CLARK, Circuit Judge:

This appeal caps a series of protracted postconviction proceedings designed to determine whether Robert Bruce was legally competent to stand trial for the murder of his wife in 1965. Pursuant to our mandate in Bruce's last appeal, the district court conducted a nunc pro tunc competency hearing and concluded that petitioner did not suffer from any mental illness which would interfere with his ability to assist in his own defense and factually and rationally understand the proceedings against him. Since we are unable to agree with that crucial determination, we must reverse and order that the writ of habeas corpus be issued subject to the right of the State of Texas to retry petitioner within a reasonable time.

As we recognized in our prior opinion, the factual background pertinent to petitioner's constitutional challenge is far from routine. Retelling the story properly begins this analysis. On the morning of December 22, 1964, Bruce shot and killed his wife of 12 years, the mother of his three children. When the police arrived in response to Bruce's telephone call, they found him partially crying and asking over and over again if his wife was dead. While confined in the Dallas County jail subsequent to arrest, Bruce attempted to commit suicide. Bruce's father initially retained attorney Joe McNicholas to represent his son. At first McNicholas was unable to interview his client because he appeared to be in a state of shock. Upon learning that Bruce had experienced psychological difficulties while serving in the Marine Corps in the 1950's, McNicholas obtained his client's military treatment file. The records recounted two disturbing episodes which precipitated Bruce's medical discharge from the service. Each time Bruce had become extremely belligerent after drinking heavily. On one occasion he commandeered a barracks for 20 minutes before he was finally subdued and taken to the hospital by six Marines. Based on this medical history, Bruce's counsel arranged for his client to be examined by the county health officer and Dr. Holbrook, a psychiatrist. They concluded that Bruce was suffering from a severe chronic paranoid personality disorder and that he might be dangerous to himself or to others if not confined. Dr. Holbrook predicted long-term psychiatric hospitalization.

Thereafter, the Dallas County Grand Jury which had been investigating the shooting decided to no-bill the case and order Bruce held for a lunacy commitment. On March 3, 1965, the Probate Court adjudged Bruce "mentally ill" and directed that he be involuntarily committed to Terrell State Hospital for his own welfare and the protection of others. From the record it appears that this adjudication has never been set aside.

While at Terrell, Bruce was prescribed large daily dosages of an antipsychotic medication by the staff physician. At that time he related to his father that he had seen his dead wife coming down the walk and accused his parents of hiding her from him, claiming that she was probably now working on the second shift at a local factory. After being confined for a little over a month, Bruce's discharge was authorized by an imposter who was neither a psychiatrist nor a physician. The "doctor" was eventually convicted and imprisoned for perjury and practicing medicine without a license. Approximately a month later, Bruce voluntarily recommitted himself to Terrell but soon thereafter escaped. He spent the next few weeks at his parents' home, a motel and in an alcohol treatment center.

Bruce's legal difficulties commenced when the district attorney's office learned that Bruce was no longer at Terrell. His case was represented to the grand jury; this time Bruce was indicted for murder with malice aforethought. Attorney McNicholas advised that an insanity defense be urged. However, both Bruce and his father insisted that insanity not be raised, apparently because they believed such a plea would "reflect on his children." McNicholas was dismissed and another attorney, Mr. Snodgrass, was retained under instructions not to make an issue of petitioner's mental state.

The only defense offered at the murder trial was that the shooting was an accident. Bruce testified to his version of the event in a logical and understandable fashion and was able to withstand thorough-going cross-examination. When his daughter was on the stand, however, Bruce interrupted the proceedings twice with statements disputing her testimony and urging her to tell the truth. At one point the judge threatened to bind and gag petitioner if he continued to disrupt the trial with his outbursts. Neither the state nor the defense explored the subject of Bruce's mental condition. The jury did not learn of the prior commitment; in fact, Bruce denied on cross examination that he had ever been treated for a nervous condition. Rejecting the accident defense, the jury returned a guilty verdict and Bruce was sentenced to life imprisonment. His conviction was affirmed on direct appeal. Bruce v. State, 402 S.W.2d 919 (Tex.Cr.App. 1966).

