dissenting:
This case once again raises the troubling question of the extent, if any, to which our Brawner1 decision should be given retrospective effect.2 In Brawner we adopted an ALI-derived formulation for the responsibility defense. The only fault in the Durham formulation— upon which we all agreed — was that its “product” phase had been construed, wittingly or not, in a manner that permitted almost total reliance on the experts to the exclusion of the jury.3 The underlying substantive reason for our reformulation of the insanity defense in Brawner was a prophylactic one, to protect the jury’s role by minimizing this distorting effect on the responsibility inquiry.4
At trial in the instant case, appellant relied exclusively on an insanity defense; the expert witnesses agreed that he suffered from an abnormal condition of the mind but disagreed as to whether his criminal act was a product of his impairment. Productivity was effectively the only issue in dispute and the experts conclusory testimony on that subject clearly raised the spectre of expert domination. The very danger of distortion we sought to eliminate in Brawner permeated Marshall’s trial.5
In Brawner the majority announced that its reformulation of the responsibility test would be effective prospectively,6 and remanded the case for a determination of whether there was a substantial possibility that Brawner’s insanity defense would have succeeded under our altered standard. Consistent with this procedure, the Brawner court specifically rejected the approach of the Second Circuit in United States v. Tarrago,7 in which that court applied its adoption of the ALI standard retroactively to automatically reverse eases in the appellate pipeline.
Similarly, in Gaither v. United States,8 this court adopted a rule requiring the approval by twelve grand jurors of the actual indictment returned, noting that all indictments brought after the date of the decision not in conformity *1055with this rule would automatically be dismissed without further inquiry. The court refused to apply this rule retroactively to automatically dismiss indictments returned prior to the announcement of its decision, but, at least initially, did give the litigants before it “an opportunity to achieve dismissal of their indictments on a showing of a reasonable possibility” of prejudice. Recognizing that there was an injustice in affording this remedy to the appellants but not to those similarly situated, the Gaither court extended its availability. The court observed that
. such injustice in case of need is amply authorized by Supreme Court precedent, . . . [but] it seemed to us better avoided if another solution could be found.9
On petition for rehearing, the court modified the application of its decision because of circumstances not relevant in the Brawner remand context.10
The Brawner court never directly addressed itself to the question of whether the limited-purpose remand afforded Brawner should be extended to litigants whose convictions were on appeal when Brawner was decided. Nor did it consider the possibility of “another solution”. I would find such a solution in giving the Brawner remand procedure “normal judicial operation”.11 I would make the remand available to all pending appeals that patently present the same danger of distortion that led us to our reformulation of Durham.12 I cannot assume that the Court ruled out every alternative solution without addressing the *1056issue or suggesting any need — recognized by current case-law — which would support absolute prospectivity.13
My position is entirely consistent not only with the “sound principles of Stovall v. Denno”,14 but also with this court’s express rejection of the kind of retrospectivity suggested by Tarrago. In Stovall the Supreme Court set out a three part test to determine what retrospective application its decision would be given: (1) the purpose of its new rule, (2) the extent of reliance by law enforcement authorities on the former law, and (3) the effect on the administration of justice of the proposed retrospectivity.15
In Stovall, as in those other cases where the Supreme Court’s decisions have been given completely prospective effect, a “major change in the law was announced, overruling settled precedent.” 16 The Brawner court, however, characterizes its decision as “not . a change in substantive law from outmoded doctrine . . . [but meant merely to] be helpful to the jury [although not] . . . indispensable to integrity in the fact-finding process.” 17 Brawner was not meant to effect a substantive change in the responsibility defense but only to alter its formulation in order to prevent the misuse of expert testimony.18
This court long ago recognized the danger of expert oversteering19 and Brawner is but another attempt to deal with it. The effect of Brawner’s remand is merely to give him an opportunity to show a substantial possibility that his insanity defense had been rejected because of the distorting effect of improper expert testimony. The court has found it unnecessary, given its limited purpose in Brawner, to reverse any conviction absent some showing that this purpose had been contravened. To automatically reverse convictions rendered under the superseded standard might, as in Stovall, overturn many convictions unnecessarily.20 This reasoning may have prompted our rejection of the Tarrago example. But remanding only those cases in which the very danger sought to be avoided in Brawner is patent on the record would clearly not be so overly broad an application of our decision.
