFowler at the lineup as the man with the shotgun. The State thus urges us to indulge the inference that Meredith saw a second intruder in addition to the man with the shotgun.

However, there is not a shred of evidence that Meredith did see more than one of the intruders. Indeed, the clear import of the testimony is that Meredith only saw one person. Meredith's testimony about the shooting indicated that she saw only one intruder (the man with the shotgun) and heard a second (the voice which said, "if you are going to shoot him; shoot him" (Tr. G22)). Her testimony about the lineup likewise indicates that she was identifying one person and one person only — the man with the shotgun. Foley's stipulated testimony too refers to the identification of a single person. The trier of fact thus had no basis from which to conclude that Meredith might have seen two people at the scene of the crime, thus enabling her to have identified both McFowler and Byndum.

We cannot say it was impossible for Meredith to have seen two intruders; but the mere possibility that she did is not sufficient to rehabilitate her testimony about the lineup. The State is entitled to every reasonable inference that may be drawn from the record, but to be reasonable the inference must find at least some support in the facts. See United States v. Lewis, 787 F.2d 1318, 1323 (9th Cir. 1986) ("mere suspicion or speculation cannot be the basis for creation of logical inferences"), amended in other respects, 798 F.2d 1250 (9th Cir. 1986); United States v. Brown, 763 F.2d 984, 990 (8th Cir. 1985) ("a conviction based on evidence that requires speculation or mere conjecture by the [factfinder] is invalid"). Such support is lacking here. All of the evidence points to Meredith having seen one and only one intruder. Indeed, had she seen two, surely the State would have elicited that information, particularly after Meredith announced on the witness stand that she had picked "both" Byndum and McFowler from the lineup. As things stand, however, we would have to venture beyond the record and into the realm of speculation in order to suppose that Meredith was identifying two different intruders when she identified "both" Byndum and McFowler. It is equally possible that Meredith was confused or flummoxed, or that her memory was faulty. The evidence does not permit us to choose one of these possibilities over another.

Meredith's own testimony thus rendered it impossible for the factfinder to determine whom it was that she identified at the lineup. The Illinois Appellate Court recognized the inconsistencies in the evidence as to Meredith's lineup identification (App.Ct. Order at 33), and as the Appellate Court noted, it is difficult to tell whether the trial court disregarded altogether the possibility that Meredith had identified McFowler at the lineup or simply gave Meredith's (initial) testimony that she did decreased weight ( id.). The Appellate Court's suggestion that the trier of fact could have given her purported lineup identification of McFowler at least some weight ( id.) was objectively unreasonable. Once Meredith testified that she had identified "both" McFowler and Byndum at the lineup, the trier of fact could not have found with any degree of confidence that it was McFowler that Meredith actually had identified. Whom Meredith identified at the lineup is simply an unsolvable mystery.

This leaves Meredith's in-court identification as the sole evidentiary support for a finding that McFowler stood next to Logan's body with a shotgun in hand. The Illinois Appellate Court reasoned that "[e]ven if Meredith's lineup identification were given no weight whatsoever," the factfinder could still make this finding based on her "unwavering in-court identification of the defendant" alone. App.Ct. Order at 34. Had there never been a lineup, we have little doubt that Meredith's positive in-court identification would have sufficed as proof of McFowler's culpable involvement in Logan's murder. But a lineup did take place, and Meredith made an identification at that lineup. So what we must consider is whether Meredith's inconsistent testimony about the lineup casts such doubt on her courtroom identification as to render the latter insufficiently reliable to support McFowler's conviction.

As a threshold matter, we agree with the district court that Meredith's in-court identification of McFowler cannot be considered in isolation, apart from her testimony about the lineup identification. 2003 WL 76861, at *7. For all of the reasons we have discussed, the evidence as to the lineup precludes any reliable determination as to whom Meredith identified in the lineup. That does not mean, however, that the lineup (and the testimony about that lineup) should be disregarded as if it had never occurred. Meredith made two identifications in this case, one at the lineup and one in court. Whether her in-court identification of McFowler was reliable enough for the trier of fact to find that McFowler was the person she saw in her home armed with a shotgun is a question that can only be answered with reference to the lineup and the reliability of that identification as well. See Kennaugh, 289 F.3d at 46 (witness's inability to identify defendant in prior lineup and several photo arrays "inevitably heightens the risk that her in-court identification was induced by the suggestiveness of the setting in which it occurred").