Upon conviction, Bruce was assigned to the Wynne Treatment Unit of the Texas Department of Corrections. Two staff psychiatrists diagnosed his condition as "schizophrenic, chronic, undifferentiated" and again Bruce was treated with antipsychotic medication.

The anfractuous ten-year postconviction route began with an application for a writ of habeas corpus to the Texas Court of Criminal Appeals. One of the grounds urged by Bruce was "insanity" at the time of trial. The state court denied the writ without a hearing. Similarly, the federal district court refused all relief. On appeal this court vacated and remanded to allow Bruce to reapply to the state convicting court under recently enacted postconviction provisions. Bruce v. Beto, 396 F.2d 212 (5th Cir. 1968).

To resolve the several competency and sanity questions related to Bruce's conviction, the state convicting court empaneled an advisory jury and conducted an extensive evidentiary hearing in 1969. Four witnesses provided testimony supporting Bruce's claim of incompetence. Bruce's father recounted how his son had never performed well in school and had constantly complained that his teachers and classmates were mistreating him. Bruce's employment record was equally poor and unstable. He could not hold a steady job and a pattern emerged whereby he would either get mad and quit or be fired. Two physicians who had performed mental status examinations on Bruce prior to the hearing and had access to petitioner's medical history testified that they believed that Bruce was incompetent at his 1965 trial. Both experts concluded that Bruce suffered from a severe form of schizophrenia. Dr. Tauber had seen Bruce daily over a six-week period; his diagnosis of paranoid schizophrenia was confirmed by a battery of tests administered to Bruce by a consulting psychologist. Finally, attorney McNicholas related that while he was employed as counsel, Bruce had not provided any assistance in his defense; he expressed the opinion that his former client was a very sick man.

The state's case revolved around the testimony of two witnesses. Bruce's trial attorney stated that he believed that petitioner was able to reasonably confer with and assist him at trial, except for two occasions (coinciding with the outbursts) when Bruce became so deeply, emotionally involved that he was incapable of communication. The evidence most damaging to petitioner's incompetency claim was provided by Dr. Grigson, the state's sole expert witness. Dr. Grigson explained that he had detected intentional falsification in some of Bruce's responses when he conducted an hour-long mental status examination of petitioner prior to the hearing. On the basis of this faking determination and a study of petitioner's medical history, Dr. Grigson concluded that the other doctors' diagnoses of schizophrenia were wrong, and that Bruce was merely a malingerer who was certainly able to function well enough to assist his attorney in his defense.

The jury returned a verdict deciding all competency and sanity questions adversely to Bruce and the trial court entered judgment in accordance with the verdict. When Bruce returned to federal court with his second habeas petition, the district judge dismissed the action on the basis of the state court findings. However, on appeal this court again reversed, characterizing the state proceedings as fundamentally unfair. The major deficiencies cited were the failure to separate the various competency and sanity questions, prejudicial closing remarks and application of an improper standard to determine competency to stand trial. Bruce v. Estelle, 483 F.2d 1031 (5th Cir. 1973). This time we remanded to the federal district court to resolve the single question of petitioner's competency in 1965. The district court was instructed to initially determine whether a meaningful nunc pro tunc hearing could be conducted at this late date and, if so, to reach and decide the competency issue.

Two psychiatrists were appointed by the district court to examine Bruce, furnish a medical report and testify at the 1974 retrospective hearing. The experts' reports revealed irreconcilable differences with respect to the crucial threshold question of whether Bruce had ever suffered from any clinically recognized psychiatric disorder. Dr. Grigson, the same physician who testified at the 1969 hearing, was one of those appointed. He persisted in believing that Bruce had never been psychotic and was at all times legally competent. This time he diagnosed petitioner as a sociopath, a descriptive term used for otherwise mentally well persons who frequently commit antisocial aggressive acts. In marked contrast, the other appointed psychiatrist, Dr. Cannon, looking at the same history and the same subject, concurred with the physicians who had previously diagnosed Bruce as schizophrenic and went on to list 16 reasons to support her conclusion that at the time of trial Bruce's disease caused him to be incompetent.

