OPINION OF THE COURT

VAN DUSEN, Circuit Judge

Defendants-appellants contend that the District Court of the Virgin Islands, Division of St. Croix, erred in denying their motion for a new trial. 1 Rejecting this contention, we affirm the district court.

On August 13, 1973, defendants were found guilty of first degree murder, first degree assault, and robbery. 2 The *218 jury which returned the guilty verdicts had deliberated for nine days. 3 The jurors were polled individually and . each acknowledge the verdict as his own. Two days later, defendants filed a motion requesting a new trial on the ground that the verdict had not been freely assented to by all the jurors. The motion was supported by the affidavits of two jurors, Rodgers and Allick, who represented that their verdict was the result of certain “pressures,” 4 but the trial judge found that these affidavits were “involuntarily made out of fear” engendered by the President of the Virgin Islands Senate (formerly related by marriage to Allick and “friendly with all of the defendants, save Raphael Joseph” — p. 24 of opinion cited at note 1), Mario Moorhead and John Ross, who were “sympathizers of the defendants.” See X-B at page 8 below, and page 26 of Memorandum Opinion cited at note 1 above. These same *219 two affidavits were presented to Chief District Judge Almeric Christian, who ordered a post-trial hearing to. inquire into the allegations that the verdict was influenced by “unauthorized communications with the jury during its sequestration and deliberation.” Report of the Special Master, Crim. No. 97/1972 at 1. Since neither the trial judge nor Chief Judge Christian was available to conduct the hearing, 5 the parties consented to the appointment of a special master. 6 Upon conclusion of the hearing, 7 the master submitted to the trial judge a report containing both findings of fact and conclusions of law. On September 24, 1973, the trial judge, determining that the special master’s findings were not clearly erroneous, entered an order denying the motion for a new trial. The September 24, 1973, order was appealed, along with the judgments of conviction. Government of the Virgin Islands v. Gereau, et al., 502 F.2d 914 (3d Cir. 1974). This court then vacated the order denying a new trial and remanded so that the trial judge could review the record of the hearing de novo “and enter findings without reliance on those made by” the special master. Id. at 937. Pursuant to the mandate of this court, the trial judge proceeded to make his own findings from the record compiled at the hearing before the special *220 master. No additional hearing was had. 8 On the basis of his de novo review, the trial judge entered an opinion and order which set forth his factual findings and again denied defendants’ motion for a new trial. This appeal followed.

Defendants press two arguments in seeking to persuade us that the trial court’s refusal to grant a new trial was an abuse of “its sound discretion.” 9 First they attack certain of the trial judge’s fact findings as unsupported by the evidence adduced before .the special master. 10 Second they contend that the Government had the burden of proving that none of the incidents which formed the basis of the new trial motion was prejudicial to the defendants. This burden, they claim, was not sustained.

I. Findings of the Trial Judge

A. Standard of Review

Where, as here, the findings of the trial court are based *221 on non-demeanor evidence, there is a diversity of views as to the proper scope of appellate review. Some adopt the position that the “clearly erroneous” standard of F. R. Civ, P. 52(a) should apply to all findings of fact; others espouse the practice of de novo review on the theory that the trial court has no advantage over the appeals court in assessing inanimate evidence. Compare 9 C. Wright & A. Miller, Federal Practice and Procedure, §§ 2585-2587 (1971), with 5A J. Moore, Federal Practice, ¶ 52.04 (2d ed. 1974). Our circuit has not, however, taken either of these approaches.

Some of the past decisions of this court have reviewed cases tried solely on papers, without oral testimony, differently, depending on whether the facts set forth in the papers are stipulated or disputed. In the former circumstance, it was recognized that “the Court of Appeals may, within certain limits, substitute its factual conclusions and inferences for those of the” trial court. Demirjian v. C.I.R., 457 F.2d 1, 4 (3d Cir. 1972). See also Consolidated Sun Ray, Inc. v. Lea, 401 F.2d 650, 659 N. 34 (3d Cir. 1968), cert. denied, 393 U.S. 1050 (1969). In the latter situation, the court has scrutinized facts not dependent on demeanor evidence according to the clearly erroneous standard. United States v. United Steelworkers of America, 271 F.2d 676, 685 & 688 (3d Cir.), aff’d, 361 U.S. 39 (1959). 11

The present case is sufficiently distinct from both Demirjian and United Steelworkers that neither case is *222 necessarily, controlling. 12 Cf. Orient Mid-East Lines, Inc. v. A Shipment of Rice, 496 F.2d 1032, 1037-38 and n. 8 (5th Cir. 1974); Gulf Shores Leasing Corp. v. Avis Rent-A-Car System Inc., 441 F.2d 1385, 1388 n. 1 (5th Cir. 1971). On the basis of Universal Athletic Sales Co. v. Salkeld, 511 F.2d 904, 907 (3d Cir. 1975), and Congoleum Industries, Inc. v. Armstrong Cork Co., 510 F.2d 334, 336 at n. 3 (3d Cir. 1975), which are our two most recent decisions on this issue, we affirm all the challenged findings of the trial judge except two 13 on the ground that we are “in as good a *223 position to determine the question as is the district court.” 13a

B. Affidavits pf Allick and Rodgers

After careful review of the record, we have concluded that these findings of the trial judge are fully supported by the evidence:

“It is my findings that jurors Allick and Rodgers were coerced by Messrs. Ross and Moorhead into making the affidavits, that the affidavits were involuntarily made out of fear by which both Allick ánd Rodgers were possessed.”

See trial judge at 26 and notes 4 and 18 above.