As we have already discussed, the factors relevant to the reliability of the two identifications point to the lineup identification as the more reliable of the two. That identification took place within hours of the crime, called upon Meredith to choose from six individuals, and, so far as the record reveals, was free of any circumstance that would have made it suggestive. The later identification, on the other hand, took place nearly two and one-half years after the shooting and within the inherently suggestive environs of a courtroom. The fact that Meredith reaffirmed her identification of McFowler after seeing Byndum and McFowler stand side by side adds little weight to the strength of her identification. Byndum was not summoned to the courtroom until after Meredith already identified McFowler during her direct testimony, and given the evident differences in their height alone, there would have been no mixing the two up. Indeed, the fact that Meredith apparently had no difficulty distinguishing between the two men in court years after the crime merely reinforces the mystery as to how she could have chosen both of them from the lineup, when the event was still fresh in her mind.

Of course, it is not unheard of for an eyewitness to fail to identify the defendant in a lineup, to pick the wrong person, or to express some ambivalence about her identification, only to later make a positive identification of him in court. Often the apparent inconsistency is explained by such things as the witness's emotional state, her failure to view the lineup with sufficient care, or an unexpected change in the defendant's appearance between the time of the crime and the lineup. See, e.g., United States ex rel. Kosik v. Napoli, 814 F.2d 1151, 1159 (7th Cir. 1987) (defendant had cut his hair and shaved his mustache); Cole v. State, 588 N.E.2d 1316, 1317-18 (Ind.Ct.App. 1992) (witness confused by defendant's longer hair); Commonwealth v. Hurd, 268 Pa.Super. 24, 407 A.2d 418, 422 (1979) (witness did not stand close enough to lineup on first occasion); cf. Taylor v. State, 232 Ga.App. 383, 501 S.E.2d 875, 876 (1998) (witness unable to make in-court identification because her vision was poor, but was able to identify him from photo array). So long as the record indicates that the witness had a sufficient basis for identifying the defendant ( e.g., an adequate opportunity to observe the defendant at the time of the offense), her inability to identify him prior to trial normally will not detract from her in-court identification. E.g., United States v. Matthews, 20 F.3d at 547-48 (lack of pretrial identification did not undermine witnesses' in-court identifications, where their testimony was "sufficient to establish that their identifications had an origin independent of their viewing [defendant] in the courtroom and hence was sufficient to meet the threshold of reliability needed to permit them to attempt in-court identifications of the men they had seen"); see also United States v. Davies, 768 F.2d 893, 904 (7th Cir. 1985) ("[g]enerally, the question of the suggestiveness or credibility of the in-court identification is to be resolved ultimately by the jury after the defendant has had an opportunity to test the accuracy of an identification through cross-examination"). By contrast, Meredith's initial inability to identify McFowler was not explained by any such circumstance. Indeed, Meredith did not profess to have had any inability to make an identification, or uncertainty about her ability, in viewing the lineup. Only when she was confronted with the possibility that she had picked Byndum rather than McFowler did she say that she had picked both. Nothing in the record explains why at the lineup, on the very day of the offense, Meredith could have picked two people as the one person she saw holding the shotgun and then two years later picked McFowler alone. Nothing, that is, except the inherently suggestive environs of the courtroom. See Kennaugh, 289 F.3d at 46.

Indeed, Meredith's testimony itself calls into doubt the reliability of her identification. In testifying on cross-examination that she identified both Byndum and McFowler at the lineup, Meredith not only contradicted her own initial testimony, but gave an answer that makes no sense on the facts. Inconsistencies in a witness's testimony are not unusual either, and normally these are left for the factfinder to assess. See, e.g., United States v. Gutman, 725 F.2d 417, 421 (7th Cir. 1984) (trial court not compelled to instruct jury to disregard testimony of mentally ill key witness notwithstanding witness's inconsistency and confession that he had lied; witness's testimony did not reflect "abnormal degree of internal inconsistency" and "it is common for witnesses to change their stories on the stand"). But Meredith's belated assertion that she identified both McFowler and Byndum at the lineup is irreconcilable with the facts. Moreover, it relates not to a collateral point but to the most material aspect of her testimony. Cf. United States v. Spain, 536 F.2d 170, 173 (7th Cir. 1976) ("[e]ven if the jury found that an agent had deliberately testified falsely on a collateral matter, it could still accept the substance of his testimony on the issues in the case"); United States v. Tropiano, 418 F.2d 1069, 1074 (2d Cir. 1969) ("[a] witness may be inaccurate, contradictory and even untruthful in some respects and yet be entirely credible in the essentials of his testimony"). Meredith's self-contradictory testimony casts doubt on the value of her in-court identification, which was the centerpiece of her testimony.

