James Martorano appeals from the judgment of the district court, 457 F.Supp. 803 (D.Mass.1978), denying his motion for new trial on the ground of ineffective assistance of counsel due to joint representation.1 His appeal was originally decided by a panel of this court, which reversed and remanded for a new trial after holding the government had not demonstrated that Martorano was not prejudiced by the failure of the trial court to conduct the inquiry required in United States v. Foster, 469 F.2d 1, 5 (1st Cir. 1972). Appellee then moved for rehearing en banc, and that motion having been allowed by vote of a majority of this court, the panel’s judgment was vacated and an en banc hearing held, resulting in the present opinion. See 28 U.S.C. § 46(c); Fed.R.App.P. 35(a); Part IV infra.
The facts are more fully set forth in the panel opinion, to which we shall refer, United States v. Martorano, 610 F.2d 36 (1st Cir. 1979). See also United States v. Martorano, 557 F.2d 1 (1st Cir.), reh. denied, No. 76-1372 (1st Cir. May 23, 1977) (unpub.) and 561 F.2d 406 (1st Cir. 1977), cert. denied, 435 U.S. 922, 98 S.Ct. 1484, 55 L.Ed.2d 515 (1978) (affirming appellant’s conviction on direct appeal). In the present opinion, we repeat the facts only as necessary to our analysis of the issues.
I.
Our decision in United States v. Foster, 469 F.2d 1 (1st Cir. 1972), requires the *915district courts of this circuit to warn on the record every defendant who indicates that he contemplates being jointly represented at trial of the risks inherent in such representation, so that the court can be assured that his waiver of the right to separate counsel is knowing and voluntary. 469 F.2d at 5. In the present case, appellant Marto-rano received no direct warning by the court, although in response to instructions from the magistrate, his attorneys provided certain advice orally and in writing, and secured Martorano’s signature to a letter to the court, which they had prepared, stating that he was aware of the risks of joint representation but nonetheless wished to retain the same law firm which represented his co-defendant Brian Halloran.2 We agree with the panel opinion that this letter, and any advice preceding it, did not relieve the government from the consequences of the district court’s failure to comply materially (or indeed at all) with the Foster rule. We made no provision in Foster for an alternative to direct inquiry in open court, and are not inclined to establish one now. Letters prepared and signed outside the court’s purview are not a substitute for the in-court colloquy mandated by Foster.3
II.
We thus turn to the principal issue in this case and the one upon which our decision to grant rehearing en banc was based.4 In Foster we said that the failure of the trial court to make adequate inquiry would shift to the goverñment the burden of proving the unlikelihood of any prejudice arising from joint representation. While we described the government’s burden in somewhat different ways according to whether the issue arose on direct appeal or on collateral attack, 469 F.2d at 5, we did not mean to announce as between the two situations any significant difference in the ultimate burden of persuasion. Rather we merely recognized that in the case of a direct appeal an appellate court’s focus would be perforce on the trial record alone. Where, as here, an evidentiary hearing has been held, whether in connection with a motion for new trial, a motion to vacate sentence, or the like, such a hearing may be a “useful supplement” to the trial record, see United States v. Donahue, 560 F.2d 1039, 1044 (1st Cir. 1977), providing further insight into the actuality of any conflict between the interests of various defendants and illuminating the degree to which the choice of trial strategy may have been influenced by conflicting concerns. Still the ultimate question is the same in either context, namely, whether the government has carried its burden of showing it is more likely than not that no prejudice stemmed *916from the joint representation. We have not adopted a rule of per se reversal, compare Holloway v. Arkansas, 435 U.S. 475, 489, 98 S.Ct. 1173, 1181, 55 L.Ed.2d 426 (1978) (automatic reversal where defendant was compelled, over his objection, to accept joint representation), nor is our standard as stringent as the “harmless beyond a reasonable doubt” rule followed where errors of constitutional magnitude are present, compare Chapman v. California, 386 U.S. 18, 24, 87 S.Ct. 824, 828, 17 L.Ed.2d 705 (1967). Here we are not dealing with a court’s unconstitutional refusal to allow separate representation, see Holloway v. Arkansas, 435 U.S. 475, 98 S.Ct. 1173, 55 L.Ed.2d 426 (1978), but only with its failure to provide a defendant with an auxiliary procedure required by this circuit under our supervisory powers, which is aimed at making absolutely certain the defendant understands all the pitfalls of joint representation. While we assuredly do not encourage violations of our supervisory rule, see United States v. Waldman, 579 F.2d 649, 653 (1st Cir. 1978), there are undoubtedly situations in which such violations cause little or no material prejudice to defendants; we must take care not to provide a windfall to a defendant, who has been found guilty of a serious crime, simply because of some attenuated hypothesis of prejudice woven by counsel after conviction. Joint representation, after all, may sometimes be quite beneficial to a defendant and may at other times be of little consequence to the adequacy of a defendant’s representation. See Holloway v. Arkansas, 435 U.S. 475, 482-83, 98 S.Ct. 1173, 1177-78, 55 L.Ed.2d 426; Glasser v. United States, 315 U.S. 60, 92, 62 S.Ct. 457, 475, 86 L.Ed. 680 (1942) (Frankfurter, J., dissenting). To overturn a defendant’s conviction for violation of a supervisory rule where it is more likely than not that conflicting interests did not hamper counsel’s pursuit of a potentially effective defense would be both wasteful and pointless.