In evaluating the testimony of Allick, the trial judge made these comments, inter alia, concerning pressures on Allick to change his verdict and his fears about returning to the community at the end of the trial:

(1) “. . . I give considerable importance Allick’s post-verdict contacts with friends of the defendants. I find, as will be more fully discussed below, that Allick was'subjected to pressures to change his verdict. His testimony must therefore be scrutinized, and I give *224 considerable weight to his recently acquired interest to have the verdict changed.”

(trial judge at 17)

(2) ... I find that Allick also had fears about returning to the community and particularly to his friends in Frederiksted.” 13b (trial judge at 20)

*225 With reference to Rodgers’ credibility, the trial judge stated:

(1) “Throughout the hearing for a new trial, juror Rodgers displayed a poor power of recollection, and much of his testimony was either vague, incomprehensible, or inconsistent. For that reason plus the suspicious circumstances surrounding the taking of his affidavit (to be discussed later) I find him less than credible.”

(trial judge at 11)

(2) “It is my finding that juror Rodgers did not understand what the affidavit was all about at the time he signed it and that he did not understand it at the time of the hearing.”

(trial judge at 25) (footnote omitted)

C. Findings Without Adequate Support in the Record

Juror Agneta Cappin 14 testified that one of the jury attendants, Matron Foye, spoke to her about the case.

“She just asked me how everything is going and I tell her not so good. And I say two of them that don’t understand, they don’t come in yet. And she say to me she want them to hurry up so she can get to go home, that is all.”

H.T. 152.

Matron Foye denied the conversation. H.T. 299. The trial judge, finding both women to be credible witnesses, chose to believe Foye rather than Cappin because he knew that Foye “was grateful for the opportunity to earn extra income as a jury matron.” Trial judge at 19. We do not consider these credibility findings to lack adequate support in the record. However, we do hold that the trial judge’s reliance on his personal, subjective belief about the needs and motives of Matron Foye was an improper ground for rejecting Cappin’s concededly credible testimony.

*226 In basing his fact-finding on personal knowledge, the trial judge was, in effect, taking judicial notice of extra-record, adjudicative facts. See generally K. Davis, An Approach to Problems of Evidence in the Administrative Process, 55 Harv. L. Rev. 364, 404-07 (1942). “With respect to judicial notice of adjudicative facts,, the tradition has been one of caution in requiring that the matter be beyond reasonable controversy.” Advisory Committee’s Notes to F.R.E. (Federal Rules of Evidence) Rule 201(b); cf. F.R.E. Rule 201(a) and (b). A second hallmark of facts properly the subject of judicial notice is that they be either matters of common knowledge or “capable of immediate and accurate determination by resort to easily accessible sources of indisputable accuracy . . . .” Weaver v. United States, 298 F.2d 496, 498 (5th Cir. 1962); 9 Wright & Miller, supra, § 2410 at 339. Facts possessing these characteristics are entitled to be considered by a judge without first being proved through the routine processes of introducing evidence. 15 The' necessary cachet is not, however, bestowed merely by a judge’s knowledge of a particular fact.

“There is a real but elusive line between the judge’s .personal knowledge as a private man and these matters of which he takes judicial notice as a judge. The latter does not necessarily include the former; as a judge, indeed, he may have to ignore what he knows as a man, and contrariwise.

“It is therefore plainly accepted that the judge is not to use from the bench, under the guise of judicial knowledge, that which he knows only as an individual observer outside of court. The former is in truth ‘known’ to him merely in the fictional sense that it is known and notorious to all men, and the dilemma is only the result of using the term ‘knowledge’ in two senses. Where to draw the' line *227 between knowledge by notoriety and knowledge by personal observation may sometimes be difficult, but the principle is plain.”

(Footnotes omitted.)

J. Wigmore, Evidence, § 2569 at 539-40 (3d ed. 1940). It is apparent that the trial, judge’s knowledge about Matron Foye falls, into this latter category of personal knowledge and,' therefore, does not qualify for judicial notice. It follows that the trial judge erred in rejecting Cappin’s testimony on the ground stated. ''

Similarly untenable is the trial judge’s finding that juror Torres could not have heard any rumors while she was resting in the judge’s chambers, since that finding was based solely on the judge’s personal knowledge of the soundproofing in his chambers. Trial judge at 10.

As an antidote to these errors in the fact-finding process, our inquiry into the validity of the verdict will assume both that Cappin’s testimony was accurate, see United States v. Brumbaugh, 471 F.2d 1128, 1130 (6th Cir.), cert. denied, 412 U.S. 918 (1973), and that Torres could hear discussions outside the judge’s chambers. 16

II. Impeachment of the Verdict

Any attempt to impeach a jury verdict initially encounters two evidentiary obstacles: (1) producing evidence competent to attack the verdict, and (2) establishing the existence of grounds recognized as adequate to overturn the verdict. And even where both obstacles are cleared, there must be a finding that the party seeking to impeach the verdict has suffered prejudice from the misconduct of the j ury.

A. The Evidentiary Obstacles:' Competency and Sufficiency

*228 1. Competent Testimony.

