We have no quarrel with the panel's articulation of the standard for deciding whether the jury's exposure to documents not in evidence requires a new trial; it is whether there is a "reasonable possibility" that the documents may have affected the verdict. 662 F.2d at 457; see, e.g., United States v. Dressler, 112 F.2d 972, 978 (7th Cir. 1940); Llewellyn v. Stynchcombe, 609 F.2d 194 (5th Cir. 1980). We part company with the panel only on the question of the scope of appellate review. The district judge articulated the correct standard for deciding the issue of jury prejudice, and the appellants' argument is only that he applied it to the facts incorrectly. He stated that "once it is shown that the jurors were exposed to extraneous materials it becomes incumbent upon the Court to determine whether the error may have operated to the substantial injury of the defendants" (emphasis added), citing among other cases Dressler, supra ; and this is equivalent to the "reasonable possibility" standard articulated in Dressler and in the panel's opinion in this case, and to the "harmless error beyond a reasonable doubt" approach often used interchangeably with the "reasonable possibility" standard. See, e.g., Llewellyn, supra, 609 F.2d at 196; United States v. Ackerman, 393 F.2d 121, 123 (7th Cir. 1968). In any event the appellants do not contend that the judge used an improper standard.

The proper standard of appellate review of the district court's determination of prejudice is different, however; it is "abuse of discretion." The panel stated the issue before it, correctly, as "whether the district court abused its discretion in denying defendant's motion for a new trial," 662 F.2d at 457; but then it conducted a de novo inquiry into whether the documents were prejudicial in the circumstances, and we cannot agree with this approach, though we acknowledge a paucity of relevant decisions. The foremost case is still Holt v. United States, 218 U.S. 245, 250, 31 S. Ct. 2, 5, 54 L. Ed. 1021 (1910), an opinion by Justice Holmes which states that the denial of a motion for a new trial based on the fact that jurors read newspaper articles about the trial while it was going on "cannot be treated as more than matter of discretion or as ground for reversal, except in very plain circumstances indeed." Holt was followed and its rule paraphrased in Marshall v. United States, 360 U.S. 310, 312, 79 S. Ct. 1171, 1172, 3 L. Ed. 2d 1250 (1959) (per curiam ): "The trial judge has a large discretion in ruling on the issue of prejudice resulting from the reading by jurors of news articles concerning the trial." At oral argument the appellants' counsel acknowledged that they knew of no authority contrary to Holt, and our independent research has not discovered any either.

Despite the authoritative pronouncements of the Supreme Court there has been a tendency, criticized in Judge (now Justice) Stevens' dissenting opinion in United States v. Thomas, 463 F.2d 1061, 1065-66 (7th Cir. 1972), to review the district courts' determinations of jury prejudice under a much more liberal standard than abuse of discretion. But we have decided that we must adhere to the abuse of discretion standard. Not only is it binding on us because there is no indication that the Supreme Court is prepared to overrule Holt and Marshall ; it makes good sense to us as an original matter. Because the judge is prohibited from questioning the jurors, after they have given their verdict, to determine whether their deliberations were in fact prejudiced by the introduction of documents not in evidence, Fed.R.Evid. 606(b), the inquiry into prejudice becomes a matter of assessing the probabilities that a particular jury was prejudiced by particular documents. The trial judge will always be in a better position than the appellate judges to assess the probable reactions of jurors in a case over which he has presided. He has had the opportunity to observe the jurors' demeanor and gauge their attentiveness, as the appellate judges have not. He can also judge, as we cannot, whether the atmosphere of the trial-that congeries of intangibles that no stenographic transcript can convey-might have made the jury receptive to being prejudiced by the documents in question, or, for that matter, might have inoculated it against such prejudice. As we cannot put ourselves in the district judge's shoes in these matters we ought to accept his judgment unless we have a very strong conviction of error.

