We consider a single, difficult issue in this case: may the prosecution, to rebut a defendant's claim of insanity, present the opinion testimony of a police officer that a defendant was sane at the time of the offense, when that opinion is based on observations of the defendant during an interrogation in violation of Miranda because the defendant did not intelligently waive his rights?

Terry Sonier and Kathy Collis were driving in the northwest section of Miami on the night of March 18, 1977 when their car developed engine trouble. They stopped at a closed service station to check the car when Oliver Jones, the petitioner, approached them displaying a firearm. Jones demanded money, which the women gave him. He ordered them to get back into the car and to drive to a nearby church. After directing the women to get out of the car and to lie on the ground, he committed sexual battery against them.

Jones first told the women that he would have to kill them because they could identify him; he then ordered them to dress and drive to a nearby shopping center. The car broke down again, and as the three walked away from it, they were approached by police officer Robert Barnett. When Collis told Barnett that Jones had a gun, Jones dropped the gun and fled. He was apprehended a few blocks away by another officer, taken into custody, and identified by both women.

At about 3:30 a.m., Detective Raymond Holsberry began to question Jones. Holsberry asked Jones what grade he had finished in school and whether he could read and write; Jones replied that he had completed the ninth grade and that he could read and write. Jones denied involvement in the crime until, over two hours later, he started to cry and confessed. Jones refused to repeat his statements for stenography, and Holsberry did not videotape the interrogation.

A month after his arrest, a state trial court adjudicated Jones incompetent to stand trial and committed him for treatment. He was adjudicated incompetent again in November 1977. A trial court eventually found him competent to stand trial, and the trial took place in December 1981.

Immediately before the beginning of trial, the defense moved to suppress the statements made by Jones to Detective Holsberry on the ground that Jones had not knowingly and intelligently waived his Miranda rights.1 The court heard the testimony of three psychiatrists as well as Detective Holsberry before ordering the suppression of Jones' post-arrest statements. The court also refused to allow the prosecution to use Jones' statements in the course of questioning its experts to elicit the experts' opinions of Jones' sanity.

Sanity was the central issue at trial. The state sought to introduce testimony by Detective Holsberry that Jones appeared to be rational and well oriented at the time of his questioning, and that Holsberry believed Jones to have understood the difference between right and wrong that evening. The trial court overruled defense objections, holding that Holsberry could testify as to his observations of the defendant on the night of the questioning but could not "go into the text of the statements."

[Q.] How much time did you spend with Mr. Jones[?]

A. Approximately three hours.

Q. Did you have some conversation with him?

A. Yes, I did.

Q. Were his answers responsive to your questions?

A. Yes, they were.

Q. Did he appear to be well oriented as to where he was?

A. Yes.

Q. Did you have any difficulty understanding his responses?

A. No, I did not.

Q. Based upon what he was saying, did you get the feeling that he understood what you were saying?

A. Yes.

....

Q. Was he aware of where he was?

A. Yes.

Q. Based upon your observations of him, and based upon your conversations with him, were you able to form an opinion as to whether or not he was aware of the consequences of being arrested?

A. Yes, he was. At the time he was crying.

Q. Were you able to form an opinion as to whether or not he knew right from wrong?

....

[A.] Yes. I felt he knew right from wrong.

Jones was convicted on eight counts and sentenced to six consecutive 100-year terms and one consecutive 15-year term. The Florida District Court of Appeal affirmed the judgment of conviction. The federal district court denied Jones' petition for habeas corpus, and this appeal followed.

This case lies at the intersection of two lines of authority. In Estelle v. Smith, 451 U.S. 454, 101 S. Ct. 1866, 68 L. Ed. 2d 359 (1981), the Supreme Court held inadmissible the expert opinion of a psychiatrist as to the defendant's future dangerousness because that opinion was based on a compelled psychiatric examination not requested by the state to rebut a defense of insanity2 and at which no Miranda warnings had been administered to the defendant. The Supreme Court rejected the state's contention that the fifth amendment privilege was inapplicable to exclude from evidence Smith's admissions to the state psychiatrist because the communications were nontestimonial in nature. The Court noted that the psychiatrist's diagnosis, "as detailed in his testimony, was not based simply on his observation of [Smith]. Rather, Dr. Grigson drew his conclusions largely from [Smith's] account of the crime during their interview, and he placed particular emphasis on what he considered to be [Smith's] lack of remorse. Dr. Grigson's prognosis as to future dangerousness rested on statements [Smith] made, and remarks he omitted, in reciting the details of his crime.3 The Fifth Amendment privilege, therefore, is directly involved here because the State used as evidence against [Smith] the substance of his disclosures during the pretrial psychiatric examination." Estelle v. Smith, 451 U.S. at 464-65, 101 S.Ct. at 1873-74 (emphasis added) (citation omitted).

