concurring:
I concur in the result and all of the opinion except Part III, which discusses an issue that we need not now decide.
GOLDBERG, Circuit Judge, with whom POLITZ, JOHNSON and WILLIAMS, Circuit Judges, join dissenting:Today the majority chooses to reject this Circuit’s longstanding test for determining whether a suspect is in custody.1 Judge *602Clark’s fine opinion has convinced me that the Supreme Court has specifically rejected each element of our old four part test, and mandated an objective standard. I dissent, not because I would disinter the old test, but instead because I disagree with the majority’s application of their newly fashioned standard. The rejection of the four part test is merely a change in technique. My concern is that the new standard be applied in a manner that is consistent with the underlying constitutional and prophylactic principles stated in Miranda v. Arizona, 384 U.S. 436, 86 S.Ct. 1602, 16 L.Ed.2d 694 (1966).
The majority states, and I agree, that a person is in custody when the police have imposed a restraint on liberty of a degree a reasonable person would associate with an arrest. To my mind, however, applying this standard, Ms. Bengivenga was in custody from the moment that she crossed the threshold of the checkpoint trailer. From that moment forward, a reasonable person would have found the circumstances indistinguishable from a formal arrest, and the inherently coercive aspects of custodial questioning came into play. I also disagree with the majority’s conclusions that the Fifth Amendment exclusionary rule does not apply because the baggage claim stubs were non-testimonial in and of themselves, and, that they cannot be excluded as fruit of the poisoned tree, because there was only a Miranda violation, not a constitutional violation. It was the act of producing the ticket and the baggage stubs that was testimonial, not the documents themselves, and that act of production occurred in an inherently coercive context. Without the act of production to link the ticket stubs to Ms. Bengivenga, the stubs themselves were meaningless.
I. THE HALLMARKS OF CUSTODY
A. Inherent Coercion
The Fifth Amendment guards only against coerced confessions. In Miranda v. Arizona, 384 U.S. 436, 86 S.Ct. 1602, 16 L.Ed.2d 694 (1966), however, the Court noted the natural tendency of custodial interrogation to overbear the will of suspects, saying, “Even without employing brutality, [or] the ‘third degree]]]’ ... the very fact of custodial interrogation exacts a heavy toll on individual liberty and trades on the weakness of individuals.” 384 U.S. at 455, 86 S.Ct. at 1617-18. Because of this inherent coercion we now apply a conclusive presumption that evidence produced by custodial interrogation was coerced, unless the suspect was informed of his rights. 384 U.S. at 478, 86 S.Ct. at 1630.
Today the majority fashions and applies a new standard for determining whether a suspect is in custody. In determining how to apply that standard, a page of history may be worth a pound of logic. We must look to the concerns that informed Miranda to determine when a factual situation contains those elements of inherent coercion that led the court to require Miranda warnings.
B. Privacy and Duration
The Miranda Court carefully examined police practices and focused on certain elements inherent in custody that rendered it coercive. In police manuals current at the time, two elements were noted to be particularly effective at overbearing the will of a suspect: privacy and the prospect of questioning for an indefinite period.
First, the Court noted the efficacy of privacy saying:
The officers are told by the manuals that the “principal psychological factor contributing to successful interrogation is privacy — being alone with the person under interrogation.”
Id. at 449, 86 S.Ct. at 1615 (quoting Inbau & Reid, Criminal Interrogation and Confessions 1 (1962)). The Court noted that the manuals further advised:
If at all practicable, the interrogation should take place in the investigator’s office or at least in a room of his own choice. The subject should be deprived of every psychological advantage. In his own home he may be confident, indignant, or recalcitrant. He is more keenly aware of his rights and more reluctant to tell of his indiscretions or criminal behav*603ior within the walls of his own home. Moreover his family and other friends are nearby, their presence lending moral support. In his office, the investigator possesses all the advantages. The atmosphere suggests the invincibility of the forces of the law.
