Douglas P. Woodlock, Asst. U.S. Atty., Boston, Mass., with whom Edward F. Harrington, U.S. Atty., Boston, Mass., was on brief, for petitioner.

Gael Mahony, with whom Robert G. Dreher, and Hill & Barlow, Boston, Mass., were on brief, for respondent.

John S. Leonard and The McLaughlin Brothers, Boston, Mass., on brief, for James A. Kelly, Jr.

Before COFFIN, Chief Judge, BOWNES, Circuit Judge, BREYER, Circuit Judge.

COFFIN, Chief Judge.

This petition for mandamus is brought by a United States Attorney and seeks the enforced recusal of a United States district judge from further proceedings in a federal criminal prosecution.

During the spring of 1981, former Massachusetts state Senator James A. Kelly, Jr., was tried in Boston for extortion in violation of federal law. 18 U.S.C. § 1951. The government charged that he had wielded his power as a state senator and Chairman of the Senate Ways and Means Committee to exact $34,000 from an architectural firm, promising to influence the award of architectural contracts. United States District Judge Joseph L. Tauro, assigned by random draw, conducted the 25-day trial which ended on April 29 in the declaration of a mistrial when the jury reached an eleven-to-one deadlock after approximately thirteen hours of deliberation.

On June 16, in the course of proceedings preparatory to a retrial, the United States Attorney responsible for the prosecution filed a motion for disqualification of the judge under 28 U.S.C. § 455(a). The judge, on July 9, issued a lengthy memorandum discussing the sufficiency of the allegations justifying recusal and denied the motion. He further suggested that this court in the exercise of its supervisory powers give prompt consideration to the propriety of his action. We declined to take this approach. Subsequently the prosecutor brought this petition.

The Facts

The essence of the prosecution's motion for disqualification and this petition is that the combination of (1) the judge's past and present close professional and personal relationship with former Massachusetts Governor John Volpe, (2) defendant Kelly's reported helpfulness to the governor during a 1966 legislative investigation chaired by Kelly, and (3) the judge's own reported involvement in that proceeding as the governor's legal counsel would compel a reasonable observer to doubt that the judge would be impartial in any future proceeding or retrial connected with this prosecution. The prosecutor makes a subsidiary argument: because of the appearance of bias arising from these factors, several rulings during the trial lend themselves to the interpretation that they were the result of the bias.

We discuss the relevant facts in several time frames-what the prosecutor was aware of before the case went to trial, what was reported by newspaper columnists soon after the mistrial was declared, and what was revealed by an FBI investigation spurred by one of the newspaper accounts. Before the case went to trial, the U.S. Attorney was aware that the judge, before ascending to the bench, had been chief legal counsel for Governor Volpe in 1966 when Senator Kelly, a Democrat, had chaired a legislative investigation of the Republican Volpe administration; that the investigation focussed on the way in which the administration awarded architectural contracts; that Kelly had supposedly handled the matter in a manner favorable to the administration; and that there were vague rumors to the effect that the judge and Kelly had known each other during that period.

Early in January, 1981, the U.S. Attorney approached the judge about the possibility that he should disqualify himself from another case. In the course of discussing this matter, the judge brought up the fact that a reporter had just told him he should not sit on the Kelly case. He went on to say that the only role he had played in the administration's defense during the 1966 investigation was to consult with attorney Walter McLaughlin who represented Governor Volpe's brother, also under investigation. At this point, the U.S. Attorney felt that he had no specific, reliable information about the roles of the judge and Kelly in the investigation or the degree of the judge's relationship with Volpe.

Subsequently, during the trial, the prosecutor received queries "from a variety of sources, including members of the bar, the judiciary, the press and politicians" about the propriety of the judge presiding over this particular trial. The issue came into public view when two articles appeared in the Boston papers. On May 20, three weeks after the declaration of mistrial, Boston Herald American columnist, Peter Lucas, wrote that "Tauro knew Kelly. State House observers believed that Kelly went in the tank to Volpe on that investigation." While publicly questioning the propriety of the judge continuing to preside over the case, the columnist did not supply the prosecutor with any new information.