In Cape v. Francis, 741 F.2d 1287 (11th Cir.1984), cert. denied, 474 U.S. 911, 106 S. Ct. 281, 88 L. Ed. 2d 245 (1985), this court held inadmissible similar psychiatric opinion testimony introduced at the guilt/innocence stage of a capital trial based on the results of a compelled psychiatric examination, at which no Miranda warnings had been administered and in advance of which Cape's defense attorney had never been notified that the examining psychiatrist would testify as to matters other than competency to stand trial.4 It was not clear to the court whether Cape had actually admitted committing the murder, or whether "the murder itself was even discussed during the interview," but it was clear that the psychiatrist had based his diagnosis of sanity on the substance of his conversations with Cape, including " 'the details of how he came there' " and " 'the details of the difficulty he was in.' " Cape, 741 F.2d at 1293-94. As we explained, Cape could well have told the doctor the same self-exculpatory story that he told the jury, but there was no doubt that the contents of Cape's discussions with the psychiatrist, as opposed to his mere demeanor, influenced the psychiatrist's conclusions. We did not consider it significant that the psychiatrist did not relate any specific statements made by Cape to the jury, for "[i]t is enough that the psychiatrist based his expert opinion on the contents of responses by a defendant who had not been given proper warnings and had not had an adequate opportunity to consult with counsel." Cape, 741 F.2d at 1294 & n. 7.

Another line of cases, however, holds that there is a range of evidence relating to the appearance and demeanor of a defendant and to the physical properties of a defendant's body that does not implicate the fifth amendment. Broadly speaking, "the distinction which has emerged, often expressed in different ways, is that the [fifth amendment] privilege is a bar against compelling 'communications' or 'testimony,' but that compulsion which makes a suspect or accused the source of 'real or physical evidence' does not violate it." Schmerber v. California, 384 U.S. 757, 764, 86 S. Ct. 1826, 1832, 16 L. Ed. 2d 908 (1966). Thus the Supreme Court has held that the government can compel a defendant to give a voice exemplar for identification purposes without running afoul of the fifth amendment. United States v. Dionisio, 410 U.S. 1, 5-7, 93 S. Ct. 764, 767-68, 35 L. Ed. 2d 67 (1973). The Sixth Circuit has ruled that a prosecutor's comments about a defendant's demeanor at the counsel table during trial did not violate the fifth amendment, because the comments referred not to the defendant's failure to testify but rather to the conduct and demeanor of the accused. See Cunningham v. Perini, 655 F.2d 98, 100 (6th Cir.1981) (per curiam), cert. denied, 455 U.S. 924, 102 S. Ct. 1286, 71 L. Ed. 2d 467 (1982).5

The appellant argues, and we agree, that the resolution of this case depends on whether Detective Holsberry's testimony was based on "testimonial" and "communicative" aspects of Jones' statements.6 The question is concededly a very close one; as the Supreme Court has observed, "the distinction between real and physical evidence, on the one hand, and communications or testimony, on the other, is not readily drawn in many cases." South Dakota v. Neville, 459 U.S. 553, 561, 103 S. Ct. 916, 921, 74 L. Ed. 2d 748 (1983); see Schmerber, 384 U.S. at 764, 86 S.Ct. at 1832. Nonetheless, we conclude that Holsberry's testimony falls on the permissible side of the line.

First, Holsberry never testified as to any specific statements made by Jones; the trial court excluded all such statements. Second, Holsberry never stated that Jones had said that he had committed the crimes; in fact, Holsberry mentioned neither the details of the crime nor the details of any exculpatory explanation that Jones might have had for his actions that evening. Third, unlike the testimony of the examining physician in Estelle v. Smith that he based his conclusions on the details of the story that Smith told him, Holsberry gave no indication that his opinion of Jones' sanity was grounded in the details of Jones' statement.7

Holsberry did testify that he had a conversation with Jones, that Jones appeared to be well oriented, that Jones' answers to his questions were responsive, and that Jones understood what Holsberry was saying. None of these observations related to the substance of Jones' story; they were more like demeanor evidence, observations that Jones was alert and comprehending. Our conclusion that this testimony was admissible also comports with the Supreme Court's retrospective look at Estelle v. Smith in South Dakota v. Neville, where the Court recalled that, in Smith, it had "specifically rejected the claim that the psychiatrist was observing the patient's communications simply to infer facts of his mind, rather than to examine the truth of the patient's statements." Neville, 103 S. Ct. at 921 n. 12 (emphasis added).8

