I dissent.
In my opinion the judgment appealed from once again must be reversed. This is so because several errors were committed by the prosecution during this trial. While recognizing that some of these errors were very serious indeed, the majority hold that they were not prejudicial. In doing so, they purport to apply the federal test of prejudicial error announced in Chapman v. California, 386 U.S. 18 [17 L.Ed.2d 705, 87 S.Ct. 824], but in my opinion do not do so.
The majority correctly state that the evidence is overwhelming that the two girls were killed, and that the evidence overwhelmingly demonstrates that defendant killed them. They therefore conclude that, in view of this record, there is no reasonable possibility that other than a guilty verdict could have been returned. If guilt or innocence were the only issue involved I would agree with this analysis. But guilt or innocence was not the only issue involved. Here there was also involved the degree of guilt. On this issue, unlike the guilt or innocence issue, the evidence was not overwhelming or conclusive. There was a very grave question as to whether the defendant was so intoxicated that the degree of the crime was reduced to second degree or manslaughter. There was also a grave question as to whether, even if defendant was not so intoxicated as to reduce the degree, he was so intoxicated that he should not suffer the death penalty. While I agree that the several errors involved could not and did not affect the determination of guilt, I cannot say that there is no “reasonable *718possibility” that they did not affect the determination of the degree of the crime.
We should have clearly in mind the federal test of prejudicial error, here applicable. It was recently restated in the Chapman case, supra, which followed Fahy v. Connecticut, 375 U.S. 85 [11 L.Ed.2d 171, 84 S.Ct. 229]. In that case it was held that error must be deemed prejudicial unless it can be said that there is no "reasonable possibility” that the error complained of might have contributed to the conviction (p. 86, 11 L.Ed.2d p. 173). Chapman, supra, specifically reaffirmed this language and then held that “before a federal constitutional error can be held harmless, the court must be able to declare a belief that it was harmless beyond a reasonable doubt.” (386 U.S. at p. 24 [17 L.Ed.2d at pp. 710-711, 87 S.Ct. at p. 828].) Under this rule, once error is found, it is prejudicial and therefore reversible unless the appellate court can affirmatively find that there is no “reasonable possibility” that the error might have contributed to the conviction. No such affirmative finding can be made here.
In the instant case, there were several serious errors. One of the most serious with which the majority show great concern, and rightly so, was the prosecutor’s repeated comments on the defendant’s failure to take the stand and explain damaging evidence against him. This necessarily included evidence that lie was not intoxicated. The majority frankly concede that in his closing argument the prosecutor “repeatedly exploited” this fact, and on at least one occasion directly violated the rule laid down a year before the present trial in Griffin v. California, 380 U.S. 609 [14 L.Ed.2d 106, 85 S.Ct. 1229], Repeatedly, he commented on this failure of defendant to explain adverse evidence against him. The prosecutor time and time again impliedly argued that defendant could have explained but did not, thus at least implying that silence constitutes an admission. That argument was clearly erroneous and most prejudicial. The tacit admission rule has been repudiated, insofar as police accusations are concerned, by the United States Supreme Court in Escobedo v. Illinois, 378 U.S. 478 [12 L.Ed.2d 977, 84 S.Ct. 1758], and by this court in People v. Dorado, 62 Cal.2d 338 [42 Cal.Rptr. 169, 398 P.2d 361] ; People v. Cockrell, 63 Cal.2d 659, 669 [47 Cal.Rptr. 788, 408 P.2d 116] ; and People v. Ridley, 63 Cal.2d 671, 676 [47 Cal.Rptr. 796, 408 P.2d 124]. (See also People v. Ellis, 65 Cal.2d 529, 536 [55 Cal.Rptr. 385, 421 P.2d 393].) The *719repeated references of the prosecutor to the defendant’s silence were clearly improper. Can we affirmatively declare that there is no “reasonable possibility” that these repeated references could not have convinced at least one member of the jury to vote for first degree rather than some other reduced degree, and could not have induced at least one member of the jury to vote for the death penalty rather than life ? Of course not.
The repeated arguments made by the prosecutor about the failure of the defendant to explain the evidence against him must have, and certainly could have, hit the defendant’s defense of diminished responsibility by reason of intoxication with devastating effect. I am unable to make an affirmative declaration that there is no “reasonable possibility” that these errors could not have affected the verdict so far as the degree of the crime is concerned.
Another most serious error is also chargeable to the prosecutor. Not only did he impress upon the jury his personal belief in the defendant’s guilt of first degree murder, but he made repeated and persistent references to the prior trials in which 24 jurors and 2 judges had already passed upon those issues, which necessarily included the defense of diminished responsibility. The majority forthrightly admit that these arguments overstepped the bounds of propriety and were erroneous, but hold the error not prejudicial because the defendant failed to object, and an admonition could have cured the error. This is most unrealistic. The erroneous argument must have had a shattering effect on defendant’s defense of diminished responsibility. Once made, no admonition could erase from the jurors’ minds the personal opinion of the prosecutor and his statements that two judges and twenty-four jurors had already decided that defendant was guilty of first degree murder, and had repudiated the defense of diminished responsibility. This ease is not unlike People v. Brock, ante, p. 645 [58 Cal.Rptr. 321, 426 P.2d 889], recently filed, in which the trial judge opined that defendant was guilty, and then admonished the jury that they were the sole judges of the facts. The admonition, given in that case in an instruction, did not cure the error. An admonition here would have been equally ineffectual.
Another serious error which I cannot dismiss as cavalierly as do the majority is the erroneous discussion of the insanity defense—that if found insane defendant would be “turned loose.” The majority frankly concede that this was error and *720most serious error, but conclude that it could not have been prejudicial because the plea of insanity was later withdrawn. Obviously the error did not affect the sanity issue, because that issue became moot when it was withdrawn from the jury. But the argument could well have had a most serious impact on the jury. It, in effect, told the jury that any other verdict than death conceivably could result in turning loose the defendant to kill again. This, of course, violated People v. Morse, 60 Cal.2d 631 [36 Cal.Rptr. 201, 388 P.2d 33], I cannot say that there is no “reasonable possibility” that the argument did not have this effect.
Thus there were errors, serious errors, committed during this trial. Although it is not reasonably possible that those errors could have affected the finding of guilt, there is far more than a reasonable possibility that such errors could have affected the finding of the degree of the crime and the imposition of the death penalty. For these reasons I would reverse the judgment.
Appellant’s petition for a rehearing was denied June 21, 1967. White, J.,* sat in place of Mosk, J., who deemed himself disqualified. Peters, J., was of the opinion that the petition should be granted.
Retired Associate Justice of the Supreme Court sitting under assignment by the Chairman of the Judicial Council.