On January 12, 1973, a clinic of the Planned Parenthood Association located near the campus of Wayne State University in Detroit, Michigan, was firebombed. Suspects fleeing the scene fired upon university police officers. Appellant, Hayward Leslie Brown, was apprehended in the vicinity of the clinic by Detroit police officers responding to a radio call that police officers were under fire. At the time of his arrest, Appellant and two others were subjects of a massive manhunt because of their suspected involvement in gun battles with police on December 4 and 29, 1972, which had left one officer dead and several wounded. In the back seat of the patrol car on the way to police headquarters, Appellant confessed to the firebombing.

On April 10, 1974, a federal grand jury in the Eastern District of Michigan indicted Appellant for possession of three "Molotov cocktails" in violation of 26 U.S.C. §§ 5845, 5861(d) and 5871 (1970), and for damaging by means of an explosive device an institution receiving federal financial assistance in violation of 18 U.S.C. § 844(f) (1970). Appellant moved to suppress the confession. After holding an evidentiary hearing, the District Court ruled that the confession was voluntary under the totality of the circumstances and denied the suppression motion. The Court also denied a motion for rehearing. At trial, Appellant was convicted by a jury on both counts and was sentenced to eight years on each offense, to run concurrently.

Prior to the federal prosecution, Appellant had been tried in state court on arson charges stemming from the firebombing incident. The arson indictment was dismissed after the state judge concluded that the confession was involuntary and ordered it suppressed.1 Appellant contends that the state court's finding of involuntariness estopped the District Court from holding an evidentiary hearing and from finding the confession voluntary. The District Court ruled that it was not bound by the state court's decision and that it was required to render an independent judgment on the voluntariness of the confession. We agree. Appellant was being prosecuted on a federal indictment for violating federal criminal statutes. The question of whether a confession should be admitted in a federal prosecution is a matter of federal law and district courts are not bound by the decisions of state courts in related cases. See Elkins v. United States, 364 U.S. 206, 223-24, 80 S. Ct. 1437, 4 L. Ed. 2d 1669 (1960); United States v. Beigel, 370 F.2d 751, 756 (2d Cir. 1967). See also Boyle v. United States, 395 F.2d 413, 415 (9th Cir. 1970). This is not a habeas corpus case where a petitioner is seeking review of a state conviction so the standards enunciated in Townsend v. Sain, 372 U.S. 293, 83 S. Ct. 745, 9 L. Ed. 2d 770 (1963), and 28 U.S.C. § 2254 (1970), are not applicable. Even in habeas corpus cases, federal courts are required to make an independent determination on the ultimate issue of the voluntariness of a confession. Davis v. North Carolina, 384 U.S. 737, 741-42, 86 S. Ct. 1761, 16 L. Ed. 2d 895 (1966). See LaVallee v. Delle Rose, 410 U.S. 690, 93 S. Ct. 1203, 35 L. Ed. 2d 637 (1973).

The District Court found that Appellant was given Miranda warnings so the admissibility of the confession turns on whether it was voluntary. Davis v. North Carolina, 384 U.S. at 740, 86 S. Ct. 1761. For standards of voluntariness we refer to a pre-Miranda line of cases beginning with Brown v. Mississippi, 297 U.S. 278, 56 S. Ct. 461, 80 L. Ed. 682 (1936), which held that the admission of an involuntary confession in a criminal prosecution violates due process.2 Appellate courts have a duty to examine the entire record and make an independent determination on the voluntariness of a confession. Davis v. North Carolina, 384 U.S. at 741-42, 86 S. Ct. 1761; United States v. Dye, 508 F.2d 1226, 1232 (6th Cir. 1974). In determining whether a confession was voluntary, courts focus on the state of mind of the accused at the time the confession was made. See, e. g., Culombe v. Connecticut, 367 U.S. 568, 602, 81 S. Ct. 1860, 6 L. Ed. 2d 1037 (1961). The central inquiry of a court considering the voluntariness of a confession is whether the confession was the product of free and rational choice:

The ultimate test remains that which has been the only clearly established test in Anglo-American courts for two hundred years: the test of voluntariness. Is the confession the product of an essentially free and unconstrained choice by its maker? If it is, if he has willed to confess, it may be used against him. If it is not, if his will has been overborne and his capacity for self-determination critically impaired, the use of his confession offends due process.

Culombe v. Connecticut, 367 U.S. at 602, 81 S.Ct. at 1879. See also Schneckloth v. Bustamonte, 412 U.S. 218, 225-26, 93 S. Ct. 2041, 36 L. Ed. 2d 854 (1973).3 In deciding whether an accused's will has been overborne, courts have looked to the totality of the circumstances surrounding the confession, both the characteristics of the accused and the details of the interrogation, and determined their psychological impact on an accused's ability to resist pressures to confess. Schneckloth v. Bustamonte, 412 U.S. at 226, 93 S. Ct. 2041; Culombe v. Connecticut, 367 U.S. at 602, 81 S. Ct. 1860.4 The voluntariness of a confession is a mixed question of law and fact. Justice Frankfurter once described the notion of voluntariness as an "amphibian" because "(i)t purports at once to describe an internal psychic state and to characterize the state for legal purposes." Culombe v. Connecticut, 367 U.S. at 605, 81 S.Ct. at 1880. The unique mixture of law and fact involved in a voluntariness determination is reflected in our standard of review. The Government contends that we may only overturn the District Court's conclusion that the confession was voluntary if we find clear error. That statement is not entirely correct. Justice Frankfurter separated a court's inquiry on voluntariness into three phases:

