delivered Parts I, II, and III of the opinion of the court.
GOODWIN, Circuit Judge, delivered Parts IV and V of the opinion of the court.
Winston McConney was convicted on stipulated facts of violating 18 U.S.C. § 922(h) *1198which prohibits a convicted felon from receiving firearms shipped in interstate commerce. His appeal challenges the denial of his timely motion to suppress evidence as illegally obtained. We took the case en banc for the specific purpose of resolving the issue of the appropriate standard of appellate review for the mixed question of law and fact of “exigent circumstances.”
I
On June 12, 1979, an indictment was filed in the United States District Court for the Northern District of California accusing McConney, and thirty-one other defendants, of violating title IX of the Organized Crime Control Act of 1970. This title, commonly known as the Racketeer Influenced and Corrupt Organizations Act (RICO), prohibits conducting or participating in the conduct of an enterprise through a pattern of racketeering activity. 18 U.S.C. § 1962. The indictment identified the RICO “enterprise” as the Hell’s Angels Motorcycle Club.
On the day after the indictment was filed, federal officers executed an “indicia” warrant authorizing a search of McConney’s residence and seizure of any indicia of membership in or association with the Hell’s Angels Motorcycle Club. In addition to the indicia warrant, the searching officers possessed an arrest warrant and a Prescott warrant.1
On the evening of June 13, 1979, federal agents executed the arrest and search warrants at McConney’s residence. When the agents approached the home at approximately 8:30 p.m., the solid front door was open but an inner screen door was closed. The lead agent, Olson, knocked on the door and announced his identity and purpose. He saw inside a person he recognized immediately as McConney. Between McConney and the door was a second person, sitting with his back to the door. Without waiting for a response or a refusal of entry, Olson opened the door and led the other agents into the living room. McConney and the other person were ordered to move several feet to a position on the floor next to a sofa in the same room. While the two men were being handcuffed, another agent discovered a loaded pistol beneath one of the sofa cushions. A subsequent search of the house made under the indicia warrant yielded a second weapon.
Following the discovery of the two firearms, a count was added in a superseding indictment charging McConney with violating 18 U.S.C. § 922(h).
After the district court denied McCon-ney’s suppression motion, the government moved to drop the RICO charges against him and proceeded on the firearms charge. McConney saved his objection to the denial of his motions, waived his right to a jury trial, and agreed to a court trial on a written stipulation of facts.2 The court found McConney guilty as charged. The appeal from the resulting judgment of confinement challenges the denial of the suppression motion.
McConney contends that the entry into his home violated the federal “knock-notice” requirement which provides that an officer, before opening a door of a house in order to enter, must give notice of his identity and purpose and be refused admittance by the occupant. 18 U.S.C. § 3109.3 As this court recently stated, “section 3109 codifies a tradition embedded in Anglo-American law and declares the reverence which the law attaches to an individual’s right of privacy in his house.” United States v. Whitney, 633 F.2d 902, 908 (9th *1199Cir.1980), cert. denied, 450 U.S. 1004, 101 S.Ct. 1717, 68 L.Ed.2d 208 (1981).
In Whitney, this court addressed the problem of officers entering a house upon announcing their identity and purpose without first awaiting a refusal or admittance. The court recognized that compliance with section 3109’s requirements may be excused by exigent circumstances. Id. at 908. The government’s claim here, as in Whitney, is that exigent circumstances justified the agent’s failure to await refusal of admittance.
The district court found that the agents had knocked and announced their identity, and that their simultaneous entry (without waiting for refusal of admittance) was justified by exigent circumstances. The district court also found that seizure of the first pistol was an incident of a lawful arrest.
We define exigent circumstances as those circumstances that would cause a reasonable person to believe that entry (or other relevant prompt action) was necessary to prevent physical harm to the officers or other persons, the destruction of relevant evidence, the escape of the suspect, or some other consequence improperly frustrating legitimate law enforcement efforts.
II
We turn first to the question of what standard of review is applicable to the district court’s determination that the federal agents’ failure to comply with the requirements of section 3109 was excused by exigent circumstances. In United States v. Flickinger, 573 F.2d 1349, 1356-57 & n. 2 (9th Cir.), cert. denied, 439 U.S. 836, 99 S.Ct. 119, 58 L.Ed.2d 132 (1978), this court held that the “mixed fact-law question” of exigent circumstances is factual in nature and therefore reviewable on appeal under the deferential, clearly erroneous standard. We took this case en bane to decide whether Flickinger should be overruled.
A
In Flickinger, the panel founded its determination that the question of exigent circumstances is subject to deferential review on two precedents it considered to be “analogous”: United States v. Hart, 546 F.2d 798 (9th Cir.1976) (en banc), cert. denied sub nom. Robles v. United States, 429 U.S. 1120, 97 S.Ct. 1155, 51 L.Ed.2d 571 (1977), and United States v. Page, 302 F.2d 81 (9th Cir.1962) (en banc). In Hart, we held that the determination that government agents had done everything “reasonably necessary and proper” to make a witness available is factual in nature and consequently subject to the clearly erroneous test. 546 F.2d at 801-02. Similarly, in Page, we held that whether consent to a search was “freely and intelligently given” is a factual issue properly subject to clearly erroneous review. 302 F.2d at 85. On the basis of these cases, the Flickinger panel concluded:
Our experience dictates that the question of exigent circumstances is fundamentally the same type of issue as the questions of voluntariness of a consent and whether officers had done everything reasonably necessary to produce a witness. Certainly, a finding of exigent circumstances is no less based on the “fact-finding tribunal’s experience with the mainsprings of human conduct.” Commissioner of Internal Revenue v. Duberstein, 363 U.S. 278, 289, 80 S.Ct. 1190, 1198, 4 L.Ed.2d 1218 (1960).