We do pause on Holsberry's affirmative response to the prosecutor's question, "Were you able to form an opinion as to whether or not he knew right from wrong?" If the prosecutor was pursuing a line of questioning based on the substance and the details of Jones' confession, Holsberry's testimony could well be indistinguishable from the testimony given by the doctors in Smith and Cape; it would have been based on Holsberry's analysis of Jones' story. In this trial, however, the question followed the prosecutor's questions about the responsiveness of Jones' answers, Jones' ability to make himself understood, Jones' ability to comprehend Holsberry, and Jones' awareness of the consequences of being arrested--questions that did not implicate the testimonial aspects of Jones' statements.

None of these questions touched on the substance of Jones' story, and the prosecutor's final question--if it referred to the substance of Jones' story--would not have followed from them. To hold Holsberry's answer to this final question inadmissible, we would have to conclude that, without any foundation, the prosecutor--and the jury--suddenly leaped from concentrating on Jones' behavior to theorizing about the substance of Jones' statement. Such a conclusion is too speculative. Even when he asked Holsberry about Jones' ability to tell right from wrong, the prosecutor did not prompt Holsberry with any details of the story, nor did he ask Holsberry if there were any particular aspects of Jones' story that supported his conclusion about Jones' sanity. Although the question may have been objectionable as a departure from the previous questioning concerning Jones' demeanor, we conclude that it did not implicate the content of Jones' statement so as to intrude on the fifth amendment.

We recognize that this conclusion places us in possible conflict with the D.C. Circuit's opinion in United States v. Hinckley, 672 F.2d 115, 132-33 (D.C.Cir.1982) (per curiam), affirming the suppression of the opinion testimony of FBI agents that John Hinckley was sane as based on statements taken in violation of Miranda.9 Distinguishing the compelled psychiatric examination of a defendant for purposes of proving sanity held permissible in the D.C. Circuit, the court wrote, "[w]e see, however, a vast difference between compelling a psychiatric examination as the only effective means to counter a defendant's assertion of insanity and allowing the use of lay testimony obtained in violation of Miranda generally to rebut a defendant's insanity defense." 672 F.2d at 133 n. 116. There is, of course, such a difference; nevertheless we are convinced that the more important difference is between testimonial communications, which are protected by the fifth amendment, and nontestimonial aspects of a defendant's behavior, which are not. Because we believe that Holsberry's testimony was based on the nontestimonial aspects of Jones' behavior, we conclude that the Constitution does not require that his testimony be suppressed.

Accordingly, the order of the district court denying the writ of habeas corpus is AFFIRMED.

I dissent. The majority's approach in this case rests upon an analogy it seeks to draw between the testimony of Detective Holsberry and certain types of real or physical evidence which do not call forth the protections of the Fifth Amendment. Estelle v. Smith, 451 U.S. 454, 101 S. Ct. 1866, 68 L. Ed. 2d 359 (1981), and related decisions, make clear that the Supreme Court has explicitly sought to prohibit the construction of such an analogy to resolve the issue presented here. A careful examination of the relevant decisions demonstrates that the majority is misguided in its attempt to parse the testimony of Detective Holsberry so as to render him a source of real evidence.

In Smith, the Supreme Court held that the Fifth Amendment was implicated where the prosecutor in a capital case sought to introduce the psychiatric testimony as to the defendant's future dangerousness. The State had argued that the Fifth Amendment was irrelevant because the defendant's communications to the examining psychiatrist were "nontestimonial in nature." 451 U.S. at 463, 101 S.Ct. at 1873. The State sought support from those decisions which hold that there is no Fifth Amendment violation when evidence acquired from a defendant is "neither related to some communicative act nor used for the testimonial content of what was said." Id. (emphasis added) (citing United States v. Dionisio, 410 U.S. 1, 93 S. Ct. 764, 35 L. Ed. 2d 67 (1973); Gilbert v. California, 388 U.S. 263, 87 S. Ct. 1951, 18 L. Ed. 2d 1178 (1967); United States v. Wade, 388 U.S. 218, 87 S. Ct. 1926, 18 L. Ed. 2d 1149 (1967); Schmerber v. California, 384 U.S. 757, 86 S. Ct. 1826, 16 L. Ed. 2d 908 (1966)). The Court rejected this contention because it was clear that the examining psychiatrist's testimony could not be "based simply on his observation of respondent." 451 U.S. at 463, 101 S. Ct. at 1873.