It is frequently said to be the rule that a juror may not impeach his own verdict once the jury has been discharged. 17 McDonald v. Pless, 238 U.S. 264, 267 (1915). The rule was formulated to foster several public policies: (1) discouraging harassment of jurors by losing parties eager to have the verdict set aside; (2) encouraging free and open discussion among jurors; (3) reducing incentives for jury tampering; (4) promoting verdict finality; (5) maintaining the viability of the jury as a judicial decision-making body. McDonald, supra; United States v. Dioguardi, 492 F.2d 70 (2d Cir.), cert. denied, 419 U.S. 829 (1974); Miller v. United States, 403 F.2d 77 (2d Cir. 1968). Courts applying the rule were, however, early aware that any flat prohibition against receiving such testimony contravened another public policy: that of “redressing the injury of the private litigant” where a verdict was reached by a jury not impartial. McDonald, supra at 267. In order to accommodate these conflicting policies, Miller, supra at 82, the courts, often continuing to recite the canon of “no impeachment,” evolved a more flexible rubric. This latter rule interdicted reception of a juror’s evidence only where it was offered “to show matters which essentially inhere in the verdict itself.” 18 Hyde v. United States, 225 U.S. 347, 384 (1912); *229 Domeracki v. Humble Oil & Refining Co., 443 F.2d 1245 (3d Cir.), cert. denied, 404 U.S. 883 (1971). Such matters include both “the mental process of any juror or of the jury in arriving at a verdict,” Capella v. Baumgartner, 59 F.R.D. 312, 315 (S.D. Fla. 1973), and the method by which the verdict is reached. 19 McDonald, supra; Hyde, supra; Domeracki, supra. See also Note: Impeachment of Jury Verdicts, 53 Marquette L. Rev. 258, 269 (1970). The same accommodation of policies produced the general rule’s major exception, which provides that “ ‘[a] juryman may testify to any facts bearing upon the question of the existence of any extraneous influence, although not as to how far that influence operated upon his mind.’ ” 20 Mattox v. United States, 146 U.S. 149 (1892), quoting Woodward v. Leavitt, 107 Mass. 453. “Extraneous influence” has been construed to cover publicity received and discussed in the jury room, 21 consideration by the jury of *230 evidence not admitted in court, 22 and communications or other contact between jurors and third persons, including contacts with the trial judge outside the presence of the defendant and his counsel. 23 By contrast, evidence of discussions among jurors, intimidation or harassment of one juror by another, 24 and other intra-jury influences on the verdict 25 is within the rule, , rather than , the exception, and is not competent to impeach a verdict.

2. Grounds sufficient to overturn a verdict

Those incidents of jury misconduct which may be attested by the jurors and those which furnish a basis for setting aside a verdict overlap extensively. Indeed, courts often seem to treat the two concepts interchangeably in deciding whether a verdict may be successfully impeached. See Jorgensen v. York Ice Machinery Corp., 160 F.2d 432 (2d Cir. 1947); Comment, Impeachment of Jury Verdicts, 25 U. Chi. L. Rev. 360, 383ff. (1958) [hereinafter Comment]. For this reason, it is not possible to abstract from the cases any neat and comprehensive list of grounds upon which a verdict may be overborne. It is possible, however, to categorize roughly common fact patterns *231 which , have been held to call a verdict into question. Other than incompetency of á juror during jury service, see United States v. Dioguardi, supra at 79, and Jorgensen v. York Ice Machinery Corp., 160 F.2d 432, 435 (2d Cir. 1947), these are all incidents of “extraneous influence:” 26 (1) exposure of jury to news items “about the matter pending before the jury,” Remmer v. United States, 347 U.S. 227, 229 (1954) ; 27 (2) consideration by the jury of extra-record facts about the case; 28 (3) communications between third parties and jurors where relevant to the case to be decided; 29 (4) pressures or partiality on the part of the court. 30 The first three types of jury conduct render a criminal verdict vulnerable because they are prima facie incompatible with the Sixth Amendment. See United States v. Thomas, 463 F.2d 1061 (7th Cir. 1972); United States v. McKinney, 429 F.2d 1019 (5th Cir. 1970) cert. denied, 401 U.S. 922 (1971). That Amendment guarantees, inter alia, “an impartial jury,” the right to confront witnesses, and the assistance of counsel at every critical stage of the trial. Coleman v. Alabama, 399 U.S. 1 (1970). Where jurors communicate with outsiders about the merits of the case they are deciding, their impartiality may be compro *232 mised unless the communication is tempered by “known rules of the court and .the instructions and directions of the court made during the trial . . . .” Remmer, supra at 229. And where jurors consider evidence, in the form of either fact or opinion, which has not been introduced in court, the confrontation and counsel rights of an accused are obviated as regards the particular evidence received. Parker v. Gladden, 385 U.S. 363, 364 (1966); Comment, Impeachment on Jury Verdicts by Jurors: A Proposal, 1969 Ill. L.F. 388.

The fourth category of cases — those in which an incident impugns the disinterestedness of the court — can also be viewed as turning on Sixth Amendment values. Thus, in Turner v. Louisiana, 379 U.S. 466 (1965), the Court held that a new trial was required because the two principal prosecution witnesses were deputies who also had charge of the sequestered jury. The Court found that in a trial where credibility was a crucial issue, the status of the two witnesses lent a credibility to their testimony which was not developed by evidence at the trial. “In the constitutional sense, trial by jury in a criminal case necessarily implies at the very least that the ‘evidence developed’ against a defendant shall come from the witness stand in a public courtroom where there is full judicial protection of the defendant’s right of confrontation, of cross-examination, and of counsel.” Id. at 472-73.

In addition to the Sixth Amendment, however, the duty of the court “to protect the integrity of its own processes,” McKinney, supra, Godbold, J., dissenting at 1032-33, requires that no verdict be the product of partiality, or its semblance, on the part of the court. This duty is the more binding because of the weight jurors are likely to accord information emanating from court personnel. See Querica v. United States, 289 U.S. 466 (1933); United States v. *233 Pittman, 449 F.2d 1284 (9th Cir. 1971); Truscott v. Chaplin, 403 F.2d 644 (3d Cir. 1968).

As normal jury pressures and intra-jury influences may not be impeached by juror evidence, so also they constitute no grounds for overturning a-verdict. See cases cited at notes 24 and 25, supra; United States v. Grieco, 261 F.2d 414 (2d Cir. 1958), cert. denied, 359 U.S. 907 (1959). Nor, indeed, is a verdict invalid merely because the jurors’ generalized knowledge about the parties, or some other aspect of the case, is an ingredient of the decision. Though “the specific guarantees of an impartial jury and of confrontation,” as well as “the more general one of due process,” proscribe consideration of specific extra-record facts about the case on trial, it is not necessary that the jurors be “totally ignorant about a” case. United States ex rel. Owen v. McMann, 435 F.2d 813, 817 (2d Cir. 1970) cert. denied, 402 U.S. 906 (1971).