Id. (quoting O’Hara, Fundamentals of Criminal Investigation 99 (1956)).
Second, the Court noted that the manuals focused on the prospect of indefinite interrogation. They encouraged the investigator to be “patien[t],” and to “persever[e]”:
Where emotional appeals are employed to no avail, [the investigator] must rely on an oppressive atmosphere of dogged persistence. He must interrogate steadily and without relent, leaving the subject no prospect of surcease.
Id. at 451, 86 S.Ct. at 1615 (quoting O’Hara at 112).
Both privacy and the prospect of indefinite interrogation were the hallmarks of custodial interrogation which led the Court to apply a conclusive presumption to custodial interrogation, saying that:
Without proper safeguards the process of in custody interrogation of persons suspected or accused of crimes contains inherently compelling pressures which work to undermine the individual’s will to resist and to compel him to speak where he would not otherwise do so freely-
Id. at 467, 86 S.Ct. at 1624. The Court therefore held that “when an individual is taken into custody or otherwise deprived of his freedom by the authorities in any significant way,” that individual must be advised of his constitutional rights. Id. at 478, 86 S.Ct. at 1630.
Determining why a suspect must receive Miranda warnings only begins the inquiry into when and under what circumstances such warnings are necessary. The majority takes note of these whys, but comes to an erroneous conclusion about when. They note the importance of privacy and indefinite duration, but fail to see that these hallmarks of inherently coercive custody are present in the factual situation of this case.
II. DEFINING CUSTODY BY REFERENCE TO INHERENT COERCION
As the majority explains, not all Fourth Amendment seizures constitute custody. For example, a traffic stop constitutes a seizure, but does not entitle a suspect to Miranda warnings. Similarly, a police officer may ask an airline passenger for his or her ticket on the main concourse of the airport or in the baggage claim area, without that individual being deemed in custody.2 A suspect’s freedom must be more than restricted. It must be restricted in a “significant way.” Miranda, 384 U.S. at 444, 86 S.Ct. at 1612. The “restraint” must be “of the degree associated with a formal arrest.” Berkemer v. McCarty, 468 U.S. at 440, 104 S.Ct. at 3150, 82 L.Ed.2d 317 (1984).
This distinction between seizure and custody creates a necessarily circular inquiry. Custody is defined not by any extrinsic characteristic. The locution “restraint of the degree associated with formal arrest” does little more than establish the distinction. It does not draw the line. The majority quite properly seeks a way out of this conundrum by looking to the dangers of custody highlighted in Miranda. They focus on privacy and the prospect of indefinite duration, but draw the line in a way that leaves the Fifth Amendment at the mercy of police discretion.
A. The Hallmarks of a Non-Custodial Seizure
To understand where the majority’s analysis goes awry requires a careful understanding of the relationship between Miranda and the more recent case of Berkemer v. McCarty, 468 U.S. at 420, 104 S.Ct. at 3138. The majority treats the questioning of Ms. Bengivenga as an ordinary traffic stop. In Berkemer, the Supreme Court noted, as we have already, that:
*604Two features of an ordinary traffic stop mitigate the danger that a person questioned will be induced to speak where he would not otherwise do so freely.... First, detention of a motorist pursuant to a traffic stop is presumptively temporary and brief. The vast majority of roadside detentions last only a few minutes.... In this respect, questioning incident to an ordinary traffic stop is quite different from stationhouse interrogation, which frequently is prolonged, and in which the detainee often is aware that questioning will continue until he provides the interrogators the answers they seek.
Second, circumstances associated with the typical traffic stop are not such that the motorist feels completely at the mercy of the police.... Most importantly, the typical traffic stop is public.... Passersby, on foot or in other cars witness the interaction of the officer and motorist. This exposure to public view both reduces the ability of an unscrupulous policeman to use illegitimate means to elicit self-incriminating statements and diminishes the motorist’s fear that, if he does not cooperate, he will be subject to abuse.