The D.C. Circuit, in an in banc opinion written by then Circuit Judge Scalia, has noted that Smith makes clear that it is improper to equate psychiatric examinations with the types of real or physical evidence with which Schmerber and related cases are concerned. The court explained that in Smith, "the State urged, as the Ninth Circuit had held in [United States v. Handy, 454 F.2d 885 (9th Cir.1971) ] that Smith's communication to the court-appointed psychiatrist were nontestimonial in character and specifically sought support by way of analogy to the Court's decisions in [Dionisio, Gilbert, Wade, and Schmerber ].... The Court dismissed the argument out of hand." United States v. Byers, 740 F.2d 1104, 1112 (D.C.Cir.1984); see also Gholson v. Estelle, 675 F.2d 734, 739 (rejecting argument that testimony resulting from psychiatric examination was "based upon physical, nontestimonial communications").

Nevertheless, the majority seeks to breathe new life into this argument through its suggestion that Schmerber and Dionisio "hold[ ] that there is a range of evidence relating to the appearance and demeanor of a defendant and to the physical properties of a defendant's body that does not implicate the fifth amendment." Maj. op. at 1443. It is true that the Supreme Court has found that a defendant may be required to submit to a blood test, see Schmerber, participate in a lineup, see Wade, provide handwriting exemplars, see Gilbert, and speak for the purposes of voice identification, see Dionisio. Yet these decisions have nothing to do with the "demeanor" of a defendant about which a witness testifies for the purpose of establishing a foundation for his or her belief that a defendant is sane. Where such "demeanor" testimony is based on an interview which constitutes unlawful interrogation, Smith requires that this testimony be suppressed.

Such was this court's understanding of Smith in Cape v. Francis, 741 F.2d 1287 (11th Cir.1984), cert. denied, 474 U.S. 911, 106 S. Ct. 281, 88 L. Ed. 2d 245 (1985), where a psychiatrist's testimony as to the defendant's sanity and criminal responsibility "went beyond the permissible bounds" insofar as it was based "on the contents of responses by a defendant who had not been given proper warnings...." Id. at 1294, n. 7. We thought it "of no import that [the psychiatrist] did not testify directly as to any specific statements to him by the [defendant]." Id. at 1294 n. 7. Moreover, we found it irrelevant that it was unclear from this testimony whether the defendant had discussed the crime at all. Id. at 1293. The psychiatrist's testimony in Cape, as noted by the majority, was limited to mentioning "that he had talked with [the defendant] about 'the details of how [the defendant] came there' and 'the details of the difficulty he was in.' " Id. at 1293-94. We found that these limited disclosures, coupled with the psychiatrist's ultimate opinion as to the defendant's criminal responsibility, compelled a conclusion that this opinion was "based on the substance of disclosures made during a custodial interrogation." 741 F.2d at 1294.

The majority attempts to distinguish Cape by explaining that "there was no doubt that the contents of Cape's discussions with the psychiatrist, as opposed to his mere demeanor, influenced the psychiatrists conclusions." Maj. op. at 1443 (emphasis added). Given the questions which the prosecutor posed to Detective Holsberry, it is simply inconceivable that he did not base his testimony on the petitioner's "communicative acts." The prosecutor asked Detective Holsberry to draw his responses specifically from the "conversations" between the two men and from "what [Jones] was saying."

The majority's misapprehension of Smith and its progeny stems from the view that testimony cannot be based on a "testimonial" communication unless it relates the substance of what one has actually said. This is consistent with the majority's erroneous assertion that the Schmerber line of cases permits the use of certain "demeanor" testimony which results from an unlawful interview. Yet, as noted above, Smith explained that the physical evidence concerned in these cases was admissible because it was "neither related to some communicative act nor used for the testimonial content of what is said." 451 U.S. at 463, 101 S. Ct. at 1873. The first portion of this disjunctive statement expresses the same thought contained in the rule that testimony must not be "based on" communicative acts or statements made during the course of unlawful interrogation.1

The case in which this court first applied the principles set forth in Smith exposes the nature of the majority's error. In Battie v. Estelle, 655 F.2d 692 (5th Cir.1981), the prosecution called a clinical psychologist to testify as to the defendant's future dangerousness. The witness had examined the defendant to determine whether he was competent to stand trial. We described this examination, which had been conducted without the benefit of Miranda warnings, as follows:

At the time he conducted the psychological testing of petitioner, Dr. Patterson was unaware of petitioner's background and the facts surrounding the crime charged against petitioner. The tests Dr. Patterson administered did not elicit any facts regarding the crime, and Dr. Patterson did not interrogate Battie about the murders. Instead, Dr. Patterson based his diagnostic evaluation of petitioner entirely upon the results of the tests given Battie. From his testing of petitioner, Dr. Patterson concluded that petitioner was suffering from a sociopathic personality disorder....