At the evidentiary hearing both experts elaborated on their divergent opinions. About the only point of agreement was that a meaningful retrospective hearing was possible provided that pertinent legal and medical records were combined with current medical evaluation to produce a hindsight picture of Bruce's mental condition in 1965. Dr. Cannon's reasons for finding Bruce incompetent were explored in detail. She explained that none of the cited incidents alone would invariably point to paranoid schizophrenia and thence to incompetency. Rather her medical diagnosis and resultant conclusion on competency were based on the total picture. According to Dr. Cannon, several of the mentioned factors ( e. g., childhood emotional problems, poor job record, law violations, attempted suicide) could also be present in the profile of a sociopathic personality. Similarly, it would not be impossible for a sociopath to have engineered the disturbances leading to Bruce's military discharge, though this kind of behavior is more easily explainable as a dissociative reaction characteristic of a primary symptom of schizophrenia. But in concluding that Dr. Grigson was wrong in his diagnosis, Dr. Cannon emphasized that Bruce had been medically treated as a schizophrenic since he entered Terrell shortly after his arrest and pointed out that there is no medically justifiable reason to prescribe anti-psychotic drugs for a sociopathic condition. Since Dr. Cannon had also found evidence of schizophrenia during her examination of Bruce and the history of Bruce's legal and personal problems tended to corroborate (or at least was not inconsistent with) that diagnosis, Dr. Cannon concluded that Dr. Grigson's medical opinion could be discounted as unsupported. Her analysis then progressed to the more difficult determination of whether Bruce's mental disorder resulted in incompetency. In medical terms, this inquiry translated into whether petitioner's illness was in remission or in an overt state at the time of trial. Dr. Cannon concluded from the medical history that Bruce had progressively deteriorated until he became overtly and openly psychotic following his wife's death and that his condition did not stabilize until after he received heavy medication and lived in a structured, rigid environment over the next few years. On cross-examination, however, Dr. Cannon conceded that mental states ( e. g., exacerbation or remission) can on occasion change quite rapidly and that it was possible that Bruce was not incompetent every moment of his trial.

The starting point for Dr. Grigson's evaluation was that he found no evidence of the primary symptoms of schizophrenia (principally thought disorganization) when he examined Bruce in 1969 and again in 1974. He then testified that Bruce's behavior pattern was not only characteristic of a sociopathic personality, but was entirely inconsistent with a diagnosis of paranoid schizophrenia. In contrast to Dr. Cannon's statement that many of the aberrant incidents in Bruce's life could suggest either diagnosis, Dr. Grigson insisted that there was absolutely no clinical similarity between the two conditions. As a sociopath, Bruce was devoid of conscience and most of his behavior could be interpreted as merely an attempt to manipulate others. For example, Bruce admitted to faking his 1969 interview with Dr. Grigson, presumably in the hopes that the psychiatrist would pronounce him incompetent. Likewise, the trial outbursts were viewed by the witness as purposefully designed to destroy harmful testimony. To Dr. Grigson, even Bruce's professed visual hallucinations were nothing more than stories designed to "manipulate" his parents. Most importantly, Dr. Grigson dismissed the other expert opinion as incorrect, claiming that he was the only one truly qualified and experienced enough to arrive at the correct evaluation.

A detailed opinion was written by the district court which has proven most helpful in our review. While recognizing that its ultimate task was to rule on the legal issue of competency, the court reasoned that it should initially resolve the conflict between the experts, for if Bruce could properly be classified as a sociopath, there would be no medical reason for believing that his "illness" could in any way interfere with his ability to understand the proceedings and assist counsel. The court in its opinion did not undertake an independent review of the background facts as they related to petitioner's competency claim; instead, it principally dealt with the testimony of the two psychiatrists at the 1974 hearing who had utilized petitioner's medical and legal records in constructing their professional opinions.