Even while giving completely prospective effect to the prophylactic rule it was adopting in Stovall, the Supreme Court left open the possibility of reversal on a showing that the very unfairness meant to be prevented had occurred so *1057as to deny the litigant due process.21 Our reformulation of the insanity defense has a similar purpose regarding expert testimony. I merely suggest that we allow litigants like appellant an opportunity to show that the very distortion we meant to prevent by our Brawner decision had determined the verdict in their cases. This application is clearly so limited in scope as to be entirely consonant with the limited purpose of Brawner.
There is no “reliance by law enforcement authorities”22 on the Durham formulation and Brawner does not, like Stovall, render inadmissable evidence secured by the police and used by prosecutors in reliance on prior judicial approval.
Finally, the Brawner Court stated that one of its reasons for rejecting the retrospective application of Tarrago was that such retrospectivity would impose a substantially greater “burden on [the] administration of justice” 23 for this Circuit than it did for the Second Circuit. But the Brawner court also observed that “if there is a case where there would be a difference in result [under the new standard] . . . [and hence a need to hold a new trial] it would seem rare . . . . ”24 Accordingly, there could be no intolerable “burden on the administration of criminal justice.”
If, as the Brawner court suggests, our new formulation will improve the adjudication of the insanity defense by minimizing the domination by experts, then I believe that appellant should have an opportunity to benefit from that fairer adjudication, at least to the extent Brawner did. I would therefore remand this case to the trial court to determine whether there is a substantial possibility that appellant’s insanity defense would have succeeded under our new formulation.
. United States v. Brawner, No. 22,714, 158 U.S.App.D.C.-, 471 F.2d 969 (D.C. Cir. June 23, 1972).
. United States v. Wilson, 153 U.S.App.D.C. -, 471 F.2d 1072 (1972) ; see United States v. Bryant & Murdock, - U.S.App.D.C. -, 471 F.2d 1040 (1972).
. Brawner, at---, - of 153 U.S.App.D.C., at 978-979, 983 of 471 F.2d; Brawner, at -, at 1023 of 471 F.2d (separate opinion).
. The Brawner Court also suggested as a reason for its decision a desire for uniformity of expression among the Circuits. Brawner, at 969, at 984-985 of 471 F.2d. This is clearly a cosmetic and not substantive purpose.
. While the expert witnesses’ testimony was not phrased in terms of “product”, the inference to the jury was nonetheless clear that, as in Brawner, “causality was the cutting edge of the responsibility test and ... at least some of the experts were convinced that causalty did not exist in this case.” Brawner at 969, at 1035 of 471 F.2d (separate opinion) ; see Tr. 441, 470.
. Brawner, at 969, at 1005 of 471 F.2d.
. 398 F.2d 621 (2d Cir. 1968) (en banc). Three other circuits have applied their adoption of the ALI criminal responsibility standards with similar retrospective effect. Wade v. United States, 426 F.2d 64 (9th Cir. 1970) (en banc) ; Blake v. United States, 407 F.2d 908 (5th Cir. 1969) (en banc) ; United States v. Smith, 404 F.2d 720 (6th Cir. 1968).
. 134 U.S.App.D.C. 154, 413 F.2d 1061 (1969).
. 413 F.2d at 1081.
. On petition for rehearing in Gaither v. United States, 134 U.S.App.D.C. 154, 413 F.2d 1061, 1081 (1969), the court abandoned its initial approach to the retrospeetivity of its decision (described above, at notes 8-9, supra, and accompanying text). Although the Court had not been deterred from its earlier decision by the knowledge that up to 1100 hearings could have resulted, it was not then aware of additional problems created by the need for Grand Jury transcripts in eacli such hearing. Because of the bottleneck in producing those transcripts, even the court’s limited retrospective application would have resulted in “serious disruption of the District Court calendar” and “serious delays in trials in the District Court.” Since its original approach to retrospeetivity threatened such an intolerable burden on the administration of criminal justice, the Court, with obvious reluctance, eliminated all retrospective application for its decision. There is no indication of any such intolerable burden being generated by the limited application of the Brawner remand suggested herein. See 413 F.2d at 1083-1084.