“We cannot expunge from jury deliberations the subjective opinions of jurors, their attitudinal expositions, or their philosophies. These involve the very human elements that constitute one of the strengths of our jury system, and we cannot and should not excommunicate them from jury deliberations.”

McKinney, supra at 1022-23.

B. Misconduct Alleged By Defendants

The trial judge determined that several rumors were circulating among the jurors during their deliberations. He found that at least some of the jurors learned of killings which had occurred on St. Croix during the trial, and concluded that “[t]he news must have come from the outside since it was true.” Trial judge at 8. The judge further found that “general rumors of FBI investigations of jurors’ families” were circulated among the jurors. Because a similar report was broadcast over a local radio station, the trial judge surmised that- the rumor was *234 common throughout the island and had “filtered into .the jury room from outside.” 30a Trial judge at 11, 10. Neither the actual source of these, rumors nor the means by which the jury learned of them was found by the trial judge. Other rumors which he identified concerned the investigation of three jurors regarding specific events in their pasts. None of these specific rumors was credited with an outside source by the trial judge. Rather, he considered that .the specific rumors were juror-generated. We find nothing in the record to refute this conclusion.

The defendants characterize all these rumors as threats. They point to the rumors and to the alleged conversations between certain jury attendants and jurors, see Part I and note 13, supra, as instances of impermissible outside influences on the verdict which require that it be set aside. We disagree.

The only rumors that can even qualify as outside influences are . those which were assigned an outside source by the trial judge. The rumors about specific jurors are clearly in the category of intra-jury discussions which meet neither the competency nor the sufficiency criteria for overriding the verdict. The rumors concerning other killings on St. Croix and the FBI investigation of persons involved in the case are not so easily categorized. It is clear that they are not “extraneous influences” merely by virtue of. having an outside source, however. See II-A, supra. It is also apparent that the rumors are dissimilar from the types of occurrences which were roughly categorized above as “extraneous influence” cases. Although it may appear, superficially, that they might qualify as communications between jurors and third parties, closer examination establishes that the rumors do not resemble that class of cases either.

*235 The third-party communication cases break down, factually, into three major subclasses: (1) cases where jurors glean from non-jurors facts or opinions concerning the liability or guilt of a. defendant; (2) cases involving attempts by outsiders to influence the verdict through intimidation or bribery of jurors; (3) cases where jurors come into personal contact with persons related to one side or another of the controversy before the jury. See generally Comment, supra at 366-69.

The trial judge made no finding that any contact, either direct or indirect, with any non-juror, brought the FBI investigation and murder rumors to the jury’s attention. In fact, the trial judge was not able to discover the origin of these rumors. See pages 233-235 above. Absent a finding as to the rumors’ origin, the element of bias, obvious in the types of third-party communications cases discussed above, is. lacking here. 31 Perhaps more importantly, absent such a finding, the rumors themselves are, at most, only tangentially germane to the jury’s decision-making. For the rumors are fundamentally unlike the communications typically found objectionable; they are not “pertinent to the disposition of an ultimate issue in the controversy, that is, to .the jury’s consideration of the guilt or innocence of the defendant.” United States v. Burke, 496 F.2d 373, 377 (5th Cir. 1974). Only if the rumors carried the coercive force of threats or bribery would we be justified in treating them, factually, as “extraneous influences.” Unlike a bribe or threat, however, the rumors did not represent either investigation or other killings as contingent upon any particular verdict outcome. Since the rumors were nebulous, had no identified source and since the jurors appeared to regard them as emanating from other jurors, we can discover no coercive or intimidating *236 effect in these rumors. 32 As the Second Circuit observed in United States v. Gersh, 328 F.2d 460, 464 (1964), “ [s] omething more than the mere fact of an unknown and here uncompleted contact with a juror is needed . . . .” 33

In determining that the rumors in this case do not qualify as “extraneous influences,” we rely not only upon factual distinctions but upon our analysis of the rationale underlying the “extraneous influence” cases. No Sixth Amendment values are implicated by these rumors. 34 Where.no coercive or biasing effect can be attributed to the statements, the impartiality of the jury is not impugned. Where the statements are in no sense “evidence developed” against the defendants, neither confrontation nor counsel rights are abused.

Nor is the integrity of the court put in jeopardy by the spread of such rumors among the jury. The rumors had no official source. See note 13, supra. Since “a complete sanitizing of the jury room is impossible,” McKinney, supra at 1022, as well as unnecessary, McMann, supra at 817, the court has not failed in its duty to sequester the jury merely because rumors such as these managed to *237 penetrate to the jury room. See also Rideau v. Louisiana, 373 U.S. 723, Clark, J., dissenting, at 733 (1963).

Finally, were we to set aside the jury verdicts because of these rumors, we would be defeating the purposes served by the competency-sufficiency rules in this area of the law. See II-A, supra. If rumors such as these, filtering into the jury from no known source, were held sufficient to impeach these verdicts, jury tampering and harassment would be encouraged and we “would add unduly to the already fragile state of criminal convictions.” McMann, supra at 817. Particularly in a locale like the Virgin Islands, where the small area and population are conducive to the rapid spread of information about anything that occurs, and “where it might be impossible to find 12 jurors who were totally ignorant about a defendant,” we find it difficult to imagine any verdict which would not be subject to challenge on the basis of claims like those raised here. McMann, supra at 817-18. In our view, we cannot allow impeachment by incidents of this type. United States v. Kafes, 214 F.2d 887, 889 (3d Cir. 1954).