Flickinger, 573 F.2d at 1356.
In relying on Hart and Page, the Flick-inger panel rejected as governing precedent another related line of cases — the Ninth Circuit decisions concerning the proper standard of review for the question of probable cause. While noting that these decisions appeared to conflict with one another, compare United States v. Thompson, 558 F.2d 522, 524 (9th Cir.1977) (probable cause reviewable under the clearly erroneous test), with United States v. One Twin Engine Beech Airplane, 533 F.2d 1106, 1108 (9th Cir.1976) (probable cause reviewable de novo), the Flickinger panel found it unnecessary to resolve this conflict because it concluded that the question of exigent circumstances is “more closely related to the *1200mixed fact-law question involved in United States v. Hart ... than the ultimate constitutional standard of probable cause.” 573 F.2d at 1356 n. 2.4
In Flickinger’s discussion of the appropriate standard of review for the question of exigent circumstances, we find a number of indications that our jurisprudence concerning appellate review of mixed questions lacks clarity and coherence. First, we are struck by the difficulty the Flickinger panel had in determining which of our standard of review precedents controlled the case before it. This difficulty doubtless reflects the absence of clear legal guideposts in this area of the law. Second, an even clearer indication of disarray is the conflict in our precedents concerning the appropriate standard of review for the question of probable cause which the Flickinger panel noted.
This disarray in standard of review jurisprudence appears to be pervasive. The Supreme Court recently stated that “there is substantial authority in the circuits on both sides of [the question of the applicability of the rule 52(a) clearly erroneous standard to mixed questions of law and fact].” Pullman-Standard v. Swint, 456 U.S. 273, 289-90 n. 19, 102 S.Ct. 1781, 1790-91 n. 19, 72 L.Ed.2d 66 (1982). The Court also acknowledged that, while it has usually reviewed mixed questions independently, its precedents are not entirely consistent and there is support in its decisions for clearly erroneous review of some mixed questions. Id. (contrasting Bogardus v. Commissioner, 302 U.S. 34, 58 S.Ct. 61, 82 L.Ed. 32 (1937), and Helvering v. Tex-Penn Oil Co., 300 U.S. 481, 57 S.Ct. 569, 81 L.Ed. 755 (1937), with Commissioner v. Duberstein, 363 U.S. 278, 80 S.Ct. 1190, 4 L.Ed.2d 1218 (1960), and Commissioner v. Heininger, 320 U.S. 467, 64 S.Ct. 249, 88 L.Ed. 171 (1943)).
In light of the disarray noted in Pullman-Standard, we begin our reexamination of Flickinger by returning to first principles of standard of review jurisprudence in an effort to formulate an analytical framework for dealing with mixed questions.
B
The Supreme Court has defined mixed questions as those in which “the historical facts are admitted or established, the rule of law is undisputed, and the issue is whether the facts satisfy the [relevant] statutory [or constitutional] standard, or to put it another way, whether the rule of law as applied to the established facts is or is not violated.” Pullman-Standard v. Swint, 456 U.S. 273, 289 n. 19, 102 S.Ct. 1781, 1790 n. 19, 72 L.Ed.2d 66 (1982). Thus, there are three distinct steps in deciding a mixed fact-law question. The first step is the establishment of the “basic, primary, or historical facts: facts ‘in the sense of a recital of external events and the credibility of their narrators .. . Townsend v. Sain, 372 U.S. 293, 309 n. 6, 83 S.Ct. 745, 755 n. 6, 9 L.Ed.2d 770 (1963) (quoting Brown v. Allen, 344 U.S. 443, 506, 73 S.Ct. 397, 446, 97 L.Ed. 469 (1953) (opinion of Frankfurter, J.)). The second step is the selection of the applicable rule of law. The third step — and the most troublesome for standard of review purposes — is the application of law to fact or, in other words, the determination “whether the rule of law as applied to the established facts is or is not violated.” Pullman-Standard, 456 U.S. at 289 n. 19, 102 S.Ct. at 1790 n. 19.
The district court’s resolution of each of these inquiries is, of course, subject to appellate review. The appropriate standard of review for the first two of the district court’s determinations — its establishment of historical facts and its selection of the relevant legal principle — has long been settled. Questions of fact are reviewed under the deferential, clearly erroneous standard. See Fed.R.Civ.P. 52(a).5 *1201Questions of law are reviewed under the non-deferential, de novo standard. See, e.g., United States v. One Twin Engine Beech Airplane, 533 F.2d 1106, 1108 (9th Cir.1976); Lundgren v. Freeman, 307 F.2d 104, 115 (9th Cir.1962). These established rules reflect the policy concerns that properly underlie standard of review jurisprudence generally.