The district judge dealt in detail with the issue of jury prejudice in his opinion denying the appellants' post-trial motions, thus exercising his discretion-and using the proper standard, too, as we have already pointed out. So we have only to decide whether he abused his discretion in ruling against the defendants. We begin with the Bureau of Prisons report, with its accusation that Bruscino was a member of the Mexican Mafia, whatever exactly that is, and its statement that an investigation had failed to substantiate the accusation. We do not agree with the government that the second statement made the net effect of the report exculpatory, for even if the average reader would conclude that Bruscino was not a member of the Mexican Mafia this would not show that Bruscino had not killed Martinez: not all killers are Mexican Mafiosi. But there is an additional point to be considered. Bruscino's counsel at trial used the Bureau of Prisons report (without introducing it into evidence) in an unsuccessful attempt to show, through cross-examination of one of the prosecution's witnesses, that Bruscino had never wanted to be at Terre Haute, which if true would undermine the prosecution's theory that Bruscino had been out to "get" Martinez. In fact the report makes clear that Bruscino never wanted to be transferred from McNeil to Terre Haute, and to that extent the report should actually have helped Bruscino and his partner Kell with the jury.

This point to one side, we cannot say that the district judge abused his discretion in concluding that the report was highly unlikely to have prejudiced the jury against Bruscino and Kell. Nothing in the testimony at trial connected Bruscino to any Mexican Mafia; a pretrial stipulation, not alleged to have been violated, forbade mention at trial of any relationship between Bruscino and the Mexican Mafia. We cannot agree with the appellants that the connection insinuated by the report supplied a motive for the murder, for there was no indication either in the report or in the evidence at trial of what the Mexican Mafia is or does or how the other conspirators or Martinez, the victim, might have been connected with it; and, as we have noted, the report makes clear that Bruscino had not wanted to be transferred to Terre Haute, either to clean out the "rats" there or for any other reason. No doubt just calling a man a suspected member of the Mexican Mafia invests him with a sinister character even if you add that you have not been able to substantiate your suspicion; but this is a prison killing; the defendants were all inmates, as the jury of course well knew. Conceivably the fact that Bruscino, besides being an inmate, had been suspected-and provisionally exonerated-of being a member of the Mexican Mafia, an organization nowhere described or linked to the murder or the victim, may have caused the jury to think him even more likely than was otherwise the case to have murdered a fellow inmate; but the district judge thought not, and we cannot say that his conclusion, based as it was on the Bureau of Prisons report as a whole, read by the judge in light of his first-hand knowledge of the case and of the jury, was so unreasonable as to amount to an abuse of discretion.

We turn now to the newspaper article. In reporting that the appellants had been indicted along with Barron, Howell, and Norman, the article told the jury nothing it did not know already from the reading of the indictment. The jury also knew that Howell had pleaded guilty, as this was brought out (without objection) in his direct examination. The only new information in the article was that Barron and Norman also had pleaded guilty. This could hardly have come as a surprise to the jury, since Howell's testimony had linked Barron and Norman to the crime and no witness had tried to exculpate them. More important, we do not see how Bruscino and Kell could have been harmed, when their theory of the murder was that either Barron or Norman had done it-a theory that was, if anything, supported by the guilty pleas. The article was drably descriptive; it was not inflammatory. The district court did not abuse its discretion in concluding that it was highly unlikely that the jury had been prejudiced either by the article or by the cumulative impact of the two documents, since the article seems not to have been prejudicial at all.

The panel rejected the defendants' other contentions; so do we, relying on the panel's discussion. The judgments of conviction are therefore

Affirmed.

I cannot accept the conclusion of the majority that the district court in this case appropriately exercised its discretion in determining that the presence, in the jury room, of these extraneous and improper documents could not have affected the jury verdict. The interest of the jury here in the "Mexican Mafia" report was so intense that it asked the trial judge for another opportunity to examine the prejudicial document in the course of its deliberations. Whatever criteria of review may be invoked, it seems to me to depart wholly from reality to suggest that there was not a "reasonable possibility" that this wayward document influenced the verdict.