In finding prejudice, the panel opinion focused on the following language in United States v. Donahue, 560 F.2d 1039, 1045 (1st Cir. 1977):
“Where, as here, an alternative strategy — whatever its ultimate merit — plainly existed, and where such a strategy involved, as between the joint defendants, some potential for conflict of interest, we cannot say for sure that the potential conflict did not influence the choices made by those representing both defendants.”
In applying that language to the very different facts of the present case, the panel derived two key elements necessary to establish prejudice: (1) the existence of an alternative defense strategy, and (2) the potential for conflict of interest between the defendants. The panel then found that at least one alternative strategy plainly existed, namely having Louis “Bugsy” Pallot-ta, brother of the chief prosecution witness Peter Pallotta, testify on appellant’s behalf. The panel also found a potential for conflict in that the decision not to use Louis Pallot-ta as a witness may have been influenced by the possibility that Louis would damage appellant’s co-defendant Brian Halloran by giving testimony as to Halloran’s reputation and character.
As for the first element, the panel may well be correct that calling Louis Pallotta to testify should be regarded as an existing alternative defense strategy even though the district court, for rather substantial reasons, found that Louis had overwhelming disabilities as a defense witness. It was represented that Louis Pallotta, the brother of the chief prosecution witness, would have testified that Peter Pallotta’s account of the loan transaction was a fabrication. Peter Pallotta himself had testified that Louis was present when the allegedly extortionate loan was negotiated: Louis, it is asserted, would have corroborated Martorano’s version of this event, confirming that the loan was made to Louis rather than to Peter and that it was paid back by Louis prior to the time of the alleged extortionate collection efforts. Louis’ hoped-for testimony would thus, on its face, have been very favorable to Martorano. The problem with Louis as a potential witness was that he had previously given a contradictory statement to the FBI (which, depending on its con*917tents, might have gone some distance to destroy Martorano’s as well as Louis’ own credibility). Louis also had a history of very serious mental problems. Under these circumstances, the district court felt Louis would have been worse than useless — and so, according to themselves, did the defense attorneys. On the other hand, after-the-fact reconstruction of what an attorney might have done had he been completely free from any sense of divided loyalties is a difficult feat. It makes sense, therefore, to assume, at least for purposes of argument, that calling Louis to the stand was a conceivable alternative strategy, and then to proceed to the key issue — which is whether there was in fact any such conflict of interest between the two defendants as might have led separate attorneys to make a different choice in Martorano’s case than in Halloran’s case with regard to the question of calling Louis Pallotta to the stand.
The foregoing is, of course, the key question in any case where Foster warnings have been omitted. The mere existence of alternative strategies is not by itself demonstrative of prejudice so long as there was no inherent conflict between the considerations that weighed upon rejection or acceptance of the strategies in both defendants’ cases.
Appellant contends that his attorneys balked at putting Louis Pallotta on the stand at least in part because Pallotta might have given testimony on cross-examination which would have blackened Hallo-ran’s reputation. Since the evidence against Halloran was relatively weak, it is argued, the attorneys were concerned to prevent any corroboration of Peter Pallotta’s testimony identifying Halloran as a “loan shark” and “enforcer” and implicating him in the “holdup” of Pallotta’a nightclub.5 This concern for Halloran, so the argument goes, made it impossible for counsel to give proper consideration to the value of calling Louis on behalf of Martorano, for whom Louis would be a valuable corroborative witness.