Other evidence corroborates McFowler's presence at the scene of the crime, but none of that evidence supplies the proof necessary to establish his complicity in the crime. Only Meredith's testimony supplies that proof by placing McFowler next to Logan's body with a shotgun in his hand. The balance of the evidence bolsters Meredith's identification of McFowler only in the limited sense that it confirms the possibility that it was McFowler that she saw. But Byndum was present at the scene of the crime as well (as evidenced by both his guilty plea and the circumstantial evidence). Foley's stipulated testimony that Meredith had identified Byndum at the lineup as the man with the shotgun, and Meredith's own testimony that she identified "both" Byndum and McFowler, render it just as possible that it was Byndum that Meredith saw with the shotgun in his hand.

The identity of the defendant as a perpetrator of the crime is one of essential elements of the offense that must be proved beyond a reasonable doubt. United States v. Weed, 689 F.2d 752, 754 (7th Cir. 1982); People v. Dante, 35 Ill.2d 538, 221 N.E.2d 409, 410 (1966); People v. Daniels, 331 Ill.App.3d 380, 264 Ill.Dec. 447, 770 N.E.2d 1143, 1152 (2002). Because Meredith's identification testimony alone addresses this element without any corroboration, a reasonable doubt about her identification would necessitate McFowler's acquittal. See United States v. Jackson,

509 F.2d 499, 507 (D.C. Cir. 1974) ("[t]he problem of witness-competence is accentuated in cases wherein the guilt of the accused is sought to be rested on an uncorroborated identification by a single witness"); see also, e.g., Commonwealth v. Wiley, 288 Pa.Super. 397, 432 A.2d 220 (1981) (finding witness's tentative identification of defendant insufficient to support conviction where no other evidence corroborated identification).

For all of these reasons, were we reviewing McFowler's conviction on direct appeal, we might have reversed on the ground that no factfinder could find beyond a reasonable doubt that it was McFowler whom Meredith saw next to Logan's body with a shotgun in his hand. Meredith's self-contradictory and nonsensical testimony about the lineup makes it impossible to determine whom she identified as the intruder at the lineup. Her in-court identification some twenty-nine months later is suspect given the substantial passage of time, the inherently suggestive character of the in-court identification, and Meredith's own inconsistent testimony about the lineup.

But we do not sit in direct review of McFowler's conviction — it was the Illinois Appellate Court that filled that role. And the AEDPA demands that we give appropriate deference to the decisionmaking authority of the state courts. We may not use the writ of habeas corpus to effectively overrule the decisions of those courts simply because we disagree with them. Woodford v. Visciotti, supra, 537 U.S. at 27, 123 S.Ct. at 361. Only if we find the state court's application of Supreme Court precedent objectively unreasonable ( 28 U.S.C. § 2254(d)(1)) — that is, "well outside the boundaries of permissible differences of opinion" ( Hardaway v. Young, supra, 302 F.3d at 762) — may we uphold the district court's decision to grant the writ. Visciotti, 123 S.Ct. at 361. Notwithstanding our doubts, we do not believe we can characterize the Illinois Appellate Court's decision sustaining the sufficiency of the evidence underlying McFowler's conviction as objectively unreasonable.

Meredith's in-court identification of McFowler was, as the Illinois Appellate Court described it, unwavering. She first described the man that she had seen in her apartment next to Logan's body in terms consistent with McFowler's appearance. When shown a photograph of the lineup in court, she circled McFowler's picture. She then pointed him out in the courtroom. Later, after testifying that she had picked "both" McFowler and Byndum out of the lineup, she was asked to view the two men standing beside one another in the courtroom. Once again she identified McFowler as the man she had seen in her apartment, arguably dispelling any doubt about her ability to distinguish between McFowler and Byndum. At no time did Meredith profess any uncertainty in making her courtroom identification, nor does the record otherwise suggest any hesitation or confusion on her part in doing so.