Dr. Grigson's analysis was found the more convincing. Three circumstances were cited as particularly supportive of Dr. Grigson's approach: Bruce's early antisocial behavioral pattern, the trial transcript and Bruce's propensity to fake his symptoms. The court was persuaded by Dr. Grigson's statement that early antisocial behavior is characteristic of a sociopathic rather than a preschizophrenic type and its view was bolstered somewhat by Dr. Cannon's concession that such behavior is compatible with either diagnosis. For the court, the outbursts took on less significance when viewed in the context of an "emotion laden" portion of a difficult trial, especially since they might also be indicative of purposefulness and rationality. Finally, though admitting that true schizophrenics may also be fakers, the court considered Bruce's pretense to be consistent with a manipulative personality and corroborative of Dr. Grigson's overall evaluation. After finding that petitioner had never suffered from any potentially incapacitating ailment, the court had little difficulty in pronouncing him retrospectively competent. In the alternative, the court mentioned in a footnote that even if Dr. Cannon's diagnosis were correct, it had found no clear and convincing evidence that Bruce's mental disorder was manifest in such a degree as to cause him to fall below the Dusky standard.

I. Meaningfulness of Nunc Pro Tunc Competency Hearing

In our prior opinion we commented upon the possibility of holding a meaningful retrospective hearing despite the long interval since the trial. While leaving this determination as a threshold question for the district court, we expressed the view that "the evidence tends to support the conclusion that a meaningful hearing can now be conducted." 483 F.2d at 1043. The central factor cited was the abundance of expert testimony, particularly contemporaneous observations by the staff at Terrell. On remand, the district court addressed the meaningfulness question to both testifying experts. The doctors agreed that a reliable after-the-fact determination could be made if the records pertaining to Bruce's legal and medical history were taken into account. Since there was ample evidence that both experts had extensively utilized the record in developing their opinions, the court held that it was possible to hold an adequate hearing.

On appeal Bruce makes three arguments for overturning the district court's meaningfulness ruling, none of which persuades us to do so. First, Bruce contends that the more than nine-year gap between the trial and the competency hearing precludes any intelligent retrospective resolution of the issue. Such a per se argument is not appropriate here. Recently in United States v. Makris, 535 F.2d 699, 904 (5th Cir. 1976), we remarked that "the passage of even a considerable amount of time may not be an insurmountable obstacle if there is sufficient evidence in the record derived from knowledge contemporaneous to trial." A reliable reconstruction of petitioner's mental status in 1965 depends less on time than on the state of the record. Especially where medical information substantially contemporaneous to trial is available, the chances for an accurate assessment increase. Holloway v. United States, 343 F.2d 265 (D.C. Cir. 1964). As the voluminous record in this case suggests, time alone did not vitiate the opportunity for a meaningful hearing.

The remaining challenges concern the district court's treatment of expert opinion and were also adequately discussed and dismissed in Makris. Specifically, Bruce urges that it is inconsistent to allow the existence of contemporaneous medical evidence to contribute to an affirmative ruling on meaningfulness whenever the conclusions of those experts on the competency issue are ultimately rejected. Although not specifically mentioned by the district court, the existence of the medical record developed in connection with Bruce's pretrial confinement at Terrell undoubtedly aided the court in deciding that it was possible to hold a retrospective hearing. Nor can it be disputed that in finding Bruce competent (and nonpsychotic), the court took a position at variance with the diagnosis and treatment prescribed by those physicians who had the opportunity to observe and examine petitioner close to the trial date.

As Makris demonstrates, however, this apparent inconsistency does not necessarily vitiate the court's ruling on meaningfulness. The meaningfulness determination is a threshold legal inquiry which is not measured against the same scale as the later ruling on the merits. It is the existence of contemporaneous data (both medical and lay), not the experts' interpretation of that data, which is the critical element at this stage of the inquiry. In Makris, the contemporaneous lay evidence pointed to competency, while the contemporaneous medical evidence suggested incompetency. The district court was faced with a dilemma, for unless it "rejected" the view of either the laymen or the experts, no determination could be made at all. In affirming the lower court's ruling that a meaningful retrospective hearing was possible, we emphasized that the court was not bound by the experts' inferences, but must decide for itself if the quantity and quality of available evidence was adequate to arrive at an assessment that could be labeled as more than mere speculation. Although the contemporaneous lay evidence favoring competency in today's case (the trial transcript and the opinion of attorney Snodgrass) is not nearly as strong as that present in Makris, the district court was nevertheless justified in holding the retrospective hearing here. The court had been assured by the experts that the contemporaneous medical evidence had been used by them in arriving at their present diagnoses and retrospective opinions on competency. Both the transcript of the trial and the record of the state post-conviction hearing were available. The abundance of record evidence indicates that the nunc pro tunc substitute hearing was as reliable as if it had been held prior to trial. This is not to say that the judicial task of evaluating the contemporaneous data to determine competency was therefore an easy one. On the contrary, we believe that despite a careful review, the district court erred in pronouncing Bruce competent. Affirmance of the district court's meaningfulness ruling declares only that the record was complete enough to make an intelligent determination, not that the ruling made was correct. Similarly, Bruce contends that the conflict among the experts in their ultimate opinions as to competency is alone enough to destroy the chance of conducting a meaningful hearing. See Clark v. Beto, 359 F.2d 554 (5th Cir. 1966). We disagree. The district court had enough probative evidence before it to resolve the conflict. That it did so incorrectly is irrelevant to the primary meaningful hearing inquiry.