As the original panel opinion in this case noted, the prior decisions of this circuit strongly indicate that, in reviewing a district court's determination that no prejudice resulted from a jury's exposure to materials not properly within its consideration, an appellate court is entitled, and may indeed be required, to examine for itself the question whether there was a "reasonable possibility" that the extraneous material affected the jury's verdict. See, e.g., United States v. Thomas, 463 F.2d 1061, 1065 (7th Cir. 1972); United States v. Grady, 185 F.2d 273, 275 (7th Cir. 1950); United States v. Dressler, 112 F.2d 972, 978 (7th Cir. 1940). See also Llewellyn v. Stynchcombe, 609 F.2d 194, 195 (5th Cir. 1980); United States v. Vasquez, 597 F.2d 192, 193 & n.1 (9th Cir. 1979) (citing cases from other circuits). The majority's failure to distinguish, or even deal with, these cases in its discussion of the "paucity of relevant decisions," ante at 940, is puzzling.

Equally disturbing is the majority's reliance on Marshall v. United States, 360 U.S. 310, 79 S. Ct. 1171, 3 L. Ed. 2d 1250 (1959). In Marshall, the Supreme Court reversed, per curiam, the "no prejudice" determinations of both a trial court and a court of appeals, and ruled that the possibility of prejudice resulting from several jurors' exposure to newspaper articles about the defendant necessitated the grant of a new trial. In my view, both the analysis and the result reached in Marshall support the panel's original decision in this case.1

I, therefore, incorporate by reference in my dissent the panel opinion2 (which I originally joined) authored by Senior Circuit Judge Fairchild, formerly Chief Judge of this Circuit, which meticulously and with a salutory regard for the ordinary requirements of a fair trial, reversed these convictions. Senior Circuit Judge Gibson dissented from that opinion, in large part, on the ground that defense counsel was in some way responsible for the submission of the "Mexican Mafia" document to the jury-a theory on which the majority here does not rely.

ESCHBACH, Circuit Judge, dissenting.

Notwithstanding the well reasoned majority opinion, I respectfully dissent.

In my view there was more than a "reasonable possibility" that the so-called "Mexican Mafia" report influenced the jury. The suspicion, planted in the jurors' minds early in the trial, that defendant Bruscino was a member of something called the "Mafia," a word which connotes violent, sinister, conspiratorial, and habitual criminal activity, quite possibly colored their perception of all the evidence which followed. The taint caused by such an accusation is difficult to dispel, and the fact that the report stated that authorities had been unable to find evidence of a substantial nature to verify the charge may have seemed typical to the jurors. The risk is high that the jury believed that individuals are often publicly known as "Mafiosi," but authorities are unable to substantiate the charge because people are afraid to talk.

Through the newspaper article, at least some of the jurors learned that all of the other alleged conspirators had pleaded guilty. This was especially damaging to Kell. The jurors may have reasoned that if the government was correct in their identification of all those other conspirators, as evidenced by their guilty pleas, the government was probably right about Kell as well. The repeated admonitions of the trial judge to disregard press accounts were not heeded. If active disregard of such instructions, the importance of which is attested to by the frequency with which they are given, does not constitute grounds for a mistrial, the rule against such exposure has little force.

While the damage caused by the documents at issue could possibly have been alleviated by strongly worded curative instructions, there of course was no practical opportunity for such instructions in this case. Having served as a federal trial judge for twenty years, it is impossible for me to conclude under the circumstances of this case that the documents which improperly found their way into the jury room did not influence the verdict. While I do not fault the able trial judge who made every effort to provide a fair trial, I am forced to conclude that it was an abuse of discretion to deny Bruscino and Kell a new trial. Exactly what occurred which resulted in the documents entering the jury room and being considered by the jury may never be known. However, the crucial issue to me is the existence of a reasonable possibility that the material influenced the verdict.