Id. at 695 (emphasis added). On the basis of these test results, the witness testified that Battie was likely to commit future criminal acts of violence. Id. Noting that Dr. Patterson "based his analysis on petitioner's answers to the test questions ... and upon 'the pertinent information that [Battie] produce[d] in relation to the test protocol or test battery,' " Id. at 700 n. 18, we rejected the State's argument that "Dr. Patterson's tests constituted real rather than testimonial evidence." Id. at 700 (emphasis added).

We held that Battie "was not simply 'required to use his voice as an identifying physical characteristic' but instead was required 'to speak his guilt' by responding to Dr. Patterson's test questions." Id. at 700 (quoting Wade, 388 U.S. at 223, 87 S.Ct. at 1930). This was so regardless of the fact that Dr. Patterson did not describe in any way Battie's responses or statements. The key to the decision in Battie was that the defendant was compelled, through communicative acts, to assist the state in proving elements necessary to support the imposition of a criminal punishment. See 655 F.2d at 700-01. The same reasoning applies here, since the State has availed itself of unlawfully obtained communications to support a guilty verdict.2 This is true regardless of whether Holsberry described--directly or indirectly--the substance of what Jones said.3

More recent cases in the Fifth Circuit have continued to follow this approach. In Jones v. McCotter, 767 F.2d 101 (5th Cir.), cert. denied, 474 U.S. 947, 106 S. Ct. 345, 88 L. Ed. 2d 292 (1985), two psychiatrists who conducted examinations of the defendant testified as to his future dangerousness during the penalty phase of the defendant's capital trial. The trial judge prohibited the psychiatrists from testifying "(1) as to the contents of any conversation that they had with [the defendant] while he was in custody and (2) any opinions by the psychiatrists based on a conversation regarding the facts of the case against [him]." Id. at 103. The court, applying Smith, found that the introduction of the psychiatrists' testimony nevertheless violated the defendant's Fifth Amendment rights. Id. at 103, n. 6 (citing Muniz v. Procunier, 760 F.2d 588 (5th Cir.) (same result), cert. denied, 474 U.S. 934, 106 S. Ct. 267, 88 L. Ed. 2d 274 (1985)); see also Gholson, 675 F.2d at 739-42 (same result); White v. Estelle, 720 F.2d 415 (5th Cir.1983) (hypothetical questions directed at examining psychiatrist interfered with defendant's Fifth Amendment rights because "they reasonably indicated that the psychiatrist's prognosis ... was influenced by and derived from" examination conducted without Miranda warnings).

The only material distinction between the facts of this case and those presented in the foregoing series of cases is that here the testimony was that of one with no experience in the mental health field. This distinction, however, strengthens the grounds upon which the court should find that Holsberry's testimony could not have been based solely on his "observations." To understand why this is so, it is important to consider what psychiatrists do when they conduct an examination.

The products of psychiatric examinations fall along the entire continuum from real to testimonial evidence. On one end of the spectrum are the clearly physical parts of the examination: x-rays; blood and urine tests; and electro-encephalograms. Standardized psychological tests, which probe patterns of association, require vocal responses and more active cooperation from the individual examined.

Byers, 740 F.2d at 1148 (dissenting opinion) (footnotes omitted). Where a psychiatric examination is composed principally of an oral examination, such an interview "falls squarely at the testimonial end of the spectrum." Id. Yet when the person conducting the examination of a defendant has no training in the interpretation of personality disorders, his views must necessarily be based in large part upon what that person is saying. This will be particularly true of one whose purpose in conducting an interview is to discover the details of a crime. Schmerber, discussing the difficulties which inhere in drawing distinctions between testimonial and nontestimonial communications, addresses this point:

There will be many cases in which such a distinction is not readily drawn. Some tests seemingly directed to obtain "physical evidence," for example, lie detector tests measuring changes in body function during interrogation, may actually be directed to eliciting responses which are essentially testimonial. To compel a person to submit to testing in which an effort will be made to determine his guilt or innocence on the basis of physiological responses, whether willed or not, is to evoke the spirit and history of the Fifth Amendment. Such situations call to mind that the protection of the privilege "is as broad as the mischief against which it seeks to guard."

