Raymond Young and Michael Richardson were convicted of two counts each of armed robbery under 18 U.S.C. §§ 2, 2113. Young was sentenced to terms of 15 years and 5 years to run concurrently. Richardson received two concurrent 15 year sentences. In separate appeals both Young and Richardson allege that the trial court1 erred in refusing to grant a mistrial because of prejudicial publicity during the trial. Young also alleges error in that the trial court failed to suppress a photographic identification of him because it was unduly suggestive. In addition, Richardson charges that the trial court erred in permitting the government's witness to testify when, due to prejudicial publicity surrounding the shooting incident in which the witness was involved, her appearance was likely to unduly influence and prejudice the jury. These cases have been consolidated for purposes of appeal. We reverse the convictions for reasons stated below but affirm the trial court's refusal to suppress the photospread identification.

The joint trial of Young and Richardson began April 28, 1980. On the morning of April 29 defense counsel moved for a mistrial on the basis of prejudicial publicity. The publicity in question included broadcast media and newspaper reports of a shooting incident in which an intended witness in the trial was wounded. It was reported that Consuela Brown, a prospective government witness in the trial who was granted immunity for her testimony, had been shot and wounded outside of her home by unknown assailants in the early morning hours of April 28. The reports went on to state that Brown had received numerous threats because of her expected testimony and that she had been placed under protection by the FBI. It was also specifically stated in the reports that although Ms. Brown had received numerous threats because of her expected testimony, officials were unable to connect the shooting to the trial. The trial judge questioned jurors collectively in open court whether any of them had seen or heard any publicity regarding the trial. Five jurors and one alternate indicated that they had had some exposure to the news stories concerning the trial. The judge then questioned each of these jurors separately in open court, but out of the hearing of the other jurors. Counsel for defendants were given opportunity to question each juror individually. Five of the jurors questioned responded that they could disregard the publicity and deliberate only on the evidence introduced at trial. Juror Dennis Jones responded affirmatively to the trial judge's question regarding his ability to disregard the publicity, but displayed some nervousness and hesitation. He was ultimately excused.2 The trial judge denied the defendants' motion for new trial.

The trial judge has broad discretion in ruling on an issue of prejudice resulting from the exposure of jurors to publicity regarding the trial. Marshall v. United States, 360 U.S. 310, 312, 79 S. Ct. 1171, 1172, 3 L. Ed. 2d 1250 (1959); United States v. Williams, 604 F.2d 1102, 1114 (8th Cir. 1979). The trial court's findings on this point may not be set aside unless error is manifest. Mastrian v. McManus, 554 F.2d 813, 818 (8th Cir.), cert. denied, 433 U.S. 913, 97 S. Ct. 2985, 53 L. Ed. 2d 1099 (1977).

A number of courts have held that where publicity is determined to be potentially prejudicial, the trial judge is obliged to conduct an inquiry to determine whether jurors have actually been exposed to that publicity and to what extent. In United States v. Herring, 568 F.2d 1099, 1106 (5th Cir. 1978), the Fifth Circuit reversed a conviction because the trial court failed to conduct an inquiry into the possibility of prejudice where newspaper headlines reported death threats against a key prosecution witness and relied instead upon its preliminary instructions to the jury to disregard any mention of the case outside the courtroom. There the appellate court found that the trial court should have examined each juror separately, in the presence of counsel, to determine 1) how much contact the jury members had had with the damaging publicity, and 2) how much prejudice, if any, resulted from that contact. Id. at 1106. The Second Circuit reversed a conviction under similar circumstances in United States v. Lord, 565 F.2d 831, 838 (2d Cir. 1977). There the trial court relied only on repetitive admonitions to members of the jury despite the publication of two newspaper articles detailing a previous stabbing incident involving a government witness and mentioning that one of the defendants had been on the FBI's "ten most wanted" list.

It is clear that in the instant case there existed a sufficient threat of prejudice to require determination of the amount of juror exposure to the publicity. It is equally clear that the trial judge here conducted such an inquiry. He polled the jury as a whole using general references to news stories regarding the case. He questioned each juror individually out of the hearing of the others and admonished each privately and as a group from the bench to consider only evidence introduced at trial.

