This appeal presents an issue of first impression in this circuit: whether an in-custody, unsolicited statement, not made in response to any police interrogation, must be suppressed because it followed an earlier voluntary statement made in violation of Miranda v. Arizona, 384 U.S. 436, 86 S. Ct. 1602, 16 L. Ed. 2d 694 (1966).1I. Factual and Procedural History

On June 13, 1979, Thompson Myers was shot at close range with a flare gun outside the Wonder Bar in Honolulu, Hawaii. Shortly thereafter, Officer Trela stopped the appellant, Harold C. Medeiros, who was driving an automobile which matched the description given by witnesses to the shooting. Medeiros' eyes were red and glassy and he had an odor of alcohol about him. Medeiros asked Officer Trela why he had been stopped. Officer Trela replied that "there had been a shooting at the Wonder Bar and that [Medeiros'] automobile matched the description of one identified as leaving the scene." Officer Trela then asked Medeiros where he was coming from, without first advising him of his Miranda rights. Medeiros replied that he had come from the Wonder Bar and then spontaneously incriminated himself with respect to the shooting. (The "first statement").2

Medeiros was arrested and taken to the police station. After being "booked," Officers Silva and Miyashiro were instructed to take Medeiros to the Pawaa Annex for medical treatment of a laceration over Medeiros' left eye. Officer Silva testified that Medeiros' eyes were red and glassy, his voice was loud, and he was unsteady on his feet. The parties stipulated that Medeiros had a blood alcohol level of 0.19 approximately one hour after his arrest.

Prior to and during treatment, and without any prompting from either officer, Medeiros made several more inculpatory statements to these officers, collectively referred to as the "second statement." Medeiros exclaimed:

He went hit me and the gun went off. I killed him. Good for him. How you like he take my 15 year old daughter.3

Officer Silva responded by telling Medeiros "You shouldn't give any statements at this time," but stopped short of reading Medeiros the Miranda warnings. Not heeding Silva's suggestion, Medeiros then exclaimed:

He hit me and I shot him. He was selling her ass. Officer Silva again told Medeiros to say nothing further, but did not read Medeiros the Miranda warnings. Immediately thereafter, Medeiros continued:

I killed that------I killed that nigger. Good for him. I hope he's dead. I shot him in the fucking head. They beat me up and kick me. Good for him and I hope that black bastard dies, that black son-of-a-bitch.

Medeiros made these statements approximately 30 minutes after he had made the first incriminating statement to Officer Trela.

On the following day, Officer Chung advised Medeiros of his Miranda rights. After waiving his rights, Medeiros requested the officer provide him with his prescription pain medication for his back condition. After the officer refused, Medeiros made a third inculpatory statement. (The "third statement").

The state court trial judge granted Medeiros' motion to suppress the first statement as the product of an unwarned custodial interrogation, which violated Miranda. The trial judge also granted Medeiros' motion to suppress the third statement because the failure of the police to give Medeiros his pain medication, once he requested it, rendered his third statement involuntary. However, the trial judge denied Medeiros' motion to suppress the second statement given to Officers Silva and Miyashiro at the Pawaa Annex. Ultimately, Medeiros was convicted of manslaughter at a bench trial and sentenced to twenty years incarceration.

On appeal to the Intermediate Court of Appeals for Hawaii, Medeiros' conviction was affirmed. State v. Medeiros, 4 Haw.App. 248, 251, 665 P.2d 181, 183 (1983). The state appellate court rejected Medeiros' argument that his second statement was either involuntary or the "fruit of the poisonous tree" of the prior inadmissible confession given to officer Trela (the first statement). The court concluded that the second statement was spontaneous, voluntary, unsolicited, and neither the product of exploiting the first statement nor the fruit of the poisonous tree.