A central flaw in this argument is that to the extent Louis’ testimony would have helped Martorano, it would also, in virtually equal measure, have helped Halloran — so that whatever considerations induced the attorneys not to call Louis cut across both clients’ cases. This is so even though, as the dissent observes, Louis would not have testified “directly” about Halloran. See page 921 infra. If the jury had credited Louis’ (and thus Martorano’s) version of events, it would necessarily have rejected Peter Pallotta’s contrary testimony and would have found that there had never been a $2,000 loan to Peter — indeed, that Peter had fabricated the entire episode. Not only would such a belief by the jury have led to acquittal for both defendants on those counts of the indictment directly concerned with the $2,000 loan (counts one to three), it would have so totally discredited Peter Pallotta as to undercut his remaining testimony regarding Halloran’s alleged collection — at Martorano’s direction — of interest payments at gunpoint (count four). In short, Louis Pallotta, if at all helpful to Martorano, as a witness, would necessarily have helped Halloran, and to a degree far exceeding any detriment to Halloran from Louis’ reputation testimony. By the same token, if Louis, having once taken the stand, was discredited as a witness, the consequences would have fallen at least as heavily on Martorano as on Halloran. Louis’ expected testimony dealt primarily with transactions in which Martorano was a *918direct participant and about which Martora-no himself testified. If Louis was shown to have made a prior contradictory statement to the FBI, or if on cross-examination he changed his testimony, or if he made a conspicuously poor impression, Martorano’s own credibility as a witness would be in jeopardy.
Similarly, the danger that Louis would admit on cross-examination that he knew the reputation of Brian Halloran to be that of an “enforcer” for loan sharking activities did not represent a conflicting interest as between the two defendants. Martorano had no more interest in bringing out such testimony than Halloran did, for the evidence clearly showed that Halloran and Martorano were close associates, and that, if Peter Pallotta were to be believed, Hallo-ran was a “collector” for Martorano. From the uncontradicted evidence at trial, it was apparent that Halloran and Martorano had often been seen together at Martorano’s restaurant; that Halloran on occasion took telephone calls for Martorano; that Hallo-ran and Martorano visited Peter Pallotta’s nightclub together on October 30, 1974, the night before the alleged “holdup” of the club by Halloran and an unidentified associate; and that Martorano was sufficiently familiar with Halloran and Peter Pallotta to know that there was ill will between the two stemming from an incident that occurred in federal prison. Were the jury in doubt as to whether Martorano had made an extortionate loan to Peter Pallotta — as opposed to a friendly, no-interest loan to Louis — testimony from Louis himself, a defense witness, that a man to whom Marto-rano was this closely tied was a reputed “enforcer” of extortionate loans could have tipped the scales.
An attorney representing only the interests of James Martorano would plainly have considered it disadvantageous to his client to have such testimony admitted unless he was persuaded that somehow, notwithstanding all the problems, those parts of Louis’ testimony corroborating Martorano would come through to the jury as credible — in which event, as stated, Louis would be a useful witness for both defendants, not just for Martorano.6 We conclude that any difference between the possible impact of Louis’ testimony on Martorano and on Hal-loran is too small and speculative to create a meaningful conflict. In large measure, the factors to be weighed in deciding whether to call Louis as a witness were the same in both cases. They boiled down to the pros and cons of Louis’ credibility — if he was credible, both would benefit; if not, he was better left uncalled. This is not a case where certain clear benefits might have been available to one defendant by creating a risk of detriment to the other. Compare United States v. Donahue, 560 F.2d 1039 (1st Cir. 1977). Here, whatever benefits Louis Pallotta’s testimony could confer were likely to accrue to both defendants; whatever risks Pallotta’s testimony might create likewise endangered each defendant. We see no reason to infer that separate counsel would have calculated differently the pluses and minuses of Louis’ testimony. Coupled with evidence indicating appellant was aware that his choice of joint representation might raise a risk of some conflict, and the absence of any evidence of overreaching by others, see note 3 supra, we think this showing sufficient to discharge the government’s burden of disproving that prejudice resulted from the court’s failure to follow Foster procedures. As stated at the outset, the government is not required to show the error harmless beyond a reasonable doubt, only that it is more likely than not that no material prejudice occurred as *919the result of noncompliance with a prophylactic procedure this circuit now requires. This standard has been met.