Shortly thereafter another columnist, David Wilson, writing for the Boston Globe, brought the issue to a head. In a June 1 column he noted that in 1966 the judge had been in the inner circle of Governor Volpe's Republican campaign forces, that the judge owed his appointment as a judge to Volpe, that Volpe's campaign had been threatened by the investigation, and that Kelly was the chairman of the investigatory committee. "(W)hen the Volpe organization communicated its wishes to Chairman Kelly, the communicator, so far as anyone in Kelly's hearing room could tell, was Atty. Joseph L. Tauro. Frequent Tauro-Kelly conferences had Democratic members seething." Wilson stated that Kelly had wanted to withhold publication of the committee's report until after the election. The impression conveyed was that the judge appeared to have reason for gratitude for favors rendered to him as the result of his personal communications with Kelly.

The Wilson article caused the prosecutor to instigate an FBI investigation to probe the facts of the reported relationships. He felt that inaction was no longer appropriate: specific allegations-in particular the labelling of the judge as "communicator"-created a previously unknown cause for concern. The FBI investigation, which involved interviews with some fifteen persons who had been involved in or close to the 1966 hearings, turned up no evidence whatsoever that the judge had ever communicated with Kelly concerning the investigation or had ever had any conferences with him. Although the investigation provided no support for the triggering allegation that the judge was "the communicator", it did reveal in detail the roles of various individuals in the 1966 hearings.

One focus of the investigation was the nature of the judge's involvement in the 1966 hearings. He did not directly participate, there being other attorneys representing the specific targets of the contract award inquiry, but he did play a substantial role by preparing witnesses, planning strategy, consulting closely with Walter McLaughlin, the attorney representing the governor's brother, attending hearings as an observer, and reporting to the governor. There was no evidence that he ever conferred with any of the committee members.

A second focus of the FBI investigation was the issue of bias in favor of the Volpe administration on the part of Senator Kelly. The principal source of information tending to show such bias was a conversation between FBI agents and one Beryl W. Cohen, in 1966 a senator and member of Kelly's committee and also a Democratic candidate for the office of lieutenant governor. Cohen recalled that Kelly had initially tried to control the hearings, not allowing others to ask questions and not asking probing questions himself, but soon lost this control when others gained the right to ask questions. After the hearings were over, Cohen, feeling that Kelly was not going to submit a written report, found someone to prepare a report. Cohen felt there had to be some arrangement between Kelly and the Volpe administration but never learned what it was. In addition to Cohen's views, the investigation revealed that Kelly had characterized the investigation as "political", adding that "everything is political", had leaked a copy of the report to the governor's office, and, though a Democrat, had supported the governor's sales tax proposal.

Factors in the report tending to contradict or minimize Kelly's bias include Cohen's posture as a candidate for top office in an election year who had an incentive to exploit to the maximum any legislative inquiry into an opposing administration; the fact that the committee had neither staff nor funding for help in preparing a report; and the fact that in any event Kelly not only signed the report but presented it at a press conference. The report did contain damaging material but had little impact, Volpe winning reelection overwhelmingly. One conclusion emerges clearly. Whether or not Kelly was a biased chairman, by conviction or prearrangement, he does not seem to have had much influence on the outcome of his committee's work.

A third area of information in the FBI report was the close personal and professional relationship between the judge and Governor Volpe. Professionally, the judge, when serving as legal counsel, was also part of Volpe's "kitchen cabinet" and, indeed, was the only person apart from the chief secretary to have direct access to the governor. The ties of personal friendship were close and longstanding. Volpe's wife was the attending nurse at the judge's birth. The judge's father previously had served as Volpe's legal counsel until Volpe appointed him to the Massachusetts Superior Court, and, reportedly, the judge owed his nomination for the position of district court judge to Volpe's efforts. The judge's personal regard for Volpe continues to this day.