Rule 52(a)’s mandate that appellate courts not disturb a trial court’s findings of fact unless clearly erroneous serves two policy objectives. First, it minimizes the risk of judicial error by assigning primary responsibility for resolving factual disputes to the court in the “superior position” to eval- - uate and weigh the evidence — the trial court.6 Rule 52(a) emphasizes that the trial judge’s opportunity to judge the accuracy of witnesses’ recollections and make credibility determinations in cases in which live testimony is presented gives him a significant advantage over appellate judges in evaluating and weighing the evidence: “findings of fact shall not be set aside unless clearly erroneous, and due regard shall be given to the opportunity of the trial court to judge of the credibility of the witnesses.” Fed.R.Civ.P. 52(a). Second, because under the clearly erroneous test, the reviewing court will affirm the trial court’s determinations unless it “is left with the definite and firm conviction that a mistake has been committed,” Pullman-Standard v. Swint, 456 U.S. 273, 284-85 n. 14, 102 S.Ct. 1781, 1788 n. 14, 72 L.Ed.2d 66 (1982), it is relieved of the burden of a full-scale independent review and evaluation of the evidence.7 Consequently, valuable appellate resources are conserved for those issues that appellate courts in turn are best situated to decide.
The converse rule — that conclusions of law are subject to plenary or de novo review — reflects similar concerns. Structurally, appellate courts have several advantages over trial courts in deciding questions of law. First, appellate judges are freer to concentrate on legal questions because they are not encumbered, as are trial judges, by the vital, but time-consuming, process of hearing evidence. Second, the judgment of at least three members of an appellate panel is brought to bear on every case.8 It stands to reason that the collaborative, deliberative process of appellate courts reduces the risk of judicial error on questions of law. Thus, de novo review of questions of law, like clearly erroneous review of questions of fact, serves to minimize judicial error by assigning to the court best positioned to decide the issue the primary responsibility for doing so.
De novo review of questions of law, however, is dictated by still another concern. Under the doctrine of stare decisis, appellate rulings of law become controlling precedent and, consequently, affect the rights of future litigants. Rulings on factual issues, on the other hand, are generally of concern only to the immediate litigants. From the standpoint of sound judicial administration, therefore, it makes sense to concentrate appellate resources on ensuring the correctness of determinations of law.
*1202Thus, we have a well developed standard of review jurisprudence for issues of fact and issues of law. Yet, when we review the third of the district court’s determinations — its application of law to fact — we confront “a much-mooted issue” with “substantial authority in the circuits on both sides of th[e] question.” Pullman-Standard, 456 U.S. at 289-90 n. 19, 102 S.Ct. at 1790-91 n. 19. We believe, however, that the well developed jurisprudence relating to questions of pure law and pure fact offers guideposts for working our way out of this confusion.
The appropriate standard of review for a district judge’s application of law to fact may be determined, in our view, by reference to the sound principles which underlie the settled rules of appellate review just discussed. If the concerns of judicial administration — efficiency, accuracy, and precedential weight — make it more appropriate for a district judge to determine whether the established facts fall within the relevant legal definition, we should subject his determination to deferential, clearly erroneous review. If, on the other hand, the concerns of judicial administration favor the appellate court, we should subject the district judge’s finding to de novo review. Thus, in each case, the pivotal question is do the concerns of judicial administration favor the district court or do they favor the appellate court.
In our view, the key to the resolution of this question is the nature of the inquiry that is required to decide “whether the rule of law as applied to the established facts is or is not violated.” Id. If application of the rule of law to the facts requires an inquiry that is “essentially factual,” id. at 288, 102 S.Ct. at 1790 — one that is founded “on the application of the fact-finding tribunal’s experience with the mainsprings of human conduct,” Commissioner v. Duberstein, 363 U.S. 278, 289, 80 S.Ct. 1190, 1198, 4 L.Ed.2d 1218 (1960) — the concerns of judicial administration will favor the district court, and the district court’s determination should be classified as one of fact reviewable under the clearly erroneous standard. If, on the other hand, the question requires us to consider legal concepts in the mix of fact and law and to exercise judgment about the values that animate legal principles, then the concerns of judicial administration will favor the appellate court, and the question should be classified as one of law and reviewed de novo.