Id. at 437-39, 104 S.Ct. at 3149.
The Court was particularly concerned that if these two factors should evaporate the traffic stop would lose its non-custodial character. Otherwise, traffic stops “might open the way to widespread abuse. Policemen [might] simply delay formally arresting detained motorists, and ... subject them to sustained and intimidating interrogation at the site of their initial detention.” Id. at 440, 104 S.Ct. at 3150. To prevent this, the court emphasized that “If a motorist who has been detained pursuant to a traffic stop thereafter is subjected to treatment that renders him ‘in custody’ for practical purposes, he will be entitled to the full panoply of protection prescribed by Miranda. ” Id. The Court specifically eschewed a per se rule that traffic stops are non-custodial, and pointed out that, “police and lower courts will continue occasionally to have difficulty deciding when a suspect has been taken into custody.” Id. at 441, 104 S.Ct. at 3151.
B. A Non-Custodial Seizure With All the Hallmarks of Custody?
The majority takes great pains to demonstrate that this particular border stop was both public and brief. I cannot agree. The Berkemer Court pointed out that when a traffic stop takes place on the street, the presence of passersby guards against actual police misbehavior and undercuts the police dominated environment. This characteristic of a traffic stop disappeared as soon as Ms. Bengivenga was escorted to the trailer. Ms. Bengivenga was isolated from view of her fellow travelers on the bus and from other travelers who might pass through the checkpoint. It was midnight, and if the bus left, she would be left in an unfamiliar place at least till morning. Granted, she was not taken into an interrogation room, and the bus driver was present when Ms. Bengivenga produced the ticket. Still, the questioning in the trailer was in no way akin in its public character to questioning, on a public thoroughfare open to the plain view of all who should happen to pass by. Furthermore, Ms. Bengivenga was not driving her own car, nor was she capable of leaving under her own power at the close of questioning.
The majority also argues that this interrogation took only a minute and a half, and was therefore brief. To rely upon a stopwatch to make this sensitive and subtle determination is not the correct approach. The phrase used by the Supreme Court in Berkemer, was “presumptively temporary and brief.” Berkemer, 468 U.S. at 437, 104 S.Ct. at 3149. I cannot see how the fact that it only took ninety seconds for Ms. Bengivenga to produce incriminating evidence after being brought to the trailer suggests that a reasonable person once brought to a checkpoint trailer would expect brief detention. Although the actual length of questioning is relevant to whether Ms. Bengivenga’s will was actually over-born, it has no relevance to the inherent coerciveness of the situation. Indeed, it may indicate just how coercive the situation was. The Constitution does not give police officers a free minute and a half to *605do whatever they wish. As the Court pointed out in both Berkemer and Miranda, the coercion of custodial interrogation comes not from the actual length of the interrogation, but from a recognition that the police will hold a suspect until they get answers to their questions. 468 U.S. at 438, 104 S.Ct. at 3149; 384 U.S. at 451, 86 S.Ct. at 1615-16.
The majority notes that citizenship checks, even once the individual is moved to a secondary checkpoint, take, on the average, three and a half minutes. They make, however, the unwarranted further assumption that the same presumption of brevity applies to border searches for drugs. Unlike a citizenship check, which is necessarily brief, either the individual can prove citizenship, or he cannot, a drug related detention may take much longer, even in the absence of probable cause. At the time of this case, plenary border searches were allowed in this Circuit,3 and the Supreme Court has held that extended detention at the border without sanitary facilities is permissible when based on reasonable suspicion of drug smuggling. United States v. Montoya de Hernandez, 473 U.S. 531, 105 S.Ct. 3304, 87 L.Ed.2d 381 (1985) (sixteen hours). See also, United States v. Oyekan, 786 F.2d 832, 833-34, 836 (8th Cir.1986) (four hours). A reasonable, though innocent, person in Ms. Bengiven-ga’s position could therefore have feared that her stay in the checkpoint trailer would last quite a long time.