In addition to the FBI report, the prosecutor was influenced by what he called the "parallelism" of the 1966 investigation and the Kelly trial. The subject matter at issue-extortion in the granting of architectural contracts-was the same. The cast of characters was similar, although their positions had changed. The judge had been legal counsel in 1966. Kelly, then "prosecutor", was now defendant. Moreover, Kelly was represented at trial by George A. McLaughlin, nephew and then-partner of the Walter McLaughlin who has represented Peter Volpe in the 1966 investigation, although now engaged in a separate practice. Finally, two key government witnesses against Kelly, Frank and William Masiello of the firm from which Kelly allegedly extorted money, had recently testified before an investigating commission that people formerly associated with the Volpe administration, including members of the "kitchen cabinet", were involved in arranging awards of architectural contracts in exchange for political contributions.

In sum, the prosecutor feared that the judge's involvement in the 1966 investigation would have made him aware of Kelly's apparently favorable actions in a proceeding potentially damaging to his close personal, political, and professional associate, Volpe, and that this incentive to return a favor, together with the "parallelisms", could give the public reason to believe that the judge might favor Kelly in the present proceeding whenever possible. In addition, there was the possibility that the Masiellos' previous tarnishing of the Volpe reputation might create the suspicion that the judge would be unduly harsh upon the Masiellos when appearing as witnesses in this case. The government points to rulings made during Kelly's trial that it says substantiate the appearance of partiality. All of this evidence the government found sufficient to justify the filing of a motion for disqualification.

Before beginning our analysis of the propriety of the judge's denial of that motion, we note that these relationships and parallelisms were in large outline obvious long before the Kelly trial began. Some of the information provided by the FBI investigation may have given the prosecutor new detail, but much duplicated what had already been known. These circumstances illustrate the problem that arises when a litigant files a motion for recusal after the trial is concluded. He may have proceeded with the first trial to test the court's reaction only later to marshal previously known information in an attempt to get the proverbial second bite at the apple. Even if a litigant were not consciously attempting to manipulate the circumstances to his benefit, the potential waste of judicial resources alone requires that a motion for disqualification be timely filed. See generally In re International Business Machines Corp., 618 F.2d 923, 932-34 (2d Cir. 1980). While we accept the prosecutor's position that he had not felt that an earlier motion to disqualify would have been justified by the facts then available to him and do not find this motion to be untimely, the issue as to timeliness is close. A ruling that the motion to disqualify was not timely might arguably have been sustainable.

Our Standard of Review

Although we maintain a standing watch over attempted piecemeal review, whether by interlocutory appeal or petitions for writs of mandamus, see In re Continental Investment Corp., 637 F.2d 1 (1st Cir. 1980), the issue of judicial disqualification presents an extraordinary situation suitable for the exercise of our mandamus jurisdiction. See, e.g., In re International Business Machines Corp., supra, 618 F.2d at 927; In re Corrugated Container Antitrust Litigation, 614 F.2d 958, 961 n.4 (5th Cir.), cert. denied, 449 U.S. 888 , 101 S.Ct. 244, 66 L.Ed.2d 114 (1980); SCA Services, Inc. v. Morgan, 557 F.2d 110, 117 (7th Cir. 1977) (per curiam); In re Rodgers, 537 F.2d 1196, 1197 n.1 (4th Cir. 1976) (per curiam); cf. Pfizer, Inc. v. Lord, 456 F.2d 532, 536 (8th Cir. 1972) (per curiam). A case involving a motion for disqualification is clearly distinguishable from those where a party alleges an error of law that, despite the hardship of delay, may be fully addressed and remedied on appeal. See United States v. Kane, 646 F.2d 4, 9-10 (1st Cir. 1981). In the case at bar, the issue of partiality has been broadly publicized, and the claim of bias cannot be labelled as frivolous and deferred until final appeal. "(P)ublic confidence in the courts (requires) that such a question be disposed of at the earliest possible opportunity." In re Union Leader Corp., 292 F.2d 381, 384 (1st Cir.), cert. denied, 368 U.S. 927 , 82 S.Ct. 361, 7 L.Ed.2d 190 (1961).