The pertinent inquiry, then, given Meredith's (apparent) uncertainty about whom she identified at the lineup, is whether she had a basis independent from the suggestive setting of the courtroom for making her in-court identification of McFowler. United States v. Matthews, 20 F.3d at 547-48. The record discloses that she did. Meredith's testimony indicates that she was able to get a clear look at the intruder from a relatively short distance, under circumstances where her degree of attention would have been high. At the same time, there is no evidence that the lineup conducted on the day of the shooting was in any way suggestive, such that it might have tainted Meredith's subsequent in-court identification.

The reliability of Meredith's in-court identification is also bolstered, at least to some degree, by other evidence. There is, of course, quite strong circumstantial evidence placing McFowler at the scene of the crime, which supports the likelihood that it indeed was McFowler whom Meredith saw. The evidence also confirms the accuracy of Meredith's observations in other respects: Meredith testified that she saw a shotgun in the intruder's hand, and a shotgun was found at the Flournoy building where McFowler was arrested; she testified that the intruder wore blue jeans, and McFowler was wearing blue jeans at the time of his arrest; she described the intruder she saw as somewhat short, and the record indicates that McFowler indeed is relatively short (five feet, six inches tall).

Moreover, Meredith was subject to full cross-examination as to her ability to identify the man she had seen on the day of Logan's murder, and all of the inconsistencies and vulnerabilities in her identification of McFowler were put before the trier of fact. It bears repeating that the credibility and reliability of an eyewitness's in-court identification is a question for the finder of fact "after the defendant has had an opportunity to test the accuracy of an identification through cross-examination." United States v. Davies, supra, 768 F.2d at 904; see also, e.g., United States v. Torres, 191 F.3d 799, 811 (7th Cir. 1999); United States v. Hall, 165 F.3d 1095, 1107 (7th Cir. 1999); United States v. Allen, 930 F.2d 1270, 1273 (7th Cir. 1991). The finder of fact is uniquely situated to observe how well a witness's demeanor withstands the rigors of cross-examination and the exposure of inaccuracies and inconsistencies in her story. See Anderson v. City of Bessemer City, N.C., 470 U.S. 564, 574-75, 105 S.Ct. 1504, 1512, 84 L.Ed.2d 518 (1985); United States v. Mancillas, 183 F.3d 682, 701 n. 22 (7th Cir. 1999), quoting United States v. Garcia, 66 F.3d 851, 856 (7th Cir. 1995). The trial judge in this case was, of course, aware of the circumstances that cast doubt on Meredith's identification. Despite those circumstances, he found Meredith's in-court identification of McFowler to be credible. Tr. K56. That assessment commands great deference. E.g., Murrell v. Frank, 332 F.3d 1102, 1112 (7th Cir. 2003).

To hold McFowler entitled to relief in habeas corpus, we would have to say not only that Meredith's in-court identification was insufficiently credible and reliable as a matter of law, but that no court could reasonably think otherwise. This we cannot do. McFowler's petition has considerable intuitive appeal, given the extent to which the State's case against him depended upon Meredith's identification and the degree to which Meredith's ability to identify the intruder was called into doubt not only by the police officer who arranged the lineup she viewed hours after the crime but by her own contradictory testimony about whom she picked at the lineup. Yet, witnesses (including eyewitnesses) often make surprising, contradictory, and nonsensical statements under the stress of courtroom examination, and although such gaffes would permit a factfinder to discredit their testimony, they normally do not command reversal of a conviction on appeal. See United States v. Frederick, 78 F.3d 1370, 1374 (9th Cir. 1996) (testimony which is confused, inconsistent, and uncertain can still support a conviction). We cannot know whether Meredith simply became flummoxed and contradicted herself for that or for some other reason. But given that she had a basis from which she could identify the man who stood over Logan's body with a shotgun in his hand, the Illinois Appellate Court was within its rights to conclude that her identification of McFowler as that man was not unreliable as a matter of law and was sufficient, coupled with the circumstantial evidence placing him at the scene of the shooting, to support McFowler's conviction. We can find nothing in the holdings or rationale of Jackson, Biggers, or other Supreme Court precedents that forecloses the Illinois Appellate Court's conclusion.

III.

For all of the foregoing reasons, we REVERSE the district court's judgment granting McFowler's petition for a writ of habeas corpus. We thank McFowler's appointed attorneys for their service on his behalf.