JOHN R. GIBSON, Circuit Judge.

These appeals arise from the roadside assault and death of Eddie Peltier and the resulting convictions of Richard LaFuente for first-degree murder, John Perez for second-degree murder, Jesse Cavanaugh, Loren Grey Bear and Perez for witness tampering, and Grey Bear and Leonard Fox for perjury. In our earlier panel decision, we held the evidence sufficient to sustain these convictions, but determined that the evidence was insufficient to sustain several other judgments of conviction entered against these appellants and others. See United States v. Grey Bear, 828 F.2d 1286 (8th Cir. 1987). The panel also ruled that joinder was improper, leading to a rehearing en banc in which the court, by an equally divided vote, affirmed the district court's ruling holding joinder to be proper. See United States v. Grey Bear, 863 F.2d 572 (8th Cir. 1988) (en banc). We now consider the remaining issues not yet decided by either the original panel or the court en banc. The appellants primarily argue that the district court erred in framing jury instructions, in foreclosing cross-examination of one witness, and in permitting improper closing arguments. We have carefully considered all issues raised by appellants, and affirm the remaining judgments of conviction against each of them.

I.

Appellants argue that the district court improperly charged the jury with respect to the offenses of assault resulting in serious bodily injury, 18 U.S.C. § 113(f), and murder, 18 U.S.C. § 1111. In particular, they contend that Instruction 34 was constitutionally defective in that it directed the jury to escalate a conviction on the lesser charge of assault to one of murder merely upon a showing that the assault resulted in death. The instruction stated that if an assault inflicted serious bodily injury, and the victim died as a result of the injury, the crime of assault merges with that of homicide. The jury was told that if it found such injury and resulting death occurred, it was to determine "the form of homicide, if any, that the defendant may have committed." The jury was further instructed that it could not find a defendant guilty of both homicide and assault. Appellants maintain that this instruction sidestepped the required element of criminal intent for murder. Appellants similarly argue that Instruction 37 was defective in that, with regard to "an assault which results in serious bodily injury or death," it provided that "it is not necessary for the government to prove" that a defendant intended to cause death.

Appellants failed to object to Instruction 34, and, while objection was made in passing to Instruction 37, the particular arguments now presented regarding that instruction were not presented to the district court. We therefore review any challenge to those instructions on a plain error basis. See Fed.R.Crim.P. 30; United States v. Martin, 751 F.2d 258, 261 (8th Cir. 1984); United States v. Hecht, 705 F.2d 976, 978-79 (8th Cir. 1983); United States v. Parisien, 574 F.2d 974, 976 (8th Cir.), cert. denied, 439 U.S. 850, 99 S.Ct. 154, 58 L.Ed.2d 154 (1978). Moreover, if there were plain error, it affected only Richard LaFuente, who was convicted of first-degree murder, and John Perez, who was convicted of second-degree murder. The error, if any, had no prejudicial effect with respect to the convictions of Cavanaugh, Grey Bear and Perez for witness tampering or Grey Bear or Fox for perjury. Finally, jury instructions must be viewed as a whole, and we will not find error in a portion of one instruction without considering the instructions in their entirety. United States v. Kabat, 797 F.2d 580, 588 (8th Cir. 1986), cert. denied, 481 U.S. 1030, 107 S.Ct. 1958, 95 L.Ed.2d 530 (1987).

Here, the role of the challenged instructions is apparent. Instruction 34 dealt with the interaction between charges of assault and homicide, and with respect to a conviction on homicide required the jury to refer to other instructions "heretofor [sic] given pertaining to first degree murder and the lesser included offenses of second degree murder and voluntary manslaughter." Appellants do not submit that these other instructions erred in defining the various degrees of homicide, and we observe that in each instance these instructions expressly required the jury to find an appropriate level of intent.

The purpose of Instruction 34 was to inform the jury that it could not return a verdict finding the same defendant guilty of both assault and homicide. The instruction's specific reference to other instructions on assault and homicide directed the jury to requirements in those instructions that it find intent in order to convict. That every paragraph of Instruction 34 did not refer to intent, therefore, does not render it plainly erroneous. And, contrary to appellants' argument, the instruction did not require the jury to find each of the defendants guilty of some form of murder; it stated that where an assault results in death, the jury must determine the form of homicide, "if any," that may have been committed. This clearly allowed the jury to find a defendant guilty of murder, of manslaughter, or of nothing at all. The instruction concluded that the jury may find defendants guilty of homicide, of assault, "or for that matter not guilty of any offense" if that result was consistent with the facts "and these instructions." In guiding the deliberations of the jury, the instruction thus simply informed it of the available choices.