Thereafter, Medeiros filed a petition for a writ of habeas corpus in the United States District Court for the District of Hawaii. This petition raised the same arguments that the state appellate court had rejected. The court made an independent determination of the voluntariness of the second confession, Miller v. Fenton, 474 U.S. 104, 110-12, 106 S. Ct. 445, 449-51, 88 L. Ed. 2d 405 (1985), and concluded that the second statement was "purged of the primary taint," under Wong Sun v. United States, 371 U.S. 471, 487-88, 83 S. Ct. 407, 417, 9 L. Ed. 2d 441 (1963), because Officers Silva and Miyashiro "took no action remotely designed to encourage [Medeiros] to speak at the Annex." Additionally, the district court rejected Medeiros' argument that the psychological coercion of having made the first statement let the cat out of the bag and forced him to make the second statement. Relying on Oregon v. Elstad, 470 U.S. 298, 309, 105 S. Ct. 1285, 1293, 84 L. Ed. 2d 222 (1985), the district court found that Medeiros' second statement was "volunteered" rather than the result of the previous unwarned interrogation, and thus was not irreparably tainted by the inadmissible first statement. Finally, apart from the potential impact of having made the first statement, the district court concluded that the second statement was not rendered involuntary because of any claimed diminished capacity due to intoxication, drugs, or the laceration. On these grounds, the district court denied Medeiros' petition for a writ of habeas corpus. Medeiros timely appealed.

We review a district court's decision to deny a habeas corpus petition de novo. Weygandt v. Ducharme, 774 F.2d 1491, 1492 (9th Cir.1985). State court factual conclusions are entitled to a presumption of correctness under 28 U.S.C. Sec. 2254(d). However, the state court's conclusion regarding whether Medeiros' second confession was voluntary is a legal conclusion; therefore, it is not entitled to a presumption of correctness and merits independent de novo consideration. Miller, 474 U.S. at 110-12, 106 S.Ct. at 449-51; United States v. Wolf, 813 F.2d 970, 974-75 (9th Cir.1987).

The district court concluded that the second statement was not the result of the prior inadmissible confession. Because the district court acknowledged that the first statement was voluntary, the district court appears to have presumed that the first statement was inadmissible because it was the product of an un-Mirandized custodial interrogation. The district court's conclusion that Medeiros was subjected to custodial interrogation is essentially a factual determination, United States v. Wauneka, 770 F.2d 1434, 1438 (9th Cir.1985) (citing United States v. Crisco, 725 F.2d 1228, 1230 (9th Cir.), cert. denied, 466 U.S. 977, 104 S. Ct. 2360, 80 L. Ed. 2d 832 (1984) (citation omitted)), reviewed under the clearly erroneous standard. United States v. McConney, 728 F.2d 1195 (9th Cir.), cert. denied, 469 U.S. 824, 105 S. Ct. 101, 83 L. Ed. 2d 46 (1984). Officer Trela told Medeiros that his vehicle fit the description of one seen by witnesses leaving the scene of the shooting and then asked him where he had come from. We conclude that while there is room for debate, the district court's conclusion that Medeiros was "in custody" is not clearly erroneous because "a reasonable innocent person in such circumstances would conclude that ... he or she would not be free to leave." United States v. Booth, 669 F.2d 1231, 1235 (9th Cir.1981). Similarly, because it is not clearly erroneous, we uphold the district court's conclusion that Medeiros was subjected to "interrogation" because the police should have known that their words and actions were "reasonably likely to elicit an incriminating response from the suspect." Rhode Island v. Innis, 446 U.S. 291, 301, 100 S. Ct. 1682, 1689, 64 L. Ed. 2d 297 (1980) (footnotes omitted).

We review the district court's determination that the first statement was voluntarily made under the de novo standard, and also find that it was given voluntarily. A confession is voluntary if it is " 'the product of a rational intellect and a free will' ... whether [or not] a confession is the product of physical intimidation or psychological pressure [or] a drug- [alcohol-] induced statement." Townsend v. Sain, 372 U.S. 293, 307, 83 S. Ct. 745, 754, 9 L. Ed. 2d 770 (1963) (quoting Blackburn v. Alabama, 361 U.S. 199, 208, 80 S. Ct. 274, 280-81, 4 L. Ed. 2d 242 (1960)); Gladden v. Unsworth, 396 F.2d 373, 380-81 (9th Cir.1968). See also Colorado v. Connelly, 479 U.S. 157, 164, 107 S. Ct. 515, 520, 93 L. Ed. 2d 473 (1986) (the mental condition of the defendant is the key factor in determining voluntariness). The district court examined the circumstances surrounding Medeiros' first statement and concluded that although he was intoxicated, he was not incapacitated. He was able to drive an automobile, obey the officers' orders prior to and during the initial stop and to cooperate in conversing with them. Based on these findings, we conclude that the surrounding circumstances, including Medeiros' intoxication, were insufficient to overcome his free will and cause his first statement to be the product of anything other than a rational mind.