II. Burden of Proof-Scope of Review

Before reaching the merits of the case, we must briefly address preliminary questions concerning the proper allocation of the burden of proof and the scope of appellate review in this federal habeas proceeding. Although referring only to direct criminal appeals, much of the discussion devoted to these two threshold issues in Makris is also relevant here. In that case, we rejected the contention that the Supreme Court's recent decision in Mullaney v. Wilbur, 421 U.S. 684, 95 S.Ct. 1881, 44 L.Ed.2d 508 (1975) required the prosecution to prove an accused's competency beyond a reasonable doubt and instead allowed the government to discharge its burden by the less stringent preponderance standard. On the related question of this court's scope of review, we read the Court's latest competency ruling in Drope v. Missouri, 420 U.S. 162, 95 S.Ct. 896, 43 L.Ed.2d 103 (1975) to require that while we must take a hard look at the district court's ultimate determination, the clearly erroneous rule could still be applied in connection with the lower court's evidentiary or primary factfindings.

The standard enunciated in Makris for appellate review is equally applicable to appeals from the district court in habeas proceedings. As long as a federal court has initially engaged in independent factfinding, this court need not review de novo. The combination hard look/clearly erroneous formulation is as suitable to protect the rights of state prisoners as those charged with federal offenses. A different appellate standard for state habeas cases could compound the confusion in this already difficult area.

In contrast, the burden of proof question cannot be answered solely by reference to Makris because of the substantial difference between direct and collateral attacks. As Bruce readily acknowledges, the burden of proving a constitutional violation ordinarily lies with the petitioner in a habeas proceeding. Hawk v. Olson, 326 U.S. 271, 279, 66 S.Ct. 116, 120, 90 L.Ed. 61 (1945); Johnson v. Zerbst, 304 U.S. 458, 468, 58 S.Ct. 1019, 1025, 82 L.Ed. 1461 (1938); Conner v. Wingo, 429 F.2d 630, 637 (6th Cir. 1970), cert. denied, 406 U.S. 921, 92 S.Ct. 1779, 32 L.Ed.2d 121 (1972). Unless he proves the facts necessary to establish his claim to relief by a preponderance of the evidence, the collateral attack will fail. Where the asserted ground for relief is incompetency at trial, the habeas court will require a certain quantum of evidence before it even entertains the claim. As we stated in Bruce's second appeal:

We consider it appropriate to add a caveat with respect to cases of this type. Courts in habeas corpus proceedings should not consider claims of mental incompetence to stand trial where the facts are not sufficient to positively, unequivocally and clearly generate a real, substantial and legitimate doubt as to the mental capacity of the petitioner to meaningfully participate and cooperate with counsel during a criminal trial. While the factual pattern will vary from case to case, the instant case illustrates the standard which should be met to sustain such a claim, viz. a history of mental illness, substantial evidence of mental incompetence at or near the time of trial supported by the opinions of qualified physicians and the testimony of laymen. The burden is on the petitioner to prove his allegations; such proof should be clear and convincing.