Schmerber, 384 U.S. at 764, 86 S.Ct. at 1832 (emphasis added) (citation omitted). These concerns are particularly acute in this case for we know Holsberry's interview4 with Jones was part of a criminal investigation, in which it was the detective's purpose--indeed it was his duty--to elicit from the defendant information regarding his crimes.5 During the course of his three-hour and ten-minute interview with Jones, it was not until two hours and fifteen minutes had elapsed that Jones began to confess.6 Cf. Gholson, 675 F.2d at 741 (Fifth Amendment violation where psychiatrist's examination "aimed at procuring evidence by measuring physical behavioral reactions to interrogation").

The majority's portrait of Detective Holsberry is that of a detached, clinical observer.7 Yet "[t]his case does not present the kind of situation [where one has] drawn [a] conclusion from the [defendant's] manner or deportment, his attention spans or facial expressions." Gholson, 675 F.2d at 740. "A neutral and detached observation of a person's physiological characteristics is something altogether different from an examination directed and calculated to subtly extract a defendant's thoughts." Id. at 741.

The court ought to follow the lead of the D.C. Circuit's decision in United States v. Hinckley, 672 F.2d 115 (D.C.Cir.1982), and hold that lay testimony as to a criminal defendant's sanity is inadmissible when such testimony is the result of an unlawfully conducted interrogation. In Hinckley, FBI agents ignored the defendant's request for assistance of counsel and interrogated him for one-half hour before providing him with access to an attorney. Id. at 118. The government characterized its interrogation of the defendant as a " 'background' interview." But the court rejected the notion that the government ought to be able to use "demeanor" testimony which resulted from this interview because the "whole process" was tainted by a Miranda violation. Id. at 125-26.8

Ultimately, the majority finds that it is "too speculative" to conclude that the jury's attention was drawn to the substance of the conversation between Jones and Holsberry. Maj. op. at 1446. As is explained above, this approach misapprehends the task before the court. The majority should recognize the presence of constitutional error and proceed to determine whether this error was harmless. See Cape, 741 F.2d at 1295 (citing Harryman v. Estelle, 616 F.2d 870 (5th Cir.) (admission of statements obtained in violation of Miranda may be harmless error), cert. denied, 449 U.S. 860, 101 S. Ct. 161, 66 L. Ed. 2d 76 (1980)).

Were it to do so, it could not conclude, beyond a reasonable doubt, that the error in this case had no effect on the outcome of Jones's trial. See Cape, 741 F.2d at 1294-95 (harmless error analysis). As the majority notes, "[s]anity was the central issue at trial." Maj. op. at 1442. Yet the evidence on this question conflicted substantially. The defense called four expert witnesses. The first concluded that Jones was mildly retarded, that he had symptoms of brain damage, and that he suffered from paranoid schizophrenia. He could not conclude whether petitioner was sane at the time of his offenses, but thought that he was suffering from a major mental illness which impaired his judgment.9 The second of petitioner's experts testified that Jones was legally insane at the time of his offenses.10 The third thought he was not sane, but could not make a definitive determination without further testing.11 The fourth agreed that he suffered from paranoid schizophrenia, but could not conclude that Jones was not sane at the time of his offenses.12 The State's expert testified that petitioner suffered from chronic schizophrenia and was borderline retarded. Transcript at 419. He was unable to draw a conclusion as to Jones's sanity13 but in responses to hypothetical questions, he suggested that Jones "knew the nature and consequences of his actions" at the time of the offenses. Transcript at 424. The only other testimony on the question of sanity was that of the two complaining witnesses14 and that of Holsberry.

Given the unanimous view of the five experts that Jones suffered from serious mental illness, the testimony of a single lay witness may easily have made a difference in the jury's determination that Jones was legally sane. Indeed, the testimony of police officers is often the key to a prosecution's attempt to rebut the insanity defense. Such testimony based on a defendant's statements and conduct arising soon after the commission of a crime may appear particularly persuasive to a jury. "[M]edical testimony after the fact may be less persuasive ... than immediate on-the-scene observations by lay witnesses." Hinckley, 672 F.2d at 125.

Under the circumstances presented here, the trial court's error could not be harmless beyond a reasonable doubt. The court should grant a writ of habeas corpus.15