Although Medeiros was in police custody when he made the second statement at the Pawaa Annex, the police clearly were not interrogating him at that time. Therefore, because Miranda warnings were not necessary, the key issue surrounding the admissibility of the second statement is whether Medeiros made it voluntarily. Considering there was no police interrogation and that Medeiros' intoxication was insufficient to overcome his free will and rational intellect (see Section III. discussion above), the only source of coercion which could have made this statement involuntary is the psychological impact of having let "the cat out of the bag" with the first statement.

In United States v. Bayer, 331 U.S. 532, 540-41, 67 S. Ct. 1394, 1398, 91 L. Ed. 1654 (1947), the Supreme Court recognized:

[A]fter an accused has once let the cat out of the bag by confessing, no matter what the inducement, he is never thereafter free of the psychological and practical disadvantages of having confessed. He can never get the cat back in the bag. The secret is out for good. In such a sense, a later confession may always be looked upon as fruit of the first. But this Court has never gone so far as to hold that making a confession under circumstances which preclude its use, perpetually disables the confessor from making a usable one after those conditions have been removed. (emphasis added).

The Supreme Court applied Bayer's cat out of the bag analysis in Elstad. In Elstad, the suspect made his first incriminating statement voluntarily, but without first being given the requisite Miranda warnings. One hour later, he was advised of and waived his Miranda rights and executed a written confession. The Court held that "a suspect who has once responded to unwarned yet uncoercive questioning is not thereby disabled from waiving his rights and confessing after he has been given the requisite Miranda warnings." Elstad, 470 U.S. at 318, 105 S.Ct. at 1298. While we recognize that the facts upon which Elstad is based differ from the facts of this case, we are convinced that the Elstad decision is grounded on the same fundamental constitutional principles that control this case. Therefore, we look to Elstad for guidance in determining the voluntariness of Medeiros' second confession.

Elstad recognized that a "[f]ailure to administer Miranda warnings creates a presumption of compulsion" and "[c]onsequently, unwarned statements that are otherwise voluntary within the meaning of the Fifth Amendment must nevertheless be excluded from evidence under Miranda." Elstad, 470 U.S. at 307, 105 S.Ct. at 1292. However, the Elstad court went on to examine " 'how sweeping the judicially imposed consequences' of a failure to administer Miranda warnings should be," Elstad, 470 U.S. at 308, 105 S.Ct. at 1292 (quoting Michigan v. Tucker, 417 U.S. 433, 445, 94 S. Ct. 2357, 2364, 41 L. Ed. 2d 182 (1974)), and concluded:

It is an unwarranted extension of Miranda to hold that a simple failure to administer the warnings, unaccompanied by any actual coercion or other circumstances calculated to undermine the suspect's ability to exercise his free will, so taints the investigatory process that a subsequent voluntary and informed waiver is ineffective for some indeterminate period.... [T]he admissibility of any subsequent statement should turn in these circumstances solely on whether it is knowingly and voluntarily made.

Elstad, 470 U.S. at 309, 105 S.Ct. at 1293 (emphasis added). While the Court recognized that having let the cat out of the bag may create a subtle form of lingering psychological compulsion, the Court concluded that "there is no warrant for presuming coercive effect where the suspect's initial inculpatory statement, though technically in violation of Miranda, was voluntary." Elstad, 470 U.S. at 318, 105 S.Ct. at 1297 (emphasis added) (footnote omitted).4 Instead, the voluntariness of any subsequent statement depends upon an evaluation of the "entire course of police conduct" and the "surrounding circumstances," Elstad, 470 U.S. at 318, 105 S.Ct. at 1298, including consideration of whether the conditions that made the first statement inadmissible have been removed. Bayer, 331 U.S. at 541, 67 S.Ct. at 1398. Although the Court in Elstad indicated that "[t]he fact that a suspect chooses to speak after being informed of his rights" is "highly probative" of voluntariness, Elstad, 470 U.S. at 318, 105 S.Ct. at 1298, the Court stopped short of holding that a suspect can make a voluntary statement, after having previously made a voluntary but unwarned admission, only after first being informed of his rights. Quite to the contrary, the Court stated that "[f]ar from establishing a rigid rule, [e.g., requiring a passage of time or a break in events], we direct courts to avoid one;" Elstad, 470 U.S. at 318, 105 S.Ct. at 1297, the fact finder's determination of voluntariness must turn on an evaluation of the surrounding circumstances and the entire course of police conduct.

