(specially concurring):
The proper interpretation of Section 144 is a vexed matter with which I have long struggled. The belief of the parties that they are receiving even-handed justice, the apparency of justice to those not parties, the importance of both perceptions in maintaining the legitimacy of the judicial institution, the difficult decisions faced by a judge called upon to stand recused, and the practical implications of § 144 for the continued efficient functioning of the district courts in our circuit are some of the competing considerations. They are not easily harmonized, and, indeed, there may be no entirely satisfactory manner of implementing Section 144. And though I concur fully in the opinion of the court on the assumption that United States v. Berger, 255 U.S. 22, 41 S.Ct. 230, 65 L.Ed.2d 481 (1921), remains good law, I feel obliged to express my doubt that it does or should. For, in my respectful view, Berger represents an outdated rule which has been made tolerable in present circumstances only by engraftment of dubious exceptions.
The majority opinion reaffirms Berger’s antique rule that whatever “facts” the recusal affidavit may assert cannot be questioned but must be accepted as gospel.1 Such an approach gives free play to the unscrupulous or reckless affiant, willing to run his chance of a ponderous and unlikely prosecution for perjury — and perhaps in little danger, since Berger seems to say that “affidavit” assertions made on mere information and belief will suffice for § 144 purposes.2 Perjury charges have traditionally been based on falsely stated physical facts, and are rarely extended to representation of opinions. Suffice to say, establishing beyond reasonable doubt bad faith in assertion of a belief purportedly *106held at the time of an affidavit’s filing is no light task.3 For similar reasons, counsel will not likely fear disciplinary proceedings initiated by the local bar. I am reluctant to join in mandating a procedure which envisions, for example, that a judge must take as true an affidavit asserting, perhaps on “information and belief,” that he has recently engaged in an acrimonious personal dispute with a defendant — complete with particulars^ — and is therefore disqualified to sit in his case, when the judge well knows that the affidavit has misidentified him and is mistaken. We go far enough when we read § 144 as withdrawing from the judge decision of the final fact, his own actual bias. We should not require him to conduct such a curious and hypothetical proceeding as deciding whether an apprehension or bias is reasonably supported by whatever suppositious state of facts a daring and unscrupulous, or perhaps .merely misadvised and agitated, party may be willing to swear to.
Factual matters necessary to decision of preliminary questions, of which recusal is a prickly example, are routinely resolved by weighing and evaluating affidavits. See, e. g., Wright & Miller, Federal Practice and Procedure § 1373, at 714. Only three workable modes of deciding this particular question occur: (1) peremptory disqualification upon the mere filing of an affidavit; (2) decision of the issue by another magistrate than the one accused; or (3) decision either of actual bias or the reasonable appearance of it by the magistrate sought to be disqualified.
There are indications in the legislative history that peremptory disqualification was the legislative intent;4 but this construction has never been adopted by any court. Reference to another magistrate is utterly foreign to the statutory scheme and raises its own problems of administrative inconvenience and delay. The statute’s language gives fair support to the construction that the judge is not to determine bias-in-fact, and common sense supports the view that few if any humans can fairly decide whether they themselves are or are not biased in any given matter. But it is not too much to ask that a conscientious magistrate determine whether a given affidavit contains enough truth to fairly support a reasonable apprehension that he may be biased, or that an appellate court review that decision effectively. There is, therefore, no need to discern in § 144 a rule by which a party who really wants to do so and has the nerve can at pleasure disqualify any federal judge in a given proceeding by presenting to him a spurious set of ex parte “facts” which he cannot question — and by which his opponent can disqualify his first replacement by the same means.5