Appellant’s other assertions of prejudice need not detain us long. We agree with the panel’s conclusion that the decision to have Martorano testify on his own behalf was compelled by the strength of the government’s tape-recorded evidence against him. In any event, Martorano’s testimony, if anything, hurt Halloran more than it hurt him, since Martorano testified on direct examination both that Halloran had been in prison (corroborating Peter Pallotta on that point) and that Peter Pallotta feared physical violence from Halloran. Introduction of Martorano’s criminal record for impeachment purposes merely confirmed Peter Pal-lotta’s testimony on direct suggesting Pal-lotta feared violence might accompany any nonrepayment of a debt owed to Martorano because Martorano had been convicted as an accessory after the fact to murder. In other respects, Martorano’s testimony was exculpatory as to both himself and Hallo-ran. Calling Brian Halloran as a witness plainly was not a viable alternative defense strategy for Martorano. Even had Marto-rano been represented by separate counsel, he could not have compelled Halloran to forego his fifth amendment right not to testify if Halloran’s interests dictated that he remain silent. Martorano could not have been prejudiced by either of these tactical decisions.
III.
In light of its resolution of the case, the original panel did not consider appellant’s further argument that the district court erred in denying appellant’s motion to re-cuse. We are inclined to believe that this issue is now effectively moot, since we have not relied to any significant extent on the findings of the district court, and since it would plainly be pointless to remand this case for a new hearing before a different judge. However, lest it be thought that we are ignoring any possible appearance of impropriety in the handling of this matter, we offer the following comments.
Appellant moved to have the district court judge recuse himself pursuant to 28 U.S.C. § 455(a) both because the judge had presided at trial and might be compromised in passing on his own asserted errors, and because certain remarks made by the judge at another trial assertedly indicated he had a personal bias against appellant. As we stated in United States v. Cowden, 545 F.2d 257, 265 (1st Cir. 1976), cert. denied, 430 U.S. 909, 97 S.Ct. 1181, 51 L.Ed.2d 585 (1977), the standard by which such a motion is judged is,
“whether the charge of lack of impartiality is grounded on facts that would create a reasonable doubt concerning the judge’s impartiality, not in the mind of the judge himself or even necessarily in the mind of the litigant filing the motion under 28 U.S.C. § 455, but rather in the mind of the reasonable man.”
Applying this standard, we think the mere fact that a judge entertains a motion for new trial in a case over which he presided initially does not reasonably call into question his impartiality. Indeed, we have indicated it may often be advantageous to have the original judge continue on a case because of his familiarity with earlier proceedings. See O’Shea v. United States, 491 F.2d 774, 779 (1st Cir. 1974); cf. Halliday v. United States, 380 F.2d 270, 272 (1st Cir. 1967) (contrasting hearing on motion for new trial with second trial on facts previously found by court).
Having thoroughly reviewed the record, we see no indication here that the district judge was influenced by personal animus or bias either at the original trial or in considering appellant’s collateral attack thereon. Although the court in reviewing the Foster issue was inclined to find that appellant had waived his right to separate counsel despite the absence of an on-the-record warning, the court went on to consider carefully whether, assuming a violation of Foster, the government had adequately disproved prejudice. As for the remarks the district court made at another trial intimating that unnamed individuals might cause harm to Peter Pallotta if his *920usual whereabouts were revealed in open court, we think the court sufficiently disavowed any connection between these remarks and appellant Martorano, at the opening of Martorano’s trial, so that no reasonable person could believe they indicated bias against appellant. We thus discern no error in the district court’s denial of the motion to recuse.
IV.
In conclusion, we turn briefly to appellant’s challenge to our authority to hear this appeal en banc. The power of this court to entertain cases en banc is governed by 28 U.S.C. § 46(c) and by Rule 35 of the Federal Rules of Appellate Procedure. Section 46(c) provides:
“Cases and controversies shall be heard and determined by a court or panel of not more than three judges, unless a hearing or rehearing before the court in banc is ordered by a majority of the circuit judges of the circuit who are in regular active service. A court in banc shall consist of all circuit judges in regular active service.”