After evaluating the surrounding circumstances and the entire course of police conduct in this case, we conclude that Medeiros made his second statement voluntarily. The circumstances which had made his first statement inadmissible--custodial interrogation without having first been advised of his Miranda rights--no longer existed at the time he made his second statement. Although Medeiros was in custody at the Pawaa Annex, Medeiros spontaneously made his second incriminating statement, without any prompting or questioning by the officers present. At no point did either Officer Silva or Miyashiro ever interrogate Medeiros. Quite to the contrary, the officers repeatedly told Medeiros not to say anything further.5 Additionally, these were not the same officers who initially stopped Medeiros and heard his first incriminating statement. These officers did not know that Medeiros had previously incriminated himself, served merely as escorts, and had no role in the investigation or in any interrogation of the suspect. Therefore, there could not have been even a covert attempt by the police to exploit the first statement to obtain another incriminating remark.

Although the one-half hour time lapse between Medeiros' first and second incriminating statements may not seem long enough to remove any lingering psychological compulsion from having already confessed, the time lapse between the suspect's two incriminating statements in Elstad was only one hour. In this case, like Elstad, the second statement was made at a different location that cannot be considered "coercive." See Elstad, 470 U.S. at 315, 105 S.Ct. at 1296. Furthermore, while it is true that Medeiros, like the suspect in Elstad, remained in police custody from the time he made his first incriminating statement to the time he made his second incriminating statement, custody alone is not sufficient to demonstrate involuntariness. United States v. Watson, 423 U.S. 411, 424, 96 S. Ct. 820, 828, 46 L. Ed. 2d 598 (1976).

Finally, the rationale underlying the Fifth Amendment privilege against self-incrimination supports our conclusion that Medeiros made his second statement voluntarily. "The fundamental import of the [Fifth Amendment] privilege while an individual is in custody is not whether he is allowed to talk to the police without the benefit of warnings and counsel, but whether he can be interrogated." Miranda, 384 U.S. at 478, 86 S.Ct. at 1630 (emphasis added). See also Connelly, 479 U.S. at 164, 107 S.Ct. at 520 ("Absent police conduct causally related to the confession, there is simply no basis for concluding that any state actor has deprived a criminal defendant of due process of law." (footnote omitted)). Medeiros was not interrogated at the Pawaa Annex and the surrounding circumstances, as well as the entire course of police conduct indicate that the conditions which made the first statement inadmissible were no longer present. Thus, because Medeiros chose to speak voluntarily, we conclude that his second statement is admissible.

There may be some question whether Elstad can be applied retroactively to this appeal. The Intermediate Court of Appeals of Hawaii affirmed Medeiros' conviction in 1983, two years before the Supreme Court decided Elstad. Thus, this appeal presents the question of whether Elstad can be applied retroactively to a case seeking federal habeas review of a state conviction that has become final.6