As the Supreme Court appeared to indicate in Pullman-Standard, 456 U.S. at 289 n. 19, 102 S.Ct. at 1790 n. 19, the concerns of judicial administration will generally favor the appellate court, justifying de novo review. This is so because usually the application of law to fact will require the consideration of legal concepts and involve the exercise of judgment about the values underlying legal principles. In United States v. General Motors Corp., 384 U.S. 127, 141 n. 16, 86 S.Ct. 1321, 1328 n. 16, 16 L.Ed.2d 415 (1966), for instance, the Supreme Court ruled that “the ultimate conclusion by the trial judge, that the defendants’ conduct did not constitute a combination or conspiracy in violation of the Sherman Act, is not to be shielded by the ‘clearly erroneous test’.” In explanation, the Court noted that “the question here is not one of ‘fact,’ but consists rather of the legal standard required to be applied to the undisputed facts of the case.” Id. Helvering v. Tex-Penn Oil Co., 300 U.S. 481, 57 S.Ct. 569, 81 L.Ed. 755 (1937), offers another instructive example. In Tex-Penn Oil, the Court treated the factual findings of the Board of Tax Appeals as established because they were supported by substantial evidence. Id. at 490-91, 57 S.Ct. at 573. In contrast, the determination by the Board that a transaction did not fall within the non-recognition of gains provision of the Internal Revenue Code was classified as “a conclusion of law or at least a determination of a mixed question of law and fact.” Id. at 491, 57 S.Ct. at 574. The Court thus felt that “it [was] to be distinguished from the findings of primary, evidentiary or circumstantial facts ... [and was] subject to judicial review [during which] the court [could] substitute its judgment for that of the board.” Id. This assessment of tax *1203consequences did not turn on a strictly factual inquiry. Rather, it required an analysis of the intricacies of the corporate reorganization provisions of the Internal Revenue Code and the legislative purpose in exempting certain gains from immediate taxation.9
The predominance of factors favoring de novo review is even more striking when the mixed question implicates constitutional rights. In cases involving such questions, the application of law to fact will usually require that the court look to the well defined body of law concerning the relevant constitutional provision. Thus, the Supreme Court has held that the mixed question of probable cause is treated as a question of law and reviewed de novo. See, e.g., Ker v. California, 374 U.S. 23, 83 S.Ct. 1623, 10 L.Ed.2d 726 (1963). This holding implicitly recognizes that the inquiry involved in applying the probable cause standard goes well beyond the facts of the case and requires consideration of the abstract legal principles that inform constitutional jurisprudence. A similar recognition, though it may be unarticulated, seems to underlie cases in which our court has reviewed as questions of law such mixed questions as the reasonableness of certain restrictions on first amendment rights, Ellwest Stereo Theatres, Inc. v. Wenner, 681 F.2d 1243 (9th Cir.1982); whether there was a “search” in an INS sweep of a factory, International Ladies’ Garment Workers’ Union v. Sureck, 681 F.2d 624 (9th Cir.1982), cert. granted sub nom. Immigration and Naturalization Service v. Delgado, — U.S. —, 103 S.Ct. 1872, 76 L.Ed.2d 805 (1983); and whether a post-deprivation hearing within forty-eight hours following the seizure of property conformed to fourteenth amendment due process requirements, Goichman v. Rheuban Motors, Inc., 682 F.2d 1320 (9th Cir.1982).
There are, however, some types of mixed questions that are exceptions to this general predominance of factors favoring de novo review. First, there are those mixed questions in which the applicable legal standard provides for a strictly factual test, such as state of mind, and the application of law to fact, consequently, involves an “essentially factual” inquiry. Pullman-Standard, 456 U.S. at 288, 102 S.Ct. at 1790. In Pullman-Standard, for example, the mixed question before the Court was whether the established facts demonstrated the intent to discriminate required by section 703(h) of title VII of the Civil Rights Act of 1964. Since, for the purposes of section 703(h) of title VII, intent means subjective intent, the relevant legal standard was that of “actual motive.” Id. at 290,102 S.Ct. at 1791. The Court distinguished this legal standard from “some legal concept of discriminatory intent,” id. at 289,102 S.Ct. at 1790 (emphasis added), noting that actual motive “appears ... to be a pure question of fact,” id. On this basis, the Court concluded that in the case before it the concerns of judicial administration favored the district court and that the mixed question under consideration thus “[was] not ... a mixed question of law and fact of the kind that in some cases may allow an appellate court .to review the facts.” Id. The Court subjected the lower court’s determination to clearly erroneous review.10
*1204A trial court’s determination whether established facts constitute negligence offers us a second situation in which the concerns of judicial administration favor the district court. Because the legal standard for judging whether conduct is negligent requires us to determine, by reference to the “data of practical human experience,” Duberstein, 363 U.S. at 289, 80 S.Ct. at 1199, whether an individual acted “reasonably” by community standards, the trial court’s findings of fact effectively determine our legal conclusions. Consequently, clearly erroneous review is appropriate. See Hasbro Industries, Inc. v. M/S “St. Constantine,” 705 F.2d 339 (9th Cir.1983), cert. denied, — U.S. —, 104 S.Ct. 537, 78 L.Ed.2d 717 (1983).11
Thus, because the application of law to fact will generally require the consideration of legal principles, the concerns of judicial administration will usually favor the appellate court, and most mixed questions will be reviewed independently. This is particularly true when the mixed question involves constitutional rights. But we identify from existing precedent at least two exceptions to this general rule: mixed questions in which the applicable legal standard provides for a strictly factual test and the mixed question of negligence.
C
In summary, to classify mixed questions of law and fact for standard of review purposes, we adopt a functional analysis that focuses on the nature of the inquiry required when we apply the relevant rule of law to the facts as established. The analysis is not a precise one and does not offer any litmus test by which all mixed questions can be neatly categorized. It does not “unerringly distinguish between findings of fact and conclusions of law.” Pullman-Standard v. Swint, 456 U.S. at 289 n. 19, 102 S.Ct. at 1790 n. 19. Nonetheless, we think that if we focus on the nature of the inquiry required in determining whether the established facts fall within the relevant legal definition, we employ a neutral test that accurately reflects the concerns that properly underlie standard of review jurisprudence.
Ill
Applying our functional analysis to the case at hand, we conclude that the *1205mixed question of exigent circumstances is reviewable de novo as a question of law, and, consequently, we overrule Flickinger. Our reason is this: to decide if the facts satisfy the legal test of exigency — whether, judged by an objective standard, “there [was] a likelihood that the occupants [would] attempt to escape, resist, destroy evidence, or harm someone within,” United States v. Bustamante-Gamez, 488 F.2d 4,12 (9th Cir.1973), cert. denied, 416 U.S. 970, 94 S.Ct. 1993, 40 L.Ed.2d 559 (1974),— necessarily involves us in an inquiry that goes beyond the historical facts.