The inquiry whether, in a particular case, a confession was voluntarily or involuntarily made involves, at the least, a three-phased process. First, there is the business of finding the crude historical facts, the external, "phenomenological" occurrences and events surrounding the confession. Second, because the concept of "voluntariness" is one which concerns a mental state, there is the imaginative recreation, largely inferential, of internal "psychological" fact. Third, there is the application to this psychological fact of standards for judgment informed by the larger legal conceptions ordinarily characterized as rules of law but which, also, comprehend both induction from, and anticipation of, factual circumstances.

Culombe v. Connecticut, 367 U.S. at 603, 81 S.Ct. at 1879.

As to the first phase, findings of historical fact, great deference is afforded the trier of fact because of its ability to observe the witnesses' demeanor. Resolution of testimonial conflicts and specific findings of fact by the trial court will not be disturbed on appeal unless it is clear from the record that an error has been committed. Id. at 603, 81 S. Ct. 1860. In reviewing the second and third phases of the inquiry, determining the mental state of the accused and assessing its legal significance, appellate courts are granted greater leeway:

The second and third phases of the inquiry determination of how the accused reacted to the external facts, and of the legal significance of how he reacted although distinct as a matter of abstract analysis, become in practical operation inextricably interwoven. This is so, in part, because the concepts by which language expresses an otherwise unrepresentable mental reality are themselves generalizations importing preconceptions about the reality to be expressed. It is so, also, because the apprehension of mental states is almost invariably a matter of induction, more or less imprecise, and the margin of error which is thus introduced into the finding of "fact" must be accounted for in the formulation and application of the "rule" designed to cope with such classes of facts. The notion of "voluntariness" is itself an amphibian. It purports at once to describe an internal psychic state and to characterize that state for legal purposes. Since the characterization is the very issue "to review which this Court sits," Watts v. Indiana, 338 U.S. 49, 51, (69 S. Ct. 1347, 1348, 93 L. Ed. 1801) (opinion of Frankfurter, J.), the matter of description, too, is necessarily open here. See Lisenba v. California, 314 U.S. 219, 237-238, (62 S. Ct. 280, 290, 86 L. Ed. 166); Ward v. Texas, 316 U.S. 547, 550, (62 S. Ct. 1139, 1141, 86 L. Ed. 1663); Haley v. Ohio, 332 U.S. 596, 599, (68 S. Ct. 302, 303, 92 L. Ed. 224); Malinski v. New York, 324 U.S. 401, 404, 417, (65 S. Ct. 781, 783, 89 L. Ed. 1029).No more restricted scope of review would suffice adequately to protect federal constitutional rights. For the mental state of involuntariness upon which the due process question turns can never be affirmatively established other than circumstantially that is, by inference; and it cannot be competent to the trier of fact to preclude our review simply by declining to draw inferences which the historical facts compel.

Id. at 604-05, 81 S.Ct. at 1880.

With the above in mind, we now proceed to an independent determination of the voluntariness of Appellant's confession.

Appellant's claims at the suppression hearing were twofold: he denied making the incriminating statements attributed to him and he contended that any statements he did make were coerced by beatings and the threat of beatings. The District Court found that Appellant had made the statements attributed to him; that he was not beaten by police either at the time of his arrest or in the patrol car where he confessed; and that his confession, viewed in the totality of the circumstances, was voluntary. Whether Appellant made the statements and whether he was beaten by police are questions of fact which will not be disturbed on appeal unless they are clearly erroneous. See United States v. Montos, 421 F.2d 215, 219 n. 1 (5th Cir. 1970). After reviewing the record, we are not convinced that the Court's findings on these issues are clearly erroneous and they are hereby affirmed.

We cannot affirm, however, the District Court's finding on the ultimate issue of voluntariness. On review of the record, we are inescapably led to the conclusion that the confession was not voluntary, in the sense that it was not the product of a free and rational choice, but was induced by Appellant's overwhelming fear that he would be beaten by police. A number of factors dictate this conclusion, among which are the manifest hostility of the police toward Appellant; his age; his physical condition and emotional state at the time of the confession; the proximity of the confession to a violent arrest; his expressed fears that he would be beaten by police; the inherent coerciveness of the back seat of a patrol car as a setting for a confession; and the fact that Appellant was struck by one of the officers in the car at the time he made the incriminating statements. Under the totality of the circumstances, the statements cannot be deemed "the product of a rational intellect and a free will."5 Blackburn v. Alabama, 361 U.S. 199, 208, 80 S. Ct. 274, 280, 4 L. Ed. 2d 242 (1960). See also Schneckloth v. Bustamonte, 412 U.S. at 225, 93 S. Ct. 2041; Culombe v. Connecticut, 367 U.S. at 602, 81 S. Ct. 1860.