Despite the careful and deliberate attempt by the trial judge to insure jury impartiality, we believe the "collective impact," United States v. Titsworth, 422 F. Supp. 587, 591 (D.Neb.1976), of prejudicial in-trial publicity and the appearance of the wounded witness testifying from the wheelchair adversely affected the case for the defense, and the trial court abused its discretion in failing to grant a mistrial.

The Supreme Court found similar abuse of discretion in Marshall, supra, where seven jurors were exposed to in-trial publicity which reported that the defendant had two previous felony convictions. Although the trial judge polled the jurors individually and was satisfied with the assurances that each would decide the case only on the evidence introduced at trial, the court found the information involved so inherently prejudicial as to necessitate a new trial. Id. 360 U.S. at 312, 79 S. Ct. at 1172. The Fifth Circuit in United States v. Williams, 568 F.2d 464, 471 (5th Cir. 1978) held that where two jurors had seen a television newscast reporting that the defendant had been convicted by another jury on the same charges, but that a new trial had been ordered because of "erroneous testimony" neither the fact that the two said they could decide the case solely on the evidence adduced in court, nor the court's admonition to disregard everything not heard in court, cured the prejudice. Following a lengthy discussion of pretrial, in-trial, and "overlap" publicity cases, the Williams court stated:

Principles from the pretrial and "overlap" cases must not be haphazardly applied to cases involving only publicity that occurred during the trial. The "during trial" cases though fewer in number, contain greater opportunities for prejudice. For example, information reported during the trial seems far more likely to remain in the mind of a juror exposed to it, and he may be more inclined to seek out this information when he is personally involved in the case. See Jaffe (Trial by Newspaper, 40 N.Y.U.L.Rev. 504 (1965)), supra, at 522. Moreover, exposure of potential jurors to news accounts before trial need not result in an aborted proceeding, since the problem can be cured by a continuance or change of venue. If the exposure occurs during the trial, however, the trial judge must squarely face the question of whether a fair trial is still possible. Consequently, a stricter standard should be employed in during-trial cases than in pre-trial situations.

Id. at 468.

This court in United States v. Williams, 604 F.2d 1102, 1115 (8th Cir. 1979) found no abuse in discretion in a pretrial publicity case but qualified its holding by suggesting that "the exposure of the jury to the same newspaper article during the trial of those individuals named in the article would necessarily present a different question." (Emphasis supplied). There the news article reported that the infant son of a confidential informant in an unrelated narcotics case had been killed and the mother of the child had been wounded in a shooting which police believed to have been in retaliation for the informant's cooperation. We held that the newspaper article was not "inherently prejudicial" to appellant in the case even though an informant in the case had testified that one of the defendants had threatened him and his family. The qualification is support for our holding that news stories involving parties intimately connected with a case are to be more closely scrutinized for potential prejudice.

Finally, in United States v. Titsworth, supra, 422 F.Supp. at 591, the district court granted a defense motion for a new trial on the basis of the "collective impact" of prejudicial in-trial publicity. That impact was produced by newspaper articles and media broadcasts to which jurors were exposed immediately prior to the commencement of the second day of their deliberations. Those reports disclosed the disappearance of a government witness and speculation as to whether foul play was involved and revealed a relationship between a witness and the defendant which, the court felt, hampered the jury's ability to deliberate only the evidence adduced at trial. Despite assurances from each juror that his vote was not influenced by the newscast, the court concluded that " '(h)ere the content and timing of the material went beyond anything * * * the average juror could withstand. It also went beyond anything the defendant should be forced to accept.' " Id. at 592. In support of its contention that the jury was unduly influenced, the court indicated that:

the jurors had deliberated for over five hours prior to their exposure to the newscast in question, and pursuant to the Court's examination, they admitted a difference among them as to the guilt of the defendant upon their separation for the evening. (Cite omitted.)

On the next morning, subsequent to the newscast, the jurors reached a verdict within twenty minutes.

Id. at 591.