Conversations between jurors and jury attendants, however, stand on a different footing than do the rumors just discussed. Numerous cases treat such communications as “extraneous influences.” See cases cited note 23, supra. See also United States v. Brumbaugh, supra. Thus, the occurrence of the conversation between Matron Foye and juror Cappin, see Part I, supra, would require us to set aside the verdict, if that conversation were prejudicial to the defendants. 35

C. Prejudice

Even though a party establishes, by competent evidence, an act of jury misconduct sufficient to set aside a verdict, the verdict will stand unless the party has been *238 prejudiced by the misconduct. Although there is normally a presumption that a verdict is valid, United States v. Robbins, 500 F.2d 650 (5th Cir. 1974), certain types of misconduct possess such a high potential of prejudice that they are considered prima facie prejudicial. Remarks by jury attendants to jurors, where those remarks relate to the jury’s deliberations, are usually of this latter type. See Remmer, supra; Mattox, supra; United States ex rel. Tobe v. Bensinger, 492 F.2d 232 (7th Cir. 1974). Thus, we will assume, as defendants contend, that the Government has the burden of proving that Foye’s remark to Cappin was not prejudicial. 36 We conclude that the Government carried this burden.

In United States v. Brumbaugh, supra at 1129, the following exchange took place between a juror and a court bailiff:

“Court

“Q During this time that you were with the bailiff, was there any conversation between you and him concerning the deliberations of the jury and the state of those deliberations ?

“A If I recall right, he asked me how it was going? and I answered, ‘9 to 3/

“Q Was there any other conversation between you concerning the matter I mentioned?

“A He said ‘For or against?’ And I didn’t say nothing. I didn’t even answer him. And then he said—

“Q Just one minute, please. Can you tell me, please, in terms of where you were when this conversation occurred ?

“A On the elevator — at the elevator door.

“Q You say he said to you, ‘For or against ?’ and you said you did not answer?

“A I did not answer.

“Q Then what further followed; any further conversation ?

*239 “A Well, he said that, ‘It’s always some woman or something holding it up,’ to that effect, or something like that, and I still did not answer. I didn’t say no more.

“Q Can you tell me whether this conversation in any way influenced you one way or the other as to your own participation in this case?

“A No, it did not influence me in no way.

■ “Q Did you report this to your fellow jurors, the fact that you had this conversation with the bailiff?

“A No.”

Although the juror involved was voting not guilty at the time, the trial court’s determination that no prejudice resulted from the incident was upheld on appeal. The trial court’s finding was based, in part, upon the juror’s averment that he had not been affected by the bailiff’s remarks.

In the present case, Cappin did not indicate whether she considered herself influenced by the matron’s statement. 37 However, the trial judge found that she had voted guilty from the first ballot to the last. 38 Trial judge at 6; see also H.T. 152. Since Cappin did not mention the incident to any of the other jurors, no juror could have been moved by the remark to change his vote. See Bailey v. United States, 410 F.2d 1209, 1215 (10th Cir.), cert. denied sub non. Freeman v. United States, 396 U.S. 933 (1969).

It thus appearing that no prejudice accrued to the defendants from the only occurrence which was both legally. cognizable and sufficient to impeach the jury verdict, we find no abuse of discretion in the trial judge’s *240 refusal to order a new trial. The judgment of the district court will be affirmed.

Notes

Memorandum Opinion and Order filed September 20, 1974, in Criminal No. 97/1972.

See Government of the Virgin Islands v. Gereau, et al., 502 F.2d 914 (3d Cir. 1974).

The case was submitted to the jury at 2 P.M. Saturday, August 4, 1973. A verdict was reached by 2:30 P.M. Sunday, August 12, 1973, but was not announced until the following day. In all, the jury deliberated for some 50 hours. See N.T. 6540-41; Order dated August 8, 1973, in Criminal No. 97/1972; Hearing re Motion on Tampering with Jurors (hereinafter referred to as “NT”) at 20.

“The affidavit of Mr. Allick states:

‘I’m a concerned citizen writing to you on the Verdict of Guilty voted by the twelve jurors. I was one of the twelve, and I have to say that the verdict I turned in was not from my free will. It was involuntary and due to pressure of information, being carried in and out of the jury room. Because I am a concerned about the justice system in these Virgin Islands that I am writing this letter. Sincerely.’

“That of Mr. Rodgers recites:

‘The verdict delivered after nine (9) days of deliberation does not represent my honest judgment. My verdict was not voluntary. It was the end product of pressures applied by Judge Warren Young, who refused two- verdicts and the majority of the jury who, to me seemed to have predetermined verdict. Sincerely.’ ”

Rodgers testified that the wording of' the affidavit was not his, but was dictated by Johnny Ross, who urged Rodgers to aid “civil rights” by signing an affidavit. Rodgers identified his signature on the affidavit, but stated that he had. signed “a piece of paper” in the office of one of the defendant’s attorneys without looking at -the paper he signed. H.T. 99-102; 105-111. See also 85-87; 100-11; 286-94.,

With reference to the credibility of Allick and Rodgers, see note 13 below and part I-B at pages 8-10 below. In this court’s previous decision in this case (see note 2 above), claims that (1) the trial judge was disqualified from conducting the proceedings in the district court (502 F.2d at 931-33), and (2) the trial judge had coerced the jury’s verdict through requiring continuing deliberations- and through its instructions (502 F.2d at 935-36), were rejected.

Judge Young was not on the island at the time, and Chief Judge Christian was involved in another trial.