At the time of his arrest, Appellant was nineteen. Although he had prior contact with juvenile authorities, nothing in his background could have prepared Appellant for the furor surrounding him in December of 1972. The search for the three suspects in the December shootings was intense and highly emotional.6 The shootings and the search for the suspects received a great deal of attention in the Detroit area and were widely publicized by the media. The testimony of the police officers at the suppression hearing reflected the high priority placed on locating the three suspects by the Detroit Police Department and the personal feelings of individual officers involved in the search. Officer Gregory Ciaglo, one of the arresting officers, testified that the December shootings were a matter of great concern to the Department and to him; the arrest of the three suspects, Brown, Boyd and Bethune, was his number one priority; he made himself aware of their identities and carried their photographs on a clipboard in his patrol car; he wanted them caught, he was not concerned about their versions of the incident, and he didn't care if they were brought in dead or alive. Similar sentiments were expressed by other officers who testified. Almost all of the officers testified that they carried a photograph of Appellant, either on their person or on the visor of their patrol cars, including the car in which Appellant was riding when he made the statements.

Appellant's arrest was, by all accounts, a violent one. Although the court discounted testimony that Appellant was beaten and kicked during the arrest, the descriptions of the arrest offered by police officers at the scene indicated that Appellant was grabbed by the back of the neck, had his legs kicked out from under him, and was flung to the sidewalk where he landed on his face.7 Once on the sidewalk, between three and four officers subdued the kicking and thrashing suspect and handcuffed his hands behind his back. During the struggle, he was struck with a closed fist once and "nudged" on the back of the head or neck with the butt of a shotgun. It was during arrest that the Court found that Appellant sustained most of the injuries described by witnesses who saw him later in the day. The circumstances of the arrest and the amount of force employed have relevance to determining Appellant's state of mind when he made the incriminating statements a short time afterwards. During the arrest, a number of other police officers arrived on the scene, some carrying shotguns. Although testimony as to their numbers varied, it appears that there were between fifteen and thirty officers present at the scene of the arrest. The District Court made a finding that the officers who were present did not know that the suspect being arrested was Hayward Brown.8 The Court concluded that their manifest attitude was not relevant to a determination of Appellant's mental state. However, the fact remains that the officers present at the scene had been involved in the search for a man who had fired on university police officers and they knew that the individual struggling on the sidewalk with the arresting officers was being arrested for shooting at policemen at a time when the Department was particularly sensitive about attacks on police. The manifest attitude of the police toward an accused is a relevant factor to consider in attempting to reconstruct the accused's state of mind. See Culombe v. Connecticut, 367 U.S. at 602, 81 S. Ct. 1860. Any hostility shown by police at the time of arrest is relevant to the extent that it colors the accused's appreciation of the circumstances. Thus, the number of police present at the scene of the arrest and their manifest attitude toward Appellant must be viewed from his perspective and according to his knowledge of the circumstances. Appellant was certainly aware that he was wanted for the murder of a police officer and the wounding of several others. A show of force by police at the time of his arrest could only confirm what he may already have suspected, that his was a case of special interest and concern for the Department and the individual officers.

Another salient factor which indicates the hostility of the police toward Appellant is the extraordinary precaution taken by supervisory officers to protect him from being beaten while in custody. Inspector Bensmiller, who arrived on the scene following the arrest, asked Appellant his name while he was seated in the back seat of the patrol car. Appellant answered with the last name "Brown" but he gave a different first name. The Inspector asked Appellant whether he was Hayward Brown. Appellant denied his true identity, claiming that he was Hayward Brown's brother. An officer on the other side of the patrol car was looking at a photograph of Hayward Brown and said, "that's Hayward Brown." The police officer had opened the rear door and the Inspector came around the car to shut it. The Inspector then ordered a sergeant, Sergeant Studer, to accompany Appellant to police headquarters, follow him through the arrest procedures, and "see that nothing happens to him." The Inspector explained that he wanted the sergeant to escort Appellant to headquarters because he felt it was necessary to have a "restraining influence" in the patrol car. The obvious inference from the Inspector's remark is that the sergeant's presence was intended to restrain the other officers in the car. The Inspector admitted that he was anxious to avoid brutality charges. He ordered the driver to take the prisoner straight to homicide. Another indication of unusual concern for Appellant's safety after he was taken into custody is to be found in the testimony of Sergeant Awe of the homicide division. When the patrol car bearing Appellant arrived in the garage at police headquarters, Sergeant Awe was instructed by one of his superiors to go down to the garage "to insure the safety of one Hayward Brown who was being brought up from the garage at the Police Headquarters building at the time." Once again the inference is clear that supervisory officers were fearful that Appellant would suffer violence while in custody. The extraordinary measures taken to protect Appellant after his arrest are commendable, but they also provide compelling evidence of widespread hostility toward Appellant among the officers of the Detroit Police Department.