Although the Constitution neither prohibits nor compels retroactive application of new constitutional decisions, Linkletter v. Walker, 381 U.S. 618, 629, 85 S. Ct. 1731, 1737, 14 L. Ed. 2d 601 (1965), the general rule is to apply judicial decisions retroactively. Solem v. Stumes, 465 U.S. 638, 642, 104 S. Ct. 1338, 1341, 79 L. Ed. 2d 579 (1984). The Supreme Court considered whether a new constitutional decision, Batson v. Kentucky, 476 U.S. 79, 106 S. Ct. 1712, 90 L. Ed. 2d 69 (1986), was "applicable to litigation pending on direct state or federal review or not yet final when Batson was decided." Griffith v. Kentucky, 479 U.S. 314, 316, 107 S. Ct. 708, 710, 93 L. Ed. 2d 649 (1987). The Court held that "a new rule for the conduct of criminal prosecutions is to be applied retroactively to all cases, state or federal, pending on direct review or not yet final, with no exception for cases in which the new rule constitutes a 'clear break' with the past." Griffith, 479 U.S. at 328, 107 S.Ct. at 716. The retroactivity of new constitutional decisions in federal habeas corpus cases was not an issue before the Griffith court and presented a question "carefully left open until it is squarely presented." Griffith, 479 U.S. at 329, 107 S.Ct. at 716 (Powell, J., concurring). See also Yates v. Aiken, 484 U.S. 211, 215, 108 S. Ct. 534, 537, 98 L. Ed. 2d 546 (1988) (Griffith 's retroactive analysis applies only to cases pending on direct appeal). This case squarely presents this question.7

The three criteria delineated in Stovall v. Denno, 388 U.S. 293, 297, 87 S. Ct. 1967, 1970, 18 L. Ed. 2d 1199 (1967), guide our analysis of the retroactivity question. Allen v. Hardy, 478 U.S. 255, 258, 106 S. Ct. 2878, 2880, 92 L. Ed. 2d 199 (1986); Solem, 465 U.S. at 642-43, 104 S.Ct. at 1341-42. We must examine "(a) the purpose to be served by the new standards, (b) the extent of the reliance by law enforcement authorities on the old standards, and (c) the effect on the administration of justice of a retroactive application of the new standards." Stovall, 388 U.S. at 297, 87 S.Ct. at 1970.

The purpose to be served by the Elstad decision is its interpretation and application of Miranda and Bayer to a particular set of facts. Elstad 's analysis illuminates, by way of example, the cat out of the bag doctrine as it applies to a confession which follows a voluntary, un-Mirandized statement. However, "[c]omplete retroactive effect is most appropriate where a new constitutional principle is designed to enhance the accuracy of criminal trials." Solem, 465 U.S. at 644, 104 S.Ct. at 1342. Elstad 's interpretation of Bayer contributes very little, if anything, to enhancing the accuracy of criminal trials. Rather, Elstad 's analysis acts as a prophylactic rule which protects a defendant's right not to incriminate himself. Accordingly, this factor does not support retroactive application of Elstad.

Conversely, both the second factor, the reliance by law enforcement authorities on old standards, and the third factor, the disruptive effect on the administration of justice, support retroactive application of Elstad. Law enforcement authorities could not "have justifiably relied on a prior rule of law said to be different from" Elstad or which represented a " 'clear break with the past.' " Solem, 465 U.S. at 645-46, 104 S.Ct. at 1343 (quoting Desist v. United States, 394 U.S. 244, 248, 89 S. Ct. 1030, 1033, 22 L. Ed. 2d 248 (1969)). "[T]he rule announced in Elstad is not a clear break with the past, prior precedents, which would preclude retroactive application, but is an application of Miranda principles to a particular situation." Wauneka, 770 F.2d at 1441.8 See also In Re Grand Jury Proceedings, Ayres, 653 F. Supp. 465 (D.R.I.1987) (Elstad did not change the standards for assessing the voluntariness of statements made subsequent to involuntary, incriminating statements). Additionally, because Elstad merely applies preexisting law, there is no reason to anticipate that retroactively applying Elstad to this case will disrupt the administration of justice. These two factors weigh in favor of retroactively applying Elstad to this case. Accordingly, our reliance on Elstad is proper.

We affirm the district court's denial of Medeiros' petition for a writ of habeas corpus. Although Medeiros' second statement followed a previous voluntary but unwarned admission, the second statement was made voluntarily and, therefore, is admissible into evidence.

AFFIRMED.

As I read Oregon v. Elstad, 470 U.S. 298, 105 S. Ct. 1285, 84 L. Ed. 2d 222 (1985), it reaffirms the rebuttable presumption that an initial confession, even if voluntary, has a coercive effect tending to undermine the voluntariness of subsequent confessions as articulated by the Supreme Court in United States v. Bayer, 331 U.S. 532, 67 S. Ct. 1394, 91 L. Ed. 1654 (1947). Because the presumption of coerciveness arising from Medeiros' initial confession was not rebutted by any significant intervening events, I respectfully dissent.