In a letter of October 2, 1973, from counsel for defendants to Chief Judge Christian (attached to Motion to Further Augment the Record, filed January 19, 1974, at No. 73-1775), such counsel agreed that Judge Marsh was appointed “to preside over the hearing but not to make a finding of fact.” In response Chief Judge Christian states in his October 5, 1973, letter (attached to the same Motion):

“Your recollection that my appointment of Judge Marsh contemplated no more than that he would preside over the hearing, and that the testimony taken was to have been sealed and handed over to Judge Young on his return,.is correct. You may recall that initially I had thought of having the deputy clerk-in-charge preside over the examination of jurors, but later changed my mind and asked Judge Marsh, my feeling being that with a judicial officer presiding, ■ matters would proceed in a more orderly fashion.”

The hearing took three days: August 18, August 24 and August 29. Testimony was taken from'all jurors, two court attendants and' the President of the Virgin Islands Senate.

In remanding, this court left to the trial judge the decision whether he should preside over a new hearing so as to have “the opportunity to observe witnesses and assess the credibility of those giving conflicting testimony.” 502 F.2d at 937, n. 14.

See, e.g., Mattox v. United States, 146 U.S. 140, 147 (1892); 6A J. Moore, Federal Practice, ¶ 59.08[4] at 59-136 to 138; 59-152 (1974).

As part of this argument, defendants propose that we remand for a hearing before the trial judge where “there is a direct conflict in the testimony and the issue is material.” Brief for Appellants at 62. Such a hearing, they claim, is necessary to the trial judge’s ability to make credibility choices among witnesses. We reject this proposal for four reasons: (1) our former decision indicated that, in our view, no such hearing was essential (see 502 F.2d at 937, n. 14); (2) the defendants requested no hearing following our remand; (3) when the special master was appointed, the understanding of the parties was that “the raw material from the testimony of the jurors would he turned over to” the trial judge so that he could make his findings from the testimony, letter from defense counsel to Chief Judge Christian dated October 2, 1973 (see note 6 above); (4) the trial judge considered the adequacy of the record developed by the special master, the “reliability and credibility problems reasonably expected to arise in a new hearing after a year’s passage of time,” his opportunities to observe the jurors during the trial, and his knowledge of the local accent, idiom and grammatical usage in deciding that no further hearing was necessary. Our own reading of the record confirms the correctness and propriety of these considerations. Particularly because of the poor recollection of many jurors at the hearing held within a week of their verdict, we see nothing to be gained from any further hearings.

The Steelworkers court did acknowledge that the clearly erroneous standard was “inapplicable so far as it has to do with the trial court’s opportunity to observe the demeanor of the witnesses and like matters.” 271 F.2d at 685. Moreover, the court discounted the importance of the disagreement among the affiants in that it found “no variances in the affidavits basically material to the findings made by the court below or to the issues with which we are immediately concerned.” Id. at 688.

Demirjian’s standard is inapplicable both because the facts here are not stipulated,, but disputed, and because our earlier remand would be ren- • dered nugatory were we to ignore the findings which we directed the trial judge to make. Yet, since at least some of the fácts in issue here are material, we are reluctant to employ in this case the clearly erroneous standard of the Steelworkers case. Nor are we satisfied that the trial judge’s credibility assessments come within the exception to the clearly erroneous standard spelled out in Steelworkers, premised as it is on the proposition that the appeals court is as favorably situated as the trial court in "appraising the evidence. “Credibility involves more than demeanor. It apprehends the over-all evaluation of testimony in the light of 'its rationality or internal consistency and the manner in which it hangs together with other evidence.” Carbo v. United States, 314 F.2d 718, 749 (9th Cir. 1963); Indiana Metal Products v. N.L.R.B., 442 F.2d 46, 52 (7th Cir. 1971). In this case the trial judge’s thorough acquaintance with the circumstances surrounding the hearing into the validity of the verdict give special weight to his credibility findings. The trial judge’s observation of certain incidents related to the jurors’ testimony, his knowledge of the circumstances under which the jury was sequestered, and his familiarity with the local accent, idiom and grammatical usage, as well as with other peculiarities of the vicinage, give him a distinct advantage over this court in evaluating the jurors’ testimony. See also note 10 supra. Moreover, insofar as demeanor becomes important in assessing credibility, the trial judge did have the opportunity to observe the jurors’ demeanor during three stages of the trial, though not at the hearing: (1) when he questioned them individually on the voir dire preceding trial; (2) during the course of the trial; and (3) when he. polled the jurors individually upon the return of the verdict. Such demeanor evidence does not_ carry the weight assigned to direct observation of a witness while testifying, but we do not consider it a negligible factor in determining how much deference we should allot to the trial court’s findings.

We have examined the-record carefully in view of the defendants’ claim that other evidence in the record is so contradictory that the trial judge could not properly make these challenged findings. _ Much of the other evidence cited by defendants is irrelevant to the findings attacked; some is, in our estimation, of lesser quality, than that relied upon- by the trial judge. Still other evidence referred to by defendants is not present in the record at all, such as the contention that matters overheard by juror Torres while she was resting in the judge’s chambers had an official cast. It is clear from the record that none of the jurors, including Torres, knew where she was resting during her illness. Our'- reading of the record convinces us that the following findings attacked by defendants are *223 sustainable on the grounds set forth in the trial judge’s opinion:

(1) “[N]o discussion concerning a possible perjury prosecution ever took place within the confines of the jury room” [Memorandum Opinion and Order, filed Sept. 20, 1974, Crim. No. 97/1972 (hereinafter referred to as “trial judge”) at 15];

(2) That Matron Foye had not made certain statements to juror Rodgers [trial judge at 17-18];

(3) The- vote changed to 11-1 for conviction on Thursday, August 9 [trial judge at 6];

(4) That Marshal Sutterfield did not-make any statement to juror Allick to the effect that the trial was costing the Government a lot of money [trial judge at 16-17] ;

(5) That the credibility of jurors Allick, Isaac and Rodgers was diminished because they were pressured to impeach their verdicts after the jury was dismissed [trial judge at 26].