Appellant admitted to the firebombing while placed between two officers in the back seat of the patrol car transporting him to police headquarters.9 The back seat of a patrol car is an inherently coercive setting for a confession. Cf. Brewer v. Williams, 430 U.S. 387, 97 S. Ct. 1232, 51 L. Ed. 2d 424 (1977).10 The prisoner and police officers are in close contact within a confined area. Often, the inside door handles are removed and the front and back seats are separated by wire mesh or a plastic divider. Invariably, the prisoner is handcuffed. He is effectively cut off from the world outside the patrol car. As a practical matter, he has no access to friends or counsel. If the prisoner has just been arrested, he may still be disoriented and apprehensive in an often hostile and alien setting. In short, the back seat of a patrol car as the setting for a confession conforms in all respect to the "incommunicado, police-dominated" atmosphere which led the Supreme Court in Miranda v. Arizona, 384 U.S. 436, 456, 86 S. Ct. 1602, 16 L. Ed. 2d 694 (1966), to recognize the need for special procedures to minimize the inherent coerciveness of custodial interrogation. A confession made in such a setting, even where Miranda warnings are given, is suspect, and courts called on to determine its voluntariness must carefully sift the surrounding circumstances to discern any signs that the statement supposedly "volunteered" by a prisoner was actually obtained under duress.

The evidence is clear that Appellant was emotionally distraught when he was placed in the patrol car following his arrest. According to Officer Thomas Arutoff, Appellant was crying with "tears coming down his face . . . He was sobbing and calling out something. It wasn't coherent." Officer David Gilbeau, who rode with Appellant in the patrol car, agreed that Appellant was " sobbing" immediately after he was placed in the car. Officer Steven Dest added that Appellant appeared quite nervous and was either crying or perspiring. On the trip to the stationhouse, Officer Ciaglo was driving. He was separated by a thick plastic divider from the back seat where Appellant was seated between Sergeant Studer and Officer Gilbeau. Appellant's hands were handcuffed behind his back. There is no doubt that the officers in the patrol car knew that their prisoner was Hayward Brown. Officer Ciaglo testified that he had asked Appellant his name before the patrol car had left the scene and Appellant had blurted out: "My name is Hayward Brown and I'll tell you everything you want to know."11 Officer Gilbeau confirmed Appellant's identity by examining a flyer with a photograph of Hayward Brown which was in his jacket pocket. Sergeant Studer, who testified that he was unaware of Appellant's identity when he entered the car, asked Appellant his name and he answered: "My name is Hayward Brown and I will tell you everything you want to know."12 It was about this time, Sergeant Studer testified, that Appellant began to cry. The Sergeant described Appellant as breathing heavily, sweating profusely, and that he appeared nervous or excited. The only injury that the Sergeant noticed was an abrasion on Appellant's right cheek that was covered with smeared blood. After the Sergeant became aware that Appellant was Hayward Brown, he gave Appellant Miranda warnings which Appellant interrupted at least twice. According to the Sergeant's testimony, after he gave the warnings but before Appellant made any statements, Appellant's facial expression looked as if he were crying and he shouted: "Don't let them beat me Sarge." The Sergeant replied, "No one is going to hit you." It was about this time, however, that Officer Gilbeau admits that he struck Appellant. The exact sequence of events is not clear from the record. Officer Gilbeau testified that at some point he struck Appellant with the back of his hand. He said that he was reasonably sure, but not positive, that it was after Appellant admitted the firebombing. He also testified that he thought that he struck Appellant before he screamed to the Sergeant not to let anybody get him, but he wasn't certain. Neither of the other officers in the car saw Officer Gilbeau strike Appellant so the exact timing of the event cannot be determined. Sergeant Studer testified that Appellant expressed the fear that he would be beaten before he admitted the firebombing. If that is true and if Officer Gilbeau's recollection that the incident occurred before Appellant asked the Sergeant not to let anyone get him is accurate, then Appellant was struck before he made the incriminating statements. In any event, Appellant was struck at about the time he was making the statements. When asked why he struck Appellant, Officer Gilbeau replied that Appellant "was screaming, screaming real loud and I just wanted to shut him up." Officer Gilbeau testified that during this period Appellant was "screaming and moving around considerably in the back seat." He also admitted to "moving around considerably in the back seat" with Appellant. We assume that he was seeking to restrain Appellant from moving about. Sergeant Studer also testified that Appellant was thrashing around the back seat of the patrol car. He stated that Appellant "thrashed around periodically in the car, moving around quite a bit throwing his shoulders around." At the police garage, Appellant resisted getting out of the patrol car. He had to be pulled from the car and wrestled to the elevator where he was forcibly subdued. Sergeant Awe testified that as Appellant was being brought up to homicide from the garage he began screaming for no apparent reason, "don't beat me, don't beat me." The Sergeant commented that "(h)is entire demeanor changed suddenly . . . for a moment, I had the impression he was going to go beserk or something." Appellant's conduct in the patrol car and in the police garage is not that of a man eager to cooperate with police.