483 F.2d at 1043.

Bruce argues that the above quotation should be interpreted as holding that he has already discharged his clear and convincing burden and that the burden now shifts to the state to prove his competency. On remand, the same language was relied upon by the district court as authority for ruling that the petitioner had still to prove his incompetency claim by clear and convincing evidence. We disagree with both these constructions, because we read the prior panel's caveat as referring only to the petitioner's threshold burden of proof which must be satisfied before the habeas court has a duty to investigate the constitutional challenge further. Once petitioner has come forward with enough probative evidence to raise a substantial doubt as to competency, however, his task is not complete. He must then go further and prove the fact of incompetency, at least by a preponderance of the evidence.

Although particularly tailored to competency claims, this allocation is essentially the traditional one. As the habeas cases indicate, it is entirely proper to place the burden on the petitioner. See Drope v. Missouri, 420 U.S. 162, 174, 95 S.Ct. 896, 905, 43 L.Ed.2d 103 (1975). Similarly, proof by a preponderance (especially after establishing clear doubt) is all that is required. See Lego v. Twomey, 404 U.S. 477, 92 S.Ct. 619, 30 L.Ed.2d 618 (1972); United States v. Makris, 535 F.2d 899, 906 (1976). To place a greater burden on the petitioner might bring up due process considerations. In sum, at the federal nunc pro tunc hearing, Bruce had the burden of proving that he was most probably incompetent at the time of his 1965 trial.

III. Competency

The ultimate focus in a retrospective competency hearing must be whether at the time of trial the accused had sufficient ability to consult with his attorney with a reasonable degree of rational understanding and whether he had a rational as well as factual understanding of the proceedings against him. Dusky v. United States, 362 U.S. 402, 80 S.Ct. 788, 4 L.Ed.2d 824 (1960). Before the court can meaningfully apply this legal standard, however, it must often ascertain the nature of petitioner's allegedly incapacitating illness. It is at this initial juncture that expert testimony is particularly valuable, for the existence of even a severe psychiatric defect is not always apparent to laymen. Because of this difficulty in detecting medical diseases, the trial court may find it necessary to make an initial factfinding on whether the accused suffers from a mental defect at all. Although sometimes dispositive of the ultimate competency question, this medical inquiry is properly classified as pure factfinding and reviewable only under the clearly erroneous standard.

Once it is established that an individual suffers from a clinically recognized disorder, the court must decide whether such condition rendered the accused incompetent under the Dusky formulation. As enunciated in Makris, this second stage determination of legal incompetency is subject to a review more stringent than the clearly erroneous rule. To insure protection of valuable constitutional rights, this court is bound to take a hard look at the ultimate competency "finding."

The district court correctly determined that its first task was to decide whether Bruce suffered from a form of schizophrenia or was simply a sociopathic personality. This primary factfinding was crucial, for unless the underlying schizophrenic disorder were present, there could be little doubt of competency. As we understand the experts' description, a sociopath suffers from no disability which could affect competency. The medical term solely describes manipulative, egocentric persons who frequently commit antisocial acts without feelings of remorse. The expert testimony established that if Bruce should be classified as a sociopath, he would have no medical basis for claiming a diminished capacity to understand his trial or assist his attorney. A finding that Bruce's actions established he was a sociopath would thus be equivalent to finding no medical defect at all.

While we agree with the court's method of first resolving the conflicting diagnoses, we must declare its finding that Bruce is a sociopath clearly erroneous. The factfinder is not lightly to be disregarded, and the disagreement here is not easily reached. Neither the court's three reasons for crediting Dr. Grigson's evaluation nor anything else in this extensive record dispels our definite and firm conviction that the court made a mistake in failing to find that the petitioner's mental condition at the time of trial was at the very least potentially disabling. Indeed, the bizarre and tragic facts of this particular case compel the conclusion of clear error.

Of the three reasons given by the court for accepting Dr. Grigson's diagnosis, the first two do not support the court's findings; the third, while more probative, pales in significance when viewed in the proper context of the entire record.

(1) Bruce's early antisocial behavior.

Contrary to Dr. Grigson's position, the preponderant medical view is that Bruce's early antisocial behavior is compatible with either of the two competing diagnoses. The other physicians who examined Bruce and detected schizophrenia were aware of this history, yet this knowledge served to bolster rather than to change their professional opinions. Even the district court concluded that much of Bruce's "scrutinized behavior is consistent with either schizophrenia or a sociopath." In view of the lack of a definite correlation between an antisocial behavioral pattern and a given medical classification, this factor cannot be listed as supportative of either physician's view.