Instruction 37 did not refer specifically to other instructions. In defining the offense of assault resulting in serious bodily injury or death, however, it expressly required the jury to find that a defendant was a participant and not simply a knowing spectator. The instruction accurately reflects the case law interpreting 18 U.S.C. § 113(f), which appellants were alleged to have violated under Count Two of their indictment. See United States v. Knife, 592 F.2d 472, 483 (8th Cir. 1979) (no specific intent to cause serious bodily injury need be proven); United States v. Eagle, 586 F.2d 1193, 1196 (8th Cir. 1978). We need not determine that the instruction is a model to ascertain that it is free from plain error. Here, Instruction 37 placed sufficient limits on the jury's deliberations with regard to the charges of assault.

Although these two instructions perhaps could have been more carefully drafted, the factual circumstances of this case made their submission understandable. The evidence established that a series of confrontations occurred between Peltier and a group of individuals. On two occasions, Peltier broke away, only to be pursued and caught. Testimony identified various appellants as participating in some of these confrontations, but not in others. The assault was ongoing and took place over some distance. Under the doctrine of merger, however, an individual once found guilty of murder could not also be separately punished for assault for the same acts. See generally 40 C.J.S. Homicide § 20 (1944 Supp. 1989) (citing cases). While the nature of these events therefore justified charging all defendants with both assault resulting in serious bodily injury and homicide, it was apparent that some could be found guilty of assault while others could be convicted on the more serious charges of murder or manslaughter. Explicit instruction concerning the relationship between these charges undoubtedly was helpful to the jury, and we conclude there is no plain error in the submission of Instructions 34 and 37.

II.

Appellants also urge error in the district court's failure to adopt their tendered instruction on eyewitness identification. Although the jury was instructed as to issues of credibility, appellants contend that the district court's charge was inadequate in failing to conform with the instruction on eyewitness identification established in United States v. Telfaire, 469 F.2d 552, 558-59 (D.C. Cir. 1972), as it is reversible error to refuse such an instruction where the government's case rests solely on questionable eyewitness identification. United States v. Mays, 822 F.2d 793, 798 (8th Cir. 1987). In particular, appellants assert that Instruction 54 was defective in that it did not expressly command the jury to consider the condition of each witness while allegedly observing Eddie Peltier's killing. They stress that one witness, Fred Peltier, had consumed fifteen cans of beer and smoked two or three joints of marijuana on the day of Peltier's death, and that other witnesses similarly suffered from diminished capacity from drinking alcohol or smoking marijuana.

The jury, however, was specifically directed to consider the ability of each witness to observe matters on which the witness had testified. It was told to question whether each witness possessed an accurate recollection of these matters. It was further told that, in evaluating identification testimony, it should consider "whether the witness had an adequate opportunity to observe the person in question at the time or times about which the witness testified," and also "such matters as the length of time the witness had to observe the person in question, the prevailing conditions at that time in terms of visibility or distance and the like, and whether the witness had known or observed the person at earlier times."

This instruction adequately pointed out the relevant considerations to be weighed in gauging eyewitness testimony. The jury was instructed to consider each witness's intelligence, motive, memory, and state of mind, along with the conditions prevailing at the time of observation. With regard to the drunkenness of some witnesses, the district court appropriately instructed the jury to consider each witness's ability to observe matters testified to, as well as his opportunity to observe the persons in question. We are satisfied that Instruction 54 afforded defense counsel ample latitude to argue that the observational capacity of certain witnesses was impaired by their drunken state. The instruction given properly guided the jury's deliberation, and it was not error to refuse the tendered instruction.

III.

Appellants next assert that they were denied their sixth amendment right to confrontation because the district court prohibited defense counsel from cross-examining witness Billy Fox, and that the government's use of Fox at trial was an abuse of Fed.R.Evid. 607. Fox had originally testified before the grand jury that he knew nothing about the suspected murder. After other witnesses came forward, however, and Fox was indicted for perjury, he gave a statement to the FBI in which he indicated that he was present at the party which took place at the residence of Bernice Juarez, and later that night watched Peltier being beaten and run over on the highway. Fox listed the names of the individuals he saw at the party and those participating in the beating of Peltier. Fox later recanted the statement.