Appellant's physical condition is another important factor to weigh in determining his mental state. Testimony about his physical condition came from four sources: police officers who were present at the arrest and in the patrol car; police officers and two attorneys who observed him at police headquarters; two doctors who examined him at Wayne County Jail; photographs and a clip from a television news film taken at the arraignment. The officers at the scene of the arrest testified that Appellant had an abrasion on his right cheek. Two of the arresting officers, Boice and Ciaglo, testified that they noticed the abrasion before Appellant was forced to the ground. The District Court found that Appellant suffered the abrasion when he fell while running from police. The abrasion on his cheek was the only injury testified to by police officers at the scene of arrest. The injuries observed by witnesses at police headquarters were more extensive. The defense offered the testimony of two police officers, Lieutenant Robert Taylor and Sergeant Gilbert Hill, and two attorneys, Richard Soble and Geoffrey Taft, concerning Appellant's physical and emotional condition at police headquarters. The two attorneys and Sergeant Hill testified that Appellant was crying and distraught at various times and appeared to have been severely beaten. Lieutenant Taylor recalled that Appellant was nervous and excited with abrasions on his face and some swelling. Sergeant Hill testified that Appellant's whole face appeared to have swelling and bruises and little specks of blood in different places. According to Sergeant Hill, Appellant's lip was split, and something was wrong with his tongue because he could hardly speak and had trouble getting a cigarette in his mouth. Richard Soble, one of Appellant's attorneys, arrived at police headquarters shortly after Appellant was brought in. Mr. Soble testified that Appellant's face was generally puffed, his cheekbones puffed and almost distorted, and that it looked like his nose had been broken. He said that there were several abrasions and blood on Appellant's face and that he was crying. He described Appellant's lips and jaw as puffed and swollen so that he had difficulty speaking. He testified that Appellant was bent over and sobbing, his body moving up and down, and when a police officer offered him a cigarette, Appellant had great difficulty closing his fingers because his hands and body were shaking. Geoffrey Taft corroborated the description given by Mr. Soble.

Medical testimony concerning Appellant's physical condition came from two doctors, Dr. Mounir Guindi and Dr. Milas W. Lebedovych, both of whom had treated Appellant the day of the arrest. Dr. Guindi, the first doctor to examine Appellant after the arrest, is employed at the Wayne County Jail and testified for the prosecution. Dr. Guindi testified that Appellant had abrasions on the cheek, nose, chin, upper lips and a laceration of the wrist and tongue. He characterized the injuries as minor. He gave Appellant some pain medication. Dr. Lebedovych, who examined Appellant later in the day, was called by the defense. Dr. Lebedovych testified that he gave Appellant a tetanus shot and sutured a laceration on his chin. He recalled Appellant's face as "badly beaten," swollen with multiple abrasions and contusions. He described Appellant as "frightened" and "shook up." Although Dr. Lebedovych believed that the injuries were the result of a beating, he conceded that they might have occurred during a violent struggle where Appellant's face was pushed and scraped along the sidewalk. He rejected the possibility that the injuries could have been sustained in a fall because if that had been the case the swelling and lacerations would have been localized and he remembered that "most of the face was badly beaten." When asked whether he agreed with Dr. Guindi's description of the injuries as "minor." Dr. Lebedovych said that he did, defining minor injuries as those that are not "life endangering," in that they would not lead to death without medical treatment. He also stated his belief that the injuries were not self-inflicted. The District Court characterized the injuries as "minimal" and "relatively minor."13 An examination of the photographs included in the record reveals abrasions of Appellant's right cheek and nose, and a marked swelling of the right side of the face and lips. While the injuries appear relatively minor, they are extensive, covering half the face and were undoubtedly painful. The description of Appellant's emotional state is consistent throughout the period of custody. The prisoner described as crying, screaming and thrashing about the back seat of the patrol car is also described as crying, screaming and shaking at police headquarters. The consistency in the description of Appellant's emotional state at the time of arrest, in the patrol car and at the police headquarters belies the Government's contention that for a brief period in the patrol car while he was being transported to headquarters Appellant regained sufficient composure to rationally and freely choose to incriminate himself.

From the undisputed testimony in the record, it is clear that Appellant was extremely distraught at the time he made the incriminating statements. The depiction of Appellant's conduct in the rear of the patrol car is that of someone in near hysterics. He is described as crying, screaming and thrashing about. To dismiss this conduct as the normal nervousness and behavior of someone recently taken into custody is to indulge in understatement and to ignore the totality of the circumstances. We also cannot accept the Government's theory that Appellant was overcome by remorse for his past misdeeds. The more reasonable explanation for Appellant's behavior is that he was seized by an overwhelming fear that he would suffer violence at the hands of police. This was not an ordinary arrest. Appellant was wanted for the murder and wounding of Detroit police officers. His capture was the culmination of an intense and highly emotional manhunt that was widely publicized in the community. Feelings against Appellant in the Detroit Police Department ran high as amply demonstrated by the officers' testimony at the suppression hearing. In this milieu, apprehensions Appellant may have harbored about his safety in custody were reasonable, and apparently shared by some officers in the Department. That Appellant actually was fearful of being beaten by police is evident from the statements made to Sergeants Studer and Awe, with the former statement occurring immediately prior to the confession. Other factors appear in the record which may have influenced Appellant's belief that he would be beaten by police. He had just received injury at the hands of police during a violent arrest. His injuries, while described as minor, were extensive and obviously painful. His expressed fear that he would be beaten in the patrol car was confirmed when he was struck by Officer Gilbeau. Although a single blow would not necessarily require the suppression of a confession absent additional circumstances, the infliction of physical violence during an interrogation is a weighty factor to consider in determining whether a contemporaneous confession was voluntary. See Stein v. New York, 346 U.S. 156, 182-84, 73 S. Ct. 1077, 97 L. Ed. 1522 (1953). The blow struck by Officer Gilbeau, in conjunction with all the attendant circumstances, could well have influenced Appellant to make statements intended to mollify the officers in the patrol car which, given the opportunity for reflection, he would not have made. The setting of the confession was conducive to the overbearance of Appellant's will. He was situated between two officers in the back seat of a patrol car with his hands cuffed behind his back. If he had any doubts that he was at the mercy of the officers escorting him, they were dispelled when he was struck by Officer Gilbeau.