For these reasons, I believe that the character of the questioning changed as soon as Ms. Bengivenga was moved from the area outside the bus to the checkpoint trailer. At that point the questioning was no longer public, and there was the prospect of extended interrogation. The majority’s analysis takes a long first step toward allowing police officers to blur the line, between a brief traffic stop and custodial detention, that was central to the Court’s opinion in Berkemer.
I also object for another reason. By declaring this search non-custodial, the majority misses an opportunity to clarify the law. Every police officer must determine for himself when a traffic stop becomes custodial. The burden is always on the police to show that evidence is admissible and that a confession is voluntary. However, Berkemer tells us that so long as the stop is on the street, and short, it is presumptively non-custodial. To my mind, that presumption ends as soon as the suspect is removed from the public thoroughfare to a squad car, checkpoint trailer or station house. To reestablish a presumption of voluntariness, the police must inform the suspect of his rights, or of the fact that he is not under arrest.4
The majority, however, leaves it unclear what factors are dispositive in this case. Does entering the trailer matter at all? Is the presence of the bus driver in the trailer dispositive? I would simplify matters and hold that when a suspect is moved to a secondary checkpoint on suspicion of drug smuggling (as distinct from a citizenship check), he or she is in custody and should be informed of his or her rights. Even if one does not adopt such a per se rule, I believe that Ms. Bengivenga, purely on the facts of this case, was in custody.5
*606III. TESTIMONY?
Because the majority finds that Ms. Ben-givenga was not in custody, they needed go no further to affirm the conviction. Still, they reach out to redefine testimony to exclude the production of baggage stubs. In so doing they misread the primary case they cite for support. The majority argues that production of the ticket was non-testimonial, because all passengers carry tickets. In so doing, they focus on the content of the requested papers. This is completely inconsistent with the approach adopted in Fisher v. United States, 425 U.S. 391, 96 S.Ct. 1569, 48 L.Ed.2d 39 (1976), and followed in United States v. Doe, 465 U.S. 605, 104 S.Ct. 1237, 79 L.Ed.2d 552 (1984).
Fisher rejected the approach followed in Boyd v. United States, 116 U.S. 616, 6 S.Ct. 524, 29 L.Ed. 746 (1886), which fo-cussed on the content of a requested document. Fisher held that the Fifth Amendment applies only to the act of production. The issue is whether the act of producing a document conveys sufficient information to be considered testimonial. The Court said:
The act of producing evidence in response to a subpoena nevertheless has communicative aspects of its own, wholly aside from the content of the papers produced. Compliance with the subpoena tacitly concedes the existence of the papers demanded and their possession or control by the taxpayer. It also would indicate the taxpayer’s belief that the papers are those described in the subpoena.... The elements of compulsion are clearly present but the more difficult issues are whether the tacit averment of the taxpayer are both “testimonial” and “incriminating” for purposes of applying the Fifth Amendment. These questions perhaps do not lend themselves to categorical answers; their resolution may instead depend on the facts and circumstances of particular cases.
Id. 425 U.S. at 410, 96 S.Ct. at 1581. In Fisher the existence, possession and control of the requested documents were not an issue. Production of the documents conveyed no information in and of itself. However, the Court stated that an act of production that did convey information would be testimonial.
This approach was followed in United States v. Doe, 465 U.S. 605, 104 S.Ct. 1237, 79 L.Ed.2d 552 (1984), where the district court held that requiring defendant to produce the business records of a sole proprietorship would force him to admit that the documents existed, that they were in his possession and that they were authentic. The Supreme Court concluded that defendant could not be compelled to produce those records.
This case is indistinguishable from Doe. The police knew that the seized luggage contained drugs, but they needed something to link them to Ms. Bengivenga. Baggage stubs alone do not suffice to link a person to a bag. The baggage stubs must be in possession of the suspect to convey any information at all. Where as here, the physical evidence would be meaningless without the physical act of production, that act must be deemed testimonial.