(2) Bruce's propensity to fake his symptoms.

The court recognized that Bruce's admitted faking would not "be dispositive of whether petitioner has a sociopathic personality or suffers from schizophrenia." Convinced that Bruce's faking could not change her diagnosis, Dr. Cannon testified that "anyone can attempt to fake. But at the time the schizophrenic would be attempting to fake mental illness, his real symptoms of schizophrenia would be present." Fakery by Bruce permits two equally permissible inferences. It could be that as a sociopath he faked his 1969 interview in an attempt to fool Dr. Grigson into believing him schizophrenic. Or the attempt at deception could be viewed as a manifestation of Bruce's paranoid personality, i. e., Bruce dissembled because he did not trust the doctors with the truth. The only certainty is that like the antisocial behavior factor, the ambivalence of the faking factor eliminates its utility in the balancing process.

(3) Reliance on trial transcript.

Viewed in isolation, the court's reliance on the significance of the 1965 trial transcript seems to be a more weighty matter. Putting to one side Bruce's abnormal medical history, his testimony has the appearance of being the product of a rational mind and the much-discussed outbursts may be explained away as "normal" emotional releases in the context of this intrafamily tragedy. Judged in this light, Dr. Grigson's insistence on Bruce's capacity to think rationally appears corroborated.

However, as soon as the trial transcript is perceived in the context of petitioner's entire medical and personal background, as it must be, it cannot serve as the basis for concluding that Bruce suffered from no recognizable medical defect. The weight of the entirety of the remaining probative evidence simply overwhelms the opposite conclusions that might be drawn if only the transcript were considered.

A summary of the extensive medical and lay evidence adduced over a nine-year span convincingly demonstrates that Bruce was schizophrenic in 1965 and serves to discredit Dr. Grigson's analysis. Except for Dr. Grigson, all the physicians who examined Bruce detected an underlying schizophrenic disorder. Most significantly, the experts who observed Bruce prior to the trial treated him as a dangerous psychotic who needed large dosages of medication to keep under control. After Bruce was convicted, the diagnosis of the doctors at the Wynne Unit was again "schizophrenic" and drug treatment was accordingly prescribed. The two physicians who conducted extensive testing in 1969 in preparation for the state postconviction hearing arrived at the same result. Finally, almost ten years after trial, Dr. Cannon examined Bruce and classified him as a paranoid schizophrenic.

In contrast, the only dissenting expert, Dr. Grigson, conducted his first examination three and one-half years after trial. When compared to the conclusions of those conducting more contemporaneous examinations, this lapse alone diminishes the force of Dr. Grigson's contrary views. United States v. Makris, 483 F.2d 1082, 1090 (5th Cir. 1973). In today's case, the danger of putting too much weight on any expert's after-the-fact medical assessments is acute, for by 1969 Bruce had already undergone extensive treatment, including taking antipsychotic medication and being confined in a structured, supervised environment. Nor did Dr. Grigson keep Bruce under lengthy observation. In contrast to six weeks of study by Dr. Tauber, one of the treating psychiatrists in 1969, Dr. Grigson's exposure to the petitioner both in 1969 and 1974 totaled less than four hours. When asked how it was possible that the other experts who had examined Bruce over a nine-year period had arrived at a radically different diagnosis, Dr. Grigson's sole explanation was that he was better qualified than they to determine Bruce's condition, a fact not established in the record.

Moreover, the accounts of petitioner's conduct related by lay observers during the relevant time-frame do not serve to erode the majority medical diagnosis. The strange facts of this case establish that Bruce often acted like a man who suffered from a severe psychotic disorder. The doctors' opinions aside, we are convinced that Bruce's continuing aberrant behavior cannot be uniformly dismissed as attempts to manipulate. To mention only the most striking events — the military episodes, the killing of his wife, the visual hallucinations and the escape from Terrell clearly appear to be products of a diseased mind. Had it not been for his release by a bogus doctor, Bruce might never have stood trial and the grand jury's initial decision to defer to the medical authorities might well have proven to be the wisest course.