The voluntariness inquiry in this case requires us to revisit the "cat out of the bag" doctrine as applied to successive confessions in light of Elstad. Elstad first incriminated himself in response to police interrogation at his parents' home. After being read his Miranda rights an hour later at the police station, he gave and signed a full confession. As in this case, the state trial court suppressed Elstad's first statement on the ground it was obtained in violation of Miranda, but rejected Elstad's argument that the first statement, in letting the cat out of the bag, undermined the voluntariness of his later confession because it created the impression that his fate was already sealed. On appeal, the Oregon Court of Appeals held that the second confession must also be suppressed because its voluntariness was "tainted" by the first confession. Citing Bayer, the Court of Appeals reasoned that the first confession "let the cat out of the bag" and that there was not a "sufficient break in the stream of events between [the first] statement and the written confession to insulate the latter statement from the effect of what went before." Oregon v. Elstad, 61 Or.App. 673, 658 P.2d 552, 554 (1983).

The Supreme Court granted certiorari to "decide whether an initial failure of law enforcement officers to administer [Miranda warnings], without more, 'taints' subsequent admissions made after a suspect has been fully advised of and has waived his Miranda rights." Elstad, 470 U.S. at 300, 105 S.Ct. at 1288. The Court held that when, as in the instant case, an initial confession was not actually coerced in violation of due process even though obtained in violation of Miranda, the traditional "fruit of the poisonous tree" doctrine of Wong Sun v. United States, 371 U.S. 471, 83 S. Ct. 407, 9 L. Ed. 2d 441 (1963) and its progeny is inapplicable. Elstad, 470 U.S. at 304-09, 105 S.Ct. at 1290-93. Rather, "the admissibility of any subsequent statement should turn in these circumstances solely on whether it is knowingly and voluntarily made." Id. at 309, 105 S. Ct. at 1293. See generally United States v. Wauneka, 770 F.2d 1434, 1439-40 (9th Cir.1985).

The Supreme Court then considered the Oregon Court of Appeals' decision that Elstad's second confession was rendered involuntary by "a subtle form of lingering compulsion, the psychological impact of the suspect's conviction that he has let the cat out of the bag and, in so doing, has sealed his own fate." Elstad, 470 U.S. at 311, 105 S.Ct. at 1294. The Supreme Court did not question the Oregon court's point that an initial confession has a psychologically coercive impact on later confessions, nor did the Court repudiate Bayer and its progeny establishing the "cat out of the bag" doctrine.1 Indeed, the Court endorsed Bayer 's original explanation of this phenomenon:

[o]f course, after an accused has once let the cat out of the bag by confessing, no matter what the inducement, he is never thereafter free of the psychological and practical disadvantages of having confessed. He can never get the cat back in the bag. The secret is out for good. In such a sense, a later confession always may be looked upon as a fruit of the first.

Id. (quoting Bayer, 331 U.S. at 540-41, 67 S. Ct. at 1398).

I acknowledge that some ambiguous comments in Elstad, read in isolation, appear to cast some doubt on the continuing vitality of the "cat out of the bag" doctrine. The Court said that "[i]t is difficult to tell with certainty what motivates a suspect to speak," id. 470 U.S. at 314, 105 S. Ct. at 1296, and that "absent deliberatively coercive or improper tactics in obtaining the initial statement, the mere fact that a suspect has made an unwarned admission does not warrant a presumption of compulsion." Id. However, a close reading of the Court's opinion indicates that these statements do not purport to challenge the continuing vitality of of Bayer 's presumption that an initial confession has a coercive impact that tends to undermine the voluntariness of later confessions. Rather, the Court's statements merely make the point that Bayer 's presumption of psychological coercion is rebuttable.