In the final analysis, we simply cannot conclude that incriminating statements made by a crying, screaming nineteen-year old, thrashing about the back seat of a patrol car, were the product of a free and rational choice. See Culombe v. Connecticut, 367 U.S. at 465, 81 S. Ct. 1860. "(A)ny questioning by police officers which in fact produces a confession which is not the product of a free intellect renders that confession inadmissible." Townsend v. Sain, 372 U.S. at 308, 83 S.Ct. at 754. (emphasis in original).14 Under the totality of the circumstances, the Government has not borne the burden of demonstrating that the confession was voluntary. See Lego v. Twomey, 404 U.S. 477, 92 S. Ct. 619, 30 L. Ed. 2d 618 (1972).

Appellant raises two additional issues which we will address due to the likelihood of retrial on remand. Appellant contests the admission of expert testimony comparing a sample of his hair with three hairs of unknown origin found on a bottleneck discovered at the site of the firebombing. Testimony comparing the hair samples was given by three prosecution witnesses: Kenneth Snow of the Bureau of Alcohol, Tobacco and Firearms, Michael Bayard and Dr. Robert Muggli of the McCrone Institute in Chicago, Illinois. Mr. Snow is a forensic chemist at the Bureau's laboratory in Washington, D. C. He was qualified by the Court as an expert in hair comparison. Mr. Snow conducted the initial tests on the samples using an optical microscope to compare the physical appearance of the hair. He testified on direct examination that the samples were similar in eleven characteristics, although on cross-examination he conceded that the samples were also dissimilar in eleven characteristics. He admitted that visual comparison of hair is a subjective test, but he explained that the sample of unknown hairs was too small to conduct meaningful tests using neutron activation analysis. Appellant does not dispute the admissibility of Mr. Snow's testimony. The controversy surrounds tests on the hair samples conducted by Mr. Bayard and Dr. Muggli using a technique known as ion microprobic analysis. Ion microprobic analysis is a technique for measuring the trace element content of a sample matrix. Both ion microprobic analysis and neutron activation analysis proceed from the theory that matrices from a common source will have similar trace element contents and that matrices from different sources will exhibit dissimilar trace element contents. The techniques are quite different however. In neutron activation analysis the sample matrix is irradiated by bombarding it with neutrons. The trace element content is determined by measuring gamma rays emitted from the now-radioactive sample. See United States v. Stifel, 433 F.2d 431, 436 (6th Cir. 1970).15 See also Note, The Evidentiary Uses of Neutron Activation Analysis, 59 Cal.L.Rev. 997, 998-1010 (1971). In ion microprobic analysis the trace element content of a sample matrix is measured by the number and mass of the ions it releases when pierced by an ion beam. The device for conducting ion microprobic analysis, the ion microprobe mass analyzer, essentially consists of two mass spectrometers. The first mass spectrometer generates a magnetic field which separates ions according to their mass. A beam composed of one type of ion is then projected toward the sample matrix. The ion beam's striking the matrix causes it to cast off ions of varying mass and numbers which are collected and measured by a second mass spectrometer. The data collected by the ion microprobe mass analyzer indicates the concentration of trace elements in the matrix. Although the initial use of ion microprobic analysis was to test metals and inorganic materials, experiments have been conducted using organic matrices as well.16 At the McCrone Institute, Mr. Bayard and Dr. Muggli have been experimenting with human hair as a matrix. In the course of their experiments they have used the ion microprobe mass analyzer to trace the element content of approximately 130 samples of human hair. Each sample was tested for 25 to 40 elements. Comparisons of the samples were made by charting the data supplied by the mass analyzer and visually comparing the charts of elements for each sample against charts of other samples tested. Mr. Bayard and Dr. Muggli were commissioned to test the hair samples in this case. Based on a comparison of charts prepared on the hair samples supplied by the prosecution and the samples they had previously tested, Mr. Bayard and Dr. Muggli reached the conclusion that the hairs were of common origin. After holding a special evidentiary hearing, the District Court ruled that the technique of ion microprobic analysis was well accepted in the field and that expert testimony based thereon was admissible. The District Court allowed the witnesses to testify on the findings of their experiments on human hair and the results of the tests conducted on the hair samples in this case, but the Court would not permit the prosecution's experts to state their ultimate conclusion that the unknown hairs belonged to Appellant. Appellant contends that it was error to admit any testimony based on ion microprobic analysis of human hair.