Turning now to the question of the standard governing our decision, we start with the statutory provision that a judge "shall disqualify himself in any proceeding in which his impartiality might reasonably be questioned". 28 U.S.C. § 455(a). We deem it of first importance to recognize at the outset the twin-and sometimes competing-policies that bear on the application of this standard. The first and most obvious policy is that courts must not only be, but must seem to be, free of bias or prejudice. To ensure that the proceedings appear to the public to be impartial and hence worthy of their confidence, the situation must be viewed through the eyes of the objective person. See H.Rep.No. 1453, 93d Cong., 2d Sess., 1974 U.S.Code Cong. & Admin.News 6351, 6355.

A second and less obvious policy is that a judge once having drawn a case should not recuse himself on a unsupported, irrational, or highly tenuous speculation; were he or she to do so, the price of maintaining the purity of appearance would be the power of litigants or third parties to exercise a negative veto over the assignment of judges. Because there exists this second policy, our inquiry cannot stop with the questions: have a number of people thought or said that a judge should not preside over a given case? has the judge's failure to recuse himself been a subject of unfavorable comment in the media? or, would the judge have avoided controversy and the need for appellate review if he had stepped aside? Instead, we must conduct our review in accordance with ground rules designed to determine when the fear of partiality is real and strong enough to require disqualification.

First, a charge of partiality must be supported by a factual basis. See, e.g., United States v. Mirkin, 649 F.2d 78, 82 (1st Cir. 1981); 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); accord, H.Rep.No. 1453, 1974 U.S.Code Cong. & Admin.News, supra, at 6355. Although public confidence may be as much shaken by publicized inferences of bias that are false as by those that are true, a judge considering whether to disqualify himself must ignore rumors, innuendos, and erroneous information published as fact in the newspapers. Compare United States v. Cepeda Penes, 577 F.2d 754, 758 (1st Cir. 1978) with Spires v. Hearst Corp., 420 F.Supp. 304, 307 (D.C.Cal.1976). To find otherwise would allow an irresponsible, vindictive or self-interested press informant and/or an irresponsible, misinformed or careless reporter to control the choice of judge.* Second, disqualification is appropriate only if the facts provide what an objective, knowledgeable member of the public would find to be a reasonable basis for doubting the judge's impartiality. Were less required, a judge could abdicate in difficult cases at the mere sound of controversy or a litigant could avoid adverse decisions by alleging the slightest of factual bases for bias. See H.Rep.No. 1453, 1974 U.S.Code Cong. & Admin.News, supra, at 6355. This restricted mandate to disqualify is calculated to induce a judge to tread the narrow path between timidity and tenacity.

We recognize that the analysis of allegations, the balancing of policies, and the resulting decision whether to disqualify are in the first instance committed to the district judge. And, since in many cases reasonable deciders may disagree, the district judge is allowed a range of discretion. The appellate court, therefore, must ask itself not whether it would have decided as did the trial court, but whether that decision cannot be defended as a rational conclusion supported by reasonable reading of the record. Moreover, in this case, which comes to us by a petition for mandamus, the party seeking the writ of mandamus must show a "clear and indisputable" right to relief. Kerr v. United States District Court, 426 U.S. 394, 403, 96 S.Ct. 2119, 2124, 48 L.Ed.2d 725 (1976); In re International Business Machines Corp., supra, 618 F.2d at 926; In re Corrugated Container Antitrust Litigation, supra, 614 F.2d at 962.

Applying the Standard

We now turn to consideration of whether on the facts before us a reasonable person would clearly have doubted the judge's impartiality. Certainly some of the public and press did question his impartiality, but to the extent that the doubts were created by representations of the press shown to be not grounded in fact, they cannot require disqualification. What we have referred to as the event triggering the FBI investigation, the Wilson article, aroused fears by publishing the claim that Tauro has been "the communicator" with Kelly, holding frequent conferences with Kelly that caused resentment. This centerpiece claim never having been supported in the slightest degree, we must disregard it and any impressions of partiality it may have created. Attempting to exorcise this factor from our analysis is difficult, but our analysis must be limited to more indirect relationships and potentials for bias.