The mixed question of exigency is rooted in constitutional principles and policies.12 Like many such mixed questions,13 its resolution requires us to consider abstract legal doctrines, to weigh underlying policy considerations, and to balance competing legal interests. In particular, its resolution requires that we strike a balance between two sometimes conflicting societal values — the safety of law enforcement officers and fourth amendment privacy interests. The essential and difficult question raised by this balancing is how much risk police officers can reasonably be expected to assume before disregarding the rules society has adopted to otherwise circumscribe the exercise of their considerable discretionary authority in carrying out their vital-law enforcement duties.
This is a question that no amount of factfinding will answer. The inquiry requires us to ask not merely whether the trial judge was correct in finding that Olson in fact had reason to believe, for instance, that McConney was a member of the Hell’s Angels Motorcycle Club or that he had a prior felony conviction; it also requires a determination whether these facts and any other information Olson had about McCon-ney’s past are sufficient to satisfy the legal standard for exigency. We must decide, for instance, whether a reasonable belief that a suspect is a member of the Hell’s Angels Club, without demonstration of a link to criminal activity, is relevant to the issue of exigency, cf. United States v. Rubio, 727 F.2d 786 at 793 (9th Cir.1983) (membership in Hell’s Angels without a link to actual criminal activity insufficient to support a finding of probable cause to issue a search warrant). We must also decide how much weight may be given to an officer’s knowledge that a suspect has previously been convicted of an unspecified felony.
When, as here, the application of law to fact requires us to make value judgments about the law and its policy underpinnings, and when, as here, the application of law to fact is of clear precedential importance, the policy reasons for de novo review are satisfied and we should not hesitate to review the district judge’s determination independently.
In conclusion, we hold that the mixed question of exigent circumstances is not reviewable under the clearly erroneous test because it is not, like the question of intent, essentially factual. We do not at this juncture decide how any other mixed question should be reviewed. We do, it is true, adopt a functional analysis for the resolution of these questions. But our approach is an ad hoc one, permitting individual analysis and classification of each type of mixed question. Under it, we need only decide that the question of exigent circumstances is subject to de novo review.
IV
We have made the required review of the record, and we affirm the district court’s holding that exigent circumstances excused the agents’ failure to await refusal of admittance before entering McConney’s home. The “knock-notice” statute provides that an officer, before opening a door of a house in order to enter, *1206must give notice of his authority and purpose, and be refused admittance by the occupant. 18 U.S.C. § 3109. The requirements of section 3109 apply to locked and unlocked doors alike. Sabbath v. United States, 391 U.S. 585, 590, 88 S.Ct. 1755, 1758, 20 L.Ed.2d 828 (1968). The purpose of the requirement is twofold. First, it is to protect an individual’s right to privacy in the home to the extent possible and consistent with the execution of the warrant. Second, it is to protect officers themselves, who might otherwise be mistaken, upon unannounced intrusion, for someone with no right to be there. Id. at 589, 88 S.Ct. at 1757. The interval of time an officer must wait between announcement and entry depends on the circumstances of each case. See United States v. Phelps, 490 F.2d 644 (9th Cir.), cert. denied, 419 U.S. 836, 95 S.Ct. 64, 42 L.Ed.2d 63 (1974).
When police have properly knocked and announced their identity and purpose, mild exigency is sufficient to justify simultaneous entry when entry can be accomplished without physical destruction of property. United States v. Bustamante-Gamez, 488 F.2d 4, 11 (9th Cir.1973), cert. denied, 416 U.S. 970, 94 S.Ct. 1993, 40 L.Ed.2d 559 (1974). Accord United States v. Whitney, 633 F.2d at 909. Mild exigency may exist where there is a likelihood that the occupants will try to escape, resist, or destroy evidence. More specific inferences of exigency are necessary if entry may be obtained only by the physical destruction of property; an explicit refusal of admittance or lapse of a significant amount of time is necessary if the officers have no facts indicating exigency. United States v. Bustamante-Gamez, 488 F.2d at 12. But here the officers found an open door.
We emphasize, and McConney concedes, that the agents knocked and announced their entry. They simultaneously opened an unlocked wire screen door through which they observed McConney and a second then unknown person seated inside. Agent Olson had recognized McConney as he observed him through the door, and feared for his safety and that of the other officers if McConney were allowed time to arm himself.
[H] Whether the circumstances present sufficient exigency necessarily involves judgment. An unjustified but sincere fear by an officer cannot excuse noncompliance or the protection of the occupants’ privacy interest would depend on no more than an officer’s anxiety. Here, however, there is no reason to discount either the sincerity or the reasonableness of the agent’s concern.
When Agent Olson served as an undercover agent, he had a close relationship with McConney. McConney had a conviction for a violent crime, and Olson testified that he knew that McConney was a convicted felon. Olson further knew that McConney was a drug dealer and that he was charged with racketeering. Both types of offenses often involve the possession of firearms. Moreover, McConney once made a statement to Olson about the “protection of members of the club,” supporting a reasonable inference that there was danger in arresting any club member.