During trial, the government obtained immunity for Fox. When called to the stand by the government and questioned regarding his whereabouts on the night Peltier was killed, Fox answered that he was at home. In order to impeach Fox, the prosecution then read to Fox excerpts from his earlier statements to the FBI, but omitted the names of the individuals set forth in the statement. While he admitted making these statements, Fox persisted in maintaining that he was not present at the site of Peltier's death. At no point did appellants object either to the calling or to the impeachment of Fox.

Following Fox's direct examination, defense counsel began cross-examining him, attempting to explore Fox's reasons for changing testimony. The government quickly objected, arguing that the line of questioning exceeded the proper scope of cross-examination. Defense counsel responded with an offer of proof concerning the content of Fox's proposed testimony. The district court nevertheless sustained the government's objection to any cross-examination of Fox on grounds that Fox's testimony to that point, after impeachment, stood as a nullity. The district court then instructed the jury that it was not to consider Fox's prior inconsistent statements as substantive evidence against the defendants.

Although appellants later called Fox as their own witness, they suggest that they were compelled to do so in order to permit him to explain his motive for providing the statement to the FBI. Appellants argue that this did not cure the sixth amendment violation resulting from their inability to cross-examine Fox at the time he gave his initial testimony. They further contend that by instructing the jury to disregard Fox's testimony as "no evidence at all," the district court discredited a potential witness on their behalf.

In essence, appellants maintain that the government improperly called Fox for the sole purpose of presenting inadmissible evidence under the guise of impeachment. See United States v. Fay, 668 F.2d 375, 379 (8th Cir. 1981) (impeachment may not be used as a subterfuge to place otherwise inadmissible hearsay before the jury). In support of their motion for a new trial, appellants submitted an affidavit in which Fox's attorney, Joel Arnason, averred that he had notified the government before trial that Fox had renounced his statement to the FBI. When Arnason, Fox, and an assistant district attorney later met, according to the affidavit, neither Fox nor Arnason indicated that Fox, would in the future claim anything other than a complete lack of knowledge regarding Peltier's death. The assistant district attorney indicated at this meeting that Fox's testimony could not harm the government because the prosecution would impeach Fox with his prior statements to the FBI were Fox to deny knowledge of Peltier's killing. This, in appellants' view, establishes that the government called Fox as a witness merely to impeach him with his otherwise inadmissible statements.

In ruling on appellants' motion for a new trial, the district court determined that "the prosecution appears to have exercised good faith in impeaching Billy Fox." The court reasoned that although the prosecution was aware that Fox had recanted his statement to the FBI, the government's attorneys could not be certain as to how Fox would testify at trial, when he was under oath and after he had been granted immunity. The court found that the government in impeaching Fox had demonstrated its good faith by omitting all references to the names of defendants and other individuals from the impeaching statements. It further relied for evidence of good faith on the government's decision not to call two other witnesses, Kevin and Bruce Mindt, who like Fox had given inculpatory statements only later to recant them, and who like Fox were then charged with perjury.

A finding by a district court is entitled to great deference and may not be rejected unless clearly erroneous. Anderson v. City of Bessemer City, 470 U.S. 564, 573-75, 105 S.Ct. 1504, 84 L.Ed.2d 518 (1985); cf. United States v. Aviles, 337 F.2d 552, 559-60 (2d Cir. 1964) (finding that government exercised good faith in destroying interview notes reviewed under clearly erroneous standard). Here, the district court's finding as to prosecutorial good faith was supported by ample basis in fact. Fox first denied any knowledge of Peltier's death. After others later identified him at the scene and after he was then indicted, Fox gave full statements to the FBI incriminating several individuals. This statement was later formally recanted. Where Fox so waivered, first claiming no knowledge, later providing a narrative in detail, and then recanting it, the government reasonably suggests that it was unsure as to what Fox would say when tested under oath, especially under a grant of immunity after perjury charges were pending against him. The circumstances surrounding Fox's testimony establish that the government did not call Fox as a subterfuge to place otherwise inadmissible hearsay before the jury. See United States v. Rogers, 549 F.2d 490, 497 (8th Cir. 1976), cert. denied, 431 U.S. 918, 97 S.Ct. 2182, 53 L.Ed.2d 229 (1977). Although the government at one point told Fox that he would be impeached with the FBI statements if he persisted in his denials, this does not establish that the prosecution felt certain of the content of Fox's prospective testimony. Instead, this simply represents one lawful attempt by the government to induce Fox to testify truthfully.