Rule 35(a) provides, in pertinent part:
“A majority of the circuit judges who are in regular active service may order that an appeal or other proceeding be heard or reheard by the court of appeals in banc.”
Appellant argues that because the First Circuit is presently authorized to have four circuit judges, the required majority to order an en banc hearing is three, even though the newly created seat remains unfilled.7 We disagree. The statute plainly provides that the required majority must exist among the judges “in regular active service.” A judge who is yet to be appointed is not a judge in regular active service. Cf. United States v. American-Foreign Steamship Co., 363 U.S. 685, 688, 80 S.Ct. 1336, 1388, 4 L.Ed.2d 1491 (1960). As we stated in our order of January 10, 1980, a majority of the three regular circuit judges of this court have voted to grant rehearing en banc, and this, in our opinion, is all that the statute requires. Compare Zahn v. International Paper Co., 469 F.2d 1033, 1040 (2d Cir. 1972), aff’d, 414 U.S. 291, 94 S.Ct. 505, 38 L.Ed.2d 511 (1973) (four members of court having eight active members are not empowered to grant rehearing en banc).
Appellant next argues that, assuming the court is comprised at present of only three active judges, we are barred from granting an en banc rehearing because one three-judge panel is not authorized to overrule another. Reliance is placed largely on a provision in a proposed 1941 amendment to the Judicial Code — never enacted — that would have authorized en banc hearings “in a circuit where there are more than three circuit judges.” See Western Pacific Railroad Case, 345 U.S. 247, 251, 73 S.Ct. 656, 658, 97 L.Ed. 986 (1953). That this proposal was never passed by the Senate, and that instead the very" different language of present section 46(c) was enacted, should in itself suffice to rebut appellant’s argument. Moreover, the policies behind the en banc procedure, particularly the desirability of providing “that the active circuit judges shall determine the major doctrinal trends of the future for their court," American-Foreign Steamship, 363 U.S. at 690, 80 S.Ct. at 1339, apply as much to this circuit as to any other. When, as often occurs, panels contain one or more judges who are not regular members of the court, the same danger exists here as elsewhere that uniformity and stability of precedent will suffer. We have been directed to no authority indicating that the First Circuit was intended to be excluded from the provisions of section 46(c) and Rule 35, and we see no reason to refrain from employing the device of en banc rehearing where it is appropriate.
*921Appellant’s other contentions relate primarily to the internal procedures by which the court regulates the grant of en banc hearings. The Supreme Court has stated that such questions are “largely to be left to intramural determination by each of the Courts of Appeals,” American-Foreign Steamship, 363 U.S. at 688, 80 S.Ct. at 1339. Rule 35 and First Circuit Rule 15 provide sufficient notice to litigants of the process we follow to satisfy the mandate of the Western Pacific Railroad Case that such procedures be “clearly explained.” 345 U.S. at 267, 73 S.Ct. at 666. See Comment, In Banc Procedures in the United States Courts of Appeals, 43 Fordham L.Rev. 401, 403 (1974). Appellant’s procedural contentions are thus without merit.
[H] We note finally that appellant, pointing to the requirement of Fed.R. App.P. 35(a) that an en banc hearing “ordinarily” be granted only where the issue is of “exceptional importance” or where there is disuniformity in the decisions of the circuit, has questioned whether this case is an appropriate one for en banc consideration. As the word “ordinarily” suggests, the question of when to hold, or when not to «hold, an en banc hearing is within the discretion of this court. Even so, our decision to rehear en banc was demonstrably in conformity with the standards of Rule 35. As already indicated, the rehearing serves to maintain uniformity of circuit policy. The panel opinion, in our view, construed Donahue to impose a burden of proving lack of prejudice on the government so severe as to approach imposition of a rule of per se prejudice in joint representation cases where Foster warnings have not been given. Our opinion today serves to clarify the standard by which the issue of prejudice in Foster -type cases should be evaluated, a question we regard as of considerable importance, in the context of a case like this, and of evident controversy.8
Accordingly, for the reasons stated herein, and, insofar as we have specifically adopted them, the reasons stated in the panel opinion, the decision of the district court is affirmed.