Bruce's two attorneys possessed irreconcilable opinions concerning their client's mental state, though both counselors described conduct occurring at essentially the same time. Based on unsuccessful attempts to communicate and the observation that his client was in a state of shock following the killing, attorney McNicholas forcefully concluded that Bruce was a very sick man, possibly a schizophrenic. He was dismissed when he urged that an insanity defense be presented. McNicholas was succeeded by attorney Snodgrass, the only nonexpert witness who even slightly indicated that Bruce was competent and rational. We are bound to note that had Snodgrass testified that Bruce's behavior was in fact inconsistent with his deference to family instructions not to raise the issue of insanity, he would have placed himself in an awkward ethical position. See Bruce v. Beto, 396 F.2d 212, 213-14 (5th Cir. 1968) (Godbold, J., concurring). In sum, the overwhelming weight of the medical evidence pointed to schizophrenia and the clear preponderance of the lay testimony did nothing to discount that evaluation.

As the district court carefully noted, to label Bruce a paranoid schizophrenic does not end the inquiry. There still remains the ultimate question of whether Bruce's illness pushed him below the minimum level contemplated by Dusky. As expressed in Makris, our review of this portion of the district court's analysis is less restrictive; a critical reappraisal is required to guard petitioner's due process rights. At this stage, expert testimony is not so important, although the psychiatrist's "inexpert" opinion can be a factor in the court's independent decision.

This appellate review is disadvantaged because the district court essentially halted its analysis when it erroneously classified Bruce as a sociopath. Its alternative holding in the final footnote of its opinion — that even if Bruce suffered from schizophrenia in 1965, he was competent at his trial — is not fully helpful because it does not disclose the rationale for the court's conclusion. Like its initial factfinding, the court's ultimate competency determination cannot withstand testing in light of the record. Once Bruce's schizophrenia is acknowledged, the conclusion that the disease prevented him from rationally communicating with his attorney and understanding the proceedings is most compelling.

There is much record evidence that Bruce was overtly schizophrenic at the time of trial. According to the testimony of his first attorney, the trauma of the shooting left Bruce in an uncommunicative state of shock. Not long before trial, Bruce had been medically determined to be dangerous to himself or others and had been institutionalized and treated for a severe psychotic condition. Bruce also claims to have experienced visual hallucinations while at Terrell, a recognized symptom of schizophrenia in an overt state. Even Bruce's trial attorney who did not consider his client incompetent admitted that for at least short intervals, petitioner was incapable of understanding or communicating.

The case for competence really boils down to the trial transcript and the judgment expressed by attorney Snodgrass that overall Bruce was competent enough to understand and assist. Both factors are relevant, but the value of each must be discounted. First, without the aid of the observations of the presiding trial judge, a printed record should be received with caution. A transcript cannot reveal tone, speech inflections, mood and other indicia of a mental state and certainly cannot pick up subtle but crucial changes in petitioner's demeanor. Second, as before mentioned, the testimony of attorney Snodgrass cannot be treated as that of a disinterested witness.

In reaching the conclusion that the evidence did preponderate in favor of finding Bruce incompetent, we are not disregarding the uncontradicted medical testimony that mental states can change rapidly nor the indications that Bruce may have been competent for a portion of his trial. Neither are we holding that temporary medical problems are enough to cause failure of the Dusky test. Cf. United States v. Makris, 535 F.2d 899, 909 (5th Cir. 1976). Rather, what emerges from this record is a profile of a defendant with a severe psychiatric disorder which most probably caused him to misperceive important elements of the proceedings against him and likely interfered with his ability to relate the true facts to his counsel. Under the circumstances, Bruce proved as much as he could about his medical condition and certainly enough to carry his burden. Unless a psychiatrist had been at petitioner's side during the actual proceedings, there would be no truly reliable method of charting the possible changes in Bruce's mental state and establishing their effect on Bruce's legal competency. The only conclusion we can reach from this record is that Bruce's schizophrenic condition prevented him from effectively consulting with counsel and rationally understanding the proceedings.

The district court's order is therefore reversed with directions to issue the writ of habeas corpus subject to the right of the State of Texas to retry petitioner within a reasonable time.

REVERSED WITH DIRECTIONS.