We reject the claims of error.

IV.

Appellants further claim that the district court allowed prosecutorial misconduct in that the government pressed upon the jurors a "guilt by association" trial strategy. During closing argument, a prosecutor stated that the defendants "rehearsed in lockstep, and they testified the same way." The government also maintained that, if the jury found that a party at the Juarez residence had indeed occurred, the jury must find that each defendant was culpable. A prosecutor stated at one point that "if one is guilty, it's a pretty fair conclusion that all are guilty. That's the choice they made when they testified on the witness stand." The government concluded its case by telling the jurors that this was an "all or nothing case, all guilty or all not guilty."

Although stripped from context these statements seem troublesome, we discover no reversible error here. We observe, first, that no contemporaneous objection was made to any of these statements. We therefore review the prosecution's closing arguments under the plain error standard. See United States v. Young, 470 U.S. 1, 14-16, 105 S.Ct. 1038, 1045-1047, 84 L.Ed.2d 1 (1985). Here, we cannot conclude that the district court committed plain error in permitting the prosecution to argue as it did. All of the appellants testified in attempting to establish alibis. The predominant theory of the defense was that there had been no party at the Juarez residence, no fight alongside the highway, and no murder of Peltier. Against such a defense, the government was entitled to argue that the testimony of defense witnesses left no middle ground and that, if the party took place, which each defendant denied, an inference arose that each was culpable to some extent. The prosecution's arguments were properly tailored to the record before the jury, and argued permissible inferences from the evidence.

The defense adopted the position that if the government failed to prove that all defendants were guilty, the case for the prosecution collapsed. The government's closing argument merely responded to this argument, and it made plain that if the evidence was sufficient to convict some defendants but not others, the jury was obligated to find those latter defendants not guilty. It is well established that "prejudicial error does not result from the improper remarks made during closing argument when such remarks were provoked by opposing counsel." United States v. Schwartz, 655 F.2d 140, 142 (8th Cir. 1981) (per curiam) (citing Isaacs v. United States, 301 F.2d 706, 738 (8th Cir.), cert. denied, 371 U.S. 818, 83 S.Ct. 32, 9 L.Ed.2d 58 (1962)). Much evidence suggested that Eddie Peltier's death was brought about by mob action, and the government presented testimony that each of the defendants participated in the beating. The appellants' evidence similarly indicated group action of a peculiar sort — that none were present at the Juarez residence and none knew about Peltier's death. The instructions given by the court admonished the jury that it was to consider individually the evidence against each defendant and to acquit those for whom there was insufficient evidence of guilt. In light of this record we cannot conclude that plain error was committed.

Appellants also argue that the government improperly vouched for the credibility of its witnesses. At one point in closing argument, the prosecution asked, "Don't you think over the years of experience we have picked up a little bit of sense of what rings true in a criminal investigation and what rings true in a case?" In commenting upon one witness, the prosecution further argued that "she and others who testified are, from the perspective of the government, the most believable and I will tell you why."

We are satisfied that by these statements the prosecution did not improperly vouch for the testimony of its witnesses. Nothing in the record indicates that personal reputations of the government attorneys were placed behind the testimony of their witnesses. Indeed, prosecutors, as well as defense lawyers, may and must argue as to the credibility of witnesses, and in a case of this kind the issue of credibility is critical. The very nature of closing argument requires a detailed analysis of the testimony of each witness and the inferences to be drawn from the evidence. We are satisfied that the government's arguments were within permissible limits and did not imply that the government possessed knowledge concerning facts not in evidence. See United States v. Eley, 723 F.2d 1522, 1526 (11th Cir. 1984). We cannot conclude that there was plain error in allowing these closing arguments.

V.

Appellants also attack their convictions in several ways which merit no more than summary consideration. They first contend that the district court erred in failing to permit their expert witness, Dr. Jerry Baldwin, to state that the opinions of two other pathologists agreed with his own. We are persuaded that Fed.R.Evid. 703 does not permit an expert witness to circumvent the rules of hearsay by testifying that other experts, not present in the courtroom, corroborate his views.