The Court in Elstad articulated this understanding by repeating the qualification in Bayer that "[t]his Court has never gone so far as to hold that making a confession under circumstances which preclude its use, perpetually disables a confessor from making a usable one after those conditions have been removed." Id. (quoting Bayer, 331 U.S. at 541, 67 S.Ct. at 1398 (emphasis added)); see also Elstad, 470 U.S. at 311-12, 105 S.Ct. at 1294 ("the Court has assumed that the coercive effect of the [initial] confession could, with time, be dissipated"). Far from rejecting the "cat out of the bag" doctrine and holding that "the fact that the first statement was made is completely irrelevant" to the voluntariness inquiry, as did the district court below, Order at 7, Elstad simply clarified that the Bayer presumption is a rebuttable one. The Court in Elstad therefore reaffirmed its previous rule that once officials induce a suspect to let the cat out of the bag, the voluntariness of a subsequent confession is presumed to have been undermined by the first unless some significant intervening event sufficiently attenuates the psychologically coercive impact of the initial admission of guilt.

In Elstad, the presumption that the coercive effect of Elstad's first confession undermined the voluntariness of his second confession was rebutted by the administration and waiver of Miranda warnings prior to the second. The Supreme Court explained that the warnings created an environment in which "the suspect [was] free to exercise his own volition in deciding whether or not to make a statement to the authorities." Elstad, 470 U.S. at 309, 105 S.Ct. at 1293. It appears to me that the Court believed the warnings dispelled the psychologically coercive effect of Elstad's first confession because he could reasonably infer from the warnings that his cat was not irretrievably out of the bag after all. While the traditional Miranda warnings do not expressly inform a suspect that a prior confession obtained in violation of Miranda cannot be used against him,2 they implicitly suggest this result. See, e.g., United States v. Schmidt, 573 F.2d 1057, 1064 (9th Cir.), cert. denied, 439 U.S. 881, 99 S. Ct. 221, 58 L. Ed. 2d 194 (1978) ("The Miranda warnings [the suspect] received also carry the implicit statement of this fact [that a prior illegal confession would be inadmissible]."). Because the warnings implicitly suggest that the suspect's fate has not yet been sealed, "[a] subsequent administration of Miranda warnings to a suspect who has given a voluntary but unwarned statement ordinarily should suffice to remove the conditions that precluded admission of the earlier statement." Elstad, 470 U.S. at 314, 105 S. Ct. at 1296.3 Coupled with the fact that "the exchange of words [Elstad] had with his father" between the first and second confessions might have prompted Elstad to confess at the police station even had he not already let the cat out of the bag at his parents' home, id., the Court concluded that "[certainly], in [Elstad's] case, the causal connection between any psychological disadvantage created by his [initial] admission and his ultimate decision to cooperate is speculative and attenuated at best." Id. at 313-14, 105 S.Ct. at 1295-96.

Elstad 's focus on the administration of Miranda warnings and Elstad's conversation with his father as intervening events sufficient to rebut the presumption that Elstad's first confession had a lingering coercive effect reinforces my conclusion that Elstad did not repudiate Bayer's "cat out of the bag" doctrine. Rather, the Court in Elstad simply found under the circumstances that the presumption of psychological coercion was effectively rebutted: "We hold today that a suspect who has once responded to unwarned yet uncoercive questioning is not thereby disabled from waiving his rights and confessing after he has been given the requisite Miranda warnings." 470 U.S. at 318, 105 S. Ct. at 1298 (emphasis added). Indeed, rather than undermine Bayer, Elstad follows directly from it. In Bayer, the first confession was suppressed, not because it was actually coerced in violation of due process, but because it was obtained in violation of the rule of McNabb v. United States, 318 U.S. 332, 63 S. Ct. 608, 87 L. Ed. 819 (1943), that confessions obtained in violation of congressionally-prescribed interrogation procedures must be suppressed pursuant to the federal courts' supervisory powers. Bayer, 331 U.S. at 540-41, 67 S.Ct. at 1398. Because the first confessions in both Bayer and Elstad were suppressed due to noncompliance with nonconstitutional prophylactic rules (congressionally-mandated procedures in Bayer and judicially-fashioned Miranda rules in Elstad ), both cases establish that the voluntariness of a suspect's second confession is presumptively undermined by a prior one, even when the first one was entirely voluntary and not extracted through coercive interrogation techniques. Moreover, in both Bayer and Elstad the Supreme Court concluded that the psychologically coercive effect of the first confession had dissipated before the second one primarily because the suspect had been warned before the second one that any incriminating statements might be used against him.4 The only salient difference is that in Bayer the second confession was given six months, as opposed to one hour, after the first. Far from rejecting Bayer's analytical framework, therefore, Elstad simply applied the same analysis to a similar set of facts and reached the same conclusion--intervening events including administration of Miranda (or functionally equivalent) warnings served to dissipate the cat's coercive effect.