Basic to our analysis is the undisputed conclusion that the judge did in fact have a close relationship with Governor Volpe. It is from this relationship that the government infers two different causes for an appearance of partiality: it reasons that the judge's allegiance to Volpe would appear to cause him to reward Kelly for handling the 1966 investigation in a manner favorable to Volpe, and that his allegiance to Volpe might make the public doubt his ability to respond impartially to the Masiellos when they appear as witnesses at Kelly's trial. The success of the argument that the judge's gratitude to Kelly will cause an appearance of impartiality depends upon finding that a reasonable person would conclude (1) that Kelly actually did act in a manner favorable to Volpe during the 1966 investigation, (2) that the judge was aware of Kelly's actions, and (3) that the feeling of gratitude was so deep and enduring as to cause him to act in a biased manner today.

Addressing the second factor first, we think it clear that the judge would have been aware of the manner in which Kelly conducted the hearings and sensitive to any favorable treatment given the administration. The first query-whether Kelly actually did favor the administration-is more difficult. As our earlier discussion of the investigation indicates, Kelly's actions can be interpreted either as very partial toward the administration if viewed in light of then-Senator Cohen's comments, or as fair in that Kelly asked some pointed, factual questions incriminating the administration and publicly stood behind the report when it was written. Given this contradictory evidence, the linchpin of the government's case is weak. Kelly may have been no more than a chairman interested in a fair hearing. Even indulging inferences favorable to the prosecutor, we conclude that at most Kelly's actions portray someone with less than a killer instinct or someone in the circumstances unable to give effective voice to such an instinct, or both. This perception, however, might well have provided a basis for the judge and others in Volpe's camp to be thankful, at least until Kelly lost control of the committee.

The remaining factor-whether the judge's awareness of Kelly's favorable conduct fifteen years ago would prompt a reasonable observer to believe that the judge would be partial toward Kelly in a current criminal trial-is critical. As we have observed, even if Kelly's motivation may have been to give the strongest help to the Volpe administration, his actual performance provided little assistance. Whether or not this was enough to generate a reasonable expectation of gratitude in the months immediately following the hearing, it surely lacks the significance to endure a decade and a half. Even, however, if one may assume the survival of some residue of gratitude after such a period, it is beyond contemplation that such gratitude would be of the weight necessary to cause a judge to jettison his impartiality and, in open court day after day, to violate his deepest professional and ethical commitments as a judge.

We turn to the second alleged reason for disqualification, that the Masiellos' impugning of the Volpe administration during a recent investigation might give rise to suspicions that the judge would mistreat them during trial. While the Masiellos' testimony before the commission may have made the judge feel uncomfortable by reason of his association with the incriminated kitchen cabinet members and defensive out of loyalty to Volpe, we do not think it clear enough that an objective person would find either his discomfort or loyalty sufficient to cause a reasonable doubt about his ability to deal with the Masiellos today in an impartial manner. Moreover, because the Masiellos are not repeating their previous testimony at the present trial and are not a party to the trial with any stake in the evidentiary rulings, it would seem to us that both the motive and opportunity for significant reprisal would be slim.

Our conclusions with respect to both of the possible causes of bias-gratitude to Kelly for his effort to help an old friend and hostility to the Masiellos for their damaging testimony-are strengthened by the implications of an opposite outcome. If the receipt by a judge's friend of a favor long ago from one who is a present litigant should disqualify the judge, judges could hope to preside without challenge solely in communities in which they are strangers. For when a judge presides in an area where he and his family have lived for one or more generations, the numbers of people who have, directly or indirectly, helped family members, relatives, close friends, and friends of friends would form a large and indeterminate community. So also are there bound to be indefinite numbers of people who have been critical of or been on opposite sides of controversies with families, relatives, and friends. Not only would the role of judges be severely constricted by requiring disqualification under these circumstances but the result would reflect a more jaundiced view as to when there should be a reasonable doubt about a judge's impartiality than accords with the public perception.

We freely recognize that there are many instances when past associations may require disqualification and we do not lightly deny relief in this instance. We are constrained, however, by our standard of review, requiring (1) that the government show that the judge has violated his considerable discretion to balance the need to disqualify himself to preserve public faith in impartiality against the reasons for sitting when there is not sufficient reason for disqualification and (2) that this showing be strong enough to establish a "clear and indisputable" right to a writ of mandamus. Even viewing all of the facts and inferences in a manner most favorable to the government, we cannot say that the only conclusion on this record as a matter of law is that the judge should have disqualified himself. Such a ruling would go far to establish the precedent that any suspicion of partiality, though tenuous and remote in origin, would suffice to compel disqualification.