Although any one of these factors standing alone might be insufficient to create the mild exigency required, the combination of factors in this case satisfies the requirements of Bustamante-Gamez. The agents did not violate section 3109.
V
McConney contends that even if the entry were unobjectionable, the first gun seized was inadmissible as evidence because the indicia search warrant was invalid and the seizure cannot be justified as incident to his arrest. Because we hold that the first gun was seized as an incident to his arrest, we need not reach McConney’s arguments about the indicia search warrant. Because the same type of fourth amendment right is involved as in the determination of exigent circumstances, we apply the same independent standard of review.
“ Search incident to arrest” is an exception to the general rule against searches without a warrant. The justification for permitting a warrantless search is *1207the need to seize weapons or other things which might be used to assault an officer or effect an escape, as well as the need to prevent the loss or destruction of evidence. See Preston v. United States, 376 U.S. 364, 84 S.Ct. 881, 11 L.Ed.2d 777 (1964). In Chimel v. California, 395 U.S. 752, 89 S.Ct. 2034, 23 L.Ed.2d 685 (1969), legitimate search incident to arrest was limited to the arrestee’s person and the area “within his immediate control.” Id. at 763, 89 S.Ct. at 2040.
The critical inquiry, then, is whether the search that produced McConney’s pistol was properly limited to the area within his immediate control at the time of his arrest. The number of persons being arrested, the number of officers present, their physical positioning with regard to the arrestee and the place searched, the display of guns by the officers and, of course, the distance between the arrestee and the place searched are all factors to be weighed by the court. See United States v. Mason, 523 F.2d 1122 (D.C.Cir.1975); United States v. Patterson, 447 F.2d 424 (10th Cir.1971), cert. denied, 404 U.S. 1064, 92 S.Ct. 748, 30 L.Ed.2d 752 (1972).
We conclude that the search under the sofa cushion, which revealed the loaded pistol, did not exceed the limitations under Chimel. At the time the sofa was searched, McConney was only two feet from the gun, and was not yet handcuffed. Chimel does not require the police to presume that an arrestee is wholly rational. Persons under stress may attempt actions which are unlikely to succeed. The presence of the agents with drawn guns may have limited the area McConney could have reached had he made a sudden attempt to do so. This presence, however, was no guarantee that armed violence could not break out when a loaded gun was only two feet away.
For the foregoing reasons, the loaded pistol was properly admitted as evidence against McConney on the charge of violation of 18 U.S.C. § 922(h). McConney does not contend that the evidence against him, if the pistol was properly admitted, is insufficient to support his conviction. We need not reach the admissibility of evidence concerning the second gun as the finding of guilt did not depend upon it.
The decision filed on March 30, 1981, and withdrawn on January 15,1982, is superseded by this decision.
AFFIRMED.
All members of the court join in Parts I and V of the opinion.
EUGENE A. WRIGHT, TANG, FLETCHER, NELSON and CANBY, Circuit Judges, join NORRIS, Circuit Judge, in Part II.
TANG, FLETCHER, NELSON, and CANBY, Circuit Judges, join NORRIS, Circuit Judge, in Part III.
BROWNING, Chief Judge, and GOODWIN, ALARCON and BOOCHEVER, Circuit Judges, concur in the result of Part III.
BROWNING, Chief Judge, and EUGENE A. WRIGHT, SNEED, TANG, ALARCON, NELSON, CANBY and BOOCHEVER, Circuit Judges, join GOODWIN, Circuit Judge, in Part IV.
SNEED, Circuit Judge, with whom EUGENE A. WRIGHT, Circuit Judge, joins, concurring in the judgment:
I concur in the result reached by the majority, Part I, with much of what is said in Part II, and with the conclusions reached in Parts IV and V. I would not reach the issue whether Flickinger should be overruled which is dealt with in Part III.
I come to rest in this position because whatever the proper standard of review should be with respect to a trial court finding of exigent circumstances, the majority has concluded, as did the trial court, that they existed in this case. Therefore, this case should have been disposed of with an opinion that would read substantially as follows:
The entry of the agents in the appellant’s residence did not violate 18 U.S.C. § 3109. Exigent circumstances justified the entry at the same time as the knock and announcement of their identity. The *1208search incident to the arrest did not exceed the limits fixed by Chimel v. California, 395 U.S. 752 [89 S.Ct. 2034, 23 L.Ed.2d 685] (1969). These conclusions follow without regard to the standard of review we might employ in scrutinizing the record. The trial court properly denied the appellant’s suppression motions. Affirmed.
The court states that this case was taken “for the specific purpose of resolving the question of the appropriate standard of appellate review of ‘exigent circumstances’ as a mixed question of fact and law.” The statement is inaccurate in two respects. The en banc call was not cast in those precise terms and, although it is true that some members of the court undoubtedly voted to take the case en banc for that reason, it is also undoubtedly true that some who voted similarly did so for other reasons. In short, the case was taken en banc, not the question. Second, the characterization of “exigent circumstances” as a mixed question of fact and law proceeds not from the en banc call but from the majority’s analysis of the issue whether Flicking-er, which treated “exigent circumstances” as a question of fact, should be overruled.
Nonetheless, the standard-of review discussion in Part II contains much wisdom. The problem is essentially one of managing the allocation of appellate court resources. Those things the trial court does better we should not attempt to repeat. True, we should review the trial court’s performance even with respect to those things it can do better to eliminate intolerable deviations. And, of course, those things the appellate court can do better we should not delegate to the trial court.