Finding (4) was conceded by defendants to be non-material. Brief for Appellants at 15.

In addition to the reasons stated by the trial judge for discounting Alliek’s testimony at the hearing, we note that' Allick, as the one juror who consistently voted for acquittal until August 12, 1973, would have a special stake in overturning' the verdict. See, e.g., H.T. 165, 197, for testimony indicating Allick’s bias.

13a

See Salkeld, supra at 907, and note 12 above. We note that all these challenged findings which we affirm have adequate and reasonable support in the record. See Olympic Towing Corp. v. Nebel Towing Co., 419 F.2d 230, 232-33 (5th Cir. 1969).

13b

The trial judge’s opinion continued as follows:

“He was nineteen years old, a black, a school chum of defendant Joseph and understandably susceptible to peer pressure. When Allick signed the verdicts, he stood up and made the puzzling remark that now he could not go back to the local college in St. Thomas. At the time, none of the jurors questioned his remark, but I find it significant that he feared any such consequence from signing guilty verdicts. Moreover, shortly after he signed the verdicts, he sent the following note to me:

‘Dear Judge Young,

Some of us would like to know what kind of protection will be provided for all of us.

With this we mean Personal Protection.

/s/ Laura Torres

Foreman Myron Allick’

“Although, as the jury foreman, he would sign all notes to the Judge, I believe that the fear of possible reprisals was his own fear as well as that of Torres and other jurors.

“In voir dire, Allick showed concern that the estimated length of the trial might interfere with his entering the College of the Virgin Islands in August. However, at the post trial evidentiary hearing, he testified that, after signing the verdicts, he made plans with one of the U.S. Marshals to go away. After being driven around by Senator Olaude Molloy to talk with other jurors about impeaching their verdicts and after being ‘interviewed’ by Johnnie Ross and Mario Moorhead, he signed the affidavit which, in effect, told the community that the guilty verdict was really not his verdict. Significantly after those events, Allick no longer needed ‘Personal Protection’ nor did he, in fact, go through with his plan to leave the islands (as did almost all of the other jurors).

“I have selected another incident from the transcript which I find sheds some light on Allick’s post-verdict actions. At page 42, United States Attorney Brady asked Allick if he recalled when he was in court the day of the verdict. After receiving an affirmative answer, Brady asked if he recalled any one of the defendants speaking to him as he left the courtroom. The response is as follows:

‘A [Allick] Maybe so but my mind was not really clear.

‘Q [Brady] You did not hear anything?

‘A. I did not really take it down. Like I hear him say, you know, something I couldn’t say that he said this. I really, you know, I just hear some grumbling something like that.’

“That response clearly indicates to me a cover up for Allick’s post-verdict fear. It actually happened as described by Allick. From my bench I saw and heard a remark being made by one of the defendants to Allick, but I could not hear what was actually said. Since the incident went no further than the remark itself, I let it pass. Moreover, under the circumstances immediately following the dramatic jury poll, whatever remark was made by one of the defendants to Allick could hardly have *225 been an amenity. It had to be unnerving', and I do believe that it made Allick an easy subject for those who, in my opinion, improperly tampered with a jury verdict by coercive pressures exerted not only upon Allick, but upon Rodgers,....”

(trial judge at 20-23) (footnotes omitted).

Juror Cappin is referred to as “Coppin” in the trial judge’s opinion.

They may, however, be subject to rebuttal. See- 5 J. Moore, supra; ¶ 43.09 at 1371-72; F.R.E. Rule 201.

The defendants concede that this latter finding is not material. See Brief for Appellants at 20.

The origin of the rule is Lord Mansfield’s opinion in Vaise v. Delaval, 1 Term. Rep. 11, 99 Eng. Rep. 944 (K.B. 1785). For a sketch of the rule’s development and permutations, see ABA Standards Relating to Trial by Jury, Approved Draft, 1968, Commentary to § 5.7(a).

Other variations of the rule may be found in the ABA Standards, note_ 17 supra. Some federal courts have allowed a juror to testify regarding overt acts, which can be corroborated by other jurors. This circuit has not adopted this approach, which was disapproved by Congress when it adopted the Federal Rules of Evidence in January 1975. See Congressional Conference Report concerning F.R.E. Rule 606(b):

“The Conference adopts the Senate amendment [which] . . . does not permit juror testimony about any matter or statement occurring during the course of the jury’s deliberations.”

(Federal Rules of Evidence Annotated, West Pub. Co., p. 55)

Wigmore considers the parol evidence rule as the basis for excluding juror testimony. 8 J. Wigmore, Evidence, § 2346 (McNaughton ed. 1961). See also Note: Impeachment of Jury Verdicts, 53 Marquette L. Rev. 258, 262-63 (1970). A second evidentiary basis for the rule is the privilege accorded communications among deliberating jurors. See Wigmore, id.; Note, id.; Clark v. United States, 289 U.S. 1 (1933).

The rule and exception are now embodied in F.R.E. Rule 606(b):

“Upon an inquiry into the validity of a verdict or indictment, a juror may not testify as to any matter or statement occurring during the course of the jury’s deliberations or to the effect of anything upon his or any other juror’s mind or emotions as influencing him to assent to or dissent from the verdict or indictment or concerning his mental processes in connection therewith, except that a juror may testify on the question whether extraneous prejudicial information was improperly brought to the jury’s attention or whether any outside influence was improperly brought to bear upon any juror. Nor may his aflidavit or evidence of any statement by him concerning a matter about what he would be precluded from testifying be received for these purposes.”