As the Supreme Court stated in Marshall v. United States, supra, 360 U.S. at 312, 79 S. Ct. at 1172, "each case (involving such publicity) must turn on its special facts." There, as here, there existed "doubts whether exposure of some of the jurors to newspaper articles about petitioner was so prejudicial in the setting of the case as to warrant the exercise of our supervisory power to order a new trial." Id. at 311, 79 S.Ct. at 1172.

Applying the stricter standard to be employed in during-trial publicity cases as set forth by this court in United States v. Williams, supra, 604 F.2d 1102 (8th Cir. 1979) and the Fifth Circuit in United States v. Williams, supra, 568 F.2d 464 (5th Cir. 1978), we conclude as did the district court in United States v. Titsworth, supra, 422 F. Supp. 587, 591, that the "collective impact" of the content and timing went "beyond anything * * * the average juror could withstand."

Appellant Richardson maintains as a separate ground for reversal the trial court's failure to exclude the testimony of Ms. Brown. He argues that her appearance after the shooting and the fact that she testified from a wheelchair unduly prejudiced and influenced the jury. While we cannot say that Brown's appearance was in itself so prejudicial that the trial court abused its discretion in admitting her testimony under Rule 403, Federal Rules of Evidence, we do conclude that her appearance in conjunction with juror exposure to the reason for her confinement to the wheelchair may have resulted in a prejudicial assessment of her credibility. The defense was largely based on attacking the credibility of this witness who had agreed to testify only upon receiving immunity from prosecution, and an assessment of her credibility by an impartial jury was critical.

Therefore, this ground merges with that of prejudicial in-trial publicity and on the basis of their cumulative effect on the jury we reverse the district court's refusal to grant a motion for mistrial.

The final issue on appeal is whether appellant Young was deprived of his right to a fair and impartial trial in that the photospread was impermissibly suggestive. Carol Blackmon testified at trial that she had identified Raymond Young from an FBI photospread as one of the individuals involved in the robbery. The photospread had been composed exclusively of photographs of individuals believed by the FBI to have been involved in the robberies. Young was the only bald individual depicted in the photospread. Blackmon also made an in-court identification of Young and testified that she had seen him on two prior occasions: first, in the car on the drive to and from the credit union prior to the robbery for approximately 30 to 40 minutes; and second, for nearly an hour when Young sat next to her while driving to her house immediately following the robbery.

Convictions based on eyewitness identification at trial following a pretrial identification by photograph will be set aside on that basis only if procedures used in the photospread were so impermissibly suggestive that they created a substantial likelihood of irreparable misidentification. Simmons v. United States, 390 U.S. 377, 384, 88 S. Ct. 967, 971, 19 L. Ed. 2d 1247 (1968); United States v. Smith, 602 F.2d 834, 837 (8th Cir.), cert. denied, 444 U.S. 902, 100 S. Ct. 215, 62 L. Ed. 2d 139 (1979). Thus the crucial inquiry is the reliability of Blackmon's identification. That reliability is to be determined from the totality of the circumstances. United States v. Smith, supra, 602 F.2d at 837. Factors to be weighed in making that determination include: 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 descriptions and the length of time between the crime and confrontation. United States v. Anderson, 618 F.2d 487, 491-92 (8th Cir. 1980). In Anderson, we upheld an identification by several witnesses who had only ten minutes to observe the defendant. There, only one photograph was shown to witnesses approximately ten days after the crime had been committed. Here Blackmon had ample opportunity to observe defendant Young. Not only was she in a car with him on two separate occasions, each of several minutes duration, he was also present in her home immediately following the robbery. Blackmon identified Young's photograph approximately three months after the robbery occurred and later again at trial. Her testimony indicates she felt certain of her identification. In view of the circumstances we cannot say that the photospread was so suggestive as to create a "substantial likelihood of irreparable misidentification." Simmons v. United States, supra, 390 U.S. at 384, 88 S. Ct. at 971; United States v. Anderson, supra, 618 F.2d at 492; United States v. Mears, 614 F.2d 1175, 1177 (8th Cir.), cert. denied, 446 U.S. 945, 100 S. Ct. 2174, 64 L. Ed. 2d 801 (1980).

Accordingly, the judgments of conviction of appellants Richardson and Young are reversed and remanded for further proceedings consistent with this opinion.