The problem is which things go where. Part II is a wise and intelligent attempt to formulate general principles that will provide the solution. My basic reservation is that I believe the process of deciding which things go where is more tentative, experimental, and ad hoc than even Part II suggests. What is a “fact” can never be satisfactorily fixed. See L. Green, Judge and Jury 270 (1930). External phenomena can be described in ascending or descending levels of abstraction. The term “historical fact” merely indicates that the description should be at a fairly low level of abstraction. For example, whether Officer Olson knew that McConney had a prior felony conviction can be described as an “historical fact”; it is also true, however, that Officer Olson’s knowledge can be described more precisely by stating the exact circumstances under which the conviction came to his attention. The latter also can be described as the “historical facts.” The precise level of abstraction deserving of being characterized as a fact, historical or otherwise, turns on the purpose being served. In the context of the standard of review with respect to exigent circumstances, I believe the trial courts, by employing the definition of such circumstances stated by the majority, page 1199, can do a better job than we. Therefore, I would say whether a reasonable person would believe that entry was necessary to prevent “either physical harm to the officers or other persons,” etc., see Majority Opinion page 1199, is a question of fact. True, the definition employs a fairly high level of abstraction but it provides sufficient guidance to enable trial courts to bring to bear their unquestionably superior experience with police officers and defendants accused of crime. This superiority flows from their frequent opportunity to observe them face-to-face. Appellate courts can only read and classify cold records; trial courts confront the people whose actions they characterize.
For me this superior experience resolves the “which goes where” problem. Flicking-er was properly decided. An appellate court cannot do as good a job in this area as the trial courts. This is said while recognizing, as I must, that our unrelenting duty is to protect constitutional rights. To me it better serves those rights to treat “exigent circumstances,” as the majority defines them, as a question of fact subject to the “clearly erroenous” standard of review. But, as I pointed out above, I would not even reach the issue.
*1209GOODWIN, Circuit Judge, with whom BROWNING, Chief Judge, and ALARCON and BOOCHEVER, Circuit Judges, join, specially concurring in the judgment:
I concur in the majority opinion that the appellate review of a trial court’s findings on the existence of exigent circumstances requires the application of a legal standard rather than the “clearly erroneous” standard of Fed.R.Civ.P. 52 which we routinely apply to factual findings made by trial courts in the decision of purely factual questions.
I respect the scholarship, and agree with much of the reasoning of the majority. But I disagree with the majority’s decision, stated in Parts II and III of the opinion, to leave a number of categories of suppression questions open to review under standards that will have to be developed for each of the categories by the separate weighing of the preponderance of facts or law in so-called mixed questions of law and fact. This exercise is likely to prove both time-consuming and uncertain in its results.
I agree that there is no simple, or bright line, test that will always steer this court to the appropriate characterization of a question, whether of law or fact. The difficulty in defining the difference between questions of law and questions of fact in a given formulation of a suppression question itself underscores the argument that we should adopt a rule that when the court is in doubt, it should treat the question as one of law. Judicial review in the doubtful, or complex law-fact questions would then be conducted as it should be, de novo, giving proper concern to underlying constitutional principles as the court makes certain that they were correctly applied to the historical facts. If reasonable lawyers and judges can’t agree whether a given question is one of fact or one of law, then it is probably a question of law.
It is only in the close and debatable questions that arise when one invokes constitutional principles that anything important is likely to turn upon the way we characterize the question in any event. It is in precisely this type of case that we ought to proceed independently and examine the question of exigent circumstances as a question of law. Accordingly, I would be content to rest the debate that has for so long engaged this court upon a statement made by the Supreme Court, to which we look for leadership in such matters:
“While this Court does not sit as in nisi prius to appraise contradictory factual questions, it will, where necessary to the determination of constitutional rights, make an independent examination of the facts, the findings, and the record so that it can determine for itself whether in the decision as to reasonableness the fundamental — i.e., constitutional — criteria established by this Court have been respected.... ” Ker v. California, 374 U.S. at 34, 83 S.Ct. at 1630.
While Sumner v. Mata, 455 U.S. 591, 597, 102 S.Ct. 1303, 1306, 71 L.Ed.2d 480 (1982), has limited the scope of federal court review concerning facts found by the state courts in proceedings that satisfied due process, we have found no Supreme Court case holding that federal appellate courts are bound by the district court’s application of constitutional principles following the district court’s designation of certain questions as questions of fact. We do, of course, defer to the district judge’s findings when they deal with historical facts. But when the ultimate question involves both historical facts and their constitutional meaning, we have the duty to review the ultimate question independently. United States v. One Twin Engine Beech Airplane, 533 F.2d 1106, 1108 (9th Cir.1976).
With the exception of the language in Parts II-B and III, I concur in the opinion.
. A Prescott warrant authorizes the search of a home for a person named in an accompanying arrest warrant. United States v. Prescott, 581 F.2d 1343 (9th Cir.1978).
. McConney’s stipulation did not cover the factual ingredients in the ruling on the motion to suppress.
. The statute specifically states:
The officer may break open any outer or inner door or window of a house, or any part of a house, or anything therein, to execute a search warrant, if, after notice of his authority and purpose, he is refused admittance
18 U.S.C. § 3109.
. We appreciate the Flickinger panel’s concern with the question of probable cause. Probable cause and exigent circumstances implicate very similar standard of review concerns. See infra at-.