Despite our conclusion that the factual basis is not sufficient to persuade a reasonable observer of an appearance of partiality, we take one additional step and look at the judge's conduct at trial to see whether it reveals any grounds that might cause an observer to doubt his impartiality. It may well be that our inquiry should end before taking this step. See Brody v. President & Fellows of Harvard College, 664 F.2d 10 at 12 (1st Cir. Nov. 16, 1981). The government has stated that the judge's conduct at trial is significant only in the context of the appearance of bias arising from the relationships discussed above, a context already demonstrated not to support such an appearance. Recognizing, however, that this case is largely a product of interest generated by the press and that the interest continues to be widespread, and wishing to avoid the impression that we have not examined all claims that have been made, we proceed to examine the judge's challenged trial conduct. We have reviewed in detail the government's claims that certain rulings evidence an appearance of partiality and report our findings very summarily. More detail would dignify a kind of attack that should be reserved for only the most egregious rulings, the danger being that to give more detailed answers would allow the government an appeal on trial rulings to which it is not now entitled.

In the course of a twenty-five day trial we can assume that there must have been at least several hundred rulings. Of these, the government has objected primarily to only a handful of evidentiary rulings on the testimony of five witnesses and the admission of one exhibit. In addition, it cites the judge's jury instructions and his declaration of a mistrial as evidence of an appearance of bias. None of these objections, as illustrated below, have substance enough to fill the gap we have already found to exist in the government's case.

The government objects to a ruling that prevented examination of one of the witnesses' state of mind about the purpose of the "retainer" he paid to Kelly's firm when, regardless of the correctness of that ruling, the witness was still allowed to testify over Kelly's objection to a similar end that he had told another person that Kelly was trying to extort money from the corporation. Another objection is that the government was not allowed for technical reasons to rehabilitate one of its witnesses, a ruling which, again, whether correct or not, was counterbalanced by the fact that the same witness testified that his purpose in authorizing the retainer to the Kelly firm was to obtain new business. This was a basis, if believed, for finding Kelly guilty. Another example is the government's complaint about the exclusion of a prior inconsistent statement, an objection that ignores the fact that such statements are admissible only if made when the witness had no reason to testify falsely and that the witness had already been promised immunity. Still another is the objection to the exclusion of the testimony that allegedly would have affected the running of the limitations period by clearly establishing when the most recent payments had occurred. Our reading of the records, however, has revealed no specific exclusions of specific dates.

One last contention that we note is the argument that the judge appeared to be biased when he declared a mistrial after the jurors had reached an eleven-to-one deadlock following approximately thirteen hours of deliberation. The government thinks the time for deliberation too short to declare a mistrial, given a twenty-five day trial, but it had agreed to the volley of notes to the jury that probed the issue of deadlock and precipitated the declaration of mistrial. We emphasize in passing the salient fact that the division in the jury was eleven for conviction to one for acquittal. An objective observer, we think, would conclude that this result hardly points to a twenty-five day effort by an able judge to favor the defendant.

In sum, we might, if forced to make rulings now-a result that should be avoided to prevent the inevitable distortion of an appeal-find that some of the judge's rulings were indisputably correct, some marginal, and some in error, though whether reversible or not would, without more specific research, be impossible to say. This is not the kind of record in light of the total context that would compel us to require that the judge not conduct any further proceedings in this matter.

The petition for writ of mandamus is denied.

* In this respect, the standard for judicial disqualification is not quite on the same plane as the standard for judging Caesar's wife. It will be recalled that Caesar dismissed his wife Pompeia for having supposedly been the object of an amorous house breaking committed by Clodius even though he had nothing to charge Clodius with when he was summoned as a witness. Caesar dealt with the paradox by saying, "I wish my wife to be not so much as suspected." Plutarch, The Lives of the Noble Grecians and Romans 860 (Modern Library, N.Y.). In other words, Caesar was willing to base the decision as to spousal removal on suspicion alone, whether reasonable or not