I freely admit that Berger appears on its face to foreclose my reading of § 144. The Berger decision, however, has not gone unscathed — even by its authors — in the many years since 1921. It is notorious that, faced with its quixotism, courts on the firing line have, addressing other issues, limited its scope in ways perhaps dubious. As is duly observed, for example, in Comment, 57 Minn.L.Rev. 749, 755 (1973), the lower federal courts have effectively rejected a liberal reading of Berger by such anomalous and desperate devices as requiring that the affidavit establish bias-in-fact, United States v. Gilboy, 162 F.Supp. 384, 393 (M.D.Pa.1958), by restricting the types of facts which may constitute a sufficient affidavit, Chessman v. Teets, 239 F.2d 205, 215 (9th Cir. 1956), rev’d on other grounds, 354 U.S. 156, 77 S.Ct. 1127, 1 L.Ed.2d *1071253 (1957), and by requiring that the bias be directed to the party personally, Cole v. Lowe’s Inc., 76 F.Supp. 872, 876 (S.C.Cal.1948), rev’d on other grounds, 185 F.2d 641 (9th Cir. 1950), cert. denied, 340 U.S. 954, 71 S.Ct. 570, 95 L.Ed. 688 (1951). The Supreme Court has directly approved one such inroad in United States v. Grinnell Corp., 384 U.S. 563, 86 5. Ct. 1698, 16 L.Ed.2d 778 (1966), holding that the alleged bias “must stem from an extrajudicial source and result in an opinion on the merits on some basis other than what the judge learned from his participation in the case.” Although the Grinnell Court cited Berger, the proposition stated can be construed far more broadly than anything mentioned there. Also, Berger to the contrary notwithstanding, it is far from clear that the Supreme Court in Grinnell felt itself bound to accept as given fact the allegations made by the affiants: in its opinion the Court quotes from what can only be the transcript of pretrial proceedings, a type of matter which the Berger Court went out of its way to declare irrelevant. In other instances the Court has denied certiorari and left standing inroads on and narrow interpretations of Berger. See, e. g., Pfizer, Inc. v. Lord, 456 F.2d 532 (8th Cir.), cert. denied, 406 U.S. 976, 92 S.Ct. 2411, 32 L.Ed.2d 676 (1972);6 Mirra v. United States, 379 F.2d 782, 787-88 (2d Cir.), cert. denied, 389 U.S. 1022, 88 S.Ct. 593, 19 L.Ed.2d 667 (1967); Ryan v. United States, 99 F.2d 864, 871 (8th Cir. 1938), cert. denied, 306 U.S. 635, 59 S.Ct. 484, 83 L.Ed. 1037 (1939). I am all but convinced that if faced with the facts of Berger today the Supreme Court would decide it otherwise. Being so persuaded, I would not lightly expose our circuit to such risks of wholesale disruption as an untimely resurrection of Berger in its pristine and literal form threatens, to be endured until the Supreme Court — grappling with the mighty concerns which face it — is able to reconsider these questions.
Finally, I realize that the approach to § 144 I advance is likely foreclosed even by previous panel decisions in our own circuit. Davis v. Board of School Comm’rs, 517 F.2d 1044 (5th Cir. 1975); United States v. Roca-Alvarez, 451 F.2d 843, 847-48 (5th Cir. 1971), rehearing granted, 474 F.2d 1274 (5th Cir. 1973); Beland v. United States, 117 F.2d 958, 960 (5th Cir.), cert. denied, 313 U.S. 585, 61 S.Ct. 1110, 85 L.Ed. 1541 (1941); Simmons v. United States, 89 F.2d 591, 592-93 (5th Cir.), cert. denied, 302 U.S. 700, 58 S.Ct. 19, 82 L.Ed. 540 (1937), and indeed Henry v. Speer, 201 F. 869 (5th Cir. 1913), quoted with approval in Berger. But I do not think such a position wise or practical, and the court en banc is free to adopt a different one. I would grasp the nettle now.
. And though it also reaffirms the settled rule that judicial actions cannot be made the basis of an accusation of bias, it should not be difficult for an inventive affiant to assert extra-judicial bases that disguise an aversion actually grounded in judicial philosophy and approach.
. 255 U.S. at 34, 41 S.Ct. 230.
. The author of Note, 79 Harv.L.Rev. 1435 (1966), found no cases in which a perjury charge stemming from a § 144 affidavit had ever been prosecuted during the (then) forty-five years since Berger. Id. at 1442. Indeed, I have found none to date.
. The chief sponsor of the bill, asked if the judge retained any discretion after the filing of the recusal affidavit, replied: “No, it provides the judge shall proceed no further with the case.” 46 Cong.Rec. 2627 (1911).
. Section 144 limits a party to “ . . . one such affidavit in any case.”
. In Pfizer the court was interested in a demonstration of actual personal bias. In order to decide the issue the court “carefully examined the petition, the underlying joint affidavit, the responses thereto, and portions of the record pertinent to petitioners’ claims.” 456 F.2d at 533 (emphasis added). And, although accepting as true the facts recited in the affidavit, “[w]e determine the validity of petitioners’ conclusion of bias by examining the cited facts against the record presented to us." 456 F.2d at 537 (emphasis added).