. Martorano moved for a new trial on the ground of “newly discovered evidence” more than seven days after the verdict, see Fed.R. Crim.P. 33, contending he had not previously learned of the prejudice joint representation allegedly caused him. The district court ruled that ineffective assistance of counsel could not constitute newly discovered evidence for purposes of Rule 33, and treated Martorano’s motion instead as one to vacate his sentence under 28 U.S.C. § 2255. As the question whether this case comes to us on direct or collateral appeal would not affect the outcome, see text at page 915 infra, we express no opinion on the district court’s ruling.
. Evidence developed at the new trial hearing established that this letter, along with another purporting to be from Joseph S. Oteri to his client Martorano warning in general terms of the risks of joint representation, was prepared by Martin Weinberg, attorney of record for Brian Halloran and Oteri’s law partner. See 610 F.2d at 40-11.
. The ultimate aim of the Foster rule — a proper warning — could, in theory, be attained by full and careful disclosure in an attorney’s office or elsewhere, but the rule’s more immediate purpose — to prevent the uncertainty that arises when defendants make post-conviction claims of ignorance or misadvice — would still be frustrated. Thus while in United States v. Donahue, 560 F.2d 1039, 1044 (1st Cir. 1977), we said that the giving of out-of-court advice was relevant to the ultimate question of whether a defendant was prejudiced by the district court’s noncompliance with Foster, only a compelling showing of the most complete and effective out-of-court advice would ever suffice, by itself, to disprove prejudice in a case such as this. The showing made here fell short of that standard. Prejudice, therefore, must be ascertained here largely in terms of whether there was an actual conflict of interest which was likely to have subverted Martorano’s defense. Still, in our overall assessment of the prejudice caused by the court’s failure to give a Foster warning, we think it reasonable to take into account that Martorano, ■ a college-educated man, was not totally lacking in some advance warning of the pitfalls of joint representation, nor was he pressured into accepting joint representation. Cf. Donahue, 560 F.2d at 1041-42 (attorneys hired at the insistence of older co-defendant for whose benefit defense strategy allegedly tailored).
. Appellant has raised a number of challenges to our authority to grant an en banc rehearing in this case. These points will be discussed in Part IV of this opinion.
. We shall take for granted that Louis Pallotta would, in fact, have given reputation testimony adverse to Halloran, although the point is not entirely clear. Attorney Joseph Oteri testified that he was unaware of any such possibility. Attorney Martin Weinberg stated that he was aware of the “danger” this might take place, but could not recall having discussed this risk with Martorano. Martorano testified that Weinberg told him Louis Pallotta should not testify because of the possibility he would hurt Halloran on cross-examination. We will assume, for present purposes, that Louis would have given such testimony and that it would have been admissible pursuant to 18 U.S.C. §§ 892(c), 894(c). See United States v. Bowdach, 501 F.2d 220, 225-26 & n.7 (5th Cir. 1974), cert. denied, 420 U.S. 948, 95 S.Ct. 1331, 43 L.Ed.2d 426 (1975).
. It is true that if counsel regarded Halloran’s defense as 70% effective and Martorano’s as only, say, 20%, a decision might possibly be made by independent counsel for Martorano to call in even a potentially disastrous witness on the theory that a long shot was better than a clear loss. But if Martorano was in that desperate posture, the district court’s findings as to Louis’ defects as a witness give us little basis for believing that calling him would have resulted in a different outcome. If on the other hand, as seems more likely, defense counsel viewed Martorano’s situation as not without hope, their decision as to whether or not to call Louis would have been guided by essentially the same considerations as in Halloran’s case— namely, evaluation of whether or not Louis would likely be a believable witness.
. Presently there is one vacancy on the circuit. The original three judge panel deciding this case included two regular judges of this circuit plus a third outside judge. Obviously there would have been rio occasion for an en banc hearing had the original panel consisted of the three regular judges, nor, as a practical matter, would an en banc have been likely had the two regular judges on the original panel been in agreement.
. Finally, we note that appellant moved to strike the government’s supplemental brief on rehearing en banc as not authorized by the Federal Rules or by specific pérmission of the court. In the alternative, appellant requested permission to file a brief of his own. Such permission having been granted, and appellant having filed his brief, we see no possible harm to appellant. His motion to strike is denied.