. Although rule 52(a) is a rule of civil procedure, the clearly erroneous test which it sets forth is applied in both civil and criminal proceedings. See, e.g., United States v. Page, 302 F.2d 81 (9th Cir.1962) (en banc).
. As noted in Zenith Radio Corp. v. Hazeltine Research, Inc., 395 U.S. 100, 123, 89 S.Ct. 1562, 1576, 23 L.Ed.2d 129 (1969), “[t]he authority of an appellate court, when reviewing the findings of a judge as well as those of a jury, is circumscribed by the deference it must give to decisions of the trier of fact, who is usually in a superior position to appraise and weigh the evidence.”
. It can hardly be disputed that application of a non-deferential standard of review requires a greater investment of appellate resources then does application of the clearly erroneous standard. Appellate courts could do their work more quickly if they applied the clearly erroneous standard in most circumstances, because the courts then need only determine if the lower court’s decision is a reasonable one, not substitute their own judgment for that of the trial judge.
. As noted by Chief Judge Coffin,
every important appellate court decision is made by a group of equals. This fact reflects the shrewd judgment of the architects of our state and federal judicial systems that an appellate judge is no wiser than a trial judge. His only claim to superior judgment lies in numbers; three, five, seven or nine heads are usually better than one.
F. Coffin, The Ways of a Judge 58 (1980).
. See also Townsend v. Sain, 372 U.S. 293, 309 n. 6, 83 S.Ct. 745, 755 n. 6, 9 L.Ed.2d 770 (1963); United States v. Mesa, 638 F.2d 582, 589, 591 n. 3 (3d Cir.1980) (Adams, J., concurring).
. Commissioner v. Duberstein, 363 U.S. 278, 80 S.Ct. 1190, 4 L.Ed.2d 1218 (1960), is to the same effect. In Duberstein, the question under review was whether a donor, by giving an automobile to a business, acquaintance, had made a “gift” within the meaning of the Internal Revenue Code. Because the answer turned on a non-technical, fact-based inquiry — the donor’s actual intent in transferring ownership of the automobile — the Court reviewed the question under the clearly erroneous standard, explaining that when
[djecision of the issue presented in [the] case [can] be based ultimately on the application of the fact-finding tribunal’s experience with the mainsprings of human conduct to the totality of the facts of [the] case, [application of a deferential standard of review is appropriate], [Then,] the nontechnical nature or the statutory standard, the close relationship of it to the data of practical human experience, and the multiplicity of relevant factual elements, with their various combinations. *1204creating the necessity of ascribing the proper force to each, confirm us in our conclusion that the primary weight in this area must be given to the conclusions of the trier of fact.
Id. at 289, 80 S.Ct. at 1198.
Similarly, in Oregon v. Kennedy, 456 U.S. 667, 102 S.Ct. 2083, 72 L.Ed.2d 416 (1982), the Supreme Court considered the issue of when a retrial is barred, under the double jeopardy clause, by prosecutorial conduct that results in a mistrial. After holding that a retrial is barred only when the conduct giving rise to the successful motion for a mistrial was “intended” to provoke the defendant into moving for a mistrial, the Court indicated that lower court findings on this question were to be reviewed deferentially.
Since the Oregon trial court found, and the Oregon Court of Appeals accepted, that the prosecutorial conduct culminating in the termination of the first trial in this case was not so intended by the prosecutor, that is the end of the matter for the purposes of the Double Jeopardy Clause of the Fifth and Fourteenth Amendments to the United States Constitution.
Id. at 679, 102 S.Ct. at 2091. In a footnote, the Court indicated the reasons for this deference. After noting that the trial judge is “best situated intelligently to make such a decision,” id. at 677 n. 7, 102 S.Ct. at 2090 n. 7 (quoting Gori v. United States, 367 U.S. 364, 368, 81 S.Ct. 1523, 1526, 6 L.Ed.2d 901 (1961)), the Court stated:
It seems entirely reasonable to expect, therefore, that appellate judges will continue to defer to the judgment of trial judges who are “on the scene” in this area, and that they will not inexorably reach the same conclusion on a cold record at the appellate stage that they might if anyone of them had been sitting as a trial judge.
Id.
. We note the disagreement among the circuits regarding the appropriate standard of review for questions of negligence as a further indication of the confusion to which the Supreme Court referred in Pullman-Standard. Compare Hasbro Industries, 705 F.2d 339, with Great Atlantic & Pacific Tea Co. v. Brasileiro, 159 F.2d 661, 665 (2d Cir.) (non-deferential standard applied to negligence determination in admiralty case), cert. denied sub nom. Republic of the United States of Brazil v. Great Atlantic & Pacific Tea Co., 331 U.S. 836, 67 S.Ct. 1519, 91 L.Ed. 1849 (1947). ■
. There is no dispute that the requirements of § 3109 are rooted in constitutional values. See , United States v. Whitney, 633 F.2d 902, 908 (9th Cir.1980) (“[sjection 3109 codifies a tradition embedded in Anglo-American law and declares the reverence which the law attaches to an individual’s right of privacy in his house”), cert. denied, 450 U.S. 1004, 101 S.Ct. 1717, 68 L.Ed.2d 208 (1981).
. See supra at 1203.