Traditionally, the permissible limits of expert testimony are left to the sound discretion of the District Court and the decision to permit expert testimony will not be disturbed on appeal unless it is manifestly erroneous. United States v. Green, 548 F.2d 1261, 1268 (6th Cir. 1977). However, expert testimony is not permissible in every case where the witness purports to base his testimony on an ostensibly scientific principle. A necessary predicate to the admission of scientific evidence is that the principle upon which it is based "must be sufficiently established to have gained general acceptance in the particular field to which it belongs." Frye v. United States, 54 App.D.C. 46, 293 F. 1013, 1014 (1923). In United States v. Franks, 511 F.2d 25, 33 n. 12 (6th Cir. 1975), we equated general acceptance in the scientific community with a showing that the scientific principles and procedures on which expert testimony is based are reliable and sufficiently accurate. However, "(a) bsolute certainty of result or unanimity of scientific opinion is not required for admissibility." United States v. Baller, 519 F.2d 463, 466 (4th Cir. 1975). Conflicting testimony concerning the conclusions drawn by experts, so long as they are based on a generally accepted and reliable scientific principle, ordinarily go to the weight of the testimony rather than to its admissibility. United States v. Franks, 511 F.2d at 33. As we stated in United States v. Stifel, 433 F.2d at 438, " neither newness nor lack of absolute certainty in a test suffices to render it inadmissible in court. Every new development must have its first day in court." The clear trend in federal court is toward the admission of expert testimony whenever it will aid the trier of fact. See Fed.R.Evid. 702. However, a strong countervailing restraint on the admission of expert testimony is the defendant's right to a fair trial. See United States v. Green, 548 F.2d at 1268.

There are good reasons why not every ostensibly scientific technique should be recognized as the basis for expert testimony. Because of its apparent objectivity, an opinion that claims a scientific basis is apt to carry undue weight with the trier of fact. In addition, it is difficult to rebut such an opinion except by other experts or by cross-examination based on a thorough acquaintance with the underlying principles. In order to prevent deception or mistake and to allow the possibility of effective response, there must be a demonstrable, objective procedure for reaching the opinion and qualified persons who can either duplicate the result or criticize the means by which it was reached, drawing their own conclusions from the underlying facts.

United States v. Baller, 519 F.2d at 466. A courtroom is not a research laboratory. The fate of a defendant in a criminal prosecution should not hang on his ability to successfully rebut scientific evidence which bears an "aura of special reliability and trustworthiness," although, in reality the witness is testifying on the basis of an unproved hypothesis in an isolated experiment which has yet to gain general acceptance in its field. See e. g., United States v. Amaral, 488 F.2d 1148, 1152 (9th Cir. 1973); Frye v. United States, 293 F. at 1014. In recognition of the inherent danger that expert testimony admitted without proper foundation may tend to confuse or mislead the trier of fact and thus defeat a defendant's right to a fair trial, in United States v. Green, 548 F.2d at 1268, we adopted the four criteria for reviewing a district court's decision to admit expert testimony set down in United States v. Amaral, 488 F.2d at 1152. Four factors must appear in the record to uphold the admission of expert testimony: "1. qualified expert; 2. proper subject; 3. conformity to a generally accepted explanatory theory; and 4. probative value compared to prejudicial effect." United States v. Green, 548 F.2d at 1268, citing, United States v. Amaral, 488 F.2d at 1152. It is the third criteria, "conformity to a generally accepted explanatory theory", which Appellant claims was not met in this case.

Appellant admits that ion microprobic analysis is not a new technique and has attained a sufficient degree of acceptance in the field of mass spectrometry. However, Appellant contends that ion microprobic analysis when used to make hair comparisons is a new technique which has yet to be accepted by the scientific community. He also criticizes the procedures followed by the prosecution's experts in conducting their hair experiments and he claims that certain theories on the properties of human hair propounded by the witnesses during their testimony are novel and unsupported. Appellant cites the rebuttal testimony of two expert witnesses at trial, Dr. Charles A. Evans, Senior Research Chemist at the Illinois Materials Research Laboratory who was qualified as an expert in mass spectrometry in general and ion microprobic spectrometry in particular, and Dr. Adon A. Gordus, Professor of Chemistry at the University of Michigan and an expert in trace element analysis of hair. Although the testimony of rebutting experts normally goes to the weight of the evidence, their testimony concerning currently accepted views in their field and the accuracy and reliability of procedures followed by opposing experts may be of invaluable assistance to the trial court in determining the adequacy of the foundation laid for the admission of scientific evidence.17