The majority further attempts to distinguish production of the ticket from production of the ticket stubs. Without doubt, production of the ticket itself was testimonial, as defined in Fisher. The agent requested the ticket and Ms. Bengivenga produced it. Possession of a ticket to Alice confirmed that Ms. Bengivenga was indeed going to Alice, the destination of the marijuana laden suitcases. In the ticket envelope were baggage stubs that linked Ms. Bengivenga to the suitcases. The majority’s focus on the content of the documents produced causes them to mistakenly treat the stubs as a fruit of the tainted testimony rather than as the testimony itself. The act of production of the ticket envelope *607which contained both the stubs and the ticket was the testimony. It makes no difference that one testimonial act produced two pieces of paper, when the pieces of paper are meaningless if the testimonial act is suppressed.
******
It is, therefore, with great regret that I dissent from the majority opinion, the author of which has earned, over the years, my respect for both his analysis and judgment. In this case, however, I am forced to conclude that a reasonable person, in Ms. Bengivenga’s position, would have felt compelled to answer any questions put to her by the officers, unless apprised in advance of her constitutional rights, and that the act of producing the ticket and baggage stubs conveyed enough information to be deemed testimonial.
On this rationale I would reverse Ms. Bengivenga’s conviction.
. United States v. Alvarado Garcia, 781 F.2d 422 (5th Cir.1986); United States v. Warren, 612 F.2d 887 (5th Cir.) (en banc), cert. denied, 446 U.S. 956, 100 S.Ct. 2928, 64 L.Ed.2d 815 (1980); United States v. Henry, 604 F.2d 908, 915 (5th Cir.1979); Brown v. Beto, 468 F.2d 1284, 1286 (5th Cir.1972).
. Berkemer v. McCarty, 468 U.S. 420, 440, 104 S.Ct. 3138, 3150, 82 L.Ed.2d 317 (1984) (traffic stop); United States v. Gonzales, 842 F.2d 748 (5th Cir.1988) (airline ticket).
. United States v. Jackson, 825 F.2d 853, 860-62 (5th Cir.1987) (en banc), cert. denied sub nom., Ryan v. United States, — U.S.-, 108 S.Ct. 711, 98 L.Ed.2d 661, cert. denied sub nom., Browning v. United States, - U.S. -, 108 S.Ct. 730, 98 L.Ed.2d 679 (1988).
. This position does not conflict with either Oregon v. Mathiason, 429 U.S. 492, 97 S.Ct. 711, 50 L.Ed.2d 714 (1977), or California v. Beheler, 463 U.S. 1121, 103 S.Ct. 3517, 77 L.Ed.2d 1275 (1983). Both of these cases deal with station house questioning, where the suspect voluntarily agreed to accompany officers to the police station. In both cases, the suspect was specifically informed that he was not under arrest, and in both cases the suspects actually did leave the station house within half an hour. 429 U.S. at 492, 494, 97 S.Ct. at 711, 713; 463 U.S. at 1122, 103 S.Ct. at 3518.
.The Ninth Circuit applies an objective standard as well, but they have articulated some of the factors which contribute to an objectively reasonable belief that custody has begun. They examine four factors: 1) the officer’s language in summoning the suspect; 2) the physical surroundings of interrogation; 3) the extent to which the accused is confronted with evidence of his guilt; and 4) the amount of pressure exerted. Applying this standard, the Ninth Circuit has held that the objective circumstances *606surrounding a customs inspection were sufficient to constitute custody, where the suspect was confronted with evidence of illegally imported merchandise and was not told that she was free to leave. United States v. Estrada-Lucas, 651 F.2d 1261, 1266 (9th Cir.1980).
Estrada-Lucas suggests that, at least in the Ninth Circuit, the context surrounding the questioning of Ms. Bengivenga would have been held to constitute custody under an objective standard.