It is clear then, that Elstad reaffirms Bayer's rebuttable presumption that an initial, voluntary confession has a coercive effect tending to undermine the voluntariness of subsequent confessions. Therefore, the next step in our inquiry should be to determine whether the presumption has been rebutted in this case by the presence of any significant intervening events.5

Medeiros' second confession followed his first by just one-half hour. He was never outside the immediate company of police officers from his roadside arrest to the Pawaa Annex; he was heavily intoxicated during this entire period; and his only conversations were with his various police escorts. Because the State of Hawaii has failed to identify any significant intervening event during the one-half hour interval that might have dissipated the coercive psychological impact of Medeiros' first confession, the presumption that Medeiros' first confession undermined the voluntariness of his second one has not been rebutted.

Medeiros was not informed, through proper administration of Miranda warnings or otherwise, that his first confession might not be usable against him. Officer Silva commented during the middle of Medeiros' second confession that he "shouldn't give any statements at this time." R.T. at 236. However, in contrast to the Miranda warnings in Elstad and the equivalent in Bayer, Officer Silva's cautionary statement cannot fairly be construed as suggesting to Medeiros that his first confession might not be usable against him.6 See supra at 828-830. Officer Silva's comment therefore could not serve to diminish the psychologically coercive effect of the first confession in the same manner as did the warnings in Bayer and Elstad.

The majority reasons that, because (1) The Pawaa annex is not a coercive location, (2) Medeiros was not interrogated at the Pawaa annex, (3) Officer Silva told Medeiros he should not give a statement there, and (4) the officers did not "exploit" the first confession to induce the second,7 Medeiros' second confession should be characterized as voluntary. I cannot agree.

Of course Medeiros' argument that his second confession was involuntary would be strengthened if he could point to additional sources of coercion, but I fail to see how the absence of additional coercive action by the officers could have the effect of dissipating the coercive impact of his prior confession.8

The change in the particular identity of the officers and the location of the confessions likewise did nothing to dissipate the coercive effect of Medeiros' first confession. While a change of officers and location is relevant to determining the ongoing coercive effect of any interrogation techniques previously used,9 the majority fails to explain how this change operated to rebut Bayer's presumption that Medeiros' answers to the initial interrogation undermined the voluntariness of his second confession. Under both Bayer and Elstad, even a voluntary initial confession tends to undermine the voluntariness of later confessions, see supra at 830, and I fail to see how a change in identity of officers and location operates to dispel a suspect's perception that his fate has already been sealed. In Elstad, the Supreme Court did not find it significant that Elstad's second confession was given at a different location from the first and with an additional officer present. The Court mentioned only the administration of Miranda warnings and Elstad's conversation with a family member, both not present here, as events which dissipated the coercive effect of Elstad's first confession.

The essence of the majority's argument is that absent coercive police conduct at the Pawaa annex or coercion carrying over from Officer Trela's unwarned questioning, the psychologically coercive effect of Medeiros' first confession is simply not enough by itself to undermine the voluntariness of his second confession. This argument renders the "cat out of the bag" doctrine a hollow shell and is contrary to Bayer and Elstad. Bayer and Elstad teach that the coercive effect of the initial confession is sufficient to require suppression of a later confession absent some tangible intervening event that dissipates that coercive effect. In holding that Medeiros' second confession was voluntary, the majority turns a deaf ear to this teaching.

I conclude that in the circumstances of this case--the second confession followed the first by a mere half hour, Medeiros was continually in the immediate presence of police officers, the second confession was preceded by neither Miranda warnings nor consultation with family or counsel, and there was no other intervening event--Medeiros' second confession was not voluntary. Accordingly, I would hold that the district court erred in denying Medeiros's petition for writ of habeas corpus.