Claiming that the jury was possibly tainted by pre-trial publicity, appellants also argue that the district court erred in denying their motion for change of venue or, in the alternative, for individually sequestered voir dire of prospective jurors. Appellants present no evidence of actual juror prejudice, however, and absent such evidence the court was not required to grant their motion. See generally United States v. Haldeman, 559 F.2d 31, 68-69 (D.C. Cir. 1976), cert. denied, 431 U.S. 933, 97 S.Ct. 2641, 53 L.Ed.2d 250 (1977).

Because the courthouse located within the venue of appellants' charged offenses, the Northwestern Division of the District of North Dakota, lacked adequately sized courtrooms, appellants' trial was held in the Southeastern Division. Appellants argue that the district court erred in drawing the jury panel from the Southeastern Division rather than the Northeastern because in so doing the court excluded Native Americans from the pool of prospective jurors, violating the sixth amendment and the Jury Selection and Service Act of 1968, 28 U.S.C. § 1861 et seq. Although appellants do present some conclusory allegations in this regard, we conclude that they have failed to present facts sufficient to establish that Native Americans were systematically excluded from the jury selection process. See United States v. Turcotte, 558 F.2d 893, 895 (8th Cir. 1977).

Fourth, appellant LaFluente argues that the district court should have permitted broader redirect examination of defense witness Colleen Dunn. We are satisfied that the matter which appellants wished to explore was collateral, and the record reveals that the district court did not abuse its discretion in this regard.

LaFuente argues, fifth, that the district court erred in requiring defendants to turn over to the prosecution a "memorandum of interview" taken from defense witness Kevin Brownshield. Appellant contends that the memorandum did not constitute a "statement" under Fed.R.Crim.P. 26.2(f)(1). A review of the record discloses substantial support for the court's finding that Brownshield's statement was a "written statement made by the witness that is * * * adopted or approved by the witness." The district court did not err in ordering disclosure of the statement.

Sixth, appellants argue that the district court erred in failing to allow access to the medical records of witness Patricia DeMarce, which were inspected by the court in camera. Appellants have demonstrated no abuse of discretion. Indeed, to the extent these records may have been relevant, both direct and cross-examination brought out the fact that DeMarce delivered a baby several weeks prior to Peltier's murder.

Finally, appellants all argue that the district court erred in allowing the prosecution to ask leading questions during its direct examination of DeMarce, and that this, in allowing government attorneys to testify in place of DeMarce herself, violated appellants' sixth amendment right of confrontation. The record reveals that leading questions were used infrequently and judiciously in order to develop testimony given by an unusually softspoken and frightened witness. Such questioning is not improper. See Fed.R.Evid. 611(c) (permitting use of leading questions on preliminary matters essential to the development of testimony).

Accordingly, we affirm the judgments of conviction.

EYEWITNESS IDENTIFICATIONS

You heard witnesses identify the defendants as being the same persons who were at the scene of the incident. You must decide whether to believe this identification and how much weight to give it. There are a number of factors that may have an effect on the reliability of identifications. You should consider the following:

1. The conditions under which the identification was made. Lighting, distractions, weather, and other things may affect a witness' ability to observe. The distance of the witness from the events observed and the length of time that the witness observed events may affect an identification.

2. The condition of the witness at the time of any observation. Eyesight, hearing and other senses of people may vary. The witness' state of mind — whether one of excitement, fear, or emotional disturbance — may affect an identification. The witness' activity — whether the witness was paying attention to events — may also be important.

3. The relationship of the witness to the person identified. It might be important how long and how well a witness knew a person before making an identification.

4. The consistency and certainty of the identification. Initial descriptions by a witness that do not match a defendant's, or a witness' initial inability to give a description immediately after seeing an event, may detract from the reliability of a later identification.

5. The length of time between an observation and an identification. Memories may fade as time passes.

6. The nature of the identification procedure. It is more significant if a witness is able to identify a person from among a group of similar people than to identify a person who is viewed alone. The reliability of the first identification a witness makes is the most important because subsequent identifications might be affected by the first one. Suggestions made to a witness may affect the reliability of the identification.

You may find additional factors useful when you evaluate eyewitness testimony. It is for you to decide whether the identification is reliable and what weight to give it.