. Mishkin, The Supreme Court 1964 Term — Forward: The High Court, The Great Writ, and the Due Process of Time and Law, 79 Harv.L.Rev. 56, 77 (1965) ; see Linkletter v. Walker, 381 U.S. 618, 627, 85 S.Ct. 1731, 14 L.Ed.2d 601 (1965).
“Retrospeetivity” would be an inappropriate description for the limited application suggested here. In a similar context, the Supreme Court explicitly defined “retrospective” application as giving a decision retroactivity beyond “cases still pending on direct review at the time [the decision] was rendered.” Linkletter v. Walker, 381 U.S. 618, 621-622, 85 S.Ct. 1731, 1734, 14 L.Ed.2d 601 (1965) ; Cf. Williams v. United States, 401 U.S. 646, 652 n. 4, 91 S.Ct. 1148, 28 L.Ed.2d 388 (1971).
. The majority in this case does not rely exclusively on the Brawner prospectivity language but also asserts, “we see no reason to believe that the jury would have reached a different result if instructed according to the formula prescribed in Brawner’s ease.” Majority, at - of 153 U.S.App.D.C. at 1053 of 471 F.2d.
Such a conclusion, particularly in a case where conclusory expert testimony is so obvious on the record, clearly seems foreclosed by our conclusion in Brawner that it is not appropriate for this court “to determine at this juncture whether a jury which convicted under our old insanity standard might have acquitted under the new standard. . . . [W]e are remanding to the trial judge . rather than considering that question at the appellate level in the first instance, because the trial judge has a superior vantage point for assessing whether there is a substantial possibility that the jury, if instructed under our new rule, would have found that appellant should be acquitted by reason of insanity.” Brawner at - of 153 U.S.App.D.C., at 1005 of 471 F.2d.
. While there is precedent for adopting a new standard witli only prospective effect in case of need, due to the burden on the administration of justice of any retrospectivity or because of the pattern of justifiable reliance of law enforcement authorities on the former standard, no such need lias been demonstrated to justify refusing the limited retrospective application suggested here. See notes 20-24, infra, and accompanying text; Cf. Gaither v. United States, 134 U.S.App.D.C. 154, 413 F.2d 1061, 1082-1083 (1969).
. 388 U.S. 293, 87 S.Ct. 1967, 18 L.Ed.2d 1199 (1967).
. 388 U.S. at 297, 87 S.Ct. 1967.
. Gaither v. United States, 134 U.S.App.D.C. 154, 413 F.2d 1061, 1081 (1969).
. Brawner, at---, 153 of U.S.App.D.C., at 1005 of 471 F.2d n. 79.
. The Brawner court explained that “Durham-MacDonald is not significantly different in substantive content from the ALI test.” While the court’s announced purpose was to cope with the problem of expert “oversteering” posed by the Durham product formulation, its new standard “retain[s] the underlying concept of causal relationship.” In sum, observes the court, its new rule was not adopted “ . . . in the contemplation that it would affect a significant number of verdicts.” Brawner, at -, at 1005 of 471 F.2d n. 79.
. Washington v. United States, 129 U.S.App.D.C. 29, 390 F.2d 444 (1967) ; see United States v. Brawner, No. 22,714, 153 U.S.App.D.C., at---,-- -, 471 F.2d, at 1017-1021, 1022-1023 (1972), (separate opinion).
. 388 U.S. at 299, 87 S.Ct. 1967, 18 L.Ed .2d 1199.
. Id.
. 388 U.S. 299-300, 87 S.Ct. 1967, 18 L.Ed.2d 1199; see United States v. Tarrago, 398 F.2d 621, 624 (2d Cir. 1968) (en banc).
. Brawner, at - of 153 U.S.App.D.C., at 1005 of 471 F.2d n. 79.
. Brawner, at -, at 990 of 471 F.2d; accord, id. at---, at 1005 of 471 F.2d n. 79.