After extensive review of the record, we are inclined to agree with Appellant that the Government failed to fulfill the threshold requirement of demonstrating that ion microprobic analysis is a generally accepted procedure for comparing samples of human hair and that the experiments conducted by their experts carry sufficient indicia of reliability and accuracy to be said to cross "the line between the experimental and demonstrable stages". Frye v. United States, 293 F. at 1014. Research has failed to disclose a single reported case where testimony based on microprobic analysis of hair has been admitted into evidence, and the parties have not cited any cases, reported or unreported, where a court has admitted evidence based on ion microprobic analysis. Neither Mr. Bayard nor Dr. Muggli claimed to be an expert on human hair. The sole bases for their claimed expertise in the field are the experiments they performed on the 130 hair samples using the ion microprobe mass analyzer. Both concede that their test results have not been duplicated elsewhere and neither was able to cite any authority in the field in support of their positions. Mr. Snow testified that the three hairs of unknown origin found at the scene of the firebombing were too small a sampling to conduct a meaningful examination using neutron activation analysis. Mr. Bayard and Dr. Muggli stated that a valid comparison of known and unknown human hairs may be made by ion microprobe mass analysis using a smaller sample than is necessary to obtain a valid comparison using neutron activation analysis. They grounded this belief on the theory that ion microprobic analysis, unlike neutron activation analysis, measures only the sub-surface trace element composition of the hair and that this factor is relatively constant in all hairs derived from a common source. They conceded that there was no published authority to support this position, but they discounted differing spectrographic analysis results as being affected by surface contamination. Dr. Gordus in rebuttal testified according to the prevailing view that trace element concentrations vary significantly across a person's head and along the length of a hair. See Note, 59 Cal.L.Rev. at 1039 & n. 240. There thus appears to be nothing in the record, except unsupported assertions by the prosecution's experts, to indicate that ion microprobic analysis may achieve a reliable and meaningful result using a quantitatively smaller sampling of hair than that which would be needed for neutron activation analysis. The procedures employed by the prosecution's experts in testing the hair samples also militate against the admission of testimony based on the results of their experiments. The witnesses' conclusion that the unknown hairs and Appellant's hairs derived from a common source was reached by comparing charts compiled on tests of those samples against the charts prepared on the samples they had previously tested. The comparisons were subjective and based on visual calculation. Both witnesses conceded that there were no standards with which to assess the accuracy of their measurement of trace element presence, and they agreed that their measurements were relative rather than absolute. The National Standard Bureau has listed absolute standards for a number of materials tested by ion microprobic analysis, but no standard has been developed for trace element analysis of human hair. Cf. Note, 59 Cal.L.Rev. at 1038. Where there is no absolute standard by which to gauge the accuracy of the tests, it is incumbent on the proponents to demonstrate that the relative standards of comparison they employ in their experiments are sufficiently reliable and accurate. The most troubling aspect of the testing done here is that no attempt was made to match the test samples against a statistically valid test group. The witnesses admitted that their samples were gathered in no particular manner and they were not seeking to obtain a random sample of the general population or of a relevant sub-group. Of the 130 samples collected, only 20 were from black people, and two-thirds of the test group were from the metropolitan Chicago area. Demographic data about the individuals who supplied the hair samples were incomplete and, apparently, not treated as a matter of great import by the prosecution's experts. However, the prevailing view in the field is that a number of factors peculiar to the individual and his background, including diet, metabolism, occupation, locale, and environment, have a definite bearing on the trace element content of human hair. See generally Note, 59 Cal.L.Rev. at 1038, 1048. There is a clear danger of misidentification where a test compares the hair of a black defendant from Detroit with hairs taken from a predominantly white test group collected primarily in Chicago. See id. at 1049. The reliability and accuracy of the tests are further drawn in question by the fact that only four of the 130 samples tested involved comparison of hairs from the same subject. The small number of test samples where actual comparisons were made of hairs taken from verifiable sources provides scant support for the prosecution's claim that the experiments conducted by its experts prove that ion microprobic analysis can reliably and accurately match hair samples from known and unknown sources.

The District Court would not permit the prosecution experts to state their ultimate conclusion that the hair samples were of common origin, presumably because of a lack of proper foundation. This ruling did not go far enough. The clear implication from the experts' testimony was that the hairs found at the scene of the firebombing belonged to Appellant. The limitation on the scope of testimony imposed by the District Court could not cure the prejudice to Appellant from exposing the jury to scientific evidence linking him to the crime which failed to satisfy threshold requirements for admissibility. See United States v. Green, 548 F.2d at 1268; Frye v. United States, 293 F. at 1014. While it is a truism that every useful new development must have its first day in court, see United States v. Stifel,433 F.2d at 438, expert testimony on a critical fact relating to guilt or innocence is not admissible unless the principle upon which it is based has attained general acceptance in the scientific community and is not mere speculation or conjecture.18 We are not convinced, on the present state of the record, that ion microprobic analysis of human hair has yet reached the level of general acceptance in its field, or that the experiments conducted in this case have been shown to be sufficiently reliable and accurate, to provide an acceptable basis for expert identification in a criminal trial.

The final issue raised by Appellant is the constitutionality of 18 U.S.C. § 844(f), malicious destruction by means of an explosion of an institution receiving federal financial assistance. We affirm the constitutionality of § 844(f) as applied to the Planned Parenthood Association for the reasons stated in the District Court's opinion. United States v. Brown, 384 F. Supp. 1151 (E.D.Mich.1974). The judgment of the District Court is reversed.