See also Hearings on H.R. 5463 Before the Comm. on the Judiciary of the United States Senate, 93rd Cong., 2d Sess., at 357 (1974). Rule 606(b) does not, of course, apply to this case, which preceded it. However, the Congress recognized that the rule, as adopted, tracked the existing case law.

Marshall v. United States, 360 U.S. 310 (1959); Mattox v. United States, 146 U.S. 140 (1892); United States v. Kum Seng Seo, 300 F.2d 623 (3rd Cir. 1962); United States v. Thomas, 463 F.2d 1061 (7th Cir. 1972); United States v. McKinney, 429 F.2d 1019 (5th Cir. 1970).

Farese v. United States, 428 F.2d 178 (5th. Cir. 1970), and cases cited therein at 180-81; United States ex rel. Owen v. McMann, 435 F.2d 813 (2d Cir. 1970).

Parker v. Gladden, 385 U.S. 363 (1966) [bailiff expressed opinion on case to jurors]; Remmer v. United States, 347 U.S. 227 (1954) [attempted jury tampering not revealed to defendant or counsel by trial judge]; United States ex rel. Tobe v. Bensinger, 492 F.2d 232 (7th Cir. 1974) [bailiff, in answer to jury queries directed to judge, told jury they must deliberate until they reached a verdict]; United States v. Brumbaugh, 471 F.2d 1128 (6th Cir.), cert. denied, 412 U.S. 918 (1973) [improper remark by bailiff to juror]; Truscott v. Chaplin, 403 F.2d 644 (3d Cir. 1968) [inquiry by judge whether jury close to verdict]; United States v. Gersh, 328 F.2d 460 (2d Cir. 1964) [phone calls to juror]; Wheaton v. United States, 133 F.2d 522 (8th Cir. 1943) [bailiff instructed jury].

United States v. Kafes, 214 F.2d 887 (3d Cir. 1954); United States v. Blackburn, 446 F.2d 1089 (5th Cir. 1971), cert. denied, 404 U.S. 1017 (1972); United States v. Stoppelman, 406 F.2d 127 (1st Cir.), cert. denied, 395 U.S. 981 (1969).

See, e.g., United States v. Kohne, 358 F.Supp. 1046 (W.D. Pa. 1973), aff’d, 487 F.2d 1395 (3d Cir. 1973), cert. denied, 417 U.S. 918 (1974), and cases there cited at 1051-52.

See Part II-A-1 supra.

“Drunkenness, bribery, receiving incompetent documents, or privately interviewing a party, do require [setting aside a verdict]; but there are many irregularities, which, however proved, do not, and among them is an agreement to abide by the vote of the majority.”

Jorgensen v. York Ice Machinery Corp., 160 F.2d 432, 435 (2d Cir. 1947). See also Comment, supra at 366ff.

See, e.g., cases cited note 21, supra.

See, e.g., cases cited note 22, supra. See also People v. DeLucia, 20 N.Y.2d 275, 229 N.E.2d 211, 282 N.Y.S.2d 526 (1967).

See cases cited note 23, supra; Bailey v. United States, 410 F.2d 1209 (10th Cir. 1969) [jury accosted by individuals urging them to acquit defendants] ; Richardson v. United States, 360 F.2d 366 (5th Cir. 1966) [conversation between Government witness and juror].

See cases cited note 23, supra. See also Jenkins v. United States, 380 U.S. 445 (1965); Turner v. Louisiana, 379 U.S. 466 (1965) [government witnesses also jury attendants]; Snyder v. Lehigh Valley R.R. Co., 245 F.2d 112 (3d Cir. 1957); United States v. Pittman, 449 F.2d 1284 (9th Cir. 1971) [Government agent who was prominent prosecution witness was allowed in jury room to play tape].

30a

These rumors were “clearly false.” See trial judge at 9-10.

See, e.g., Remmer v. United States, note 23, supra; Richardson v. United States, note 29, supra. See also Part II-A-2, supra.

Defendants argue that Remmer, supra, establishes that FBI investigation of jurors, if known to the jurors, is inherently coercive. In Remmer, however, the juror who was being investigated had previously been' offered a “profit” if he voted to acquit. As the Court recognized, such a juror might well feel pressured by an investigation to vote for conviction in order to demonstrate he had not succumbed to bribery. Jurors who had not been similarly importuned would not be subject to this sort of pressure.

Moreover, the linchpin of Remmer was the trial court’s failure to inform the defendant and his attorney that the juror had been approached. The case was remanded for an adversary hearing to “determine the circumstances, the impact thereof upon the juror, and whether or not it was prejudicial,” since the “trial court should not decide and take final action ex parte on information such as was received in this case. . . .” 347 U.S. 229-30. See also Fillippon v. Albion Vein Slate Co., 250 U.S. 76 (1919). This crucial issue is not raised by the present case.

Gersh involved anonymous phone calls made to the forewoman of the jury. Each time the caller hung up as soon as the juror answered; nothing was ever- said. The Second Circuit considered that these calls, which were not connected to the defendants and ’ which had no articulated objective, raised no issue of jury prejudice.

For this reason, the case discussed under II-A-2 at pages 230-232 above are inapplicable.

See note 13, supra.

We recognize that not all juror-jury attendant exchanges would be presumptively prejudicial. Each case turns on its own facts. Cf. Truscott v. Chaplin, 403 F.2d 644, 645 (3d Cir. 1968).

Because there is no subjective evidence as to prejudice resulting from this remark, we need not resolve the issue, raised by defendants, whether prejudice can be established by subjective evidence. Compare Miller v. United States, 403 F.2d 77, 82 (2d Cir. 1968), with United States v. Robbins, 500 F.2d 650, 653 and n. 5 (5th Cir. 1974).

Where the verdict is impugned, jurors are “competent to testify in refutation of impugning testimony.” 6A J. Moore, supra, ¶ 59.08 [4] at 59-143.